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

[m0nt3]

00:37.99 with 100 samples

 

[MrTosky]

i7 2600k @4.6
oldie but goodie

1.33.31

 

[Caring1]

Click the + symbol top right then go down three quarters and its there. Took me awhile to lol

Correction! Once you apply the file it should be on the right hand side a third of the way down the page, you might have to reopen the program again, start over to see it.

Clearly i'm an idiot, can someone explain exactly where and what i have to look for to change the settings? "Apply the file" what does that even mean?
Edit: Apparently I should change settings prior to rendering, as the setting shows once the AMD test downloads.

 

[CAPSLOCKSTUCK]

Clearly i'm an idiot, can someone explain exactly where and what i have to look for to change the settings? "Apply the file" what does that even mean?

your not alone...i couldnt find it either.

 

[HimymCZe]

3:34:37
Core2Quad Q9550 2,8Ghz with 27,3% OC 424x8,5 = 3,6 Ghz with ADATA Extreme DDRII
https://i.imgur.com/fqFtB6v.jpg

 

[m0nt3]

there is a + sign in the upper right corner that opens a panel, you may need to be on the 3D portion to see the option.

 

[EarthDog]

Screenshot maybe would be best...... so where did you hear they changed the samples? All I see is a different version which doesn't lead to different times........

Also, what is the difference? This thread is ridiculous... can't people put together an entire coherent thought in ONE post? LOL!!!





EDIT: A quick run @ 100 samples cut the time in half for me.

 

[P4-630]

So what settings should we use then @xkm1948 ?

 

[EarthDog]

I would think the point is to compare it to what AMD did, but, if we don't know for sure what they ran the samples at, its going to be useless to compare to what AMD ran.

I think the OP just got excitable like a little puppy and created a thread without really knowing or understanding. That is a problem here with a lot of these 'benchmark this' threads. Blind leading the blind.

 

[_JP_]

Starting to seem that this isn't very standardized, but the photos show version 2.77 with 100 samples...so let's go with that?

 

[EarthDog]

There isn't a difference in render times between the two versions. If the screeny says 100 and the OP wants to match what AMD did, then make the rules to set to 100 (and start another thread... perhaps with more information for the scores too....................................)

 

[Recon-UK]

1:52:59

Xeon E5640 Westmere-EP 4200mhz 1.32v

Memory 8GB dual channel 1600mhz CL10

 

[medi01]

My dream to return to AMD now seems to be slowly fading away

This is in harsh contrast with what's going on.
AMD rolled out ReLive which has much smaller (half) negative impact on FPS than nvidias shadow play.
AMDs drivers have been better than nVidia's in 2016, and that despite quick gains (480 beats 1060 at the moment)
AMDs answer to deep learning (16 bin flops) is faster than nVidia's best.
AMD just demoed their CPU beating Intel's (greatest?) while consuming (although a tad) less power.
AMD just rolled out "fuck CUDA":
https://github.com/GPUOpen-ProfessionalCompute-Tools/HIP

Sure people can pick wrong targets and whine about this and that, but the bottom line is, AMD is showing vast improvements on main fronts.

 

[Recon-UK]

Thought FX would beat Westmere here.. multithread is quite strong?

For Reference
FX-8350 @ stock speed 4.0 GHz (4.2 GHz Turbo)

2:55.81

View attachment 81940
View attachment 81941

 

[Caring1]

1:36.08

 

[P4-630]

nevermind... It won't even run anymore for some dumb reason, tried a reinstall but didn't help either.

 

[TheHunter]

So what settings should we use then @xkm1948 ?

+1

100 samples and v 2.77?

Will do a quick test too later, curious if my new lower cpuv holds its pants on lol

Lowered 4.7ghz from 1.288v to 1.266v (cpuz reads as 1.268v)

 

[Recon-UK]

Does blend use AVX instructions? is there a reason why some people are getting wildly different results?

 

[Ferrum Master]

Does blend use AVX instructions? is there a reason why some people are getting wildly different results?

I does use pretty much anything it can use in CPU. Especially AVX.

 

[Recon-UK]

I does use pretty much anything it can use in CPU. Especially AVX.

Well a 3570k stock is 6 seconds slower than a 4790k at 4.4ghz and 8 slower than a 4770k at 4ghz

4790k at 4.4 lost to the 4770k at 4..

Work that one out.

 

[EarthDog]

Hoping to summarize here...

If the goal of the OP is to compare to what we saw with Zen, then we need to run 2.77 and 100 samples (which would render this thread useless for that idea). That said, I have now tested on TWO systems (6950x and 6700K) 2.77 and 2.78 and there is not a time difference (within a margin of error, less than .5% difference in my times - not consistently slower/faster either).

This thread is good for running 2.78 at its default settings. If you want the most apples to apples you can get, it would be how I mentioned it in the previous paragraph.

Well a 3570k stock is 6 seconds slower than a 4790k at 4.4ghz and 8 slower than a 4770k at 4ghz

4790k at 4.4 lost to the 4770k at 4..

Work that one out.

You need to compare side by side those platforms, there could be many reasons that is happening.



Either way, the OP needs to get in here and clear this mess up, LOL!

 

[Ferrum Master]

Hoping to summarize here..

When I will be home will try to run my old 3960X uber Sandy...

 

[m0nt3]

Thought FX would beat Westmere here.. multithread is quite strong?

Did you check out my score? I will gladly run them at 4.0 and I doubt there will be that much of a difference. Not sure why he is running so much slower. I am running with 4770Ks and 4790ks at similiar clock speed.

 

[Recon-UK]

Did you check out my score? I will gladly run them at 4.0 and I doubt there will be that much of a difference. Not sure why he is running so much slower.

It's why i questioned it, this whole benchmark seems like a whole load of butt fuckery and is not reliable at all.

 

[ShurikN]

Isn't it called BLENDER? Not blend? Let's get that right out of the gate... LOL! It would also be useful to note memory speeds, cache, etc without looking/trying to find the result...similar to a format like this: USER NAME/ CPU @ MHZ/Memory @ MHZ/SCORE.

Anyway, my results with a 4.2GHz 6950X

33.16s

What I actually find interesting is that you got a lower score than the guy who supposedly ran 28sec on a stock 6900
EDIT
Never mind, saw that it was on 200 sample

@ShurikN Uhm...it seems 2 minutes is what separates our CPUs...for a same-gen comparison, it is interesting

Got around 3:30 on a 100 sample run