HBM | KitGuru https://www.kitguru.net KitGuru.net - Tech News | Hardware News | Hardware Reviews | IOS | Mobile | Gaming | Graphics Cards Wed, 22 Nov 2023 11:28:48 +0000 en-US hourly 1 https://wordpress.org/?v=6.4.3 https://www.kitguru.net/wp-content/uploads/2021/06/cropped-KITGURU-Light-Background-SQUARE2-32x32.png HBM | KitGuru https://www.kitguru.net 32 32 Nvidia and SK Hynix are reportedly working on a new GPU architecture with 3D-stacked HBM memory https://www.kitguru.net/components/graphic-cards/joao-silva/nvidia-and-sk-hynix-are-reportedly-working-on-a-new-gpu-architecture-with-3d-stacked-hbm-memory/ https://www.kitguru.net/components/graphic-cards/joao-silva/nvidia-and-sk-hynix-are-reportedly-working-on-a-new-gpu-architecture-with-3d-stacked-hbm-memory/#respond Tue, 21 Nov 2023 09:00:12 +0000 https://www.kitguru.net/?p=638225 SK Hynix is reportedly hiring design staff for logic devices, such as CPUs and GPUs, to develop a stacking method to put HBM4 directly on top of them. This would revolutionize the foundry sector by changing how logic and memory components are coupled and manufactured. As per the same report, Nvidia seems interested in this …

The post Nvidia and SK Hynix are reportedly working on a new GPU architecture with 3D-stacked HBM memory first appeared on KitGuru.]]>
SK Hynix is reportedly hiring design staff for logic devices, such as CPUs and GPUs, to develop a stacking method to put HBM4 directly on top of them. This would revolutionize the foundry sector by changing how logic and memory components are coupled and manufactured. As per the same report, Nvidia seems interested in this approach and is working with SK Hynix on its development.

According to The JoongAng (via Tom's Hardware), the company hopes to eliminate the need for interposers by placing HBM4 stacks directly on CPUs, similar to AMD's direct-to-CPU 3D V-Cache implementation. While HBM will be more expensive than cache, it will have a larger capacity, making it a more cost-effective solution.

Image credit: Joongang.co.kr

HBM stacks combine a hub-like logic layer with several memory devices connected to CPUs or GPUs via a 1024-bit interface. However, SK Hynix's strategy of mounting HBM4 stacks directly on CPUs will reduce costs and simplify chip designs. The company is said to be in talks with fabless businesses, including Nvidia, about their HBM4 integrated design technique. The two companies are expected to collaborate to build the chip from the ground up and have it produced by TSMC, which will also use wafer bonding to attach SK Hynix's HBM4 device to logic chips.

However, there is one issue that needs to be addressed: thermals. GPUs like Nvidia's H100 have high power consumption and heat dissipation, and HBM memory is also known for being more power-hungry than other memory solutions. More complex cooling techniques, such as liquid cooling or submersion, may be needed to cool down a package containing logic and memory.

According to Professor Kim Jung-ho of KAIST's Department of Electrical and Electronics, HBM and GPU can function as one unit without an interposer, but only if the heating problem is solved, which should happen in two to three generations. The direct integration of memory into CPUs will also alter the architecture and manufacturing process of chips and could impact prices too.

Discuss on our Facebook page, HERE.

KitGuru says: Processor architectures have always separated memory and logic semiconductors. However, if SK Hynix can find a viable way to mass-produce chips with 3D-stack HBM, the rules may be about to change relatively soon.

The post Nvidia and SK Hynix are reportedly working on a new GPU architecture with 3D-stacked HBM memory first appeared on KitGuru.]]>
https://www.kitguru.net/components/graphic-cards/joao-silva/nvidia-and-sk-hynix-are-reportedly-working-on-a-new-gpu-architecture-with-3d-stacked-hbm-memory/feed/ 0
The European Union prepares its own prototype CPU https://www.kitguru.net/components/cpu/james-dawson/the-european-union-prepares-its-own-prototype-cpu/ https://www.kitguru.net/components/cpu/james-dawson/the-european-union-prepares-its-own-prototype-cpu/#respond Wed, 29 Jan 2020 13:30:35 +0000 https://www.kitguru.net/?p=449217 As the UK prepares to officially leave the European Union this Friday (31st January), the European Processor Initiative (EPI) is priming the first prototype of its HPC processor designed for advanced computing applications. The EPI was set up by the European Commission as part of the Framework Partnership Agreement, with an aim to produce a …

The post The European Union prepares its own prototype CPU first appeared on KitGuru.]]>
As the UK prepares to officially leave the European Union this Friday (31st January), the European Processor Initiative (EPI) is priming the first prototype of its HPC processor designed for advanced computing applications.

The EPI was set up by the European Commission as part of the Framework Partnership Agreement, with an aim to produce a homegrown family of low power processors for extreme exascale computing, machine learning and emergency applications to used by the European Union.

The initial batch of the European Union’s new exascale supercomputers are due to be introduced in 2023, by this time, the U.S, China and even Japan are expected to have similar systems of their own up and running. However, according to The Next Platform, prototypes of the EPI’s processors are already on their way.

Image credit: The Next Platform

In a recent conversation between the EPI Chairman Jean-Marc Denis and The Next Platform, details of the new prototype EPI processors were outlined. Denis confirmed that they will be manufactured using TSMC 6nm EUV process and is expected to part of a larger 2.5D interposer-based package that will include HBM memory, PCIe 5.0 and have DDR links.

The prototype EPI processor will be based on the “Zeus” iteration of a custom ARM CPU designed for general-purpose tasks such as running the OS. Special-purpose chips incorporating a Titan named RISC-V EPI Accelerator (EPAC), will be equipped with Vector processing units and Stencil/Tensor accelerators for AI and machine learning tasks. EPAC will support every standard AI format from INT8 through FP64, as well as bfloat16. According to Denis.

Discuss on our Facebook page HERE.

KitGuru says: The European Processor Initiative is forging ahead with plans to launch its own processor to power European Union extreme exascale computing by 2023. What do you guys think of this initiative to create its own CPU by the European Commission?

The post The European Union prepares its own prototype CPU first appeared on KitGuru.]]>
https://www.kitguru.net/components/cpu/james-dawson/the-european-union-prepares-its-own-prototype-cpu/feed/ 0
RAMBUS begins discussing early specs for DDR5 and HBM3 https://www.kitguru.net/components/memory/matthew-wilson/rambus-begins-discussing-early-specs-for-ddr5-and-hbm3/ https://www.kitguru.net/components/memory/matthew-wilson/rambus-begins-discussing-early-specs-for-ddr5-and-hbm3/#comments Fri, 08 Dec 2017 16:08:36 +0000 https://www.kitguru.net/?p=357233 This week, RAMBUS began unveiling its early specification outlines for DDR5 memory and HBM3. It will be several years before we see either put into action, but the early specs paint a bright picture for the future, with big performance and efficiency leaps planned for both. As part of an investor event this week, RAMBUS …

The post RAMBUS begins discussing early specs for DDR5 and HBM3 first appeared on KitGuru.]]>
This week, RAMBUS began unveiling its early specification outlines for DDR5 memory and HBM3. It will be several years before we see either put into action, but the early specs paint a bright picture for the future, with big performance and efficiency leaps planned for both.

As part of an investor event this week, RAMBUS began talking about its future plans. As Computerbase reports, development on DDR5 RAM and HBM3 is underway. For starters, the next generation of High Bandwidth Memory (HBM) will boost transfer rates up to 4 GT/s, which is a big jump from HBM2's 2 GT/s transfer speed. HBM3 will be developed on the 7nm node, so that should give you some idea as to how long it will be before we see it in action.

DDR5 memory is said to be significantly further in development but once again, 7nm is the target. We are only just starting to enter the age of 10nm, so it will be a while before we see this come to fruition. Still, decent spec bumps are expected. Currently DDR4 offers 3200Mbps bandwidth, with DDR5 RAMBUS is expecting to boost things up somewhere between 4800Mbps and 6400Mbps.

KitGuru Says: This is just a very brief early look at upcoming tech, with all the numbers being estimates from Rambus. Over time, we'll get a more concrete idea of what DDR5 and HBM3 will be capable of but so far, things look positive.

The post RAMBUS begins discussing early specs for DDR5 and HBM3 first appeared on KitGuru.]]>
https://www.kitguru.net/components/memory/matthew-wilson/rambus-begins-discussing-early-specs-for-ddr5-and-hbm3/feed/ 2
Micron to launch GDDR6 graphics memory this year https://www.kitguru.net/components/graphic-cards/matthew-wilson/micron-to-launch-gddr6-graphics-memory-this-year/ https://www.kitguru.net/components/graphic-cards/matthew-wilson/micron-to-launch-gddr6-graphics-memory-this-year/#comments Mon, 06 Feb 2017 23:52:13 +0000 http://www.kitguru.net/?p=321007 This is going to be an interesting year as far as GPU memory goes as not only are we expecting HBM2 to debut on AMD's upcoming Vega graphics card but it seems that Micron might be gearing to launch GDDR6 by the end of 2017. This new memory was initially planned to launch in 2018 …

The post Micron to launch GDDR6 graphics memory this year first appeared on KitGuru.]]>
This is going to be an interesting year as far as GPU memory goes as not only are we expecting HBM2 to debut on AMD's upcoming Vega graphics card but it seems that Micron might be gearing to launch GDDR6 by the end of 2017. This new memory was initially planned to launch in 2018 but will arrive early and will eventually be used to phase out GDDR5.

Micron announced that it plans to start producing GDDR6 in the second half of this year, which is earlier than expected. With GDDR6, bandwidth per pin will be increased to 16 GB/s, which is ten times faster than DDR3. This is also a nice jump from last year's GDDR5X, which is capable of 10 GB/s per pin. It won't quite match HBM2 in terms of overall bandwidth but it will be a good option to have.

On top of being faster, GDDR6 will also use twenty percent less power, which will be great for future laptop GPUs. It seems that the current gaming market surge has prompted Micron to bring forward its GDDR6 release, with more people upgrading every three years rather than every five.

GDDR6 is expected to begin rolling out before the end of the year and should replace GDDR5 (non-X) by 2020 according to reports.

Discuss on our Facebook page, HERE.

KitGuru Says: It looks like 2017/2018 are going to be interesting years for GPU memory. HBM2 should start showing up on gaming-grade hardware in the first half of this year with GDDR6 becoming an option for late 2017 or early 2018

The post Micron to launch GDDR6 graphics memory this year first appeared on KitGuru.]]>
https://www.kitguru.net/components/graphic-cards/matthew-wilson/micron-to-launch-gddr6-graphics-memory-this-year/feed/ 6
Nvidia could debut Pascal ‘GTX 1080’ in May https://www.kitguru.net/components/graphic-cards/matthew-wilson/nvidia-could-debut-pascal-gtx-1080-in-may/ https://www.kitguru.net/components/graphic-cards/matthew-wilson/nvidia-could-debut-pascal-gtx-1080-in-may/#comments Fri, 11 Mar 2016 17:31:54 +0000 http://www.kitguru.net/?p=286756 Nvidia is gearing up to launch its new Pascal GPU architecture this year. While early rumours pointed towards a Q1 2016 release it is clear that the timeline has shifted a bit. The latest unofficial word on the street is that Nvidia may launch the GTX 1080 in May, with the card prepped to use …

The post Nvidia could debut Pascal ‘GTX 1080’ in May first appeared on KitGuru.]]>
Nvidia is gearing up to launch its new Pascal GPU architecture this year. While early rumours pointed towards a Q1 2016 release it is clear that the timeline has shifted a bit. The latest unofficial word on the street is that Nvidia may launch the GTX 1080 in May, with the card prepped to use a fully enabled GP104 GPU core and 8GB of VRAM.

This news comes from Chinese site, Benchlife, so we don't know how valid it really is. However, according to the report, Nvidia's GTX 1080 could arrive in May, following a presentation of Pascal graphics cards at the Game Technology Conference this year. Nvidia may also launch a GTX 1070 in that same time frame.

NVIDIA-14nm-GPUs

Pascal will be based on the 16nm FinFET process and the new GP104 core may offer up to double the performance per watt and use a single 8-pin power connector. If this rumour is to be believed, though, then the GTX 1080 won't arrive with HBM and will instead use 8GB of GDDR5 VRAM. HBM 2 will likely feature on the flagship GP100 GPU-based card, like an 80Ti or Titan series product.

While we are currently going on the assumption that Nvidia's next card will be called the GTX 1080, the naming scheme is not confirmed. It is possible that Nvidia will switch its product naming up this year with the launch of Pascal.

KitGuru Says: Nvidia has been pretty quiet recently but hopefully we will hear something official on the launch of Pascal graphics cards soon. If the supposed GTX 1080 is coming in May, then it won't be long before we hear something more concrete. That said, this is all rumour for now so don't get your hopes up just yet. 

The post Nvidia could debut Pascal ‘GTX 1080’ in May first appeared on KitGuru.]]>
https://www.kitguru.net/components/graphic-cards/matthew-wilson/nvidia-could-debut-pascal-gtx-1080-in-may/feed/ 50
New AMD Zen details have begun to surface https://www.kitguru.net/components/cpu/matthew-wilson/new-amd-zen-details-have-begun-to-surface/ https://www.kitguru.net/components/cpu/matthew-wilson/new-amd-zen-details-have-begun-to-surface/#comments Wed, 20 Jan 2016 18:46:12 +0000 http://www.kitguru.net/?p=281513 With AMD set to release its Zen architecture processors towards the end of this year, the company has begun giving out more details during its quarterly earnings call. This week, AMD reported its quarterly earnings and in that call, CEO Lisa Su revealed a few more details on the Zen architecture along with what we …

The post New AMD Zen details have begun to surface first appeared on KitGuru.]]>
With AMD set to release its Zen architecture processors towards the end of this year, the company has begun giving out more details during its quarterly earnings call. This week, AMD reported its quarterly earnings and in that call, CEO Lisa Su revealed a few more details on the Zen architecture along with what we can expect and when to expect it.

Right now, AMD is in talks with PC makers to begin building systems using its codenamed Summit Ridge processors, which are based on the Zen architecture and may well be released under the ‘FX' brand during late 2016. These Summit Ridge CPUs will be the first high-performance desktop CPUs from AMD to use Zen, while chips for servers will appear in early 2017. Nothing was said about Zen APUs though we did hear some rumours recently– there is also no word on Zen for laptops just yet.

amd_zen_performance_advantages_fad-e1446672925133

We already know that AMD has high hopes for Zen and according to a previous report, the chip has managed to meet internal expectations with no significant bottlenecks found. Right now, Lisa Su is saying that AMD's Summit Ridge CPUs will be a “re-entry” into the high-performance desktop market for the company.

Lisa Su also spoke about performance a bit, saying that AMD's Zen-based CPU offers 40 percent more performance per cycle than Excavator cores, which are currently found in AMD's Carrizo chips. We still have a pretty long wait but we do know that Socket AM4 mat be launching fairly early this year around March though only APUs will be taking advantage of it at first, with Zen to follow later in the year. 

KitGuru Says: AMD is gearing up for a lot of launches in 2016, with the debut of Socket AM4, DDR4 support and the new Zen architecture. Are any of you looking forward to seeing how AMD's new CPUs perform? 

 

The post New AMD Zen details have begun to surface first appeared on KitGuru.]]>
https://www.kitguru.net/components/cpu/matthew-wilson/new-amd-zen-details-have-begun-to-surface/feed/ 107
AMD cuts the price of the R9 Nano GPU https://www.kitguru.net/components/graphic-cards/matthew-wilson/amd-cuts-the-price-of-the-r9-nano-gpu/ https://www.kitguru.net/components/graphic-cards/matthew-wilson/amd-cuts-the-price-of-the-r9-nano-gpu/#comments Mon, 11 Jan 2016 14:41:46 +0000 http://www.kitguru.net/?p=280301 The AMD R9 Nano is an impressive graphics card in many ways, it was the first Fiji card to be shrunk down to a Mini-ITX form-factor, it is very power efficient and now, it is getting a pretty nice price cut. AMD announced today that the R9 Nano will now start at $499 effective immediately, …

The post AMD cuts the price of the R9 Nano GPU first appeared on KitGuru.]]>
The AMD R9 Nano is an impressive graphics card in many ways, it was the first Fiji card to be shrunk down to a Mini-ITX form-factor, it is very power efficient and now, it is getting a pretty nice price cut.

AMD announced today that the R9 Nano will now start at $499 effective immediately, a $150 price cut from its previous $649.99 price tag.

amd_radeon_fury_nano_graphics_card1

The R9 Nano is equipped with a full Fiji core and 4GB of High-Bandwidth Memory (HBM), making it a great option for those looking to build a small but powerful ITX rig for high-resolution gaming.

We reviewed the Nano back in September, giving it our ‘WORTH BUYING' award and praiding its performance, power management and cooling perofrmance.

KitGuru Says: There has been a lot of interest in bringing more power to smaller form factors in the last couple of years. Do any of you have a Mini-ITX rig? Does this price cut tempt you in to buying a Nano? 

The post AMD cuts the price of the R9 Nano GPU first appeared on KitGuru.]]>
https://www.kitguru.net/components/graphic-cards/matthew-wilson/amd-cuts-the-price-of-the-r9-nano-gpu/feed/ 3
AMD said to be working on Zen APU with HBM https://www.kitguru.net/components/apu-components/matthew-wilson/amd-said-to-be-working-on-zen-apu-with-hbm/ https://www.kitguru.net/components/apu-components/matthew-wilson/amd-said-to-be-working-on-zen-apu-with-hbm/#comments Fri, 01 Jan 2016 22:10:21 +0000 http://www.kitguru.net/?p=279351 It looks like we can expect an AMD APU based on the Zen architecture in the future, though rather interestingly, it will include stacked High-Bandwidth Memory (HBM) with bandwidth speeds of 128GB/s along with a larger onboard GPU for better gaming performance. This information comes to light from a paper(via: wccftech), which was co-authored by …

The post AMD said to be working on Zen APU with HBM first appeared on KitGuru.]]>
It looks like we can expect an AMD APU based on the Zen architecture in the future, though rather interestingly, it will include stacked High-Bandwidth Memory (HBM) with bandwidth speeds of 128GB/s along with a larger onboard GPU for better gaming performance.

This information comes to light from a paper(via: wccftech), which was co-authored by AMD graphics engineer, Mike Mantor. The paper shows that this Zen-based APU would make use of an upgraded version of the memory technology found in Carrizo, which started off as an improved design over the memory interconnects found on the current generation consoles.

amd_zen_performance_advantages_fad

On top of that, the Zen APU is shown to use stacked HBM with 128GB/s of bandwidth. Aside from that, we don't know much more at this time, though earlier reports have suggested that Zen-based APUs won't arrive until 2017, following the release of desktop Zen CPUs in the fourth quarter of 2016.

As far as the use of HBM goes, we do know that some of next year's GPUs from both Nvidia (Pascal) and AMD (Arctic Islands) will make use of second generation High-Bandwidth Memory though it looks like this Zen-based APU will be using first generation HBM technology instead.

KitGuru Says: 2016 is going to be a big year for AMD, especially with its new Zen architecture coming out towards the end of the year. Are any of you waiting for AMD to release Zen at the moment? 

The post AMD said to be working on Zen APU with HBM first appeared on KitGuru.]]>
https://www.kitguru.net/components/apu-components/matthew-wilson/amd-said-to-be-working-on-zen-apu-with-hbm/feed/ 90
Micron comments on recent GDDR6 reports https://www.kitguru.net/components/memory/matthew-wilson/micron-comments-on-recent-gddr6-reports/ https://www.kitguru.net/components/memory/matthew-wilson/micron-comments-on-recent-gddr6-reports/#comments Wed, 16 Dec 2015 13:38:28 +0000 http://www.kitguru.net/?p=278370 Earlier this week, reports began circulating surrounding the existence of GDDR6 memory for future graphics cards. The news originated from news site, Fudzilla, which claimed to be in contact with sources familiar with the matter but it seems that right now, Micron is looking to squash any talk about GDDR6 and make its plans for …

The post Micron comments on recent GDDR6 reports first appeared on KitGuru.]]>
Earlier this week, reports began circulating surrounding the existence of GDDR6 memory for future graphics cards. The news originated from news site, Fudzilla, which claimed to be in contact with sources familiar with the matter but it seems that right now, Micron is looking to squash any talk about GDDR6 and make its plans for 2016 perfectly clear.

While the original report claimed that we would see GDDR6 in 2016, Micron has sent out a statement to various press outlets (including KitGuru) to clarify that it will only be launching GDDR5X next year, which is currently tipped to be used on future Nvidia Pascal graphics cards and some of AMD's future GPUs as well, in addition to HBM 2.

DocImage-Micron-4GB-DDR3-30nm

“The new memory advancements coming from Micron in 2016 are going to be called GDDR5X, not GDDR6. GDDR5X and GDDR6 are not the same product and Micron has not announced any plans involving GDDR6.”

The memory maker also went on to say that while GDDR5X is coming, it is not intended to be a competitor to HBM: “GDDR5X is intended to provide significant performance improvements to designs that are currently using GDDR5, therefore giving system designers the option of delivering enhanced performance without dramatically altering current architectures”.

That last part makes a lot of sense, particularly since we have been hearing rumours that HBM 2 will be reserved for future high-end graphics cards, like the next Fury or Titan, while GDDR5X will be used to provide a boost on lower-tier cards, that would have been designed for GDDR5 anyway.

You can find our initial story on GDDR6, HERE.

KitGuru Says: Micron was quick to comment on these rumours, so it looks like we will only be seeing GDDR5X next year alongside HBM 2 GPUs. 

The post Micron comments on recent GDDR6 reports first appeared on KitGuru.]]>
https://www.kitguru.net/components/memory/matthew-wilson/micron-comments-on-recent-gddr6-reports/feed/ 2
Sapphire Trixx unlocks AMD R9 Fury X voltage and HBM https://www.kitguru.net/components/graphic-cards/matthew-wilson/amd-sapphire-trixx-unlocks-amd-r9-fury-x-voltage-and-hbm/ https://www.kitguru.net/components/graphic-cards/matthew-wilson/amd-sapphire-trixx-unlocks-amd-r9-fury-x-voltage-and-hbm/#comments Wed, 18 Nov 2015 19:22:40 +0000 http://www.kitguru.net/?p=275611 When the AMD R9 Fury X launched this year, reviewers quickly found out that there would be limited overclocking potential due to voltage locks, as well as limitations placed on the speed the card's high bandwidth memory (HBM) could run at. Now, it looks like things are loosening up a bit, with Sapphire's Trixx 5.2.1 software …

The post Sapphire Trixx unlocks AMD R9 Fury X voltage and HBM first appeared on KitGuru.]]>
When the AMD R9 Fury X launched this year, reviewers quickly found out that there would be limited overclocking potential due to voltage locks, as well as limitations placed on the speed the card's high bandwidth memory (HBM) could run at. Now, it looks like things are loosening up a bit, with Sapphire's Trixx 5.2.1 software able to unlock both of those things.

This means the voltage of the R9 Fury X core can be unlocked, which means buyers may be able to access any untapped overclocking headroom and squeeze more performance out of the GPU.

Sapphire_20154161518

The Fury X shipped with a liquid cooler capable of 500 watts of thermal dissipation, this keeps the card very cool under gaming load. This also means that there is untapped thermal headroom for overclocking too, so now we just have to wait and see how much extra performance is possible with faster HBM and Fiji core speeds.

For reference, we were able to get our Fury X review sample from its base clock of 1050MHz to 1130MHz with the voltage restrictions in place. Fury X users can find the latest version of Sapphire Trixx, HERE. 

Discuss on our Facebook page, HERE.

KitGuru Says: Right now we don't know if other software suites will also offer HBM overclocking and over-volting for the Fury X. However, it will be interesting to see if Fury X owners can now squeeze some extra performance out of their cards. 

The post Sapphire Trixx unlocks AMD R9 Fury X voltage and HBM first appeared on KitGuru.]]>
https://www.kitguru.net/components/graphic-cards/matthew-wilson/amd-sapphire-trixx-unlocks-amd-r9-fury-x-voltage-and-hbm/feed/ 3
AMD rumoured to be using GDDR5x on new 14nm GPUs https://www.kitguru.net/components/graphic-cards/matthew-wilson/amd-rumoured-to-be-using-gddr5x-on-new-14nm-gpus/ https://www.kitguru.net/components/graphic-cards/matthew-wilson/amd-rumoured-to-be-using-gddr5x-on-new-14nm-gpus/#comments Wed, 28 Oct 2015 22:00:12 +0000 http://www.kitguru.net/?p=273489 Recently, we have been hearing quite a bit about GDDR5x, with rumours suggesting that Nvidia intends to use it in addition to HBM2 on its Pascal graphics cards next year. Now it would seem that AMD is also planning a similar move, with GDDR5x rumoured to appear on its new 14nm graphics cards next year …

The post AMD rumoured to be using GDDR5x on new 14nm GPUs first appeared on KitGuru.]]>
Recently, we have been hearing quite a bit about GDDR5x, with rumours suggesting that Nvidia intends to use it in addition to HBM2 on its Pascal graphics cards next year. Now it would seem that AMD is also planning a similar move, with GDDR5x rumoured to appear on its new 14nm graphics cards next year while we wait for HBM to become more widely available.

Right now, it is believed that AMD has priority access to HBM2, which may already be in fairly limited supply, which means there is still some room left in the market for GDDR5.

amd_radeon_fiji_gpu

According to a presentation by a DRAM maker, which leaked on to the web (Via: Techpowerup), GDDR5x will offer double the data-rate per memory access. GDDR5x is also believed to bring new speeds of up to 12Gbps initially, which is a significant improvement over the 7Gbps we are used to seeing on current GDDR5 chips. The big thing is that GDDR5x will use the same pins as GDDR5, which should make it easy for GPU makers to implement without needing to develop new designs.

From the way things are going, it sounds like GDDR5x will replace GDDR5 for both AMD and Nvidia on lower end graphics cards, while top-end GPUs will feature the more energy efficient HBM2. This is all rumour for now though, we should learn more about AMD and Nvidia's future architecture plans as we move in to 2016.

KitGuru Says: This is a pretty interesting turn. I wasn't really aware of GDDR5x until a few weeks ago and I imagine many were under the assumption that HBM would become more widely adopted next year, though that may only be partially true. At the very least, this move to GDDR5x will put less demand on HBM2 supplies, which means manufacturers will hopefully be able to keep up. 

The post AMD rumoured to be using GDDR5x on new 14nm GPUs first appeared on KitGuru.]]>
https://www.kitguru.net/components/graphic-cards/matthew-wilson/amd-rumoured-to-be-using-gddr5x-on-new-14nm-gpus/feed/ 17
Asus white Radeon R9 Nano spotted https://www.kitguru.net/components/graphic-cards/matthew-wilson/asus-white-radeon-r9-nano-spotted/ https://www.kitguru.net/components/graphic-cards/matthew-wilson/asus-white-radeon-r9-nano-spotted/#respond Mon, 26 Oct 2015 21:00:22 +0000 http://www.kitguru.net/?p=273118 AMD officially launched its tiny Radeon R9 Nano graphics card a while back, bringing the full power of a Fiji graphics core and High-Bandwidth Memory to a tiny 6-inch PCB. Several AIB partners have already launched their own branded models and now, it looks like Asus is planning to release a white version of the …

The post Asus white Radeon R9 Nano spotted first appeared on KitGuru.]]>
AMD officially launched its tiny Radeon R9 Nano graphics card a while back, bringing the full power of a Fiji graphics core and High-Bandwidth Memory to a tiny 6-inch PCB. Several AIB partners have already launched their own branded models and now, it looks like Asus is planning to release a white version of the card.

Dubbed as the ‘Asus Radeon R9 Nano White', the new GPU will feature the same specifications and design as previous R9 Nano graphics cards. However, this one being white will open a few doors for those that like to plan colour themed rigs.

R9 Nano White

The card was first spotted by German site, ComputerBase.de. As with the other Nano cards, specifications include a Fiji GPU with 4096 stream processors, 256 TMUs, 64 ROPs and 4GB of HBM. The base clock speed sits at 1GHz on the core and the card takes just one 8-pin PCIe connector for power.

Since press shots are already floating around, this white R9 Nano should be out fairly soon, though Asus has yet to make an official announcement. We also don't have a price just yet.

KitGuru Says: If anyone was planning an all white ITX rig, then this may be worth looking in to if you need the GPU power. Do any of you own an R9 Nano already? Or even a Fury card? What do you think of it? 

The post Asus white Radeon R9 Nano spotted first appeared on KitGuru.]]>
https://www.kitguru.net/components/graphic-cards/matthew-wilson/asus-white-radeon-r9-nano-spotted/feed/ 0
Rumour suggests Nvidia may use GDDR5X on Pascal GPUs https://www.kitguru.net/components/graphic-cards/matthew-wilson/rumour-suggests-nvidia-may-use-gddr5x-on-pascal-gpus/ https://www.kitguru.net/components/graphic-cards/matthew-wilson/rumour-suggests-nvidia-may-use-gddr5x-on-pascal-gpus/#comments Tue, 13 Oct 2015 23:36:00 +0000 http://www.kitguru.net/?p=271669 Nvidia is on track to launch its new Pascal architecture in 2016, with reports suggesting that Nvidia is already testing both its GP100 and GP104 chips. Now, new rumours circulating say that there will be a successor to GDDR5 video memory known as GDDR5X, which will début with some of Nvidia's upcoming Pascal based graphics …

The post Rumour suggests Nvidia may use GDDR5X on Pascal GPUs first appeared on KitGuru.]]>
Nvidia is on track to launch its new Pascal architecture in 2016, with reports suggesting that Nvidia is already testing both its GP100 and GP104 chips. Now, new rumours circulating say that there will be a successor to GDDR5 video memory known as GDDR5X, which will début with some of Nvidia's upcoming Pascal based graphics cards in addition to HBM2.

The GP100 GPU will be Nvidia's next ultra high-end card, like a Titan or Ti-series GPU. The GP104 will be the next step up from the GTX 970 and 980 graphics cards. Rumour has it that GDDR5X will keep the same 256-bit memory bus, but the memory bandwidth will be increased to 448GB/sec. HBM2 will also be used, likely on the higher tier GP100 graphics cards, though this has yet to be confirmed.

nvidia_pascal_module

HBM2 will include a 4096-bit memory bus, according to TweakTown, along with a 1GHz memory bandwidth capable of pushing 1TB/sec. Right now, this is all rumour so take this information with a grain of salt and keep an eye out for future information regarding Nvidia's upcoming Pascal architecture.

Discuss on our Facebook page, HERE.

KitGuru Says: HBM has been in the making for years now, so it will be interesting to see how it stacks up to the newly rumoured GDDR5X. How many of you are currently waiting on new architecture from either Nvidia or AMD? Will HBM factor in to your next GPU upgrade? 

The post Rumour suggests Nvidia may use GDDR5X on Pascal GPUs first appeared on KitGuru.]]>
https://www.kitguru.net/components/graphic-cards/matthew-wilson/rumour-suggests-nvidia-may-use-gddr5x-on-pascal-gpus/feed/ 12
AMD’s dual-GPU Radeon R9 Fury X2 ‘Gemini’ adapter spotted in India https://www.kitguru.net/components/graphic-cards/anton-shilov/amds-dual-gpu-radeon-r9-fury-x2-gemini-graphics-card-spotted-in-india/ https://www.kitguru.net/components/graphic-cards/anton-shilov/amds-dual-gpu-radeon-r9-fury-x2-gemini-graphics-card-spotted-in-india/#comments Tue, 06 Oct 2015 01:33:37 +0000 http://www.kitguru.net/?p=270710 Advanced Micro Devices has begun to test its upcoming dual-chip graphics card internally. The company shipped six samples of its dual-GPU Radeon graphics adapters to India, which means that the company is developing drivers for the solution. According to Zauba.com database, which monitors imports and exports to and from India, where high-tech companies employ a …

The post AMD’s dual-GPU Radeon R9 Fury X2 ‘Gemini’ adapter spotted in India first appeared on KitGuru.]]>
Advanced Micro Devices has begun to test its upcoming dual-chip graphics card internally. The company shipped six samples of its dual-GPU Radeon graphics adapters to India, which means that the company is developing drivers for the solution.

According to Zauba.com database, which monitors imports and exports to and from India, where high-tech companies employ a lot of software developers, AMD shipped six “Fiji Gemini” graphics cards from Canada to India. The graphics boards carry two “Fiji” graphics processing units and a cooling system designed by Cooler Master, reports WccfTech.

Typically, AMD begins to send samples of its graphics cards to India months before the announcement. If AMD shipped the very first samples of its dual-GPU Radeon R9 Fury X2 graphics boards to its Indian developers in late September, it may indicate that the product will be formally introduced very late in 2015 or even early in 2016.

amd_radeon_fiji_fury_maxx
Dual-GPU AMD Radeon R9 Fury. Image by Anshel Sag, staff technologist and technical writer at Moor Insights & Strategy.

The upcoming dual-chip AMD Radeon R9 Fury X graphics card – which may be officially called Radeon R9 Fury X2 or Radeon R9 Fury Maxx – features two code-named “Fiji XT” GPUs with 4096 stream processors and 4GB of HBM memory. In total, the graphics solution features 8192 stream processors and 8GB of high-bandwidth memory. The board has two 8-pin PCI Express power connectors, which means that it can consume up to 375W of power.

Manufacturer suggested retail price of AMD’s Radeon Fury X is $649. It is highly likely that the dual-GPU product will cost over $1000.

AMD did not comment on the news-story.

Discuss on our Facebook page, HERE.

KitGuru Says: Looks like AMD Radeon R9 Fury X2 will not be introduced this fall, but will likely hit the market only in winter.

The post AMD’s dual-GPU Radeon R9 Fury X2 ‘Gemini’ adapter spotted in India first appeared on KitGuru.]]>
https://www.kitguru.net/components/graphic-cards/anton-shilov/amds-dual-gpu-radeon-r9-fury-x2-gemini-graphics-card-spotted-in-india/feed/ 19
Nvidia’s next-gen ‘Pascal’ graphics cards will get 16GB of HBM2 memory https://www.kitguru.net/components/graphic-cards/anton-shilov/nvidias-next-gen-high-end-graphics-cards-will-get-16gb-of-hbm2-memory/ https://www.kitguru.net/components/graphic-cards/anton-shilov/nvidias-next-gen-high-end-graphics-cards-will-get-16gb-of-hbm2-memory/#comments Thu, 24 Sep 2015 01:57:48 +0000 http://www.kitguru.net/?p=269090 At the GPU Technology Conference in Japan, Nvidia Corp. once again revealed key features of its next-generation graphics processing architecture code-named “Pascal”. As a it appears, the company has slightly changed its plans concerning memory capacity supported by its upcoming GPUs. As expected, Nvidia’s high-end graphics processor that belongs to the “Pascal” family will feature an …

The post Nvidia’s next-gen ‘Pascal’ graphics cards will get 16GB of HBM2 memory first appeared on KitGuru.]]>
At the GPU Technology Conference in Japan, Nvidia Corp. once again revealed key features of its next-generation graphics processing architecture code-named “Pascal”. As a it appears, the company has slightly changed its plans concerning memory capacity supported by its upcoming GPUs.

As expected, Nvidia’s high-end graphics processor that belongs to the “Pascal” family will feature an all-new architecture with a number of exclusive innovations, including mixed precision (for the first time Nvidia’s stream processors will support FP16, FP32 and FP64 precision), NVLink interconnection technology for supercomputers and multi-GPU configurations, unified memory addressing as well as support for second-generation high-bandwidth memory (HBM generation 2).

Based on a slide that Nvidia demonstrated at the GTC Japan 2015, next-generation high-end graphics cards with “Pascal” GPUs will sport up to 16GB of HBM2 with up to 1TB/s bandwidth. Previously Nvidia expected select solutions with “Pascal” graphics processors to feature up to 32GB of HBM2.

nvidia_pascal_expectations_gtc_2015_japan
Nvidia “Pascal” highlights. Image by WccfTech

Given the fact that Nvidia does not produce high-bandwidth memory itself, but relies on supplies from companies like Samsung Electronics and SK Hynix, changes of their roadmaps can affect Nvidia’s plans. In order to install 32GB of HBM2 memory on a graphics processor with a 4096-bit memory bus, 8GB memory chips are used. While the HBM2 specification allows to build such ICs [integrated circuits], it is not easy to manufacture packages with eight vertically stacked 8Gb memory dies. As a result, such chips may be delayed from 2016 to a later date.

16GB of HBM2 memory should be enough for gaming and professional graphics cards, but high-performance computing applications could take advantage of 32GB of onboard memory even now.

Discuss on our Facebook page, HERE.

KitGuru Says: If Nvidia is not able to get 8GB HBM2 chips next year, AMD will not be able to get them as well. Therefore, expect graphics cards with up to 16GB of high-bandwidth memory from both companies.

The post Nvidia’s next-gen ‘Pascal’ graphics cards will get 16GB of HBM2 memory first appeared on KitGuru.]]>
https://www.kitguru.net/components/graphic-cards/anton-shilov/nvidias-next-gen-high-end-graphics-cards-will-get-16gb-of-hbm2-memory/feed/ 44
XFX is working on liquid-cooled Radeon R9 Fury graphics card https://www.kitguru.net/components/graphic-cards/anton-shilov/xfx-is-working-on-liquid-cooled-radeon-r9-fury-graphics-card/ https://www.kitguru.net/components/graphic-cards/anton-shilov/xfx-is-working-on-liquid-cooled-radeon-r9-fury-graphics-card/#comments Wed, 23 Sep 2015 22:28:08 +0000 http://www.kitguru.net/?p=269015 XFX is gearing up to release a Radeon R9 Fury-based graphics adapter featuring a liquid cooling system. The card will feature overclocking potential on par with AMD’s Radeon R9 Fury X, which will help to at least partially close performance gap between the “Fury” and the “Fury X” products. The XFX Radeon R9 Fury with …

The post XFX is working on liquid-cooled Radeon R9 Fury graphics card first appeared on KitGuru.]]>
XFX is gearing up to release a Radeon R9 Fury-based graphics adapter featuring a liquid cooling system. The card will feature overclocking potential on par with AMD’s Radeon R9 Fury X, which will help to at least partially close performance gap between the “Fury” and the “Fury X” products.

The XFX Radeon R9 Fury with liquid cooling will rely on AMD’s reference design for its graphics cards based on the “Fiji” graphics processing units, but will utilize a cooling system similar to that installed on the flagship AMD Radeon R9 Fury X graphics products. In fact, the whole design of the graphics adapter resembles that of AMD’s top-of-the-range offering, according to images published by VideoCardz web-site.

xfx_radeon_r9_fury_liquid_cooling

Typically, AMD Radeon R9 Fury graphics cards powered by cut-down “Fiji Pro” graphics chips with 3584 stream processors, 224 texture units, 64 raster operations pipelines and 4096 high-bandwidth memory bus up to 25 per cent slower than the Radeon R9 Fury X adapters with fully-fledged “Fiji XT” GPUs (4096 SPs, 256 Tus, 64 ROPs, etc.). However, if liquid cooling system helps to significantly overclock the “Fiji Pro” GPU, then performance difference between AMD’s enthusiast-class graphics cards will shrink.

AMD and XFX did not comment on the news-story.

Discuss on our Facebook page, HERE.

KitGuru Says: It looks like AMD is preparing to let all of its partners to sell the Radeon R9 Fury products. At present only Asustek Computer and Sapphire Technology supply graphics cards featuring “Fiji Pro” GPUs, but shortly everything is expected to change.

The post XFX is working on liquid-cooled Radeon R9 Fury graphics card first appeared on KitGuru.]]>
https://www.kitguru.net/components/graphic-cards/anton-shilov/xfx-is-working-on-liquid-cooled-radeon-r9-fury-graphics-card/feed/ 2
Nvidia gets first samples of GP100 from TSMC, begins internal tests https://www.kitguru.net/components/graphic-cards/anton-shilov/nvidia-receives-first-samples-of-gp100-chips-from-tsmc-begins-to-test-them/ https://www.kitguru.net/components/graphic-cards/anton-shilov/nvidia-receives-first-samples-of-gp100-chips-from-tsmc-begins-to-test-them/#comments Wed, 23 Sep 2015 00:18:13 +0000 http://www.kitguru.net/?p=268782 Taiwan Semiconductor Manufacturing Co. has successfully produced the first samples of Nvidia Corp.’s code-named GP100 graphics processing unit. Nvidia has already started to test the chip internally and should be on-track to release the GPU commercially in mid-2016. 3DCenter reports that Nvidia has sent the first graphics cards based on the GP100 graphics processor to its …

The post Nvidia gets first samples of GP100 from TSMC, begins internal tests first appeared on KitGuru.]]>
Taiwan Semiconductor Manufacturing Co. has successfully produced the first samples of Nvidia Corp.’s code-named GP100 graphics processing unit. Nvidia has already started to test the chip internally and should be on-track to release the GPU commercially in mid-2016.

3DCenter reports that Nvidia has sent the first graphics cards based on the GP100 graphics processor to its subsidiary in India, where it has a lot of hardware and software developers. No actual details about the chip or graphics cards on its base are known, but it is about time for the graphics giant to start testing its GP100.

Nvidia taped out the GP100 in June, 2015. Production cycle of TSMC’s 16nm FinFET process technology is about 90 days, therefore Nvidia got its GP100 from TSMC very recently. Right now the company is testing the chip and its drivers internally.

nvidia_artwork_iron

Nvidia’s GP100 graphics processing unit is based on the “Pascal” architecture and is made using 16nm FinFET+ process technology. The chip is expected to integrate up to 6000 stream processors and contain around 17 billion transistors. Graphics cards featuring the GP100 will carry up to 32GB of HBM2 memory.

Nvidia did not comment on the news-story.

Discuss on our Facebook page, HERE.

KitGuru Says: It is about time for Nvidia to start testing its GP100 now. What remains to be seen is when exactly the company plans to formally introduce its next-generation GPUs. If the first revision of the chip is fully functional, the company may move in introduction of the GP100 to the first quarter of the year.

The post Nvidia gets first samples of GP100 from TSMC, begins internal tests first appeared on KitGuru.]]>
https://www.kitguru.net/components/graphic-cards/anton-shilov/nvidia-receives-first-samples-of-gp100-chips-from-tsmc-begins-to-test-them/feed/ 20
Nvidia to fetch second-gen HBM memory from Samsung and SK Hynix – Report https://www.kitguru.net/components/graphic-cards/anton-shilov/nvidia-to-fetch-second-gen-hbm-memory-from-samsung-and-sk-hynix-report/ https://www.kitguru.net/components/graphic-cards/anton-shilov/nvidia-to-fetch-second-gen-hbm-memory-from-samsung-and-sk-hynix-report/#comments Tue, 22 Sep 2015 07:00:42 +0000 http://www.kitguru.net/?p=268605 As developers of graphics processing units start to use high-bandwidth memory (HBM), manufacturers of graphics cards cease their ability to buy memory directly from DRAM makers. A good news is that Nvidia Corp. intends to buy second-gen HBM from two suppliers, which potentially means flexible pricing. Nowadays manufacturers of graphics cards can buy GDDR5 memory …

The post Nvidia to fetch second-gen HBM memory from Samsung and SK Hynix – Report first appeared on KitGuru.]]>
As developers of graphics processing units start to use high-bandwidth memory (HBM), manufacturers of graphics cards cease their ability to buy memory directly from DRAM makers. A good news is that Nvidia Corp. intends to buy second-gen HBM from two suppliers, which potentially means flexible pricing.

Nowadays manufacturers of graphics cards can buy GDDR5 memory from different manufacturers. However, when it comes to HBM, memory needs to be placed near the chip on a special silicon interposer, which means that a GPU developer has to buy DRAM chips, install them and test the “GPU package.” Advanced Micro Devices buys HBM memory chips from SK Hynix to place it near its “Fiji” graphics processing unit and then sells the whole “package” to actual suppliers of graphics cards.

nvidia_pascal_module

The first generation of HBM is manufactured only by SK Hynix. However, the second-generation HBM will be available from both Samsung Electronics and SK Hynix. As it appears, Nvidia will buy such memory from both makers for its “Pascal” GPUs, reports Business Korea. AMD is expected to do the same, yet it has not been confirmed.

Availability of HBM from two suppliers means lower prices, which is a good news for makers of graphics cards and the end user.

Nvidia did not comment on the news-story.

Discuss on our Facebook page, HERE.

KitGuru Says: In fact, both Nvidia and AMD (ATI Technologies, to be precise) used to report sales of both GPU and GDDR memory back in the mid-2000s. Actual manufacturers of graphics cards, such as Asustek Computer, Gigabyte Technology and Sapphire Technology, then managed to persuade GPU developers to sell GPUs without memory. Nowadays, apparently, it all comes back.

The post Nvidia to fetch second-gen HBM memory from Samsung and SK Hynix – Report first appeared on KitGuru.]]>
https://www.kitguru.net/components/graphic-cards/anton-shilov/nvidia-to-fetch-second-gen-hbm-memory-from-samsung-and-sk-hynix-report/feed/ 2
Nvidia changes roadmap: ‘Volta’ is now due in 2018 https://www.kitguru.net/components/graphic-cards/anton-shilov/nvidia-changes-roadmap-volta-is-now-due-in-2018/ https://www.kitguru.net/components/graphic-cards/anton-shilov/nvidia-changes-roadmap-volta-is-now-due-in-2018/#comments Tue, 22 Sep 2015 01:14:41 +0000 http://www.kitguru.net/?p=268590 Nvidia Corp. has slightly changed its roadmap concerning GPU architectures. As it appear, its next-gen GPUs code-named “Pascal” are now due in 2016, whereas their successors will be released only in 2018. Based on a new roadmap that Nvidia showcased at a tech conference in Japan, the company will release its code-named “Pascal” GPUs in 2016 …

The post Nvidia changes roadmap: ‘Volta’ is now due in 2018 first appeared on KitGuru.]]>
Nvidia Corp. has slightly changed its roadmap concerning GPU architectures. As it appear, its next-gen GPUs code-named “Pascal” are now due in 2016, whereas their successors will be released only in 2018.

Based on a new roadmap that Nvidia showcased at a tech conference in Japan, the company will release its code-named “Pascal” GPUs in 2016 and will follow on with “Volta” graphics processors in 2018. The “Pascal” chips will be made using 16nm FinFET process technology and will be available in 2016, reports WccfTech. Previously “Volta” was expected in 2017.

NVIDIA-Pascal-GPU_Roadmap

Not a lot is known about the first “Pascal” GPU. Nvidia has reportedly taped out its GP100 graphics processor back in June. Given the timeframe of the tape-out, it is highly likely that Nvidia uses TSMC’s advanced 16nm FinFET+ (CLN16FF+) manufacturing technology. Nvidia has changed its approach to roll-out of new architectures. Instead of starting from simple GPUs and introducing biggest processors quarters after the initial chips, Nvidia will begin to roll-out 16nm “Pascal” GPUs with the largest chip in the family.

Nvidia did not comment on the news-story.

Discuss on our Facebook page, HERE.

KitGuru Says: It looks like Nvidia is pulling in “Pascal”, but slightly delays “Volta”. The reason for this is simple: 10nm process technology. At TSMC it will only be available for Nvidia in 2018.

The post Nvidia changes roadmap: ‘Volta’ is now due in 2018 first appeared on KitGuru.]]>
https://www.kitguru.net/components/graphic-cards/anton-shilov/nvidia-changes-roadmap-volta-is-now-due-in-2018/feed/ 32
PowerColor, XFX prep to start shipments of AMD Radeon R9 Fury graphics cards https://www.kitguru.net/components/graphic-cards/anton-shilov/powercolor-xfx-are-gearing-up-to-start-shipments-of-amd-radeon-r9-fury-graphics-cards/ https://www.kitguru.net/components/graphic-cards/anton-shilov/powercolor-xfx-are-gearing-up-to-start-shipments-of-amd-radeon-r9-fury-graphics-cards/#comments Sat, 19 Sep 2015 16:20:37 +0000 http://www.kitguru.net/?p=268494 PowerColor and XFX are preparing to add another graphics card into their range of adapters based on AMD’s “Fiji” graphics processing unit. In the coming weeks the two companies will start to sell their AMD Radeon R9 Fury graphics cards, thus, expanding availability of AMD’s latest product. Although all partners of Advanced Micro Devices formally …

The post PowerColor, XFX prep to start shipments of AMD Radeon R9 Fury graphics cards first appeared on KitGuru.]]>
PowerColor and XFX are preparing to add another graphics card into their range of adapters based on AMD’s “Fiji” graphics processing unit. In the coming weeks the two companies will start to sell their AMD Radeon R9 Fury graphics cards, thus, expanding availability of AMD’s latest product.

Although all partners of Advanced Micro Devices formally offer AMD Radeon R9 Fury and AMD Radeon R9 Nano graphics adapters based on the fully-fledged Fiji GPUs with 4096 stream processors, 256 texture units, 64 raster operations pipelines and 4096-bit memory bus, far not all can supply AMD Radeon R9 Fury graphics cards powered by cut-down “Fiji” chips (3584 SPs, 224 TUs, 64 ROPs). In fact, only Asustek Computer and Sapphire Technology sell AMD Radeon R9 Fury graphics cards today. Fortunately, that is going to change soon.

powercolor_radeon_r9_fury

PowerColor, a Tul Corp.’s brand, recently quietly added description of its Radeon R9 Fury graphics adapter to its web-site. The board is currently not available anywhere, but its addition into the list of products clearly indicates that the company is on-track to start selling it in the coming weeks.

XFX, a Pine Holdings Limited’s brand, is also gearing up to start shipments of its Radeon R9 Fury graphics card. VideoCardz web-site has published press images of XFX’s yet-unannounced R9 Fury, which is an indicator that the product is about to be formally introduced.

xfx_radeon_r9_fury_1

The Radeon R9 Fury graphics cards from PowerColor and XFX rely on AMD’s reference design and come with 4GB of high-bandwidth memory and massive triple-fan/three-slot cooling systems with multiple copper heat-pipes to maximize their overclocking potential. The cards require two auxiliary 8-pin PCI Express power connectors. In fact, both boards will be very similar to Sapphire’s Radeon R9 Fury Tri-X.

xfx_radeon_r9_fury

At present, it is unclear when companies like Gigabyte Technology, MicroStar International and other plan to start offering their versions of the Radeon R9 Fury.

The manufacturer suggested retail price of AMD Radeon R9 Fury graphics cards is $549/£439.

Discuss on our Facebook page, HERE.

KitGuru Says: While it is great to see that availability of AMD’s Radeon R9 Fury is improving, the lack of creative touch on those cards is evident. So far, only Asustek Computer has designed a custom printed-circuit board for its Radeon R9 Fury Strix. All the others continue to use AMD’s reference design, probably because they just buy actual cards from the GPU developer.

The post PowerColor, XFX prep to start shipments of AMD Radeon R9 Fury graphics cards first appeared on KitGuru.]]>
https://www.kitguru.net/components/graphic-cards/anton-shilov/powercolor-xfx-are-gearing-up-to-start-shipments-of-amd-radeon-r9-fury-graphics-cards/feed/ 4
Shortages of AMD Radeon R9 Fury-series graphics cards persist https://www.kitguru.net/components/graphic-cards/anton-shilov/amd-radeon-r9-fury-series-graphics-cards-continue-to-be-in-short-supply-analyst/ https://www.kitguru.net/components/graphic-cards/anton-shilov/amd-radeon-r9-fury-series-graphics-cards-continue-to-be-in-short-supply-analyst/#comments Wed, 16 Sep 2015 22:08:19 +0000 http://www.kitguru.net/?p=268067 About three months since the formal announcement, Advanced Micro Devices’ flagship products based on the code-named “Fiji” graphics processing units are still in extremely short supply, according to a market analyst and various retailers. At present, AMD offers Radeon R9 Fury X and Radeon R9 Nano based on the fully-fledged “Fiji” graphics chip with 4096 …

The post Shortages of AMD Radeon R9 Fury-series graphics cards persist first appeared on KitGuru.]]>
About three months since the formal announcement, Advanced Micro Devices’ flagship products based on the code-named “Fiji” graphics processing units are still in extremely short supply, according to a market analyst and various retailers.

At present, AMD offers Radeon R9 Fury X and Radeon R9 Nano based on the fully-fledged “Fiji” graphics chip with 4096 stream processors, 256 texture units, 64 raster operations pipelines and 4096-bit memory bus. Both the “Fury X” and the “Nano” graphics adapters are based on AMD’s design and are made under supervision of the company by a contract manufacturer. Asustek Computer and Sapphire Technology also offer AMD Radeon R9 Fury graphics cards powered by cut-down “Fiji” chips (3584 SPs, 224 TUs, 64 ROPs) and featuring custom design of cooling systems and printed circuit boards.

amd_radeon_r9_fury_card

As it appears, AMD does not sell its cut-down “Fiji” graphics processors to many of its partners and does not allow the latter to customize adapters that feature fully-fledged “Fiji” GPUs. Consequently, for suppliers of graphics cards it makes more sense to sell Nvidia-based solutions right now.

“AMD’s flagship Fury X and Fury continue to have a very limited number of SKUs and are largely out of stock, suggesting that the pace of availability and ecosystem support is discouraging add-in-board (AIB) manufacture,” wrote Ian Ing, an analyst with MKM Partners, in an article at Barron’s.

amd_radeon_r9_fury_x_exploded

Overclockers UK, a major online store in the U.K., offers to pre-order Radeon R9 Fury X, but has Radeon R9 Fury from Asustek Computer and Sapphire Technology in stock. AMD Radeon R9 Nano graphics cards are available from multiple suppliers. In fact, OcUK even offers a small discount on the Sapphire Radeon R9 Nano and sells them for £515.

Newegg.com, a major online store in the U.S., at press time only had Radeon R9 Fury X from VisionTek in stock. The store also has Radeon R9 Fury from Asustek Computer and Sapphire Technology available. AMD Radeon R9 Nano graphics cards are available from several brands, but all are sold at prices that exceed MSRP [manufacturer suggested retail price].

CaseKing.de, a major German retailer, does not have a single Radeon R9 Fury X in stock, but has multiple Radeon R9 Nano from different suppliers as well as Radeon R9 Fury from Asustek and Sapphire.

amd_radeon_fiji

Insufficient availability of the AMD Radeon R9 Fury-series products have challenged AMD for about a quarter. The reasons for tight supply of the “Fiji” graphics processor are unclear. If AMD ran into yield issues with its largest graphics processor ever, then there would be products based on cut-down versions of “Fiji” available (e.g., Radeon R9 Fury or Radeon R9 Fury LE). However, the R9 Fury sold only by Asus and Sapphire, which means that there are not a lot of such chips available from AMD. The Radeon R9 Nano graphics cards are in stock of retailers, but it is the first week of its availability.

Limited availability of AMD Radeon R9 Fury-series graphics cards keeps prices of such graphics cards high and does not allow AMD to recapture market share it lost to Nvidia Corp. in the recent quarters.

amd_radeon_r9_nano_1

AMD is in its quiet period and could not comment on the news-story.

Discuss on our Facebook page, HERE.

KitGuru Says: It looks like the amount of “Fiji” graphics processors in production is limited. As a result, if you are planning to get a Radeon R9 Fury graphics card, grab one while you can. For example, the Radeon R9 Nano can be acquired in the first week of availability. Perhaps, AMD will solve the issues with availability of its “Fiji” in the coming months, but not in the coming weeks or days.

The post Shortages of AMD Radeon R9 Fury-series graphics cards persist first appeared on KitGuru.]]>
https://www.kitguru.net/components/graphic-cards/anton-shilov/amd-radeon-r9-fury-series-graphics-cards-continue-to-be-in-short-supply-analyst/feed/ 9
AquaComputer unveils full-cover water-block for Radeon R9 Nano https://www.kitguru.net/components/graphic-cards/anton-shilov/aquacomputer-unveils-full-cover-water-block-for-amd-radeon-r9-nano/ https://www.kitguru.net/components/graphic-cards/anton-shilov/aquacomputer-unveils-full-cover-water-block-for-amd-radeon-r9-nano/#comments Fri, 11 Sep 2015 22:12:05 +0000 http://www.kitguru.net/?p=267398 AquaComputer, a leading maker of components for custom liquid-cooling solutions, on Friday introduced its full-cover water-block for AMD Radeon R9 Nano graphics cards, thus making high-performance liquid-cooled gaming PCs in mini-ITX form-factor a reality. The Kryographics R9 Nano single-slot full-cover water-block for AMD Radeon R9 Nano graphics boards is made of pure electrolytic copper with …

The post AquaComputer unveils full-cover water-block for Radeon R9 Nano first appeared on KitGuru.]]>
AquaComputer, a leading maker of components for custom liquid-cooling solutions, on Friday introduced its full-cover water-block for AMD Radeon R9 Nano graphics cards, thus making high-performance liquid-cooled gaming PCs in mini-ITX form-factor a reality.

The Kryographics R9 Nano single-slot full-cover water-block for AMD Radeon R9 Nano graphics boards is made of pure electrolytic copper with either acrylic or aluminum top cover. Water channels in the block pass over the GPU+HBM package as well as voltage regulator module to cool-down the inductors, which get rather hot with the stock cooling solution. The GPU+HBM package is covered by a micro structure with 0.5mm grid to ensure maximum cooling performance.

aquacomputer_liquid_cooler_water_radeon_r9_nano

According to AquaComputer, the GPU temperature of the Radeon R9 Nano remains at below 35°C in typical conditions (with room/liquid temperature of around 27°C) when its water-block is used, which is massively cooler compared to AMD’s stock cooling system.

The Kryographics R9 Nano water-block is designed for custom-built liquid-cooled solutions and is aimed at experienced enthusiasts who are seeking for either additional performance from AMD’s smallest and most powerful graphics card, or want to make their PCs completely silent. What should be noted is that just a few of mini-ITX PC cases can support custom liquid cooling systems.

Traditionally, water-blocks from AquaComputer are easy to install and are compatible with various tubes and various pumps from the company and its competitors.

aquacomputer_liquid_cooler_water_radeon_r9_nano_1

According to TechPowerUp, different versions of the AquaComputer Kryographics R9 Nano full-cover water-block will be available shortly. Pricing looks as follows:

  • Kryographics R9 Nano – €94.9
  • Kryographics R9 Nano acrylic glass edition – €104.9
  • Kryographics R9 Nano black edition – €104.9
  • Kryographics R9 Nano, nickel plated version – €109.9
  • Kryographics R9 Nano acrylic glass edition, nickel plated version – €119.9
  • Kryographics R9 Nano black edition, nickel plated version – €119.9

Discuss on our Facebook page, HERE.

KitGuru Says: While the Kryographics R9 Nano single-slot full-cover water-block for AMD Radeon R9 Nano is a unique product, its market prospects are completely unclear. Those, who want to have extreme performance, will naturally buy AMD Radeon R9 Fury X. Meanwhile, small form-factor systems do not use liquid cooling. On the other hand, maybe the Radeon R9 Nano will just redefine SFF PCs as we know them?

The post AquaComputer unveils full-cover water-block for Radeon R9 Nano first appeared on KitGuru.]]>
https://www.kitguru.net/components/graphic-cards/anton-shilov/aquacomputer-unveils-full-cover-water-block-for-amd-radeon-r9-nano/feed/ 6
Micron to unveil third-gen hybrid memory cube in 2016 https://www.kitguru.net/components/memory/anton-shilov/micron-to-unveil-third-gen-hybrid-memory-cube-in-2016/ https://www.kitguru.net/components/memory/anton-shilov/micron-to-unveil-third-gen-hybrid-memory-cube-in-2016/#comments Tue, 08 Sep 2015 19:42:08 +0000 http://www.kitguru.net/?p=266781 Micron Technology is working on the third generation of its hybrid memory cube technology, which will be unveiled next year. If the third-gen HMC doubles data rate of the memory from 15Gb/s today, then the new memory technology will offer unprecedented bandwidth. HMC uses advanced through-silicon vias (TSVs) – vertical conduits that electrically connect a …

The post Micron to unveil third-gen hybrid memory cube in 2016 first appeared on KitGuru.]]>
Micron Technology is working on the third generation of its hybrid memory cube technology, which will be unveiled next year. If the third-gen HMC doubles data rate of the memory from 15Gb/s today, then the new memory technology will offer unprecedented bandwidth.

HMC uses advanced through-silicon vias (TSVs) – vertical conduits that electrically connect a stack of individual chips – to combine Micron’s high-performance logic with Micron's dynamic random access memory. At present Micron offers 2GB and 4GB HMC devices based on 4Gb DRAM dies with up to 15Gb/s data-rates and up to 160GB/s peak bandwidth per device.

ibm_micron_hmc

At present, Micron is working on next-generation HMC, which could improve density and bandwidth, reports HPC Wire. Nothing particular is known about the third-gen HMC, but it is logical to expect Micron to improve performance and capacity of HMC.

“To be frank, we cannot achieve the applications and system needs without developing a really good packaging technology,” said Scott Graham, Micron’s general manager of Hybrid Memory at the Intel Developer Forum last month. “We’re not going to achieve these bandwidth capabilities. We’re not going to achieve the reliability needs. We’re not going to overcome some of the scaling challenges without developing some of these new technology methods. If you look at Hybrid Memory Cube, that’s been the lead vehicle for Micron in order to develop these package technologies for future emerging memories.”

micron_altera_hmc_chip

HMC is supported by various special-purpose microprocessors. In addition, HMC will be used by Intel’s next-generation Xeon Phi co-processor code-named “Knights Landing”. HMC competes against high-bandwidth memory (HBM), but is based on different principles. Instead of using ultra-wide interface, HMC features up to 16-bit serial interconnect operating at a very high clock-rate. Maximum bandwidth supported by one HBM chip is up to 128GB/s, but second-generation HBM will have bandwidth of up to 256GB/s.

“If we have the ability to take DRAM and stack it on top of a logic layer and SoC and be able to control that DRAM with that SoC, it allows us to overcome scaling challenges,” said Mr. Graham. “Being able to combine these technologies together, gives us unprecedented memory bandwidth that keeps pace with multiple CPU cores, and DRAM alone is not going to do that. This all allows for increased savings in energy/bit, density in a small form factor, higher performance and lower energy, and compelling RAS features.”

Discuss on our Facebook page, HERE.

KitGuru Says: HMC makes a great business sense for Micron, who is the only producer of such memory and can control pricing. However, for developers of bandwidth-demanding applications it makes a lot of sense to use HBM instead of HMC because of pricing concerns. HMC has a number of advantages over HBM, but for those, who do not need features like RAS, its advantages are not important. The arrival of HBM will affect popularity of HMC.

The post Micron to unveil third-gen hybrid memory cube in 2016 first appeared on KitGuru.]]>
https://www.kitguru.net/components/memory/anton-shilov/micron-to-unveil-third-gen-hybrid-memory-cube-in-2016/feed/ 2
AMD: We are actively promoting HBM and do not collect royalties https://www.kitguru.net/components/graphic-cards/anton-shilov/amd-we-are-actively-promoting-usage-of-hbm-and-do-not-collect-royalties/ https://www.kitguru.net/components/graphic-cards/anton-shilov/amd-we-are-actively-promoting-usage-of-hbm-and-do-not-collect-royalties/#comments Fri, 04 Sep 2015 12:19:21 +0000 http://www.kitguru.net/?p=266267 Advanced Micro Devices is currently the only company to use high-bandwidth memory it co-developed with SK Hynix and other partners. While usage of HBM is clearly a competitive advantage that AMD has over its rivals, the company is encouraging others to use the new memory type and does not intend to collect any royalties for …

The post AMD: We are actively promoting HBM and do not collect royalties first appeared on KitGuru.]]>
Advanced Micro Devices is currently the only company to use high-bandwidth memory it co-developed with SK Hynix and other partners. While usage of HBM is clearly a competitive advantage that AMD has over its rivals, the company is encouraging others to use the new memory type and does not intend to collect any royalties for HBM.

It has taken AMD over 8.5 years to develop high-bandwidth memory and a set of technologies that enable new DRAM chips as well as their interface with host processors. In a bid to make HBM memory more affordable, more developers of graphics processors and other bandwidth-demanding chips have to use this technology. Once HBM gets less expensive than today, the industry in general will benefit from this.

amd_sk_hynix_hbm_implementation_1

Nvidia Corp. has publicly revealed that its next-generation “Pascal” architecture of graphics processors supports second-generation HBM memory and in 2016 graphics cards with GP100 GPUs and HBM2 DRAM will hit the market. In fact, HBM support is a key feature of Nvidia's “Pascal”, which will help the company to triple the bandwidth available to its next-gen GPUs, thus significantly improving their performance.

nvidia_pascal_3d_memory_roadmap_geforce

Earlier this week a web-site reported that Nvidia will delay adoption of HBM because of royalties demanded by AMD for its HBM-related intellectual property. In particular, it is reported that 2.5D GPU packaging used for the “Fiji” graphics processing unit is covered by AMD’s patents and Nvidia either needs to design its own package or pay licensing fees to its rival. According to AMD, this is not true.

“AMD is not involved in collecting any royalties for HBM,” said Iain Bristow, a spokesman for AMD. “We are actively encouraging widespread adoption of all HBM associated technology on [Radeon R9] Fury products and there is no IP licensing associated.”

Advanced Micro Devices owns a number of patents covering HBM, but as that intellectual property is a part of JEDEC’s JESD235 standard, it has to be licensed to applicants desiring to implement the standard “either without compensation or under reasonable terms and conditions that are free of any unfair discrimination.” Moreover, AMD and Nvidia have a broad cross-licensing agreement, which largely prevents royalty demands.

amd_sk_hynix_hbm_design

Nvidia did not comment on the news-story and called the information “rumours and speculation.”

Discuss on our Facebook page, HERE.

KitGuru Says: Adoption of HBM memory by Nvidia may be slow-downed not by AMD, but by availability of second-generation HBM. Perhaps, SK Hynix and Samsung Electronics will only release next-gen HBM in the second half of 2016, which means that HBM2 will be available in mass quantities only in 2017.

The post AMD: We are actively promoting HBM and do not collect royalties first appeared on KitGuru.]]>
https://www.kitguru.net/components/graphic-cards/anton-shilov/amd-we-are-actively-promoting-usage-of-hbm-and-do-not-collect-royalties/feed/ 50
AMD will not allow partners to modify specs of Radeon R9 Nano https://www.kitguru.net/components/graphic-cards/anton-shilov/amd-will-not-allow-partners-to-modify-specs-of-radeon-r9-nano/ https://www.kitguru.net/components/graphic-cards/anton-shilov/amd-will-not-allow-partners-to-modify-specs-of-radeon-r9-nano/#comments Tue, 01 Sep 2015 15:22:04 +0000 http://www.kitguru.net/?p=265767 Even though Advanced Micro Devices will allow its partners to build their own versions of AMD Radeon R9 Nano graphics cards, it will not let them significantly alter specifications of such graphics adapters. As a result, the difference between the original Radeon R9 Nano and custom versions from AMD’s partners will be minimal. In about …

The post AMD will not allow partners to modify specs of Radeon R9 Nano first appeared on KitGuru.]]>
Even though Advanced Micro Devices will allow its partners to build their own versions of AMD Radeon R9 Nano graphics cards, it will not let them significantly alter specifications of such graphics adapters. As a result, the difference between the original Radeon R9 Nano and custom versions from AMD’s partners will be minimal.

In about three months from now Advanced Micro Devices will allow its partners to build custom versions of Radeon R9 Nano graphics adapters. The only thing that AMD’s partners will be allowed to modify is the cooling system of the product. Producers of graphics adapters will not be able to increase clock-rates of their Radeon R9 Nano or significantly adjust printed-circuit boards, reports Expreview web-site.

amd_radeon_r9_nano_1

Custom versions of AMD Radeon R9 Nano will have to preserve form-factor of the graphics card and should be compliant with mini-ITX standards. Thermal design power of partner’s Radeon R9 Nano graphics cards will have to be the same as TDP of AMD’s version.

AMD Radeon R9 Nano-based graphics boards from AMD’s allies will use fully-fledged “Fiji” graphics processing units with 4096 stream processors, 256 texture mapping units, 64 raster operations pipelines and 4096-bit HBM memory bus.

amd_radeon_r9_nano

Discuss on our Facebook page, HERE.

KitGuru Says: It looks like custom versions of AMD Radeon R9 Nano will not be significantly different compared to AMD’s own graphics adapter. What we can expect are graphics adapters with hybrid or liquid cooling systems, which will have higher overclocking potential than AMD’s own Radeon R9 Nano. Keeping in mind that “Nano” graphics cards have different voltage regulator module compared to AMD Radeon R9 Fury X, even modified “R9 Nano” graphics adapters with advanced cooling systems will not have overclocking potential on par with the top-of-the-range product.

The post AMD will not allow partners to modify specs of Radeon R9 Nano first appeared on KitGuru.]]>
https://www.kitguru.net/components/graphic-cards/anton-shilov/amd-will-not-allow-partners-to-modify-specs-of-radeon-r9-nano/feed/ 18
AMD started to work on HBM technology nearly a decade ago https://www.kitguru.net/components/graphic-cards/anton-shilov/amd-started-to-work-on-hbm-technology-nearly-a-decade-ago/ https://www.kitguru.net/components/graphic-cards/anton-shilov/amd-started-to-work-on-hbm-technology-nearly-a-decade-ago/#comments Sat, 29 Aug 2015 12:57:50 +0000 http://www.kitguru.net/?p=265468 The high-bandwidth memory (HBM) introduced along with AMD’s code-named “Fiji” graphics processing unit radically changes the way graphics adapters are built and also dramatically improves potential performance of future graphics processing units. But while HBM looks ingeniously simple on paper, it was extremely hard to develop and is not easy to build. In fact, AMD …

The post AMD started to work on HBM technology nearly a decade ago first appeared on KitGuru.]]>
The high-bandwidth memory (HBM) introduced along with AMD’s code-named “Fiji” graphics processing unit radically changes the way graphics adapters are built and also dramatically improves potential performance of future graphics processing units. But while HBM looks ingeniously simple on paper, it was extremely hard to develop and is not easy to build. In fact, AMD started to work on what is now known as HBM as early as in 2006 – 2007.

The need for speed

Memory bandwidth has been a performance-limiting factor for graphics processors since the introduction of the first gaming-grade graphics cards, three-dimensional games and 32-bit colour back in the nineties. In a bid to considerably increase performance of a graphics adapters, IHVs [independent hardware vendors] had to bolster bandwidth of their DRAM [dynamic random access memory], which was not always easy.

ati_radeon_9700_pro
ATI Radeon 9700 Pro: World's first graphics card with 256-bit memory bus. Image by myhard.com.

There are several ways to increase memory bandwidth on a graphics card: to rise memory clock-rate, to widen memory interface, or to use a more efficient memory technology. After increasing frequencies of graphics DRAM to rather high levels in 1997 – 1998, Nvidia Corp. was the first company to start using the double data rate (DDR) memory (which transfers data on both the rising and falling edges of the clock signal, the technology known as double-pumping) on its GeForce 256 DDR graphics cards in 1999 and doubled bandwidth available to the GPU. ATI Technologies introduced the world’s first graphics card with 256-bit memory bus in 2002 and doubled memory bandwidth of graphics processors once again. In 2002 – 2003 new memory technologies – GDDR2 and GDDR3 – designed specifically for GPUs and supporting quad-pumping were introduced and doubled available bandwidth another time.

But memory bandwidth improvements in the early 2000s did not come for free. Increases of clock-rates and data rates amplified power consumption of memory chips. Wider memory interfaces required more memory ICs, which also increased power requirements of add-in-boards.

By 2006 – 2007, when the work on the ATI R600 graphics processor with 512-bit memory bus as well as on the GDDR4 and the GDDR5 memory technologies was essentially completed, it became clear that memory consumed a lot of power already and would consume even more over time. Since ATI and Nvidia planned to use their GPUs for high-performance computing (HPC) applications, which require a lot of local memory, it was obvious that power consumption of GDDR was going to become a problem.

amd_sk_hynix_hbm_history_graph

At the time, new memory tech development team at ATI Technologies led by Joe Macri came up with an idea of brand-new memory technology, which could provide extreme bandwidth while consuming a low amount of energy. The key elements of the new technology were multi-layer memory devices with an ultra-wide interfaces that used silicon interposer to connect to a processing device.

Brief history of HBM

Modern technologies take a long time to develop. For example, the work on DDR4 started back in 2005, a couple of years before DDR3 was commercially launched. Similarly, ATI Technologies (which AMD acquired in 2006) started to think about high-bandwidth memory with low power consumption about a decade ago, before the company helped to commercialize GDDR4 in 2006 and GDDR5 in 2008. The work on what is now known as HBM began sometimes in 2006 – 2007 and in 2013 the technology became an industry standard.

Architecturally, the first-generation high-bandwidth memory (JESD235) uses a protocol similar to that of the original DDR, which development kicked off in 1996 and was concluded in mid-2000. But a in a bid to finish the new standard, AMD, SK Hynix and other developers had to create a massive amount of additional technologies that ultimately facilitated creation of graphics processors like AMD’s “Fiji”.

amd_sk_hynix_hbm_implementation_1

There are several key technologies that empower HBM:

  • Memory chips with multiple vertically stacked memory devices interconnected using through-silicon-vias (TSVs) and microbumps and then placed on a base logic die.
  • Silicon interposer that connects memory ICs to host processor using an ultra-wide interface. Silicon interposer is made using a photolithography technology in a semiconductor fabrication plant.
  • Host processor with ultra-wide memory interface.

amd_sk_hynix_hbm_design

Development of new technologies requires a lot of prototyping activities in general. Before AMD and SK Hynix proceeded to standardize their HBM memory with JEDEC in 2010, the companies had to design multiple implementations of their new technologies and learn how they operated in real-life.

amd_sk_hynix_hbm_history_graph_proto

AMD started to experiment with interposers and processors back in 2007. The first GPU to connect to memory using an interposer was the RV635, which powered ATI Radeon HD 3650/3670 graphics adapters. Eventually, AMD experimented with interposers and the “Cypress”, which was the world’s first DirectX 11-supporting graphics processor. Both the RV635 and the “Cypress” were based on the TeraScale architecture (gen 1 and gen 2), which was succeeded by the GCN [graphics core next] architecture in 2012.

amd_sk_hynix_hbm_history_graph_proto_1

The JESD235 standard was published in October, 2013, when the work on AMD’s “Fiji” was well underway and the graphics processing unit was months away from tape-out.

The HBM saga continues

The first-generation HBM (HBM1) stacks four DRAM dies with two independent 128-bit channels per die on a base logic die, creating a memory device with a 1024-bit interface. Each channel supports 1Gb capacities (2Gb per die), features 8 banks and can operate at 1Gb/s data-rate (1GHz effective DDR clock-rate). As a result, each HBM 4Hi stack (4 high stack) package can provide 1GB capacity and 128GB/s memory bandwidth. AMD’s Radeon R9 Fury X flagship graphics adapter features 4GB of HBM memory with unprecedented bandwidth of 512GB/s. While the first-gen HBM has limitations when it comes to capacity, it allows to create very small and very high performance graphics solutions thanks to the fact that HBM chips are smaller than GDDR5 ICs.

amd_sk_hynix_hbm_implementation

The second-generation HBM (HBM2) utilizes 8Gb dies with two 128-bit channels featuring 16 banks and sporting up to 2Gb/s data-rates (2GHz effective DDR frequency). The architecture of the HBM2 will let manufacturers built not only 4Hi stack (4 high stack) packages, but also 2Hi stack and 8Hi stack devices. As a result, memory producers will be able to assemble HBM2 memory chips with up to 8GB capacity (8Hi stack) and up to 256GB/s bandwidth (2Gb/s data rate, 1024-bit bus).
sk_hynix_tsv_roadmap_hbm

Architectural advantages of HBM2 will allow GPU developers to use it not only for ultra-high-end applications with 4096-bit memory bus, but also for adapters that do not require extreme performance. Next-generation enthusiast-class graphics cards based on AMD’s “Greenland” graphics processors as well as Nvidia’s GP100 (“Pascal”) GPUs will feature 8GB – 16GB of HBM memory with up to 1TB/s bandwidth. Samsung Electronics forecasts that over time HBM will enable add-in-boards with up to 48GB of memory.

amd_sk_hynix_hbm_pcb

The third-generation HBM is in development and engineers currently do not share any information about it. It is logical to expect further increases of capacities as well as performance. While we have no idea how additional capacities and performance will be achieved, we are pretty sure that engineers at companies like AMD are already playing not only with prototypes of future implementations of HBM, but also with something that will succeed them a long time down the road.

Discuss on our Facebook page, HERE.

KitGuru Says: The evolution of GDDR memory slowed down in 2007 (partly because Joe Macri and the team went on to develop HBM) and HBM promises to bring back rapid increase of memory bandwidth in the coming years. How far will HBM go? Will we see graphics cards with 2TB/s or 4TB/s memory sub-systems in the next five to seven years? We do not know for sure, but it looks like exciting times are coming back!

The post AMD started to work on HBM technology nearly a decade ago first appeared on KitGuru.]]>
https://www.kitguru.net/components/graphic-cards/anton-shilov/amd-started-to-work-on-hbm-technology-nearly-a-decade-ago/feed/ 10
AMD Radeon R9 Nano: Extreme performance in mini-ITX form-factor https://www.kitguru.net/components/graphic-cards/anton-shilov/amd-unveils-radeon-r9-nano-leading-edge-performance-in-mini-itx-form-factor/ https://www.kitguru.net/components/graphic-cards/anton-shilov/amd-unveils-radeon-r9-nano-leading-edge-performance-in-mini-itx-form-factor/#comments Thu, 27 Aug 2015 14:58:11 +0000 http://www.kitguru.net/?p=265121 Advanced Micro Devices on Thursday introduced its highly-anticipated AMD Radeon R9 Nano graphics card, which combines massive performance with relatively low power consumption and ultra-small form-factor. The new graphics card uses fully-fledged “Fiji” graphics processor and its peak compute performance even exceeds that of Nvidia Corp.’s GeForce GTX Titan X. Small form-factor, unbeatable performance The …

The post AMD Radeon R9 Nano: Extreme performance in mini-ITX form-factor first appeared on KitGuru.]]>
Advanced Micro Devices on Thursday introduced its highly-anticipated AMD Radeon R9 Nano graphics card, which combines massive performance with relatively low power consumption and ultra-small form-factor. The new graphics card uses fully-fledged “Fiji” graphics processor and its peak compute performance even exceeds that of Nvidia Corp.’s GeForce GTX Titan X.

Small form-factor, unbeatable performance

The AMD Radeon R9 Nano is not a small brother to the flagship Radeon R9 Fury X, but a graphics card that offers nearly similar performance, which means that it is essentially a competitor. The miniature graphics adapter sports “Fiji” graphics processing unit with 4096 stream processors, 256 texture mapping units, 64 raster operations pipelines and 4096-bit interface for high-bandwidth memory operating at 1GHz. The board carries 4GB of HBM memory with 512GB/s bandwidth.

amd_radeon_r9_nano

Compute performance of the AMD Radeon R9 Nano is 8.2TFLOPS, which just 5 per cent below that of the top-of-the-line Radeon R9 Fury X and which is considerably higher than compute performance of Nvidia’s flagship offering, the GeForce GTX Titan X (6.2TFLOPS).

According to AMD, the Radeon R9 Nano is at least 30 per cent faster than any mini-ITX graphics card available today (i.e., Nvidia GeForce GTX 970) in various games in ultra-high-definition 4K (3840*2160) resolution. Thanks to ultimate compute performance of the novelty, that difference will only increase over time as games take advantage of compute capabilities of GPUs.

amd_radeon_r9_nano_perf

AMD’s “Fiji” graphics processing unit based on the GCN 1.2 architecture fully supports all modern technologies and application programming interfaces, including DirectX 12, OpenGL 4.5, OpenCL 2.1, Vulkan, Mantle, FreeSync and other.

“With the Radeon R9 Nano graphics card, AMD is enabling 4K class gaming in your living room in an exceptionally quiet, ultra-small design built to excel in today's games and on the latest APIs like DirectX 12 and Vulkan,” said Matt Skynner, corporate vice president and general manager of product, computing and graphics business unit at AMD. “There simply is nothing else like it.”

State-of-the-art design

The miniature AMD Radeon R9 Nano uses 6” printed circuit board specially developed for high-end graphics cards. The card will fit into any mini-ITX system that is compatible with dual-slot graphics adapters.

amd_radeon_r9_nano_2   amd_radeon_r9_nano_3

The board features 4+1-phase voltage regulator module based on high-end solid-state chokes as well as International Rectifier’s IR3564B dual-output digital multi-phase controller used on advanced mainboards. The VRM is extremely sophisticated and features additional chokes to stabilize and clean power in a bid to ensure stable operation in rather extreme conditions. The card has one 8-pin auxiliary PCI Express power connector.

amd_radeon_r9_nano_6

Too cool-down the card, which can dissipate up to 175W of power, AMD developed a special cooling system featuring one vapour chamber, numerous heat-pipe, an aluminum radiator and a 90mm fan that can exhaust heat in multiple directions. The cooler looks similarly to the one used on the Radeon R9 Fury X and features brushed aluminium finishes and multiple metal parts in order to ensure minimal vibration. The fan reportedly produces 42dBa of noise.

The graphics adapter features three DisplayPort 1.2a outputs as well as one HDMI 1.4 connector.

Price and availability

AMD’s partners will begin to sell the Radeon R9 Nano graphics boards under their own brands starting from the 10th of September. The manufacturer suggested retail price of the Radeon R9 Nano will be similar to that of the Radeon R9 Fury X: $649 in the U.S., £509 in the U.K. and €629 in Eurozone.

amd_radeon_r9_nano_1

In the first three months of availability, all AMD Radeon R9 Nano will be manufactured by a contract maker under supervision of Advanced Micro Devices and then sold to partners. Three months from now select AIB suppliers may release their own mini-ITX graphics cards based on the “Fiji” graphics processing units similar to AMD’s Radeon R9 Nano.

Discuss on our Facebook page, HERE.

KitGuru Says: Small form-factor, limited power consumption, extreme compute performance and great performance in games make AMD’s Radeon R9 Nano a very attractive product. Its MSRP is rather high, but its competitors – AMD Radeon R9 Fury X and Nvidia GeForce GTX 980 Ti – consume more power and are larger. Therefore, if you are building a mini-ITX PC to play video games in 4K UHD resolution, you really don’t have a lot of choice here.

The post AMD Radeon R9 Nano: Extreme performance in mini-ITX form-factor first appeared on KitGuru.]]>
https://www.kitguru.net/components/graphic-cards/anton-shilov/amd-unveils-radeon-r9-nano-leading-edge-performance-in-mini-itx-form-factor/feed/ 15
Pricing and launch date of AMD Radeon R9 Nano leaked https://www.kitguru.net/components/graphic-cards/anton-shilov/pricing-and-launch-date-of-amd-radeon-r9-nano-leaked/ https://www.kitguru.net/components/graphic-cards/anton-shilov/pricing-and-launch-date-of-amd-radeon-r9-nano-leaked/#comments Thu, 27 Aug 2015 06:50:54 +0000 http://www.kitguru.net/?p=265048 Specifications and capabilities of AMD’s upcoming Radeon R9 Nano graphics card have been largely known for quite some time now. The new graphics adapter will be the highest-performing mini-ITX add-in board ever built and will be among the fastest graphics cards available today. According to a new leak, the product will not be too expensive. AMD’s partners …

The post Pricing and launch date of AMD Radeon R9 Nano leaked first appeared on KitGuru.]]>
Specifications and capabilities of AMD’s upcoming Radeon R9 Nano graphics card have been largely known for quite some time now. The new graphics adapter will be the highest-performing mini-ITX add-in board ever built and will be among the fastest graphics cards available today. According to a new leak, the product will not be too expensive.

AMD’s partners will start to sell the Radeon R9 Nano graphics card on the 10th of September, 2015. The manufacturer suggested retail price of the Radeon R9 Nano will be similar to that of the Radeon R9 Fury X: $649 in the U.S., £509 in the U.K. and €629 in Eurozone, according to WccfTech, which cites AMD’s documents and sources close to the company.

amd_radeon_fury_nano

The AMD Radeon R9 Nano will be built by a contract manufacturer under supervision of Advanced Micro Devices and sold to partners as card, not as a graphics processing unit. Only three months later some manufacturers of graphics boards may be granted rights to build their own Radeon R9 Nano graphics adapters.

The AMD Radeon R9 Nano graphics card for mini-ITX personal computers is powered by AMD’s “Fiji” graphics processing unit with 4096 stream processors, 256 texture mapping units, 64 raster operations pipelines and 4096-bit memory interface. The Radeon R9 Nano will have compute performance of around 8.2TFLOPS, which is only 5 per cent below that of the Radeon R9 Fury X. Just like the flagship graphics solution from AMD, the miniature graphics adapter carries 4GB of HBM [high-bandwidth memory] operating at 1000MHz.

amd_radeon_fury_nano_1

The AMD Radeon R9 Nano will be the fastest graphics adapter for small form-factor systems ever designed. Its performance will be comparable to that of AMD’s Radeon R9 Fury X and Nvidia’s GeForce GTX 980 Ti. It is unlikely that the Radeon R9 Nano will get a strong rival in its category any time soon.

Discuss on our Facebook page, HERE.

KitGuru Says: Keeping in mind that the AMD Radeon R9 Nano graphics card is unique, its $649/£509 price tag does not seem to be excessive. What remains to be seen is whether AMD will be able to satisfy demand for the miniature graphics adapter. In fact, it is easier to build the flagship AMD Radeon R9 Fury X than the Radeon R9 Nano. For the for former, AMD needs to get one “Fiji” chip with 4096 stream processors, but for the latter the company needs a GPU with 4096 SPs, low power consumption and low heat dissipation, a combination that may be hard to find.

The post Pricing and launch date of AMD Radeon R9 Nano leaked first appeared on KitGuru.]]>
https://www.kitguru.net/components/graphic-cards/anton-shilov/pricing-and-launch-date-of-amd-radeon-r9-nano-leaked/feed/ 16
Specs of AMD Radeon R9 Nano confirmed: 4096 SPs, 8.19TFLOPS https://www.kitguru.net/components/graphic-cards/anton-shilov/specs-of-amd-radeon-r9-nano-confirmed-4096-sps-8-19tflops/ https://www.kitguru.net/components/graphic-cards/anton-shilov/specs-of-amd-radeon-r9-nano-confirmed-4096-sps-8-19tflops/#comments Wed, 26 Aug 2015 01:03:58 +0000 http://www.kitguru.net/?p=264891 It is not a secret that AMD’s upcoming Radeon R9 Nano graphics card will be the most powerful adapter for mini-ITX personal computers ever produced and will have no rivals for quite some time. As it appears, the Radeon R9 Nano will not only offer high performance, but it will actually offer compute performance that will …

The post Specs of AMD Radeon R9 Nano confirmed: 4096 SPs, 8.19TFLOPS first appeared on KitGuru.]]>
It is not a secret that AMD’s upcoming Radeon R9 Nano graphics card will be the most powerful adapter for mini-ITX personal computers ever produced and will have no rivals for quite some time. As it appears, the Radeon R9 Nano will not only offer high performance, but it will actually offer compute performance that will be comparable to that of AMD’s flagship Radeon R9 Fury X.

The AMD Radeon R9 Nano graphics card – which sales are expected to begin in the coming weeks – is based on the fully fledged code-named “Fiji” graphics processing unit with 4096 stream processors, 256 texture mapping units, 64 raster operations pipelines and 4096-bit memory interface, according to slide from AMD’s presentation published by VideoCardz web-site. The graphics board will continue to carry 4GB of HBM [high-bandwidth memory] operating at 1000MHz.

amd_radeon_fury_nano_graphics_card

The miniature graphics card from AMD and its partners will have clock-rates comparable to those of the flagship Radeon R9 Fury X. The Radeon R9 Nano will have compute performance of around 8.2TFLOPS, which is only 5 per cent below that of the Radeon R9 Fury X. Actual performance of the Radeon R9 Nano will be considerably higher than that of the Radeon R9 Fury, which uses a cut-down version of “Fiji” with 3584 stream processors, and will be very close to that of the top-of-the-range product.

AMD-Radeon-R9-Nano-Final-Specifications-900x507

Thermal design power of the AMD Radeon R9 Nano will be 175W. Many small form-factor personal computers these days can cool-down such graphics cards. Makers of high-performance systems will likely release SFF PCs featuring the Radeon R9 Nano with extreme performance in games in the coming months. Moreover, thanks to the fact that the “Fiji” graphics processor supports hardware-accelerated decoding and encoding of H.265 (HEVC) video, the Radeon R9 Nano can even be used to playback of ultra-high-definition video. Unfortunately, since “Fiji” does not support HDMI 2.0 output, it may not be the best card to build home-theater personal computers for Ultra HD Blu-ray playback.

Pricing of AMD’s Radeon R9 Nano is unknown. Its performance is clearly higher than that of the AMD Radeon R9 Fury, which costs $549, but the product is not as fast as the Radeon R9 Fury X, which retails for $649.

Discuss on our Facebook page, HERE.

KitGuru Says: Keeping in mind that far not all people, even among enthusiasts, like liquid-cooling systems, the Radeon R9 Nano may actually become more popular than the Radeon R9 Fury X. The Radeon 9 Nano will be truly unique and it has all chances to become a best-seller.

The post Specs of AMD Radeon R9 Nano confirmed: 4096 SPs, 8.19TFLOPS first appeared on KitGuru.]]>
https://www.kitguru.net/components/graphic-cards/anton-shilov/specs-of-amd-radeon-r9-nano-confirmed-4096-sps-8-19tflops/feed/ 20
Samsung expects graphics cards with 6144-bit bus, 48GB of HBM memory onboard https://www.kitguru.net/components/graphic-cards/anton-shilov/samsung-expects-graphics-cards-with-6144-bit-bus-48gb-of-hbm-memory-onboard/ https://www.kitguru.net/components/graphic-cards/anton-shilov/samsung-expects-graphics-cards-with-6144-bit-bus-48gb-of-hbm-memory-onboard/#comments Fri, 21 Aug 2015 15:56:38 +0000 http://www.kitguru.net/?p=264471 Samsung Electronics indicated that it plans to start volume production of high-bandwidth memory (HBM) next year at the Intel Developer Forum this week. At the trade-show, the company revealed its current vision and expectations concerning HBM. Samsung foresees that eventually high-performance applications (such as GPUs) could feature up to six HBM devices to enable unprecedented …

The post Samsung expects graphics cards with 6144-bit bus, 48GB of HBM memory onboard first appeared on KitGuru.]]>
Samsung Electronics indicated that it plans to start volume production of high-bandwidth memory (HBM) next year at the Intel Developer Forum this week. At the trade-show, the company revealed its current vision and expectations concerning HBM. Samsung foresees that eventually high-performance applications (such as GPUs) could feature up to six HBM devices to enable unprecedented capacities and bandwidth.

As it appears, Samsung plans to skip the first-generation HBM memory and will only manufacture products compliant with the second-generation HBM specification, which offers higher densities and clock-rates. Such approach will let Samsung to address broader market segments with its HBM offerings. Right now HBM can only be used on consumer graphics cards and certain highly-custom products, but HBM2 will enable professional and high-performance computing GPU-based solutions, which require a lot of memory onboard.

Samsung is currently working on multiple HBM packages featuring two (2Hi stack), four (4Hi) and eight (8Hi) 8Gb memory devices on a base logic die with 1024-bit interface, according to a slide that the company demonstrated at the IDF (which was published by ComputerBase.de). Maximum data-rates of Samsung’s HBM products will be 2Gb/s, which will support up to 256GB/s of bandwidth per chip.

samsung_hbm_plans

Samsung believes that HBM memory will enable it to create a variety of chip offerings targeting different market segments. Designers of logic chips (GPUs, APUs, network processors, etc.) will be able to integrate the right amount of HBM controllers into their chips in order to target different applications. At present AMD’s “Fiji” graphics processing unit supports up to four HBM stacks over its 4096-bit interface. Eventually, logic chips could accommodate more HBM controllers and expand interface width to unprecedented 6144-bit, according to Samsung Electronics.

For example, a mainstream graphics card could use just one 2Hi HBM chip to accommodate a 2GB frame-buffer with 256GB/s bandwidth. More advanced graphics adapters for consumers and creative professionals could feature four 2Hi or four 4Hi HBM stacks that provide up to 1TB/s of bandwidth as well as 8GB, 16GB or 32GB of memory. Accelerators for high-performance computing as well as ultra-high-end GPU offerings will support six HBM stacks, thus enabling cards with 12GB, 24GB or 48GB of onboard memory with 1.5TB/s bandwidth.

amd_radeon_fiji_gpu

So far, neither AMD nor Nvidia have demonstrated even hypothetical product implementations (which are used to showcase potential future uses of technologies) of GPU-based solutions featuring six HBM memory chips. Intel Corp.’s Xeon Phi co-processors use HMC [hybrid memory cube] DRAMs as “near memory” and are not expected to support HBM any time soon.

nvidia_pascal_module

Samsung expects HBM memory to be used for consumer graphics cards and high-performance computing accelerators based on AMD “Arctic Islands” and Nvidia “Pascal” graphics processors next year. Sometimes in 2017, network products will also take advantage of the new memory type. Three years from now other applications could employ HBM, according to Samsung.

The world’s largest maker of dynamic random access memory (DRAM) did not unveil any details about its actual HBM chips (capacities, clock-rates, etc.), but expect a family of products with different densities and frequencies.

Samsung did not disclose which process technology it will use to manufacture HBM.

Discuss on our Facebook page, HERE.

KitGuru Says: Samsung is definitely not the first to offer HBM, but it is pretty clear that the company takes the new type of memory very seriously. What is interesting is that the company decided not to give a glimpse into the future of HBM, which indicates that the third-generation HBM is still a work in progress.

The post Samsung expects graphics cards with 6144-bit bus, 48GB of HBM memory onboard first appeared on KitGuru.]]>
https://www.kitguru.net/components/graphic-cards/anton-shilov/samsung-expects-graphics-cards-with-6144-bit-bus-48gb-of-hbm-memory-onboard/feed/ 31