AMD Polaris 10 and Polaris 11 OpenGL Benchmarks Leaked, Polaris 11 Having Two SKUs
Back at the GDC 2016, AMD showed a brief demo of their upcoming Polaris 10 and Polaris 11, revealing the 14nm based GPUs deliver 2.5x perf/watt improvement over the predecessor. As for their market positioning, Polaris 11 will target the notebook market, while Polaris 10 is aimed at the mainstream desktop and high-end gaming notebook segment.
That being said, both next gen GPUs have just been spotted on GFXBench, with scores revealing the performance of two different Polaris 11 SKUs, the 67FF and 67E8, along with Polaris 10 67DF. The new OpenGL benchmark scores won’t reveal much about the new GPUs, except that Polaris 11 SKU 67E8 is faster than Polaris 11 67FF.
Also, strange it may seem but the Polaris 10 67DF appears to be as fast as Polaris 11 67E8. However, do note that it’s unclear if these chips were running at full speed and full silicons, and moreover, OpenGL benchmarks are often capped at 60FPS. So we’d advise you to take these values with the usual grain of salt.
If previous reports are to be believed, Polaris 10 could feature 2304 stream processors across 36 CUs, and support up to 8GB of GDDR5(X) memory on a 256-bit memory interface. The GPU is set to replace the Radeon M300 high-end segment based on the Tonga silicon.
Polaris 11, on the other hand, is to succeed the “Curacao” GPU which powers various mid-range cards. It features 1024 stream processors over 16 CUs, and is expected to be be capable of having 4GB of GDDR5 memory on a 128 bit memory interface.
In terms of the power consumption, the Polaris 10 GPU will have an impressive TDP of slightly above 100W (110-135W), and should have “no problems” with running the latest DirectX 12 games “at a resolution of 1440p with a stable 60 frames per second.” The other GPU will be the power efficient version of AMD’s lineup with a TDP of just 50 watts.
Both Polaris based GPUs are slated to debut at Computex 2016 which runs from 31st May to the 4th June 2016 in Taipei.