Radeon Pro SSG

[Mike Halper]

Our's are (water cooled, undervolted, overclocked 1600 MHz).
The good thing about Davinci Resolve is, that it is not a game, it just needs RAW compute power (FP32) and only that.

Happy gaming with the Geforce cards, I just love the OpenGL overlay at 10-bit color depth and the 16 GB of HBM2.

Vega FE does 11.321 TFlops at 1382 MHz(Base Clock) and 13.1 TFlops at 1600 MHz(ours).
GTX1080ti does 10.6 TFlops at 1480 MHz(Base Clock) 50% faster than Vega FE base clock makes 11.321 x 1.5 ~ 17 TFlops ~ 2370 MHz (That is a hell of an over clock).

Titan V does 12.288 Tflops at 1200 MHz(Base Clock).

https://www.slashcam.de/artikel/Test/Auf-Augenhoehe-mit--Ti-tanen---AMD-Vega-56-unter-DaVinci-Resolve-14--Was-bringts--63--Vega56-unter-Resolve-14.html#Was_b

Those are just specs. Real world use will show differently. I just tested Vega FE vs 1080TI last week with actual playback and rendering of video in Resolve, Premiere, and FCPX. The 1080TI was faster.

 

[DJ Meyer]

It was a marketing gimmick then and it still is now.

It certainly smelled like one at the time. I wanted to test it for myself though given the implied endorsement from Red on stage. Who knows maybe the magic software update is indeed still coming....someday.

 

[Misha Engel]

Those are just specs. Real world use will show differently. I just tested Vega FE vs 1080TI last week with actual playback and rendering of video in Resolve, Premiere, and FCPX. The 1080TI was faster.

Premiere won't use much more than a RX580 or a 1060ti.

No, it is a real test done by real Germans and Germans only fuck up Diesel tests.

 

[DJ Meyer]

Premiere won't use much more than a RX580 or a 1060ti.

No, it is a real test done by real Germans and Germans only fuck up Diesel tests.

Ha.

Everyone knows that Vega's run faster in Germany than they do in the rest of the world.

 

[Mike Halper]

Well here's a comparison of Vega FE vs 1080Ti in Resolve Studio. X79 motherboard, 4930K 6-core 4.5GHZ overclock, 64GB RAM. The grade is 2 nodes. First node is primary corrections only to lift/gamma/gain, contrast, and saturation. Second node is temporal noise reduction set to 3 frames, best, and ranging around 15-20 on luma and chroma.

GH5 10bit H264 4K DCI -> ProResHQ 4K DCI, 1:53 clip
EVGA 1080TI FTW3 Gaming CUDA: 3:53 at 11.5fps
Vega Frontier Edition OpenCL: 8:18 at 5-5.5fps
Vega Frontier Edition Metal: 6:26 at 7fps

ProRes 4444 2K -> ProResHQ 4K DCI, 4:30 clip
EVGA 1080TI FTW3 Gaming CUDA: 7:42 at 14fps
Vega Frontier Edition OpenCL: 18:50 at 5-5.5fps
Vega Frontier Edition Metal: 13:05 at 8fps

ProResHQ 4K -> ProResHQ 4K DCI, 2:00 clip
EVGA 1080TI FTW3 Gaming CUDA: 3:34 at 13.5fps
Vega Frontier Edition OpenCL: 8:30 at 5.5-6 fps
Vega Frontier Edition Metal: 5:45 at 8fps

4K R3D -> ProResHQ 4K DCI, 8:50 clip
EVGA 1080TI FTW3 Gaming CUDA: 21:46 at 9.5-10fps
Vega Frontier Edition OpenCL: About an hour 4.5-5fps
Vega Frontier Edition Metal: about 29 minutes at 6.5-7fps

 

[Sabovic Adis]

What does it mean? Resolve is not optimized for Vega, or Vega is not optimized for Resolve? NVIDIA is cool with INTEL? What?

 

[Misha Engel]

What does it mean? Resolve is not optimized for Vega, or Vega is not optimized for Resolve? NVIDIA is cool with INTEL? What?

3 hours of testing within 2 hours on a hackintosh?.

 

[Sabovic Adis]

3 hours of testing within 2 hours on a hackintosh?.

Yep, that's it! :rofl:

:thumbsup:

 

[Mike Halper]

Who said it was within 2 hours? Doesn't matter how long it took anyway. The fact is that the tests show the Vega FE is slower than 1080Ti. Simple as that. It's not about specs. Specs don't tell you actual real-world performance.

 

[Sabovic Adis]

Now make some test that show that Vega is faster.
Thanks.

 

[Misha Engel]

Specs don't tell you actual real-world performance.

That's why I waited for an independent test https://www.slashcam.de/artikel/Test/Auf-Augenhoehe-mit--Ti-tanen---AMD-Vega-56-unter-DaVinci-Resolve-14--Was-bringts--63--Vega56-unter-Resolve-14.html#Was_b and it is still the only independent test I've seen sofar with BM Davinci Resolve 14 where they compare AMD VEGA with NVidia Geforce.

We bought our VEGA FE's based on this test and the tomshardware test and we haven't been dissapointed yet. It does what it has to do and the overclock scales more or less 100%.
We don't have GTX1080ti's so I can't compare. (but with some extrapolation you can get a descent insight).

 

[Misha Engel]

It looks like Adobe is still trying to speed up their workflow with the SSG.

https://www.provideocoalition.com/amd-radeon-pro-ssg-graphics-accelerates-4k-and-8k-workflows/

 

[Mike Halper]

That's why I waited for an independent test https://www.slashcam.de/artikel/Test/Auf-Augenhoehe-mit--Ti-tanen---AMD-Vega-56-unter-DaVinci-Resolve-14--Was-bringts--63--Vega56-unter-Resolve-14.html#Was_b and it is still the only independent test I've seen sofar with BM Davinci Resolve 14 where they compare AMD VEGA with NVidia Geforce.

That is probably the only review in existence that shows a Vega 56 (flashed with a Vega 64 BIOS) meeting or beating a 1080TI in Resolve. Any mention of the comparison from other users everyone always says 1080Ti is faster than the Vega GPUs. I really wonder what the testing methodology was, what kind of computer they used, and what kind of files they were working with. I really do find it suspect. Maybe they only tested the 1080TI with OpenCL and not CUDA. In my tests the OpenCL performance vs CUDA performance was pretty similar between the two in certain codecs, but CUDA on the 1080TI really made a big difference in a lot of situations, especially with R3Ds.

We bought our VEGA FE's based on this test and the tomshardware test and we haven't been dissapointed yet. It does what it has to do and the overclock scales more or less 100%.
We don't have GTX1080ti's so I can't compare. (but with some extrapolation you can get a descent insight).

That's just it. You don't know since you haven't actually compared them. That review tells you nothing about what they tested, codec, resolution, etc. and how (OpenCL or CUDA). I've personally tested Vega FE vs 1080Ti, and my results show exactly what the files were, grade applied, and processing used. I don't know what you're extrapolating from. Specs? Specs mean nothing. The only way to know is to test side-by-side, or one after the other on the exact same system. I don't think you'd be disappointed. I wouldn't either if I hadn't been trying to go from 1080Ti to Vega FE to get away from Nvidia's crappy drivers. But I'd rather deal with driver issues and have the far superior performance that I'm used to. But if you're getting what you need from the Vega FE then you should stick with it. It's not a bad GPU by any means, just not as fast as some other options out there for a lot more money. Spending less money is always good to.