Blackmagic Forum

Thought I'd share this with the community for anyone who might be interested:

Blur:
09 Nodes: 24fps
18 Nodes: 24fps
30 Nodes: 24fps
66 Nodes: 17fps
Note: 24fps performance is maintained up to and including 45 Blur Nodes

TNR:
1: 24fps
2: 24fps
4: 24fps
6: 17fps
Note: 24fps performance is maintained up to and including 4 TNR Nodes

Edit: Clarification

Thanks for sharing those, it is definitely the fastest dual card solution at the moment.

We haven't tested the RX480 with dual cards, but the triple RX480 card benchmarks are:

Blur:
09 Nodes: 24fps
18 Nodes: 24fps
30 Nodes: 24fps
66 Nodes: 17fps

TNR:
1: 24fps
2: 24fps
4: 24fps
6: 19fps

The cards we used were USD249 each, they come with a mild overclock as 'standard' so it was USD750 for the triple card setup.

Very attractive price-to-performance ratio (RX480).
My workstation is fully-populated with PCIe cards.
With only have two available slots, and no desire for an external enclosure -- TITAN was the way to go.

Side note:
http://www.notebookcheck.net/Nvidia-Pas ... 566.0.html

laptop version of GTX 10xx seams to perform about the same as desktop, which I assume is a good news.

Andrew Kolakowski wrote:Side note:
http://www.notebookcheck.net/Nvidia-Pas ... 566.0.html

laptop version of GTX 10xx seams to perform about the same as desktop, which I assume is a good news.

And the new version of the Clevo P870 (rebranded as Sager in the US) has 2 GTX 1070's or 2 GTX 1080's

Blake LaFarm wrote:Thought I'd share this with the community for anyone who might be interested:

Blur:
09 Nodes: 24fps
18 Nodes: 24fps
30 Nodes: 24fps
66 Nodes: 17fps
Note: 24fps performance is maintained up to and including 45 Blur Nodes

TNR:
1: 24fps
2: 24fps
4: 24fps
6: 17fps
Note: 24fps performance is maintained up to and including 4 TNR Nodes

Edit: Clarification

On OS X i got the same scores on my MAXWELL Titan X's... Maybe optimisation for Pascal is still coming?

Phillip Lakis wrote:On OS X i got the same scores on my MAXWELL Titan X's... Maybe optimisation for Pascal is still coming?

That's surprising. Look at the first post in this thread for typical Titan X Maxwell numbers:

viewtopic.php?f=21&t=49713

Blake LaFarm wrote:

Phillip Lakis wrote:On OS X i got the same scores on my MAXWELL Titan X's... Maybe optimisation for Pascal is still coming?

That's surprising. Look at the first post in this thread for typical Titan X Maxwell numbers:

viewtopic.php?f=21&t=49713

Yeah but i think thats a single Titan X (Maxwell)...

I had 2x Titan X's in my mac pro with GUI set to Compute, Maybe we will see improvements on upcoming release for Pascal...

OSX can't even run pascal yet...

I ran a subsequent test with both TITAN cards mounted within the computer -- but with the DaVinci Resolve > Preferences > Video I/O and GPU > Use Display GPU For Compute option deselected.

Here is that test:

Blur:
09 Nodes: 24fps
18 Nodes: 24fps
30 Nodes: 20fps
66 Nodes: 10fps

TNR:
1: 24fps
2: 24fps
4: 20fps
6: 13fps

And once again, for comparison purposes only, here are the dual TITAN X Pascal numbers which I posted earlier in this thread (the Use Display GPU For Compute option selected).

Blur:
09 Nodes: 24fps
18 Nodes: 24fps
30 Nodes: 24fps
66 Nodes: 17fps

TNR:
1: 24fps
2: 24fps
4: 24fps
6: 17fps

So, you can see the effect of removing the Display GPU's computational capabilities on the benchmark.

Hi Blake,

I see you've got two of the new Titan X's and you get less performance when selecting to enable display compute.

Are you using one Titan just for GUI (not compute) and one discretely for compute?
Or another card for GUI and the two Titan's for compute?


Thanks
Dan

Dan_VET wrote:Are you using one Titan just for GUI (not compute) and one discretely for compute?
Or another card for GUI and the two Titan's for compute?

My system only has the two new Pascal Titans installed. So the only difference between the two tests I posted in this thread is toggling the "Use Display GPU for Compute."