AMD™ Ryzen© Blender® Benchmark Scores™©® Thread

KazeoHin

[H]F Junkie
Joined
Sep 7, 2011
Messages
9,000
EDIT!

As many people noticed, the actual download file AMD provided via their website was giving different results than the test performed at the New Horizon event. AMD has aknowledged this and updated the download. If you have not re-downloaded the benchmark in the last 24 hours, please do so and run again. Your results will look wimpy and flacid compared to the new results.


So, AMD has the balls to release the actual benchmark/render they used to show off Ryzen.

Now, this means we all have a new E-Peen measuring stick to compare. C'mon, you know you want to beat my score....

HOW DO I INSTALL THE BENCHMARK!?

1. First: install Blender:
don't wory, the whole point of Blender is that it's free.



Install the program like any other. C'mon, this is a hardware enthusiast forum, you can install a program.​


2. Now you need to grab the benchmark file!


3. LOAD the benchmark file and run it!

  • Open up Blender like any other program and navigate to the OPEN option...
  • upload_2016-12-14_12-40-14.png

  • Select the file you got from AMD using the abysmal browse window.
  • upload_2016-12-14_12-42-0.png

  • Once it opens, don't touch anything, just hit F12 or select "Render Image" from the Render Dialogue.
  • upload_2016-12-14_12-44-31.png

Show us what you got!
 
Last edited:
I'll start with mine.

New render file (150 samples)

32.33

 

Attachments

  • upload_2016-12-14_12-49-11.png
    upload_2016-12-14_12-49-11.png
    972.2 KB · Views: 804
Last edited:
2:56.81 here.

EDIT:

2:48.20 after closing background apps and disconnecting from RDP.

upload_2016-12-13_18-4-14.png
 
Last edited:
My work computer -- core i5-2500 did 3:31 my FX8320E running minor background apps did 3:43 -- might thermal throttle since its using stock heatsink though. both 64bit.
 
Almost a minute and a half on my 4770k. 4 cores doesn't help I guess. But at the same time that worries me about single core performance on the new chip. Anyone with an 8350 or other AMD 8 core want to post their results?
 
Because I am an idiot who is easily amused, I ran it on my G4400 dual core and my GPUs. Everything is stock for all four runtypes.

GTX 780 Ti - 2:52.39
i7 2600k - 2:05.15
G4400 - 4:54.02
GTX 1050 Ti - 1:59.27
 

Attachments

  • amd new horizons.png
    amd new horizons.png
    84.8 KB · Views: 376
Almost a minute and a half on my 4770k. 4 cores doesn't help I guess. But at the same time that worries me about single core performance on the new chip. Anyone with an 8350 or other AMD 8 core want to post their results?

I posted my bone-stock FX-8350 results above, before and after closing some background apps.
 
Almost a minute and a half on my 4770k. 4 cores doesn't help I guess. But at the same time that worries me about single core performance on the new chip. Anyone with an 8350 or other AMD 8 core want to post their results?

Well we won't know anything until the testers do their work but given that AMD was showing equivalent times systems both with 8Cs/16Ts the single threaded performance would have to be about the same as well.

It looks like AMD is trying to say they are at parity for now so the key will be pricing.
 
2:02.94 on my work machine: i7-4770 | 32GB DDR3-1600MHz | Crucial MX100 512GB

LOTS of things going on in the background, as you can see, around 25-35% utilization before starting the render.
upload_2016-12-13_18-45-4.png
 
2:02.94 on my work machine: i7-4770 | 32GB DDR3-1600MHz | Crucial MX100 512GB

LOTS of things going on in the background, as you can see, around 25-35% utilization before starting the render.
View attachment 12559

Wau. I'm sure you'd score a bit higher if you told some of those other apps to shut up. who need to do work?


On a side note, I'm super stoked that my $350 Engineering Sample keeps up within 25% of a 6950x. I bought this CPU specifically for rendering, so money well spent!
 
Wau. I'm sure you'd score a bit higher if you told some of those other apps to shut up. who need to do work?


On a side note, I'm super stoked that my $350 Engineering Sample keeps up within 25% of a 6950x. I bought this CPU specifically for rendering, so money well spent!

engineering samples are hard to beat in price to preformance. I will probably pick up one (or four up when intel gets the new socket out)
 
Last edited:
2 minutes and 26 seconds on my 8350 running at 4.7. Looks like Zen is quite fast overall, should be fun reading the reviews for it.
 
1:13 for me. Rig in sig. Very excited for Ryzen. Can't wait for affordable desktop 8-cores. Did I look at the previous post right, Ryzen beats a 10-core Broadwell-E?

P7lD5lknSldGv8J24p8ipWjyzGr_7C7Qtm0nFfnNMic
 
48.65 i7 5820k @ 4.7ghz.... Im sure there was shit running it background didnt close anything.

Screenshot (18).png
 
Blender test came out at 2:49:50 on the desktop rig in my sig. Stock wraith cooler, and I saw it continuously throttle to 3.96ghz.

Looks like an upgrade, one way or another, is in my future.

Just for giggles, I ran it on the wife's Skull Trail NUC: 1.35.88 @3.16Ghz
 
Last edited:
1:13 for me. Rig in sig. Very excited for Ryzen. Can't wait for affordable desktop 8-cores. Did I look at the previous post right, Ryzen beats a 10-core Broadwell-E?

P7lD5lknSldGv8J24p8ipWjyzGr_7C7Qtm0nFfnNMic

No, Ryzen plays ball with the 8-core broadwell-E, stock versus stock.
 
1:43 on my 2600k at stock speeds after fresh restart

1:45 here while browsing and several things open (doubt they'd have too much impact, though). 2600k, currently stock.

Not bad compared to some of the others I've seen. I feel I'm not doing too bad with this 'ancient' CPU.
 
1:45 here while browsing and several things open (doubt they'd have too much impact, though). 2600k, currently stock.

Not bad compared to some of the others I've seen. I feel I'm not doing too bad with this 'ancient' CPU.
that's why I still have this ancient CPU, damn thing is holding its own, still. the 6th gen i7's look like it won't be a complete waste when I am forced to upgrade due to a dead mobo, but this honestly looks like its worth seeing how it shakes out price wise.
 
Here's my 5820k at 4.1GHz

Got a 57.56 time. Now if the AMD Ryzen can beat that, I may just have to change my system out with this.

dawEBwtfVsIm6J-jgncrAz-KiROQt8-GpdQ59PprN74S49zI5nE8CpA0XtaRSAxoH7jma5G6YfejEaVm_Wog7unPmXV2GnyERA5Q6E8xNaBEdey32XzRAvyA1DvNMpxskpNkF8Qz9D7neJ24Sez3eOO0o0HXvCeU7HEP02hj49JFJagv39SQtYhc1XTkXPpzUDwNXWkqFHBLvx9uqzDEps8aeiGzq0diuVNtVbsqv9Wh-FbM77AIKoDjBCXhOxvDTzi9rCQaMvSdfeMkq8BVLOwWsGH4YJ60qdcblbxAduXNWvGQD1DcRIEN5OTTOykMSQQZVwcT36obk8pdt413yvUaxdmBGppwYUowhWdJ4rrbS915Kc7dadUEES-FRwTxI1eEoytj9bsVStSMarVSH4Yjq1d9sM4Steq6epFI6fwb-XD2yJYil9V8Bndy0NHehGRkuO8JxGrQ_INPDJdsUIVENWy03DvDQOzzd0N3HYthsQh6sftEVoYiH9sn3W5de5cxvmeGyeznQ2fLkCCBZl_0d9DTtaH43iNYIwpxPhC8CQUPqF4WEg00fE6OlcNi5Ye5JurJ53Ti7HBFk_9QNHzmsHollCK0yPndttuxr5e-QDlCQ7nl=w1761-h1252-no
 
92 seconds and change on my i7-4770 locked to 3.9 GHz.
Because I am an idiot who is easily amused, I ran it on my G4400 dual core and my GPUs. Everything is stock for all four runtypes.

GTX 780 Ti - 2:52.39
i7 2600k - 2:05.15
G4400 - 4:54.02
GTX 1050 Ti - 1:59.27
How do you get it to run on your GPU?
 
92 seconds and change on my i7-4770 locked to 3.9 GHz.

How do you get it to run on your GPU?

file/ user pref/system/ their is a drop down for devices. then go to main screen and select device on right side. you may be disapointed my 290 pair took around a min. i was goin to try it on my mining rig but at that time i couldnt motivate myself.
 
file/ user pref/system/ their is a drop down for devices. then go to main screen and select device on right side. you may be disapointed my 290 pair took around a min. i was goin to try it on my mining rig but at that time i couldnt motivate myself.
Yeah, all things equal, it runs a bit faster on CUDA than it does on the other standard (at the moment). Blender chalks it up to CUDA having more mature libraries. However, GPU rendering works quite a bit better than it used to on Blender, that is for sure.
 
The GPU rendering was quite a bit slower, only rendering a single block at a time, for me.
 
The GPU rendering was quite a bit slower, only rendering a single block at a time, for me.
That is how it works for this renderer. (unrelated) In Blender Cycles, OpenCL is a bit slower than CUDA, at the moment.

I took measurements from my laptop, because why not?

i7 5600U - 3:56.94
GT 840m - 3:26.79

That last one may be the most accurate measurement, yet, since it is an Optimus GPU not being used to render the GUI. Though anyone with a properly configured multiGPU setup can rightfully claim the same benefit.


EDIT: This is the last I'll diverge from the thread topic with GPU stuff, but changing the tile size to 128x128 (if I had to use this laptop for Blender, I would probably run with this) improved the GT 840m's time to 2:01.20. Since the output is 800x800, I decided the change the tile size to 200x200, and netted 1:58.96 from the onboard GT840m. GPUs seem to benefit from having a larger tile size, while CPUs do not (in the Cycle's renderer - my knowledge is almost all locked behind Softimage).


GTX 780 Ti with 200x200 tiles: 18.55 seconds.
GTX 780 Ti with 400x400 tiles: 16.17 seconds.

That should put everything into perspective.
 

Attachments

  • amd nh.png
    amd nh.png
    31.9 KB · Views: 345
Last edited:
Back
Top