• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

THREAD CLOSED!!! Post Your AMD RyZen Blender Benchmarks at 200 Samples!

  • Thread starter Thread starter Deleted member 50521
  • Start date Start date
Status
Not open for further replies.
alucasa said:
Bonus: 22 seconds @ 100 samples with 1060 6GB (CUDA)
Click to expand...
00:24.91 using my GPU (R9 280) through OpenCL with Samples at 100.
 
alucasa said:
Yeah, yeah. I concede, are you happy, sire? :p

In my defense, it uses less power and cost me less probably.
Click to expand...
Mine was free...This is my work rig. I am actually pretty impressed by the score yours kicked off.
 
3:59:70
Core i5-6200U on an Acer Aspire Switch Alpha 12 tablet.
I had the impression that the U processors are weaker than my old Core 2 Duo T9600 (2.8GHz, 6M Cache) under certain circumstances. Maybe in those circumstances the advanced features of CPUs (such as AVX instructions) are not utilized and the raw frequencies play a more important role.

The performance of i5-6200U, as far as this benchmark is concerned, did not improve too wildly from earlier generations of i5's (such as i5-540M's in the previous posts. If you compare the leap in gigaflops in GPUs is much greater) but energy efficiency went up.

ryzen.png
 
Last edited:
6700K
 

Attachments

  • ryzen 12-14-16.jpg
    ryzen 12-14-16.jpg
    377.3 KB · Views: 501
I get a feeling that Zen isn't very good at single-thread performance. That's probably why they are pushing so hard with Blender which loves # of cores over frequency of cores.
 
bta-blend.jpg
 
My Windows Score, is now up there with other AMD FX CPU's, not sure if it is with linux or with settings in the linux version of blender, will be checking settings between both to see what differences or not there might be. This was a 200 sample size.
blenderwindows.png

Took a screenshot of the side panel in blender and the settings seem to be the same in both windows and linux. Linux also has the same version 2.78a. Not sure what the difference is about. But linux is certainly faster, at a quick glance anyway. Difference in compiler?
 
Last edited:
Most of you don't know this but you have an option of how many threads you can let Blender use.

So, for an example, let's say you have 8 threads. You can let Blender use 6 threads to render and let 2 threads idle.

And then open a new Blender instance and select GPU compute. GPU compute uses GPU for all Blender functions but GPU compute needs few threads to function smoothly. So, with a decent GPU, you can continue working on Blender while you let render on CPU.

Vice versa doesn't work too well because GPU rendering makes everything sluggish.

So, in my case, I use 24 threads when rendering, leaving 4 threads idle. And then I use GPU compute or browse the stupud Internetz.
 
Last edited:
100 Samples;
00:36.19

YZ7fyxx.png


200 Samples;
01:11.66

RyggHHS.png
 
with 150 sample.

17.37 sec with 1060 6GB. This is interesting because, at 100 samples, it was scoring the same with CPU. At 150 samples, GPU is twice faster? Weird.

Untitled.png
 
Last edited:
Interesting to me is how much better Linux does on this program than Windows I was seeing 30-40% better performance with machines at work
 
It's widely known that Blender performs a lot better on *nix.

So, that's one way to rig the demo. Have Zen on Ubuntu and xeon on Windows. Do the same render @ same samples and Zen will excel. heh.
 
New thread posted just a few mins ago with AMD saying that it should be set to 150 Samples and using the current Version of Blender. I would personally ditch this thread and start over again since we now have the proper settings that AMD used in the horizon event.
 
alucasa said:
It's widely known that Blender performs a lot better on *nix.

So, that's one way to rig the demo. Have Zen on Ubuntu and xeon on Windows. Do the same render @ same samples and Zen will excel. heh.
Click to expand...

Or they could have just made the whole thing up? Why even bother with nix?
 
cdawall said:
Interesting to me is how much better Linux does on this program than Windows I was seeing 30-40% better performance with machines at work
Click to expand...
Long known that linux can get a more out of a CPU. This is why I like using linux for crunching.
 
Sample @ 200
i7 6900K @ 4.4 GHz
39.22
Blender.JPG


Sample @ 200
E5-2683 v3 @ 2.5 GHz
40.98
Blender 2.JPG
 
Last edited:
2:29.44 2500K 4000Mhz 2133Mhz
 

Attachments

  • RyzenGraphic_27_2500K.png
    RyzenGraphic_27_2500K.png
    350.9 KB · Views: 466
quadpixels said:
3:59:70
Core i5-6200U on an Acer Aspire Switch Alpha 12 tablet.
I had the impression that the U processors are weaker than my old Core 2 Duo T9600 (2.8GHz, 6M Cache) under certain circumstances. Maybe in those circumstances the advanced features of CPUs (such as AVX instructions) are not utilized and the raw frequencies play a more important role.

The performance of i5-6200U, as far as this benchmark is concerned, did not improve too wildly from earlier generations of i5's (such as i5-540M's in the previous posts. If you compare the leap in gigaflops in GPUs is much greater) but energy efficiency went up.
Click to expand...

IPC had increased over the generations. Since Sandy Bridge it's not a huge leap, but from previous generations it's massive. The T9600 has nothing on the i5 6200u, and neither does the i5 540m.
 
Here is my laptop score, i7 6700 with windows 10 pro 1:22:26

upload_2016-12-15_1-32-31.png
 
P4-630 said:
At what CPU temps?.....
Click to expand...
Its about 63C with .160v undervolt with auto fans on, with full fans, it does about 61 but much more noise.

upload_2016-12-15_2-1-44.png
 
Status
Not open for further replies.
Back
Top