How fast is your CPU at running the AMD Blender benchmark?

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

cbn

Lifer
Mar 27, 2009
12,968
221
106
To help clear up the misunderstanding/confusion in this thread......Blender 278.c released Feb 28 2017 (or more accurately Blender 278.b) decreased render times by 1/3 compared to the Blender 278.a used by AMD six months ago.

Therefore the reason ThreadRipper is running such a lower than expected time is solely due to software.....AMD did not get this time by instead by using the older 278.a and reducing samples to 100 (from 150).

In fact, we see the benefits of Blender 278.c being reflected in other scores as well.

For example, notice the render time of 23.86s for a stock 1800X in post #23. This compared to the render time of 36 seconds AMD got with R7 at 3.4 base six months ago. (Again, render time reduced by 1/3 due to software improvement from 278.c.....not by using 278.a and reducing samples from 150 to 100)
 
Last edited:
  • Like
Reactions: Blockheadfan

Blockheadfan

Member
Feb 23, 2017
33
55
61
23.10s
1700@3.9
3200 CL14
Waiting on a bracket for my real heatsink till I try any higher.

Edit: Both my runs have been the version from the OP (2.78c) and stock settings.
Update: Manged to get to 22.60s @3333 CL14 @3.9
 
Last edited:

bbhaag

Diamond Member
Jul 2, 2011
6,606
1,991
146
I also ran it the way cbn specified.

35.1 on a 6700K@4.5 with 3000 CL15 ram

I do remember running this bench in the last blender thread but we reduced the sample size to 100. The results I posted in this thread are with a 150 sample size.
 

DrMrLordX

Lifer
Apr 27, 2000
21,583
10,785
136
22.66 seconds
R7 1800X - 4.0 GHz
DDR4-3466 14-14-14-28 1T

edit: a second run got it down to 22.32, yay
 

Edgemeal

Senior member
Dec 8, 2007
211
57
101
If a person follows the instructions in the opening post they will be running version 2.78c and the samples will be 150.

Works for me, so seems 2.78c is ~24% faster then 2.78a on my ivy?...
i7-3770 4C/8T @4000 DDR3 1600 8-8-8-24-2T / Win7x64

Blender 2.78c: 57.49
Blender 2.78a: 75.85 (taken Jan 2017)
 

guskline

Diamond Member
Apr 17, 2006
5,338
476
126
22.93sec on the 1800x at 4 Ghz

20.84sec on 5960x @4.4Ghz

34.61sec on 6700k@4.6Ghz
 
Last edited:

otinane

Member
Oct 13, 2016
68
13
36
Ran the way OP specified in his first post

22.78 seconds 1700X@3.8 with 3200 CL14

AMD ThreadRipper 13.04 seconds(render=100)


Perfect scalability confirmed

TR will produce double the perfomance of Ryzen

The question is with the given Zen arch, is it possible to isolate CCXs and boost clocks of the cores in an uncorrelated manner?
 

rvborgh

Member
Apr 16, 2014
195
94
101
here are my previous results with 2.78a

48 core Opteron (K10 cores, only SSE2) with all cores at 3.0 GHz
24.61 seconds... samples set to 200
18.72 seconds... samples set to 150
16.23 seconds... samples set to 128
12.86 seconds... samples set to 100
16 of the K10 cores to 3.4 GHz (to match Ryzen), set threads in Blender to 16, and set thread affinity for Blender to them and reran (150 samples). Result: 46.99 seconds.

i reran with 2.78c with all cores at 3.0 GHz
19.02 seconds... samples at 150
13.00 seconds... samples at 100

so the new version runs slower for me.

Looks like 16 core/32T Threadripper approximately equals 48 core Opteron... ie each Zen core = around 2.7 K10 core in throughput with SMT... which is the same as i projected for Cinebench prior to Zen being released... i had estimated around 1600cb (my quad opty does around 3200cb in R15 so that is right about where Threadripper should perform i am guessing).
 
Last edited: