Palit RTX 5090 GameRock Review

1. Introduction2. Meet the Palit RTX 5090 GameRock3. Testing Methodology4. Game Benchmarks5. Clock Speed6. GPU Temperatures7. Acoustics8. Noise-Normalised Temperatures9. Power Draw and Efficiency10. Overclocking11. Closing Thoughts12. View All Pages

Ever since the release of Palit's RTX 4090 GameRock, I've been waiting to see what the company would do next, and today we finally get our answer. That's because we're reviewing the RTX 5090 GameRock, featuring a bold, eye-catching design and a fat – effectively quad-slot – cooler. Concerns around pricing and availability remain just as pressing as they did when the RTX 5090 first launched, but what can this partner card bring to the table?

Timestamps

00:00 Intro
00:44 5090 pricing/availability – no pre-orders at all
01:21 GameRock design impressions
03:10 PCB and heatsink analysis
04:25 Test setup
04:52 Thermals and noise
06:05 Game benchmarks
06:51 Power – 12VHPWR is a huge concern
10:07 Overclocking
10:50 Closing thoughts

I remember the first time I opened the box of the last-gen Palit RTX 4090 GameRock and couldn't quite believe my eyes. The shroud was almost entirely covered in a crystal-like plastic, and once the system turned on, the RGB lighting was like nothing I'd ever seen before.

Following up something like that can often be tricky, but Palit has gone for what it describes as a ‘chameleon' aesthetic with the RTX 5090 GameRock. It's still incredibly eye-catching, but we take a closer look at the design on the next page.

It's also worth pointing out that I was sent the regular GameRock, rather than the GameRock OC, so my sample ships without a factory overclock – i.e. at reference spec.

	RTX 5090	RTX 5080	RTX 4090	RTX 4080 Super	RTX 4080
Process	TSMC N4	TSMC N4	TSMC N4	TSMC N4	TSMC N4
SMs	170	84	128	80	76
CUDA Cores	21760	10752	16384	10240	9728
Tensor Cores	680	336	512	320	304
RT Cores	170	84	128	80	76
Texture Units	680	336	512	320	304
ROPs	176	112	176	112	112
GPU Boost Clock	2407 MHz	2617 MHz	2520 MHz	2550 MHz	2505 MHz
Memory Data Rate	28 Gbps	30 Gbps	21 Gbps	23 Gbps	22.4 Gbps
L2 Cache	98304 KB	65536 KB	73729 KB	65536 KB	65536 KB
Total Video Memory	32GB GDDR7	16GB GDDR7	24GB GDDR6X	16GB GDDR6X	16GB GDDR6X
Memory Interface	512-bit	256-bit	384-bit	256-bit	256-bit
Memory Bandwidth	1792 GB/Sec	960 GB/Sec	1008 GB/Sec	736 GB/Sec	716.8 GB/Sec
TGP	575W	360W	450W	320W	320W

First, a quick spec recap. The RTX 5090 is built on the new GB202 die, measuring 750mm², though it's not quite a full implementation of the silicon. Instead we find a total of 11 Graphics Processing Clusters (GPCs), each holding up to eight Texture Processing Clusters (TPCs), for a total of 85. Each TPC is home to two Streaming Multiprocessors (SMs), giving us 170, and each SM still holds 128 CUDA Cores, meaning the RTX 5090 has an eye-watering total of 21760 shaders. We also find 170 RT cores, 680 Tensor cores, 680 Texture Units, and 176 ROPs.

This time around, however, there's no node-shrink, and GB202 remains fabricated on TSMC's N4 node, as per the RTX 40-series. As such, rated clock speed is not increased this generation and is actually touted slightly below that of the RTX 4090, with the RTX 5090 delivering a rated 2407MHz boost clock, compared to its predecessor's 2520MHz boost.

The memory configuration has seen significant upgrades, though. The RTX 5090 now comes equipped with a super-wide 512-bit memory interface, paired with 32GB GDDR7 memory running at 28Gbps, and that puts total memory bandwidth at a staggering 1792 GB/s. L2 cache is also increased to 98MB, up from the 74MB of the RTX 4090.

Considering the large increases to die size and core count, but with no node shrink, it's perhaps unsurprising to see power draw has increased, this time boasting a 575W TGP. This is something we focus on closely in this review, using our enhanced GPU power testing methodology, so read on for our most detailed power and efficiency testing yet.

The Palit RTX 5090 GameRock ships in a colourful box, with a partial image of the graphics card visible on the front. On the back, the company highlights various key features of the card and its design.

Palit bundles various accessories with the card, including some stickers, a small mousepad, a GPU holder, as well as the quad 8-pin to 12V-2X6 power adapter.

Then we come to the card itself. It's something quite special – Palit calls the design the ‘Chameleon Panel', and it's essentially an acrylic panel that changes colour depending how the light hits it. It's certainly unlike like I've ever seen on a graphics card before, and matters are only enhanced by the RGB lighting which we'll get to shortly.

It's also packing in three of Palit's TurboFan 4.0 fans, and these measure approximately 95mm in diameter.

It's worth highlighting that the GameRock is probably best suited to be vertically mounted – the sides of the card are much less exciting, being fairly plain black plastic. I think Palit could have improved this area as it doesn't really fit the overall aesthetic.

In terms of its size, the card measures in at 331.9 x 150 x 70.4 mm, while it weighed in at 2222g on my scales.

As for the backplate, this is a full-length piece of brushed metal, but with a large flow through area to help with heat dissipation. It's certainly better than nothing, but again I can't help but feel it doesn't match the vibrancy of the shroud design – for me, the silver backplate on the 40 series GameRock cards would have worked much better here.

You can also find a BIOS switch right next to the I/O bracket, and this offers a choice of Performance or Silent modes. Both feature identical clock and power targets, but the Silent mode has a less aggressive fan curve.

Then for the RGB lighting, you get a full 360 effect around the shroud itself, while the GameRock logo on the side also lights up. Palit's ThunderMaster utility can be used to control the lighting, or you can connect the card to your motherboard via the ARGB header if you wish.

Power is of course supplied by the 12V-2X6 connector, with the adapter shown above. Display outputs consist of three DisplayPort 2.1 and one HDMI 2.1 connectors.

The first step for disassembling the card is to remove the backplate, revealing the back of the PCB. You can also see just how much ‘flow through' space Palit has built into the card – by my measurements, the PCB extends only about 20cm, leaving nearly 13cm dedicated to unobstructed airflow.

Then we get a look at the PCB itself. It's very densely packed, with no less than 22 phases for the GPU, and seven for the memory. Monolithic Power Systems MP87993 MOSFETs are used throughout, and these are rated at 50A. I could only see one controller on the PCB too, Monolithic's MP29816-A, so it would seem this is used for both memory and GPU VRMs.

As for the cooler, Palit is using a hefty heatsink array here, while the GPU and VRAM contacts with a vapour chamber. Secondary baseplates are used to cool the MOSFETs. The heatsink utilises a total of eight copper composite heatpipes.

In case you're wondering, Palit opted for what appears to be standard thermal paste, unlike the RTX 5090 Founders Edition which uses liquid metal TIM.

Driver Notes

• AMD GPUs were benchmarked with the Adrenalin 24.12.1 driver.
• Nvidia GPUs (except for RTX 5090 and RTX 5080) were benchmarked with the 566.14 driver.
• RTX 5090 was benchmarked with the 571.86 driver supplied to press.
• RTX 5080 was benchmarked with the 572.02 driver supplied to press.

Results are only directly comparable where this exact configuration has been used.

Test System:

We test using a custom built system powered by MSI, based on AMD's Zen 5 platform. You can view the Powered by MSI store on AWD-IT's website HERE.

CPU	AMD Ryzen 7 9800X3D
Motherboard	MSI MPG X870E Carbon WiFi
Memory	64GB (2x32GB) Kingston Fury Beast DDR5 6000MT/s CL30
Graphics Card	Varies
SSD	4TB Kingston NV3 Gen 4 PCIe NVMe
Chassis	MSI MPG Gungnir 300R Airflow
CPU Cooler	MSI MAG CoreLiquid i360
Power Supply	MSI MEG Ai1300P
Operating System	Windows 11 23H2
Monitor	MSI MPG 321URX QD-OLED
Resizable BAR	Enabled for all supported GPUs

Comparison Graphics Cards List

• AMD RX 7900 XTX 24GB
• AMD RX 7900 XT 20GB
• Sapphire RX 7900 GRE Nitro+ 16GB
• AMD RX 7800 XT 16GB
• Sapphire RX 7700 XT Pulse 12GB
• MSI RTX 5090 Suprim SOC 32GB
• Nvidia RTX 5090 FE 32GB
• Nvidia RTX 5080 FE 16GB
• Nvidia RTX 4090 FE 24GB
• Nvidia RTX 4080 Super FE 16GB
• MSI RTX 4070 Ti Super Ventus 3X 16GB
• Nvidia RTX 4070 Super FE 12GB
• Nvidia RTX 4070 FE 12GB
• Nvidia RTX 4060 Ti FE 8GB
• Nvidia RTX 3080 Ti FE 12GB

All cards were tested at reference specifications. For factory overclocked cards, this means we manually ‘undo' the overclock via MSI Afterburner or AMD/Intel's built-in tuning tools. Or, for cards like the Sapphire RX 7900 GRE Nitro+, we enable the reference-clocked BIOS instead of the default OC BIOS.

Software and Games List

• Alan Wake II (DX12)
• Black Myth: Wukong (DX12)
• Cyberpunk 2077 (DX12)
• F1 24 (DX12)
• Final Fantasy XVI (DX12)
• Forza Horizon 5 (DX12)
• Ghost of Tsushima (DX12)
• Horizon Forbidden West (DX12)
• The Last of Us Part 1 (DX12)
• A Plague Tale: Requiem (DX12)
• Ratchet and Clank: Rift Apart (DX12)
• Returnal (DX12)
• Senua's Saga: Hellblade 2 (DX12)
• Shadow of the Tomb Raider (DX12)
• Starfield (DX12)
• Star Wars Outlaws (DX12)
• Total War: Warhammer III (DX11)

We run each benchmark/game three times, and present mean averages in our graphs. We use FrameView to measure average frame rates as well as 1% low values (99th percentile) across our three runs.

Here we test three games, all at 3840×2160 resolution using maximum image quality settings.

We don't focus too heavily on game benchmarks in our partner cards reviews as performance doesn't tend to change a whole lot when compared to the reference models. However, given the GameRock is a stock-clocked card, it didn't quite match the MSI Suprim SOC, but the difference is typically 3-4FPS, so really nothing you'd notice in the real world.

Here we present the average clock speed for each graphics card while running A Plague Tale: Requiem for 30 minutes. We use GPU-Z to record the GPU core frequency during gameplay. We calculate the average core frequency during the 30 minute run to present here.

Despite sharing the same rated clock speed as the 5090 Founders Edition – as both use the reference spec – we actually saw the GameRock running faster during our clock speed testing. The difference isn't huge, but it was measurable as you can see here.

Averaged over the thirty minute stress test, the GameRock's Performance BIOS clocked in at 2690MHz, while the Silent BIOS was barely any slower at 2670MHz. The Performance BIOS therefore runs about 60MHz faster than Nvidia's FE model.

For our temperature testing, we measure steady-state GPU temperatures under load. A reading under load comes from running A Plague Tale: Requiem for 30 minutes.

As for out of the box thermals, the GameRock does well enough here. It's clearly not on the same level as the Suprim, but the Performance BIOS GPU temperature is on par with the Founders Edition while reducing VRAM temperatures by 6C. The Silent BIOS only runs fractionally warmer too, so I'd say these results are good enough, if not spectacular.

We take our noise measurements with the sound meter positioned 1 foot from the graphics card. I measured the noise floor to be 32 dBA, thus anything above this level can be attributed to the graphics cards. The power supply is passive for the entire power output range we tested all graphics cards in, while all CPU and system fans were disabled. A reading under load comes from running Cyberpunk 2077 for 30 minutes.

Things are complicated by the fact that the GameRock isn't particularly quiet, either. For starters, there's very little difference between the Performance and Silent modes – the former runs the fans at 1810rpm, while the latter only drops that to 1745rpm, so resulting in only a slight drop in noise levels. Even then, it's still hitting 39dBa in the Silent mode, which is the same noise output we recorded from the FE. MSI's Silent mode, for instance, dropped noise down to 33dBa, so we're not really getting any noise improvements from the GameRock.

Following on from our stock thermal and acoustic testing, here we re-test the operating temperature of the GPU, but with noise levels normalised to 40dBa. This allows us to measure the efficiency of the overall cooling solution as varying noise levels as a result of more aggressive fan curves are no longer a factor.

Given the Performance BIOS already runs at 40dBa noise output, there's no need to adjust fan speed for our noise normalised testing. That does mean the GameRock isn't overly impressive, with the GPU running 1C hotter than the Founders Edition. Thankfully the VRAM is an improvement, dropping 4C compared to the FE – but MSI's Suprim SOC is far and away the better cooler.

Power draw

We use Nvidia PCAT to measure power draw of the graphics card only, with readings from both the PCIe slot and the PCIe power cables combined into a single figure. We use A Plague Tale: Requiem (4K/2160p) for this testing.

Somewhat staggeringly, power draw for the GameRock is even higher than I was expecting. On paper the card has a 575W TGP, and Palit has not increased the power limit. Despite that, we saw real-world power draw hitting just under 620W in A Plague Tale: Requiem, nearly a 50W increase over the Founders Edition.

Now, I know what you're thinking – with all the 12VHPWR drama going on, 620W does not sound like a good idea. Upon further analysis, we found that 13W was coming over the PCIe slot, meaning the cable itself was drawing roughly 607W at any given time. I also used a clamp meter to measure current over the six 12V wires, and load balancing was fine, ranging from 7.5A to 8.6A. Total current clocked just under 50A, which checks out considering the power draw on show.

That being said, there's clearly huge concerns around the 12VHPWR, or 12V-2X6, cable itself. Our cable and PSU behave properly, but Der8auer's recent findings suggest that behaviour can vary wildly, and dangerously so. Buildzoid pointed out that Nvidia removed two shunt resistors from the power delivery configuration, meaning the GPU isn't aware of how much power is coming from each individual wire, so if 50A was coming from just two wires, it would still chug merrily away.

Nvidia is yet to respond to these incidents, but it seems obvious to me that changes are needed. Making the power delivery worse than the RTX 3090 Ti, yet power draw hitting over 600W as we can see here, just seems like a very bad idea. Additionally, the lack of any real headroom – with the cable able to deliver 684W at max, so coming within just 11% of that figure here – seems like an equally bad idea when considering a good-quality single 8-pin could do 288W (at 24A), which is almost 50% safety headroom.

Performance per Watt

Combining the power draw values shown above with the performance data, we present performance per Watt for each graphics card tested:

In any case, that increase in power draw with basically no performance improvement to go alongside means overall efficiency has dropped compared to the FE and is now only on par with the RTX 4070.

For our manual overclocking tests, we used MSI Afterburner. Our best results are as below.

Palit doesn't allow an increase to the power limit when overclocking – which seems sensible considering we were already at 620W at stock! – but we still managed to add 2000MHz to the memory and 240MHz to the GPU, bringing real-world speeds to around 3000MHz.

That overclock didn't net us huge gains – unlike the RTX 5080 and 5070 Ti, the 5090 seems a lot more power limited, so without any increase to power, we only get an extra 1-4FPS in my testing.

Actual power draw is also basically unchanged, considering we did not adjust the power limit slider.

Today we've been back at the ultra-high end, checking out Palit's RTX 5090 GameRock. It's a decent partner card, and I think its main point of appeal will come down to the aesthetics – the ‘chameleon panel' design Palit is using is very eye catching and is unlike any other card on the market right now. When you factor in the 360-degree RGB lighting too, this card is a prime contender to be vertically mounted as the showpiece component of your rig.

In fact, I'd go as far as to say that not vertically mounting it would be quite a waste, as the rest of the design is a lot more boring. There's just black plastic on the sides, while the backplate is made of black brushed metal. I think Palit could've done a lot more with the rest of the design, to bring it in-line with the psychedelic-looking shroud. Right now it feels like they only got half way designing the card and said ‘good enough'.

The cooler itself is fine but unspectacular. Noise-normalised performance was about level with the 5090 Founders Edition when looking at GPU thermals, whereas the VRAM saw just a 4C drop. Compared to the Suprim SOC, MSI's card is clearly superior in terms of its overall thermal performance, while also delivering much lower noise levels out of the box.

We also have to address the 12VHPWR situation that has reared its head again over the last week. While my own current testing today did not suggest anything out of spec, with even load balancing on show, the fact remains this is a GPU drawing over 600W across the cable, with next to no safety headroom and no way for the GPU itself to respond to dangerous current imbalances. I really hope Nvidia will be able to take the feedback on board and improve things, else I fear we may just hear more stories of melting connectors and other dangerous cable-related incidents.

The lack of availability is still a huge problem for the RTX 5090, too. I've not been able to find anywhere in the UK actually taking pre-orders for this GPU, so it will likely be months before it is actually available to buy. Curry's is listing a £2200 asking price for the GameRock, which works out as an 11% premium over the Founders Edition, and is way cheaper than the MSI Suprim SOC.

It's still incredibly expensive though, and for the money I would hope for better thermal performance, given things aren't really improved over the Founders Edition to any significant degree. It is, however, still fine overall, and the USP is clearly the shroud, so the Palit RTX 5090 GameRock would be one to consider – if you could actually buy one…

Pros

• Small performance gains over the RTX 5090 Founders Edition.
• Eye-catching ‘Chameleon' shroud design.
• Our sample hit around 3GHz when overclocking.
• Huge 32GB GDDR7 frame buffer.
• DLSS 4 has improved Ray Reconstruction and Super Resolution scaling.
• Multi Frame Generation enables higher frame rates than would otherwise be possible.

Cons

• Not available to buy anywhere.
• Very high power demands.
• It's physically huge.
• Not much of an improvement over the FE in terms of noise-normalised performance.
• 12VHPWR remains a very valid concern.

KitGuru says: It's an eye-catching design for sure, but we'd hope for a greater improvement to thermals compared to Nvidia's Founders Edition.

Become a Patron!