Compute

Unfortunately, as I mentioned earlier in my testing observations, the state of AMD's OpenCL driver stack at launch is quite poor. Most of our compute benchmarks either failed to have their OpenCL kernels compile, triggered a Windows Timeout Detection and Recovery (TDR), or would just crash. As a result, only three of our regular benchmarks were executable here, with Folding@Home, parts of CompuBench, and Blender all getting whammied.

And "executable" is the choice word here, because even though benchmarks like LuxMark would run, the scores the RX 5700 cards generated were nary better than the Radeon RX 580. This a part that they can easily beat on raw FLOPs, let alone efficiency. So even when it runs, the state of AMD's OpenCL drivers is at a point where these drivers are likely not indicative of anything about Navi or the RDNA architecture; only that AMD has a lot of work left to go with their compiler.

That said, it also serves to highlight the current state of OpenCL overall. In short, OpenCL doesn't have any good champions right now. Creator Apple is now well entrenched in its own proprietary Metal ecosystem, NVIDIA favors CUDA for obvious reasons, and even AMD's GPU compute efforts are more focused on the Linux-exclusive ROCm platform, since this is what drives their Radeon Instinct sales. As a result, the overall state of GPU computing on the Windows desktop is in a precarious place, and at this rate I wouldn't be entirely surprised if future development is centered around compute shaders instead.

Compute: LuxMark 3.1 - Hotel

Compute: CompuBench 2.0 - Level Set Segmentation 256

Compute: CompuBench 2.0 - Optical Flow

Forza Horizon 4 Synthetics
Comments Locked

135 Comments

View All Comments

  • GeoffreyA - Sunday, July 7, 2019 - link

    Many thanks, Ryan, to you and the team for all the hard work. We do appreciate it.
  • catavalon21 - Sunday, July 7, 2019 - link

    Hoping for really competitive results in the mid-range for compute, that AMD doesn't have drivers that support the new architecture is absurd. To not even run on some older computer work means this was clearly not ready for prime time. Shame on you, Lisa.

    I write this, very disappointed that the choice of a mid range GPU right now isn't much more difficult.
  • catavalon21 - Sunday, July 7, 2019 - link

    ...older COMPUTE work...<sigh>
  • just4U - Sunday, July 7, 2019 - link

    Holy crap.. I wasn't actually expecting Amd to come close to Nvidia with these. (Regardless of the hype by Amd) The 5700XT is just a smidge slower than the 2070S.. and it's quite a impressive jump over the RX580/90s they replace.
  • catavalon21 - Sunday, July 7, 2019 - link

    My whining about compute aside, you're right. The 5700XT competes very well against the 2070S - better than I hoped for.
  • DanNeely - Sunday, July 7, 2019 - link

    Yeah. AMD's showing is strong enough I'm wondering if we'll see farther NVidia price cuts in the near future.
  • Kevin G - Sunday, July 7, 2019 - link

    They are indeed impressive agains nVidia's Super cards but by pricing they're more of a Vega 56/64 replacement.
  • just4U - Sunday, July 7, 2019 - link

    I was considering it from a new norm on video card pricing as to me their upper mid range and don't appear to compete with Vega multipurpose cards to replace them.
  • tipoo - Sunday, July 7, 2019 - link

    Looks like that completely outsized Particle Physics subscore was real, from multiple results coming in. Interesting. Given AMD seems to be going for a hybrid RT approach for RDNA 2.0 in 2020, I wonder if this was a half step towards building out this portion of the chip for it.

    https://browser.geekbench.com/v4/compute/4259036

    Under OpenCL, it beats a 2080TI under CUDA, in that one subtest.
  • mildewman - Sunday, July 7, 2019 - link

    Can someone explain to me why Navi requires twice the number of transistors (10.3B) compared to Polaris (5.7B) for the same number of CU's ?

Log in

Don't have an account? Sign up now