Jump to content

Testing our GPU in OSX/Linux/Windows in pro app (Blackmagic DaVinci Resolve Studio /AE/Premiere Pro)


Recommended Posts

  • Supervisor

Blackmagic DaVinci Resolve 17.2 build 11

 

AMD Radeon Rx 6900 XT 16 GB

 

OSX 11.4 Beta (20F5065a)

 

Spoiler

 

Metal:

66 noise nodes

31

 

6TRN nodes

40

 

OpenCL:

66 noise nodes

not tested 

9

 

6TRN nodes

not tested

11

 

 

  • Ok 1
Link to post
Share on other sites

So the gap is finally filled, now we have better upgrade options than two Radeon VII 😉 

 

Can't wait to get the 6800xt here 😉

  • +1 1
Link to post
Share on other sites
Posted (edited)

Candle Benchmark in DR v17.2 with one AMD 6900XT, using Metal setting. Same results with or without WEG.

 

I also repeated, with same results, ±4G. When 4G was disabled, I tested with ResizableBar enabled; again same results. (BTW, it is not possible, at least with my mobo, to boot with both 4G and ResizableBar enabled.)

 

18 @ 66 nodes; 23 @ 6 nodes.

 

 

UDPATE:

 

The above results are incorrect. It now shows the following (4G disabled; Resizablebar enabled):

 

30 @ 66 nodes; 40 @ 6 nodes.

 

The difference is a patch I'd left engaged for the first set of results. I'd left it enabled (and just remembered this morning), as I didn't see an issue using it with the Radeon VIIs. If you enable this patch, your results are reduced by ~ 50% !!

 

The patch does not have any influence on the Octane X results discussed in the next post.

 

1869567666_ScreenShot2021-05-15at12_50_29PM.thumb.png.033929c352952abd5035762f37728e02.png

Edited by iGPU
  • Like 1
Link to post
Share on other sites
Posted (edited)

I've done some other testing using Octane (downloadable from App Store, here). You may have to sign up (register with OTOY) for a free download.

 

On the Octane (OTOY) forum, there is some discussion of testing with 6900XTs here. You may need to register on their forum to gain access.

 

When I run Octane X, I get 7 sec for the 6900XT (although my GPU is constantly listed as an unknown, "AMD Radeon HD GF10 Family Unknown Prototype"). The fastest run was 6 sec; slowest was 8 sec; avg = 7 sec (4G is disabled; ResizableBar is enabled).

 

2143704236_ScreenShot2021-05-15at10_05_33AM.thumb.png.ee8f6038ef392c19c2b693555d3555c1.png

 

Details from above image:

Spoiler

OctaveExcerpt.jpg.2ebc2434707e1c3cbe4b20b5646e4975.jpg

 

OctaveExcerpt2.jpg.96ce9ebef4163dd6f8431ffc6471a2e7.jpg

 

 

Their tabulated results would imply that this is actually pretty good. The Radeon VII gives about 19 sec (although, if the 6900XT on my setup is getting 7 sec or 4 sec faster than on their plot below, perhaps the Radeon VII would also be about 4 sec faster, or 15 sec on my setup) if the differences are linear.

 

The RTX-3080 got 8 sec, 3 sec faster than the 6900XT, so maybe 4-5 sec on my setup, assuming linearity. The 3090 is reported on some posts to get 7 sec under Linux with many scenes comparably rendered between the 3090 and the 6900; but on some renders, the 3090 is much faster than the 6900, so the results are scene dependent.

  

image.thumb.png.827eae69c64109c9a8607a74999da98b.png

 

  

Edited by iGPU
Link to post
Share on other sites
  • Supervisor

@iGPUalgrey mttr fix patch or recent shaneee and others newer one  are not usufull for trx40 users

is there are reason for you to enable it sometimes?

Your results confirm my previous one

in metal 6900xt is better than in OpenCL in windows

 

Link to post
Share on other sites
  • Supervisor

Blackmagic DaVinci Resolve 17.2 build 11

 

AMD Radeon Rx 6900 XT 16 GB

 

OSX 11.4 RC (20F71)

 

Spoiler

Metal:

66 noise nodes

32

 

6TRN nodes

42

 

  • +1 2
Link to post
Share on other sites
1 hour ago, fabiosun said:

Blackmagic DaVinci Resolve 17.2 build 11

 

AMD Radeon Rx 6900 XT 16 GB

 

OSX 11.4 RC (20F71)

 

  Hide contents

Metal:

66 noise nodes

32

 

6TRN nodes

42

 

 

Nice improvement!!!

  • Thanks 1
Link to post
Share on other sites
51 minutes ago, fabiosun said:

1574799948_Screenshot2021-05-18at3_06_10PM.png.7b3dd379e6ac998385fb7e832ecbb918.png1357354895_Screenshot2021-05-18at3_11_00PM.png.4133dfd484d92529f7ce1af2ce004587.png

 

OpenCl is not supported well from Apple but these are my 6900xt results

 

 

Strange result here since my old Radeon VII scored almost 51000 points in luxmark 😞 But in Resolve it really rocks, that's what counts!!!!

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.