Single Radeon VII faster than dual Radeon VII ?!

Hello,

I'm building a new workstation (TRX40 / 3990X / 2 x Radeon VII) running Windows / Ubuntu with latest DaVinci version. I can't get working Ubuntu yet. When I run the candle benchmark:

With a single Radeon VII, I get 16 fps for the 66 nodes
With dual Radeon VII, I get 4 fps for the 66 nodes

Strange, isn't it? I can't find any explication. If I plug each card individually, each one gives 16 fps. So they're not deffective. By the way, I've enabled the HBCC on both because it seems adviced for good performances. Both cards run simultaneously at PCI x16 3.0.

Here is my DaVinci configuration:



If I set GPU selection to "Auto", one of the card disappear:



Any idea?

Thanks.

turn [use gpu display for compute] on - try again

turn [use gpu display for compute] on - try again

I've tried it, no change.

turn [use gpu display for compute] on - try again

I've tried it, no change.

1) does shows now 2 GPU?
2) do you have any sort of SLI configuration?

Yes it shows both. No SLI configuration. The others configurations I tried, withtout any success (still 4 fps):





And I'm getting back 16 fps with this configuration (so 2 cards are plugged in, but only one activated in Resolve):



And now the results with the Blackmagic RAW test :

1 card:



2 cards:



It all seems contradictory.

Can someone at BMD explain how the Benchmark RAW speed test can give a such different results than Resolve itself?

RAW speed test take in consideration the two cards.

Thanks.

Is there a way to reduce the amount of CPU cores? Maybe down to 32 cores? Then rerun the test.

Is there a way to reduce the amount of CPU cores? Maybe down to 32 cores? Then rerun the test.

Maybe but I don't know how to do. Anyway, I need the full performances of the CPU.

Is there a way to reduce the amount of CPU cores? Maybe down to 32 cores? Then rerun the test.

Maybe but I don't know how to do. Anyway, I need the full performances of the CPU.

I know. There was a known issue with AMD CPUs with MS CPU scheduler not working well together. Just wanted to see if the CPU is causing a bottleneck with the dual GPU.

Also, what kind of cooling do you have on the GPU? Direct fan or blower style or liquid cooling?
Can you run the dual GPU test for extended loop? Does the number get progressively worse?

I've enabled the HBCC on both

Hi.

First the easy suggestion: I don't know how much RAM you got. But I expect 2 Times HBCC will require many additional GB of RAM.
So my suggestion will be, to try to disable HBCC on both.

I don't know which version of Windows 10 you are using. But the normal Windows 10 Pro, can only use max 64 threads pr processor groups. Here is a link:

https://www.anandtech.com/show/15483/am ... x-review/3

Suggest you use Windows Pro for Workstations or Enterprise.

My next suggestion is to try Linux and see how it works out?
If it work much better in Linux, are the problem related to Windows.

Regards Carsten.

I've disabled SMT (AMD hyperthreading) in the BIOS to get 64 threads, no change.

I've disabled HBCC, no change.

I'm running Windows Enterprise LTSC version 1809. I've got 64 GB of RAM.

I'm trying to make Resolve working on Ubuntu 19.1 (sorry, I gave with CentOS... really too complicated OS). That's my goal since the beginning because I bought the Linux dongle (without the advanced panel) to get ProRes export with the 3990X. My Radeon VII seem properly installed but I have a weird bug which is maybe linked to the audio device. Please check the details here: viewtopic.php?f=21&t=56878&start=1650

I'm still finding a solution...

Hi
I gaveup running dual Radeon VII's on Tr3960x MSI Creator MP600 and samsung 970 EVO m.2's windows 10 Home, after 2 month of massive full system freezes.
The free version and a single card seem fine and i got another card and the studio version

The constant system freezes and black screen during editing, idle and especialy rendering made me give
up hope.

Blackmagic support just stated you have to contact the grafic card provider after sending in logs.
Hope you get it up and running and please write down Resolve setup and Driver version. I have been through all combinations and i am now on a single RTX 2080 super (Studio driver) hoping a AMD driver release solves the issues because WHEN they are running they are super fast.

Hi
I gaveup running dual Radeon VII's on Tr3960x MSI Creator MP600 and samsung 970 EVO m.2's windows 10 Home, after 2 month of massive full system freezes.
The free version and a single card seem fine and i got another card and the studio version

The constant system freezes and black screen during editing, idle and especialy rendering made me give
up hope.

Blackmagic support just stated you have to contact the grafic card provider after sending in logs.
Hope you get it up and running and please write down Resolve setup and Driver version. I have been through all combinations and i am now on a single RTX 2080 super (Studio driver) hoping a AMD driver release solves the issues because WHEN they are running they are super fast.

Use before October 2019 drivers(before NAVI) and you're fine with the Radeon VII on Windows 10, for Linux check the Phoronix website and for Hackintosh they are pretty stable.

Hi
I gaveup running dual Radeon VII's on Tr3960x MSI Creator MP600 and samsung 970 EVO m.2's windows 10 Home, after 2 month of massive full system freezes.
The free version and a single card seem fine and i got another card and the studio version

The constant system freezes and black screen during editing, idle and especialy rendering made me give
up hope.

Blackmagic support just stated you have to contact the grafic card provider after sending in logs.
Hope you get it up and running and please write down Resolve setup and Driver version. I have been through all combinations and i am now on a single RTX 2080 super (Studio driver) hoping a AMD driver release solves the issues because WHEN they are running they are super fast.

I had the same probme with the latest drivers. The only stable ones I found are 19.11.3.

I did the AMD support route, because i have similar problems. The asked me to contact blackmagic support because they do not know the architecture of resolve to fix the problem.
They still seem to not talk to each other to figure out the problem

Again, you never replied to my questions about the GPU model and cooling.
Can you post the exact brand and model you are using?

Again, you never replied to my questions about the GPU model and cooling.
Can you post the exact brand and model you are using?

Radeon VII with default air cooling.

I've finally got more success with CentOS. The exact same hardware / BIOS configuration give me 20 fps at the same Candle Benchmark (66 nodes) using both Radeon VII cards enabled with OpenCL.

So it's defintely a bug in Resolve - Windows version.

Anyone tried the Radeon PRO drivers? Or are they the same?


Sent from my iPhone using Tapatalk

slightly off topic, but I have a query:

That's my goal since the beginning because I bought the Linux dongle (without the advanced panel) to get ProRes export with the 3990X

did you bought the "advance" linux dongle? the one you get with the panel (the 30k one)? if not, you will not be able to export ProRes (it works only with the "advance" version).

if you do, how you where able to buy an Advance dongle (those only came with the panel) and not the panel? it's 30k new....

Again, you never replied to my questions about the GPU model and cooling.
Can you post the exact brand and model you are using?

Radeon VII with default air cooling.

I've finally got more success with CentOS. The exact same hardware / BIOS configuration give me 20 fps at the same Candle Benchmark (66 nodes) using both Radeon VII cards enabled with OpenCL.

So it's defintely a bug in Resolve - Windows version.

My guess is still the combination of Windows CPU scheduler and cooling. Direct fan cooling is not great for dual GPU. It recycles hot air. You want blower style.

slightly off topic, but I have a query:

That's my goal since the beginning because I bought the Linux dongle (without the advanced panel) to get ProRes export with the 3990X

did you bought the "advance" linux dongle? the one you get with the panel (the 30k one)? if not, you will not be able to export ProRes (it works only with the "advance" version).

if you do, how you where able to buy an Advance dongle (those only came with the panel) and not the panel? it's 30k new....

I could find the Linux dongle only on eBay.

Again, you never replied to my questions about the GPU model and cooling.
Can you post the exact brand and model you are using?

Radeon VII with default air cooling.

I've finally got more success with CentOS. The exact same hardware / BIOS configuration give me 20 fps at the same Candle Benchmark (66 nodes) using both Radeon VII cards enabled with OpenCL.

So it's defintely a bug in Resolve - Windows version.

My guess is still the combination of Windows CPU scheduler and cooling. Direct fan cooling is not great for dual GPU. It recycles hot air. You want blower style.

Yes I know but I have a good case with 6 Noctua fan inside. The temps are not so bad.

slightly off topic, but I have a query:

That's my goal since the beginning because I bought the Linux dongle (without the advanced panel) to get ProRes export with the 3990X

did you bought the "advance" linux dongle? the one you get with the panel (the 30k one)? if not, you will not be able to export ProRes (it works only with the "advance" version).

if you do, how you where able to buy an Advance dongle (those only came with the panel) and not the panel? it's 30k new....

I could find the Linux dongle only on eBay.

1) be aware, there are fake ones.
2) it could be a "normal" 299$ dongle that is sold as advanced.... hence, no ProRes...

New article posted today from Puget Systems regarding 3990x test results with Adobe Creative Cloud applications comparing Windows 10 Pro to Pro Workstation with SMT both on and off.
Interesting read.

https://www.pugetsystems.com/labs/artic ... ance-1671/

Yes I know but I have a good case with 6 Noctua fan inside. The temps are not so bad.

I have 5 fans in one of our custom PCs. With 2 blower style 2080ti (250W TDP), the performance slowly decreased. Run it long enough and you will see 20% + drop.
Radeon VII is 295W TDP per card.

Just test it. Easy to confirm. Run any test in loop and track the performance. Puget, BRAW test, etc.

New article posted today from Puget Systems regarding 3990x test results with Adobe Creative Cloud applications comparing Windows 10 Pro to Pro Workstation with SMT both on and off.

Hi.

AMD recommended testing with Windows 10 Pro OS build 18362.592 (or greater)
According to this link:

https://hothardware.com/news/threadripp ... t-fast-amd

But Puget Systems did not publish, the most interesting thing. The performance benefit running
the Threadripper 3990X on Linux. In the 2nd table in this link, you can see how the number of
Threadripper cores increases the advantage for Linux, compared to Windows:

https://www.phoronix.com/scan.php?page= ... inux&num=6

But I start wonder, how the Threadripper 3990X with dual Radion VII, will perform on a Hackintosh?
Where you also, will be able to get ProRes export.

Regards Carsten.

New article posted today from Puget Systems regarding 3990x test results with Adobe Creative Cloud applications comparing Windows 10 Pro to Pro Workstation with SMT both on and off.

Where you also, will be able to get ProRes export.

Regards Carsten.

Hi Carsten,
I think you might have me confused with another commenter. I wasn't trying to get ProRes export.

Sounds like, as with all things computers, everything is inconsistent and some apps work better with one hardware and software configuration, and others work better with another, and just when you figure it out, it all changes.

Regards,
Barry

Anyone tried the Radeon PRO drivers? Or are they the same?


Sent from my iPhone using Tapatalk

I have tried the driver only install from the pro driver and Windows 10 64bit pro All Seem to Works as long as hw decode/encode and debauer is off.There is no option now for multi gpu as there is only the driver, but stability Seem to be just fine. Dual Radeon VII has the exact same rendertime as a single Radeon VII card. 5:30 in a 18:38% load split. the RTX 2080 super was faster4:09 but there is something with the video quality for h.264 that i had to redo using h.265 master and the Dual Radeon VIIs.


Sent from my iPhone using Tapatalk

Anyone tried the Radeon PRO drivers? Or are they the same?


Sent from my iPhone using Tapatalk

I have tried the driver only install from the pro driver and Windows 10 64bit pro All Seem to Works as long as hw decode/encode and debauer is off.There is no option now for multi gpu as there is only the driver, but stability Seem to be just fine. Dual Radeon VII has the exact same rendertime as a single Radeon VII card. 5:30 in a 18:38% load split. the RTX 2080 super was faster4:09 but there is something with the video quality for h.264 that i had to redo using h.265 master and the Dual Radeon VIIs.


Sent from my iPhone using Tapatalk

The 19.11.3 is stable, but the super multigpu performance form the 20.x driver in resolve are lost, but now stable. If you use blender and luxmark dual setup is super. Resolve only I would go single rtx 2080 Fir encoding and you can always use the latest drivers.


Sent from my iPhone using Tapatalk

Anyone tried the Radeon PRO drivers? Or are they the same?


Sent from my iPhone using Tapatalk

I have tried the driver only install from the pro driver and Windows 10 64bit pro All Seem to Works as long as hw decode/encode and debauer is off.There is no option now for multi gpu as there is only the driver, but stability Seem to be just fine. Dual Radeon VII has the exact same rendertime as a single Radeon VII card. 5:30 in a 18:38% load split. the RTX 2080 super was faster4:09 but there is something with the video quality for h.264 that i had to redo using h.265 master and the Dual Radeon VIIs.


Sent from my iPhone using Tapatalk

The 19.11.3 is stable, but the super multigpu performance form the 20.x driver in resolve are lost, but now stable. If you use blender and luxmark dual setup is super. Resolve only I would go single rtx 2080 Fir encoding and you can always use the latest drivers.


Sent from my iPhone using Tapatalk

The latest Project i ran out of memory 8GB rtx 2080 super. Now Running a single Radeon VII 20.2.2 latest looks and feels really fast. 10 min project rendered in 11 min. Really happy. The TNR 2 frames and lens correction 3 audio tracks 4K 59,94 fps.


Sent from my iPhone using Tapatalk

Will try to switch one of our Radeon VII's to the new drivers, let's see if it's fine wine again.

Huge improvement for the Single Radeon Vii (20.2.2) 1373 vs RTX 2080 1150 in overall score.

Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K,Overall Score,1373,Score
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K Test Average,Basic Grade,172.8,Score
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K Test Average,OpenFX - Lens Flare + Tilt-Shift Blur + Sharpen,111.2,Score
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K Test Average,Temporal NR - Better 2 Frames,124.3,Score
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K Test Average,3x Temporal NR - Better 2 Frames,114.1,Score
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K Test Average,Optimized Media,164.1,Score
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K CinemaRAW Light,Codec Average,122.3,Score
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K H264 150Mbps 8bit,Codec Average,132,Score
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K ProRes 422,Codec Average,139.8,Score
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K ProRes 4444,Codec Average,122.2,Score
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K RED,Codec Average,170.1,Score
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K CinemaRAW Light,Basic Grade,58.26,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K CinemaRAW Light,OpenFX - Lens Flare + Tilt-Shift Blur + Sharpen,19.43,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K CinemaRAW Light,Temporal NR - Better 2 Frames,19.54,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K CinemaRAW Light,3x Temporal NR - Better 2 Frames,8.64,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K CinemaRAW Light,Optimized Media,130.09,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K H264 150Mbps 8bit,Basic Grade,90.87,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K H264 150Mbps 8bit,OpenFX - Lens Flare + Tilt-Shift Blur + Sharpen,21.60,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K H264 150Mbps 8bit,Temporal NR - Better 2 Frames,22.27,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K H264 150Mbps 8bit,3x Temporal NR - Better 2 Frames,9.03,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K H264 150Mbps 8bit,Optimized Media,112.03,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K ProRes 422,Basic Grade,90.88,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K ProRes 422,OpenFX - Lens Flare + Tilt-Shift Blur + Sharpen,22.97,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K ProRes 422,Temporal NR - Better 2 Frames,23.73,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K ProRes 422,3x Temporal NR - Better 2 Frames,9.38,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K ProRes 422,Optimized Media,226.77,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K ProRes 4444,Basic Grade,40.77,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K ProRes 4444,OpenFX - Lens Flare + Tilt-Shift Blur + Sharpen,15.06,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K ProRes 4444,Temporal NR - Better 2 Frames,20.10,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K ProRes 4444,3x Temporal NR - Better 2 Frames,8.86,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K ProRes 4444,Optimized Media,139.19,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K RED,Basic Grade,53.68,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K RED,OpenFX - Lens Flare + Tilt-Shift Blur + Sharpen,20.32,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K RED,Temporal NR - Better 2 Frames,18.08,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K RED,3x Temporal NR - Better 2 Frames,8.30,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.2.26),4K RED,Optimized Media,97.05,FPS

Not bad!
Here are my scores from the puget benchmark:

I got worke results running dual Radeon VII for some reason so I switcehed back to one. Aren't you getting loading time issues when opening a project? My media takes about 10 to 15 seconds to load with the radeon vii.

-----System Specs-----
Computer Name: ZE-COLOR-WORKST
CPU: AMD Ryzen 9 3950X 16-Core Processor
GPU: AMD Radeon VII (26.20.15019.1003)
RAM: 64GB
Storage: Samsung SSD 970 EVO 500GB (C:)~ST8000VN004-2M2101 (E:)~Samsung SSD 970 EVO Plus 2TB (D:)~SanDisk SSD PLUS 240GB ()
OS Version: Windows 10 Pro (Ver. 1909)
----------

Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K,Overall Score,1107,Score
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K Test Average,Basic Grade,115.2,Score
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K Test Average,OpenFX - Lens Flare + Tilt-Shift Blur + Sharpen,103.5,Score
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K Test Average,Temporal NR - Better 2 Frames,110.6,Score
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K Test Average,3x Temporal NR - Better 2 Frames,113.6,Score
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K Test Average,Optimized Media,110.5,Score
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K CinemaRAW Light,Codec Average,108.8,Score
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K H264 150Mbps 8bit,Codec Average,127.4,Score
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K ProRes 422,Codec Average,120.7,Score
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K ProRes 4444,Codec Average,122.8,Score
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K RED,Codec Average,73.8,Score
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K CinemaRAW Light,Basic Grade,38.20,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K CinemaRAW Light,OpenFX - Lens Flare + Tilt-Shift Blur + Sharpen,18.69,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K CinemaRAW Light,Temporal NR - Better 2 Frames,18.07,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K CinemaRAW Light,3x Temporal NR - Better 2 Frames,8.54,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K CinemaRAW Light,Optimized Media,122.12,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K H264 150Mbps 8bit,Basic Grade,80.80,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K H264 150Mbps 8bit,OpenFX - Lens Flare + Tilt-Shift Blur + Sharpen,20.68,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K H264 150Mbps 8bit,Temporal NR - Better 2 Frames,22.28,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K H264 150Mbps 8bit,3x Temporal NR - Better 2 Frames,9.02,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K H264 150Mbps 8bit,Optimized Media,112.06,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K ProRes 422,Basic Grade,69.17,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K ProRes 422,OpenFX - Lens Flare + Tilt-Shift Blur + Sharpen,21.94,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K ProRes 422,Temporal NR - Better 2 Frames,22.63,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K ProRes 422,3x Temporal NR - Better 2 Frames,9.26,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K ProRes 422,Optimized Media,172.82,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K ProRes 4444,Basic Grade,36.64,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K ProRes 4444,OpenFX - Lens Flare + Tilt-Shift Blur + Sharpen,21.93,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K ProRes 4444,Temporal NR - Better 2 Frames,19.83,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K ProRes 4444,3x Temporal NR - Better 2 Frames,8.91,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K ProRes 4444,Optimized Media,121.75,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K RED,Basic Grade,11.64,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K RED,OpenFX - Lens Flare + Tilt-Shift Blur + Sharpen,11.54,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K RED,Temporal NR - Better 2 Frames,11.47,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K RED,3x Temporal NR - Better 2 Frames,8.30,FPS
Puget Systems DaVinci Resolve Studio Benchmark V0.6 BETA (DR 16.1.1.5),4K RED,Optimized Media,12.12,FPS

Overall score is nice, but the most important thing for an editor is that it can handle the given codec in realtime (faster is nice but not necessary).

Starting resolve is slow. I move the media files GoPro hevc to a RAM disk while working and the optimized media on a Corsair Mp600 disk. I only use 4K 59,94hz h.265 for export .

The single card can’t handle 8k I got BSOD with puget systems benchmark.




Sent from my iPhone using Tapatalk

Real time editing was very good on the Radeon VII 20.2.2 compared to the rtx2080 which crashed.
If they ever get the dual RADEONS VII to work it will be perfect for me.


Sent from my iPhone using Tapatalk

Looking at the project the tnr was not set on all clips somehow I missed this.

If they ever get the dual RADEONS VII to work it will be perfect for me.

Hi.

Have you ever considered to do the same as Lucius Snow. He charged from Windows to CentOS. I expect it was CentOS 8.1?

QUOTE Lucius Snow in this thread on Sun Feb 16: 'I've finally got more success with CentOS. The exact same hardware / BIOS configuration give me 20 fps at the same Candle Benchmark (66 nodes) using both Radeon VII cards enabled with OpenCL.'

Regards Carsten.

If they ever get the dual RADEONS VII to work it will be perfect for me.

Hi.

Have you ever considered to do the same as Lucius Snow. He charged from Windows to CentOS. I expect it was CentOS 8.1?

QUOTE Lucius Snow in this thread on Sun Feb 16: 'I've finally got more success with CentOS. The exact same hardware / BIOS configuration give me 20 fps at the same Candle Benchmark (66 nodes) using both Radeon VII cards enabled with OpenCL.'

Regards Carsten.

Trying to get it all in one os, but evaluate others luck with CentOS for now.

Have any one benchmarked a single vs dual Radeon VII's.
With the latest driver 20.2.2 the single card is much faster then the old 19.11.3 dual? now I am confused. I might have to get the other one in for testing again I am getting real confused. Is CentOS more performance stabel?

Can’t get 2 Radeon VII 20.2.2 running anything resolve benchmark or the latest project.


Sent from my iPhone using Tapatalk

No debauer and compression and multiguu off in amd driver: then performance is on par with a single card with driver 20.2.2.


Sent from my iPhone using Tapatalk

Have any one benchmarked a single vs dual Radeon VII's.
With the latest driver 20.2.2 the single card is much faster then the old 19.11.3 dual? now I am confused. I might have to get the other one in for testing again I am getting real confused. Is CentOS more performance stabel?

I have I don't have the results with me but the single card was faster with exception of the tests with r3d media as I had GPU decoding enabled and that did indeed provide much faster results with 2 Radeon vII. Everything else though was either negligible slower or at times much slower, ProRes 444 was much slower on dual vII than single

I ran fusion at single card 700 dual card 500.

Is there a way to reduce the amount of CPU cores? Maybe down to 32 cores? Then rerun the test.

I have the similar issues on tr3960x 24 Core 48 threads.


Sent from my iPhone using Tapatalk

If they ever get the dual RADEONS VII to work it will be perfect for me.

Hi.

Have you ever considered to do the same as Lucius Snow. He charged from Windows to CentOS. I expect it was CentOS 8.1?

Yes, CentOS 8.1. Pretty stable with AMD GPU drivers 19.50 for Linux.

Hi
I gaveup running dual Radeon VII's on Tr3960x MSI Creator MP600 and samsung 970 EVO m.2's windows 10 Home, after 2 month of massive full system freezes.
The free version and a single card seem fine and i got another card and the studio version

The constant system freezes and black screen during editing, idle and especialy rendering made me give
up hope.

Blackmagic support just stated you have to contact the grafic card provider after sending in logs.
Hope you get it up and running and please write down Resolve setup and Driver version. I have been through all combinations and i am now on a single RTX 2080 super (Studio driver) hoping a AMD driver release solves the issues because WHEN they are running they are super fast.

I had that on my newly built x570, 3950x setup and it turns out it was missing x570 drivers which I had to download from amd’s website directly. Solved the system freezes and crashes


Sent from my iPhone using Tapatalk

Latest 22.2.2 amd and chipset drivers and resolve 16.2 Dual Radeon VII stil not faster tha one card before it failed again. Looked god without using display card and having mgpu on but crash aften s minut in render h.265.
Trying without mgpu in driver and not using display card. Second card not uset at all
First card 50%. CPU 8 %





Sent from my iPhone using Tapatalk

Too quick it crashed


Sent from my iPhone using Tapatalk