GPGPU | KitGuru https://www.kitguru.net KitGuru.net - Tech News | Hardware News | Hardware Reviews | IOS | Mobile | Gaming | Graphics Cards Sat, 11 Apr 2015 14:20:11 +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 GPGPU | KitGuru https://www.kitguru.net 32 32 Concept of AMD APU for HPC leaks: 16 x86 cores, next-gen GPU, HBM memory https://www.kitguru.net/components/cpu/anton-shilov/details-about-concept-amd-apu-for-hpc-leak-16-x86-cores-next-gen-graphics-hbm-memory/ https://www.kitguru.net/components/cpu/anton-shilov/details-about-concept-amd-apu-for-hpc-leak-16-x86-cores-next-gen-graphics-hbm-memory/#comments Sat, 11 Apr 2015 12:33:46 +0000 http://www.kitguru.net/?p=244566 Advanced Micro Devices has been mulling accelerated processing unit for high-performance computing applications for many years, but software and hardware limitations have prevented the company from creating such chip. Recently AMD finally unveiled plans to release an APU for HPC market in 2017 and this week the first concept of such product leaked. At present …

The post Concept of AMD APU for HPC leaks: 16 x86 cores, next-gen GPU, HBM memory first appeared on KitGuru.]]>
Advanced Micro Devices has been mulling accelerated processing unit for high-performance computing applications for many years, but software and hardware limitations have prevented the company from creating such chip. Recently AMD finally unveiled plans to release an APU for HPC market in 2017 and this week the first concept of such product leaked.

At present supercomputers use nodes featuring traditional multi-core x86 microprocessors as well as highly-parallel compute accelerators/co-processors powered by many-core architectures, such as Nvidia Tesla, Intel Xeon Phi or AMD FirePro. While such configuration of nodes generally works very well today, performance of many operations is slowed down by limited bandwidth between CPU and accelerators. For example, many applications could benefit from uniform memory access by CPUs and accelerators/co-processors as well as from cache coherency, but bandwidth and latency limitations usually slow things down. In a bid to solve the problem, Nvidia proposes to use high-bandwidth NVLink technology to connect CPUs and Tesla accelerators, whereas AMD intends to integrate CPU and GPU cores in the same piece of silicon in a way that allows compute engines to communicate efficiently with each other.

amd_fusion_apu_chip_1

AMD’s accelerated processing units – which integrate x86 general-purpose cores and GCN graphics processing cores – available today are fine for consumer PCs, but for HPC applications a different level of integration is required. Not only bandwidth between x86 and GCN cores has to be increased and cache coherency maintained, but different processing engines have to efficiently and collaboratively use memory controller in order to realize all benefits of unified memory access. At present AMD is working on such accelerated processing units due to be available later this decade.

amd_roadmap_update_2015

Fudzilla this week published an image that resembles slides from AMD’s documents for partners. The image describes an unknown APU without a code-name, which can be a concept of AMD’s upcoming accelerated processing unit for high-performance computing applications.

Expected to be released in 2017, AMD’s APU for supercomputers is projected to integrate 16 x86 “Zen” cores with two-way simultaneous multi-threading technology and 512KB L2 cache per core, 32MB L3 cache as well as a new-generation “Greenland” graphics engine with ½ double precision compute rate. The chip will also feature a quad-channel DDR4 memory controller with enhanced ECC capabilities that supports up to 256GB of memory per channel, 64 lanes of PCI Express 3.0 that can be used for SATA Express, integrated SATA, 1GbE, USB as well as various legacy interfaces. To further speed-up bandwidth demanding applications, the APU is expected to feature on-package 16GB of HBM2 memory with 512GB/s or higher bandwidth. Given that the chip is scheduled to arrive in 2017, expect it to be made using 14nm FinFET (14LPP) or 10nm FinFET process technologies.

Since the APU described in the slide does not have a code-name, release timeframe or targeted process technology and thermal design power, it is highly likely that the chip is a concept of what AMD might develop for various market segments.

amd_apu_zen_greenland_concept_fusion

For example, HPC would benefit from 16 x86 “Zen” cores, L3 cache, full-speed DP rate (which is ½ of single precision rate in case of contemporary FirePro accelerators), HBM memory and quad-channel DDR4 with ECC. At the same time, HPC applications do not need secure boot or crypto processors, SO-DIMM or SATA Express support. By contrast, consumer-class APUs do not need 16 x86 “Zen” cores (client PC programs are not optimized for multi-core/many-core processors, hence, 16 cores will provide 10 – 15 per cent performance boost compared to four cores in real-world apps), full-speed DP rate, large L3 cache, ECC with enhancements, 256GB of memory per channel and so on. However, consumer hybrid processors need enhanced security, flexible PCI Express configuration, support for different memory module types, legacy interfaces and so on.

AMD did not comment on the news-story.

Discuss on our Facebook page, HERE.

KitGuru Says: While it will be possible to design and build the APU described above using 10nm process technology, it is really unlikely that the chip will ever see the light of day in its current configuration. Therefore, consider the story not as the first details about an unknown accelerated processing unit from Advanced Micro Devices, but as a description of what AMD might plan to create in a couple of years’ time.

The post Concept of AMD APU for HPC leaks: 16 x86 cores, next-gen GPU, HBM memory first appeared on KitGuru.]]>
https://www.kitguru.net/components/cpu/anton-shilov/details-about-concept-amd-apu-for-hpc-leak-16-x86-cores-next-gen-graphics-hbm-memory/feed/ 3
Samsung’s own CPU core took years to develop – report https://www.kitguru.net/components/graphic-cards/anton-shilov/samsungs-own-cpu-core-took-years-to-develop-report/ https://www.kitguru.net/components/graphic-cards/anton-shilov/samsungs-own-cpu-core-took-years-to-develop-report/#comments Fri, 10 Apr 2015 16:24:06 +0000 http://www.kitguru.net/?p=244513 Samsung Electronics has developed and built its Exynos application processors for years, but its chips are always powered by off-the-shelf general-purpose and graphics processing cores from ARM, which limits their advantages over competing system-on-chips. Everything is going to change next year, when Samsung finally rolls-out its new SoCs featuring its own CPU and GPU designs. …

The post Samsung’s own CPU core took years to develop – report first appeared on KitGuru.]]>
Samsung Electronics has developed and built its Exynos application processors for years, but its chips are always powered by off-the-shelf general-purpose and graphics processing cores from ARM, which limits their advantages over competing system-on-chips. Everything is going to change next year, when Samsung finally rolls-out its new SoCs featuring its own CPU and GPU designs.

In a bit to use unique application processors with exclusive capabilities inside its smartphones, tablets and other devices, Samsung is presently designing its own custom general-purpose processing cores as well as graphics processing cores, according to industry sources. Custom cores let Samsung achieve higher performance and better energy efficiency, something that gains importance nowadays.

Apparently, the work on custom general-purpose ARMv8-compatible processing core began as early as in 2011, according to BusinessKorea. Keeping in mind that the first information about Samsung hiring experts from processor developers, such as Advanced Micro Devices, emerged in 2012, it looks like the current report is correct.

samsung_exynos_7_octa

Four years is a long time and it is highly likely that Samsung’s general-purpose core will be rather powerful. Keeping in mind that at least initially Samsung planned to develop a high-performance ARMv8-A-compatible core for micro-servers, we are talking about rather powerful 64-bit CPU. Still, Samsung’s microprocessor will have very strong rivals. Smaller designers of APs will offer system-on-chips with ARM Cortex-A72 cores, Qualcomm will offer its SoCs featuring its own Kyro cores. It is unclear what Apple will offer in 2016, but the company’s Cyclone are already the fastest ARMv8 cores available today and it is highly likely that its successors will also provide extremely high performance.

What remains to be seen is how good will Samsung’s own graphics processing engine will be. If the company designs its own graphics core from scratch, it needs hundreds of specialists and many years of time. Meanwhile, there are no reports about Samsung’s hiring a lot of GPU engineers. Perhaps, when it comes to GPU, Samsung will use a semi-custom approach: it will license technologies from ARM or Imagination Technologies and will then configure the graphics processing unit in-house.

Since Samsung is a founding member of the HSA [heterogeneous system architecture] foundation, it is highly likely that its forthcoming system-on-chips will be fully HSA 1.0 compatible and will thus be able to use GPU stream processors for general-purpose tasks, saving power and providing performance improvements. Therefore, GPU performance is very important for Samsung’s own SoC.

Samsung did not comment on the news-story.

Discuss on our Facebook page, HERE.

KitGuru Says: It looks like Samsung takes development of in-house CPU and GPU cores very seriously. However, if we are talking about an HSA-enabled application processor, not only hardware design is vital, but software is crucially important as well. Hopefully, Samsung’s software department team is also getting a lot of new engineers, just like the hardware design team.

The post Samsung’s own CPU core took years to develop – report first appeared on KitGuru.]]>
https://www.kitguru.net/components/graphic-cards/anton-shilov/samsungs-own-cpu-core-took-years-to-develop-report/feed/ 2
AMD cuts prices of professional GPUs by 50% to gain market share https://www.kitguru.net/components/graphic-cards/anton-shilov/amd-cuts-price-of-professional-graphics-cards-by-50-to-gain-market-share/ https://www.kitguru.net/components/graphic-cards/anton-shilov/amd-cuts-price-of-professional-graphics-cards-by-50-to-gain-market-share/#comments Sat, 13 Dec 2014 02:34:40 +0000 http://www.kitguru.net/?p=226182 Advanced Micro Devices offers to sell its professional graphics cards for workstations or servers with a massive discount to first time buyers in an attempt to increase its revenue and market share. ATI Technologies and then Advanced Micro Devices was an underdog on the market of professional graphics processing unit (GPU) hardware for many years because of …

The post AMD cuts prices of professional GPUs by 50% to gain market share first appeared on KitGuru.]]>
Advanced Micro Devices offers to sell its professional graphics cards for workstations or servers with a massive discount to first time buyers in an attempt to increase its revenue and market share.

ATI Technologies and then Advanced Micro Devices was an underdog on the market of professional graphics processing unit (GPU) hardware for many years because of various reasons, which includes poor drivers, low performance and heavy competition from Nvidia Corp. However, in the recent years the situation began to change. AMD gradually improved its graphics drivers and worked closely with developers of CAD/CAM/CAE/DCC and other workstation-class professional software to mend performance and compatibility issues. At present AMD commands roughly one quarter of the workstation GPU market and it clearly wants to gain more share.

amd_firepro_w9000

In a bid to attract attention of professionals, AMD recently kicked off its “Experience AMD FirePro” campaign that is designed for those, who do not currently own an AMD FirePro graphics adapter, but would like to try it. For a limited time only AMD will return up to 50 per cent of a FirePro’s card's price to qualified customers who purchase a professional AMD graphics adapter, or a workstation featuring such a board, from an AMD approved reseller. AMD offers up to 50 per cent rebates, or up to £1195/$1995/€1460, on AMD FirePro W9100, AMD FirePro W8100, AMD FirePro W8000, AMD FirePro W7000 as well as on AMD FirePro S9000 and AMD FirePro S7000 cards.

Note that customers have to register for a pre-approval code first before making a purchase. The promotion is open for customers in the U.K., the U.S., Canada (excluding Quebec), Germany and France. The campaign only lasts until the 31st of December, 2014.

masthead_UK

According to Jon Peddie Research, in Q3 2014 the industry shipped approximately 1.02 million workstations, a 4.7% gain over the same period a year ago, along with a modest 2.8% quarterly decline over Q2 2014's all-time high. AMD and Nvidia sold a total of approximately 1.28 million workstation-caliber GPUs in the third quarter, including both mobile modules and desktop add-in cards.

It can be observed that only around 260 thousand of workstation-class graphics solutions were either sold not as a part of workstation computers, or were installed as secondary professional GPUs inside very powerful PCs. Therefore, while the market of retail professional hardware exists, it is rather small. Still, if AMD wants to gain market share in general, it has to address this one as well, even though it is a small fraction of the whole workstation GPU market.

Discuss on our Facebook page, HERE.

KitGuru Says: Running a rebate/discount campaign is one of the ways to increase sales and market share. However, the terms of the program seem to be rather strict and end-users or organizations will need to be approved by AMD first, which will scare off a lot of potential clients. Still, given the price of FirePro graphics boards, it is not really surprising that AMD is cautious about the cash-back promotion…

The post AMD cuts prices of professional GPUs by 50% to gain market share first appeared on KitGuru.]]>
https://www.kitguru.net/components/graphic-cards/anton-shilov/amd-cuts-price-of-professional-graphics-cards-by-50-to-gain-market-share/feed/ 13
Samsung’s own CPU and GPU cores in design stage, due next year https://www.kitguru.net/components/cpu/anton-shilov/samsungs-own-cpu-and-gpu-cores-in-design-stage-due-next-year/ https://www.kitguru.net/components/cpu/anton-shilov/samsungs-own-cpu-and-gpu-cores-in-design-stage-due-next-year/#comments Tue, 02 Dec 2014 21:02:59 +0000 http://www.kitguru.net/?p=224528 Samsung Electronics has world-class semiconductor manufacturing facilities and leading-edge process technologies. While the company designs its own system-on-chips for mobile devices, Samsung continues to use application processors from Qualcomm for its high-end smartphones and extensively uses intellectual property from ARM Holdings and Imagination Technologies. Apparently, Samsung wants to change that and design its SoCs itself …

The post Samsung’s own CPU and GPU cores in design stage, due next year first appeared on KitGuru.]]>
Samsung Electronics has world-class semiconductor manufacturing facilities and leading-edge process technologies. While the company designs its own system-on-chips for mobile devices, Samsung continues to use application processors from Qualcomm for its high-end smartphones and extensively uses intellectual property from ARM Holdings and Imagination Technologies. Apparently, Samsung wants to change that and design its SoCs itself from the ground up.

When Samsung develops its own SoC for its smartphones or tablets at present it simply licenses building blocks from ARM or Imagination, which means that its application processorsare generally similar to those from companies like MediaTek. In order to differentiate itself from its rivals, Samsung Electronics is reportedly working on a new breed of its application processors with its own general purpose and graphics processing cores inside, reports ZDNet Korea citing its own sources.

It is not a secret that Samsung has been developing its own central processing unit cores based on ARMv8-A 64-bit architecture for several years now. The company had to acquire appropriate license from ARM Holdings and hire professional developers. Given the amount of ex-AMD talent hired by Samsung in the recent years, the company should be on-track with its own CPU cores sometimes next year.

However, Samsung does not want to integrate only its own general-purpose core. According to the report, it is also working on its own graphics processing unit. The details about the project are completely unknown since if Samsung develops its own graphics core from scratch, it needs hundreds of specialists and many years of time. It is more likely that the company has licensed certain technologies from ARM and Imagination and wants to create a custom graphics processor with unique feature-set.

samsung_exynos_7_octa

Keeping in mind that Samsung is one of the founding members of the HSA [heterogeneous system architecture] foundation, it is more than likely that its forthcoming chips will be fully HSA compatible and will thus be able to use GPU stream processors for general-purpose tasks, saving power and providing performance improvements.

The first system-on-chip with Samsung’s own CPU and GPU cores is expected to emerge next year, perhaps, along with the company’s next-generation Galaxy Note premium smartphone.

Samsung did not comment on the news-story.

Discuss on our Facebook page, HERE.

KitGuru Says: It is logical for Samsung to develop its own SoCs and their components from scratch in order to make its devices better compared to competing products. It remains to be seen whether GPU cores developed by Samsung will be as competitive as those designed by ARM, Imagination, Nvidia or Qualcomm. What is clear is that even if the first-gen CPUs and GPUs from Samsung do not demonstrate truly high performance, the company will strengthen its development team and will come up with something better eventually. Competitive SoCs are just too important for Samsung. The company will do its best to make them more advanced than offerings from competing developers.

The post Samsung’s own CPU and GPU cores in design stage, due next year first appeared on KitGuru.]]>
https://www.kitguru.net/components/cpu/anton-shilov/samsungs-own-cpu-and-gpu-cores-in-design-stage-due-next-year/feed/ 9
AMD FirePro W9100: Hawaii GPU goes to work with 16GB of GDDR5 https://www.kitguru.net/components/graphic-cards/anton-shilov/amd-reveals-firepro-w9100-hawaii-gpu-goes-to-work-with-16gb-of-gddr5/ https://www.kitguru.net/components/graphic-cards/anton-shilov/amd-reveals-firepro-w9100-hawaii-gpu-goes-to-work-with-16gb-of-gddr5/#comments Thu, 27 Mar 2014 16:57:19 +0000 http://www.kitguru.net/?p=184604 Advanced Micro Devices on Wednesday introduced its first professional graphics card based on the code-named “Hawaii” graphics processing unit. The new solution for CAD, CAM, DCC and other professionals boasts unbeatable DP FP compute performance in addition to whopping 16GB of memory and six 4K display outputs. As expected, the AMD FirePro W9100 professional graphics …

The post AMD FirePro W9100: Hawaii GPU goes to work with 16GB of GDDR5 first appeared on KitGuru.]]>
Advanced Micro Devices on Wednesday introduced its first professional graphics card based on the code-named “Hawaii” graphics processing unit. The new solution for CAD, CAM, DCC and other professionals boasts unbeatable DP FP compute performance in addition to whopping 16GB of memory and six 4K display outputs.

As expected, the AMD FirePro W9100 professional graphics card is based on the code-named Hawaii XT graphics processing unit in its advanced configuration with 2816 stream processors (which also suggests that it has 176 texture units, 64 raster operating units and 512-bit memory bus, but we cannot confirm that). The graphics board is equipped with incredible amount of GDDR5 memory with ECC capability, 16GB, the world’s first for a single-chip graphics solution. The novelty also features six mDP 1.2 ports that can support up to six 4K (3840*2160) displays.

amd_fire_pro_w9100_intro

AMD states that the FirePro W9100 professional solution features over 5TFLOPS of single-precision compute power and over 2TFLOPS of double-precision compute power (AMD claims that the Hawaii XT GPU is configured in such a way that to enable half of SP rate in double-precision computations). Exact clock-rates of the FirePro W9100 are unknown, so are the exact peak performance rates.

amd_fire_pro_w9100_luxmark

Modern professional workflows include not only design itself, but also simulation as well as visualization, which is why AMD and Nvidia not only offer high performance in professional graphics applications, but also in simulation and visualization programs. As a result, GPGPU performance gains importance in general.

amd_fire_pro_w9100_sandra_fp64

According to AMD, the FirePro W9100 professional graphics solution significantly outperforms rivals like Quadro K5000, Quadro K6000 as well Tesla K20X in Luxmark 2.0 OpenCL benchmark in both single-GPU and multi-GPU configurations. The W9100 is also well ahead of the K6000 in SiSoft Sandra FP64 benchmark.

amd_fire_pro_w9100_solidworks

Based on data from AMD, the FirePro W9100 is also ahead of its rivals in SolidWorks 2014.

AMD FirePro W9100 will be available in April. Pricing will be announced at that time.

KitGuru Says: It is interesting to note that AMD decided not to introduce other new professional graphics solutions based on Hawaii graphics processor at this time. The company clearly needs a “less high-end” FirePro W8100 featuring Hawaii as well as Radeon Sky series for cloud computing. It does not look like AMD plans to refresh its FireStream family of GPGPU accelerators for technical computers as it has not done that for four years now.

The post AMD FirePro W9100: Hawaii GPU goes to work with 16GB of GDDR5 first appeared on KitGuru.]]>
https://www.kitguru.net/components/graphic-cards/anton-shilov/amd-reveals-firepro-w9100-hawaii-gpu-goes-to-work-with-16gb-of-gddr5/feed/ 1
nVidia set to seal its lock on the workstation market https://www.kitguru.net/components/graphic-cards/harrison/nvidia-set-to-seal-its-lock-on-the-workstation-market/ https://www.kitguru.net/components/graphic-cards/harrison/nvidia-set-to-seal-its-lock-on-the-workstation-market/#comments Sun, 03 Mar 2013 11:42:55 +0000 http://www.kitguru.net/?p=126888 Learning to ‘follow the money' is one of life's most important lessons. It helps answer so many questions – in every walk of life. Sure sex and drugs and rock-n-roll plus the need for weapons and violence figure into the equation, but the money trail is crucial. For graphics, that means “How will future generations …

The post nVidia set to seal its lock on the workstation market first appeared on KitGuru.]]>
Learning to ‘follow the money' is one of life's most important lessons. It helps answer so many questions – in every walk of life. Sure sex and drugs and rock-n-roll plus the need for weapons and violence figure into the equation, but the money trail is crucial. For graphics, that means “How will future generations of gaming chip be funded?”  Over at Distree 2013, KitGuru operatives recently caught up with the man leading the professional solutions drive for the UK, Middle East and Africa for Quadro.

Before landing at nVidia's professional solutions partner, PNY, Peter Butler worked his way through a lot of the industry's top tech brands – including Microsoft, Elonex and Asus.

Peter explained that the only real decision faced by customers looking for professional solutions in 2013 is whether to pick a solution created by one of the multi nationals – or to choose a local vendor.

It seems that nVidia maintains direct contact with HP and Dell, while PNY takes care of the Quadro, Tesla etc sales into systems sold by specialists like Armari. Our man was told, “Quadro has the vast majority of the workstation market these days”.

“It's not just pure graphics”, said Peter. “Products like Tesla don't really have any competition in the market”.

There's a famous story of how a journalist working for one of the UK's top design magazines was asked ‘How should you choose the best workstation card for a given application?' and started his response with “Well, the number of Cuda cores is going to be vital”.

While Radeons might vie for parity in the desktop graphics space and it looks like a clean sweep for APU into the next generation consoles, the workstation area seems saturated with a specific shade of green.

But nVidia is far from resting on its laurels.

During the unveiling of its 7 billion transistor mega chip, nVidia's CEO Jen Hsun Huang was quoted as saying “The GK110 is the most complex IC commercially available on planet”.

The k6000 workstation product variant is expected to launch this month, but when asked about details, PNY's Peter just smiled – refusing to be drawn. We'll read into that steely confidence that it's gonna be a killer product.

Yet the power being delivered by these products is still only the tip of the titanic iceberg that nVidia is floating into the market over the next four years. The Maxwell architecture should begin to make an appearance around the end of 2014, once the new 20nm process has been proven. From there, the drive to creating an ExaScale system inside 4 years is on.  128GB graphic cards anyone?

While AMD is approaching the supercomputer market with SeaMicro 'bricks', nVidia is looking to pour a few thousand tons of concrete on the problem of 'slamdunk technology for next generation mega machines'. PNY's Peter Butler is a very happy chappy.

.

KitGuru says: While AMD might have the lead in certain professional applications, like password checking, overall the connection nVidia seems to have with Hollywood and the major design houses will be hard to break. And all of that phat profit margin will be funding serious, on-going graphics research and development. What's the Radeon reply? How will AMD fund the 9000 and 10000 series high end products?

Comment below, in the forums or join us over on Facebook.

The post nVidia set to seal its lock on the workstation market first appeared on KitGuru.]]>
https://www.kitguru.net/components/graphic-cards/harrison/nvidia-set-to-seal-its-lock-on-the-workstation-market/feed/ 2
Asus launches range of ESC G2 supercomputers https://www.kitguru.net/professional/workstation/harrison/asus-launches-range-of-esc-g2-supercomputers/ https://www.kitguru.net/professional/workstation/harrison/asus-launches-range-of-esc-g2-supercomputers/#respond Fri, 25 May 2012 06:09:21 +0000 http://www.kitguru.net/?p=92358 If you’ve been sitting at home, researching GPGPU and wondering “What’s the best way for me to assemble a super computer with almost 2,000 CUDA cores, 24GB of memory and the ability to process close to 5 billion triangles a second?”, then Asus may have the answer for you. Not content with the 3,886 awards …

The post Asus launches range of ESC G2 supercomputers first appeared on KitGuru.]]>
If you’ve been sitting at home, researching GPGPU and wondering “What’s the best way for me to assemble a super computer with almost 2,000 CUDA cores, 24GB of memory and the ability to process close to 5 billion triangles a second?”, then Asus may have the answer for you.

Not content with the 3,886 awards it collected in 2011, Asus is launching even more product. Right now, the thrust is coming from the server and workstation team.

There seems to be 3 models in the new launch, all of which have G2 in the name – but we’re not sure why. Removing the G2, we get the following models:-

ESC1000:              One way graphics workstation based on Intel’s E5-1600 processor

ESC2000:              Two way workstation and GPU server with a pair of Intel Xeon E5-2600 chips

ESC4000:              Hybrid HPC server that also comes armed with two E5-2600 processors

The main difference, when it comes to GPGPU configurations, is the number of PCI Express slots available with each model.

The 1000 model can handle up to 3 graphic cards, but the ESC 4000 is ready, willing and able to take on four nVidia Quadro 6000 cards – each of which has 448 CUDA cores, 6GB of GDDR5 and a thirst that can only be quenched by 225 watts of juice.

Speaking with Miodrag Relic, head of marketing for the Open Platform Business Group, we were told, “These super computer servers and workstations offer professional users high density hybrid computing power, delivering upgrades in graphics capabilities and expandability”.

Regarding specifications, Relic said, “With support for four graphics cards and up to two Intel Xeon E5-2600 processors, they enable greater performance, presenting better reliability with Intel Ethernet and power supply redundancy. Expansion slots comply with PCI Express 3.0 x16 specifications, and all power supplies adhere to 80 PLUS standards to ensure improved energy efficiency. Additional features such as hot-swap storage, ASUS PIKE, ASUS SSD Caching, and InfiniBand round off ideal platforms for high-demand computing”.

We didn’t ask about noise levels when dealing with a fully loaded system whose requirements would be close to 1,300 watts – but we guess no one questions the fact that a Ferrari makes noise when a wide boy guns it past 150mph.

Fancy giving proteins a folding they'll never forget? Asus believes it has the workstation you need

KitGuru says: Around 10 years ago, almost no one outside Stanford was talking about GPGPU and the need for more Hybrid Computing Power. Now we have companies like Asus launching ranges of systems for just that purpose. Wonder what technology is considered rare/unusual today, but which will form part of a standard launch in 2022?

Comment below or in the KitGuru forums

The post Asus launches range of ESC G2 supercomputers first appeared on KitGuru.]]>
https://www.kitguru.net/professional/workstation/harrison/asus-launches-range-of-esc-g2-supercomputers/feed/ 0
Exclusive interview with Intel’s John Hengeveld about parallelism https://www.kitguru.net/components/cpu/faith/exclusive-interview-with-intels-john-hengeveld-about-parallelism/ https://www.kitguru.net/components/cpu/faith/exclusive-interview-with-intels-john-hengeveld-about-parallelism/#comments Tue, 12 Jul 2011 11:10:57 +0000 http://www.kitguru.net/?p=52860 KitGuru has long been fascinated with the increased levels of parallelism in modern processors. The move to Core, showed the world that doing more work per cycle is usually much more powerful than simply cranking up the clock. Intel's research teams rapidly embraced this increased parallelism and now we're moving toward a world unlike any …

The post Exclusive interview with Intel’s John Hengeveld about parallelism first appeared on KitGuru.]]>
KitGuru has long been fascinated with the increased levels of parallelism in modern processors. The move to Core, showed the world that doing more work per cycle is usually much more powerful than simply cranking up the clock. Intel's research teams rapidly embraced this increased parallelism and now we're moving toward a world unlike any we've seen before. KitGuru managed to get up close and personal with John Hengeveld, Director of Marketing inside the Data Centre Group at Intel. We'd normally re-write Q&A sessions into a stream of prose, but this one (ironically) works really well in sequence. Our questions are in black – just in case you're not sure.

Abstract concepts are often best understood with real world analogies. What analogy best highlights the difference between a computer system with very low/no parallelism and one that is extremely parallel?

I like to use the highway analogy and lanes. A highway that has only one lane, gets less people to work than a highway with 8 lanes. Computer systems with parallelism allow more work to flow through. Which is better a rush hour, a 6 lane highway with a 55mph speed limit or a 1 lane highway with a 70mph speed limit? The 6 lane highway gets 5 times more cars through (roughly) per hour.

An HPC workload is rush hour traffic. Lots of pieces of code all trying to get someplace at the same time. A traditional Xeon processor is has a small number of lanes (let’s call it 6 for this example) and the speed limit is 350 MPH. An MIC processor is a 50+ lane highway – and let’s call the speed limit 150 MPH. When you have a special assignment and have lots of small cars the MIC processor does better. If you only have fewer big or huge cars and they can all go 350mph. then the Xeon solution is better.

Haven’t computers been working in parallel for years? What’s changing with the new Intel technologies?

Yes, computers have been working in parallel for a long time. Intel has a long history of building more lanes to the highway. You can take many of today's computers and have them work in parallel and get some level of performance – like building a bypass. What is changing here is performance power efficiency. Not only does MIC give you more lanes, but it also substantially reduces the power requirement for computation. It’s like all the cars on the highway get better fuel efficiency just because they are on this road.

Another way to achieve parallelism is to build a special purpose accelerator, like a GPGPU or FPGA. The challenge there is that it takes a lot more work to develop an application for these devices. Think of this as mass transit. If you do the work to get to where it can pick you up – and do the work to get where you are going from – to where you get dropped off and you don’t mind the extra time and cost to do that work… mass transit might get you someplace. But with Intel MIC, you can use your own car (programmed with Intel standard tools and technologies). You not only gain fuel efficiency, you also get through more work, sooner.

From a hardware perspective, what are the biggest challenges as you move forward? How about from a software perspective?

The biggest challenges in getting to Exascale are power efficiency and managing programming complexity. Intel is committed to working with industry, government and research partners to try and reach, by the end of the decade, an Exascale machine with 20MW of power per system. This a big challenge that will take a lot of work and new learning on many peoples’ part to get there.

Key to Intel’s approach is preserving a straight forward programming model, so that a larger range of workloads can take advantage of such of a machine. The harder it is to program – the fewer applications will be developed.

Customers are hearing a lot of new words right now, like MIC, Knights Corner and Cilk – what’s the ‘single line explanation’ for these?

MIC is Intel Many Integrated Core: This is an adjective… followed by a noun… ‘architecture, product, products, research, programming model'. An architectural approach that solves highly parallel problems with increased number of smaller and more specific IA cores than is used for general purpose workloads and systems.

Knights Corner: Intel code name for the first commercial Intel MIC architecture hardware product.

Cilk: A programming language, based on C++, for expressing parallelism. Well worth a trip to Wikipedia.

Kirk Skaugen, vice president and general manager of Intel's Data Center Group holding the 'Knights Ferry' MIC card at the international Supercomputing Conference in Hamburg, Germany. Due to launch inside 6 months, or so we believe.

What is an intrinsic – how can a normal person understand this kind of terminology?

Think of a compiler like an automatic transmission for a car. Most of the time the automatic transmission takes its inputs and configures the car with the right gears to drive smoothly and efficiently to your destination.

Most of the time a compiler takes programming language, and pretty much knows how to do the right thing to get the work done.

But sometimes you face a sleep hill, a slippery road or a dangerous condition where you the driver want to have specific control over how the car operates. Think of an intrinsic as an on-demand manual transmission.

An intrinsic is a programming language element that directs the compiler on how to deal with the use of the underlying machine for a specific element of data or code. This is commonly used in programming for vectorization and parallel programming and allows the programmer to more specifically direct (at a high level language) the compiler on how to use the underlying hardware.

On a recent call with the Intel experts, KitGuru got the impression that you have already invested a lot of time/effort/resource toward bringing this parallel evolution to market, but it seems that more is needed from ‘outside’. Specifically in the area of development languages. With something as ubiquitous as C, where does the responsibility lie in making this language much more ‘parallel friendly’?

There is a great deal of innovation going on this space. As with most things, Intel looks at this as an industry/ecosystem challenge.

New methodologies are constantly emerging. Intel’s tools for software development create a platform for the industry to innovate upon. The whole software industry has to take this on.

Given that graphics has traditionally been a slow, but more parallel environment; is it fair to say that Intel has gained new insight about ways to develop/evolve next generation processors AFTER working through the Larrabee project?

Yes absolutely.

Intel has taken the learning from the Larrabee project [I keep promising THIS will be the last time I type Larrabee! – Interviewee note] as well as years of research at Intel labs, and learning from our mainstream HPC business to be applied to MIC activities and indeed to our whole product line. Yes, Intel and our partners have learned a lot and we’ve gained new insight about how future processors will be better mapped to high performing, and highly efficient, systems.

Last question, with Intel’s on-going advances in hardware – and this drive to new thinking/languages/environments – when will an Intel ‘CPU’ have human-brain-levels of processing power?

It’s a ways off. Inorganic computation today is organized around perfect data storage and retrieval and fast calculation. The organic computation (like the human brain) is organized around observation, intuition, inference and learned response. The logic elements of the human mind (neurons) have fascinating properties of adaptation that inorganic computation labours hard to reproduce. We have many years of research to go before we will be truly able to wrap our technology around this question. If we DO have the ability to create this level of computation, is it economically valuable to do so, such that investment will occur and profitability will follow. There is remarkable technology in both the human and inorganic computation. Each optimized for a purpose.

This is fascinating stuff for KitGuru. The legal and moral issues that a human-level of parallel computing would bring, are simply mind boggling, and yet – that is definitely something on our horizon. At KitGuru, we like to get a feel for the person behind the voice, so we presented John Hengeveld with our standard array of mind-probing questions. Infer what you will from the replies.

First up, what are your favourite bands?
John replied, “The Beatles, Matchbox 20, Cake and Lady Gaga. Yes. Lady Gaga. And in the same sentence as the Beatles!”

What about cars?
“Easy, Mustang built half way through 1964”.

If John had to star in a film, opposite one leading actress, who would that be?
“Drew Barrymore”.

KitGuru is a great lover of the food as well as the technology. If John was cooking for himself, what’s his favourite food?
“I am a serious chef. Panko and Pepper Encrusted and Pan seared ahi tuna with raspberry wasabi vinegrette”. Gosh.

What food can you only get 100% right in a restaurant ?
Molecular Gastronomy“, he replies. “It requires so much technology to deliver. Proper desserts.”

Given that Intel's kind of science did not exist when you were a kid –what did you want to be when you were 14?
“Naval Officer”.

How much parallelism would you need to be able to distinguish complex flavours in realtime?

KitGuru says: We thank John for taking the time to share his insights into the amazing future of parallelism.With the move to a 7nm process by the end of 2016 already well known – and a series of boosts to the levels of parallelism available, already lined up – we're going to be living in a very different tomorrow. One question remains with KitGuru: If parallelism truly takes off, then how many of us will actually need to know how it works in order to use it? Will the pool of hardcore programmers needed increase or decrease with time? All fascinating stuff.

Picture of John taken with courtesy from ITTV @ Siggraph 2010

The post Exclusive interview with Intel’s John Hengeveld about parallelism first appeared on KitGuru.]]>
https://www.kitguru.net/components/cpu/faith/exclusive-interview-with-intels-john-hengeveld-about-parallelism/feed/ 8
AMD releases latest (not) Stream SDK https://www.kitguru.net/components/graphic-cards/faith/amd-releases-latest-not-stream-sdk/ https://www.kitguru.net/components/graphic-cards/faith/amd-releases-latest-not-stream-sdk/#respond Tue, 25 Jan 2011 17:55:54 +0000 http://www.kitguru.net/?p=30997 Parallel processing is something which graphics cards do amazingly well and which regular CPUs are only slowly waking up to. AMD's developer kit has just been updated, so KitGuru decided to (in GPGPU's honour) multi-task by typing this story, while biting our nails, catching up on One Tree Hill and defrosting chicken. When Intel announced …

The post AMD releases latest (not) Stream SDK first appeared on KitGuru.]]>
Parallel processing is something which graphics cards do amazingly well and which regular CPUs are only slowly waking up to. AMD's developer kit has just been updated, so KitGuru decided to (in GPGPU's honour) multi-task by typing this story, while biting our nails, catching up on One Tree Hill and defrosting chicken.

When Intel announced that it's original Core processors would process video twice as fast a previous generations, the public gave a small smile, while the collective workers at ATI and nVidia laughed out loud.

Double the power sounds great in CPU terms, but around the same time there were announcements from the graphics boys that using CUDA/Avivo will give you ~18x increase in video processing performance. It's not always such a big gap, and there are definitely things that a CPU does better, but it will be a while before CPUs can cycle through as much ‘appropriately coded' parallel work as a GPU.

When AMD puts out a cry that says “Calling all software development innovators in general purpose GPU (GPGPU), data parallel and heterogeneous computing”, they are really pulling pro-punters into start smart stuff on the new Fusion platform.

If this pick makes sense to you - click it

They are trying to entice the really smart folk out there to head toward Bellevue, Washington from 13th to 16th June, where you can wear your sandals with pride and enjoy in-depth intercourse with like-minded folk.

We took this astonishing picture, using a special camera inside Terry Makedon's head when he agreed to move to the AMD Fusion team. Please file under 'hopeless romantic'

Arguably the most useful thing to come out of AMD's on-going work with developers is the creation and update of SDKs (developer kits).

The one for GPGPU (general purpose computing using graphic chips) was previously known as the ATI Steam SDK. Given AMD's love for all things open, it's now got a more mainstream (albeit not sexy) name as the AMD Accelerated Parallel Processing SDK v2.3.

If you're hot and steamy under the collar after reading that name, then click here for instant relief.

ArcSoft has embraced the APP and, apparently, Total Media Theatre 5 is very parallel indeed

How does AMD describe this new tool?

The AMD APP SDK v2.3 empowers software developers to write new applications that can take full advantage of the parallel processing power of heterogeneous computing platforms, such as those based on new AMD E-Series and C-Series APUs that combine a multi-core CPU and DirectX®11-capable GPU on a single die. In addition to support for the first AMD Fusion APUs, the updated AMD APP SDK v2.3 offers full OpenCL™ 1.1 support, improved runtime performance and math libraries for OpenCL, and support for AMD Radeon™ HD 6900 Series graphics.

Glad you asked?

Folks with a love of AMD development should go to this site and if you are of an nVidia persuasion, try this link.

KitGuru says: It will be interesting to see where all of this development work takes us over he next 5 years. Will CPUs catch up with GPUs?  WIll the Fusion-class chips create an attractive new middle ground or will all of the different types stay specialised?

Comments below or in the KitGuru forum.

The post AMD releases latest (not) Stream SDK first appeared on KitGuru.]]>
https://www.kitguru.net/components/graphic-cards/faith/amd-releases-latest-not-stream-sdk/feed/ 0