Cinebench R20 released

[UltraTaco]

Trents, Sir, it probably took you almost 2.5x longer to render it too

 

[storm-chaser]

My only complaint is the scene itself is just a boring picture of your everyday household. And I'd like to see an extreme version of R20, that would really push the latest multi core platforms to the limit.

 

[UltraTaco]

But it *does* push to the li ml it. I was at 95°c when clocked at 4.6!! 1.5vcore

 

[Witchdoctor]

It is demanding
was not watching temps but spun up the fans to full immediately

I will have to see what temps I am hitting

 

[EarthDog]

My only complaint is the scene itself is just a boring picture of your everyday household. And I'd like to see an extreme version of R20, that would really push the latest multi core platforms to the limit.

This does a really good job as is I think.

That said: https://www.tomshardware.com/news/cinebenchr15-extreme-edition,38571.html

 

[UltraTaco]

I wish it was rendering a full resolution taco!
For no reason whatsoever, it did shoot down my PC completely. as soon as I hit "run", the rig fainted(at 1.5 volts)

 

[trents]

I'm not sure what you guys are talking about when you say CB R20 is so demanding. My score on the rig in my sig was 2556 or something like that. Took me 3-4 minutes to render the image I'm guessing. My cooler fans never even ramped up. Is this because my 2700k doesn't have AVX3?

 

[mackerel]

From Maxon's own page on this:

Cinebench R20 uses a much larger and more complex test scene than R15, requiring about 8x the computational power needed to render it. The test also requires about 4x the memory. Therefore, R15 and R20 results cannot be compared.

Cinebench R20 and Cinema 4D R20 incorporate the latest rendering architectures, including integration of Intel’s Embree raytracing technology and advanced features on modern CPUs from AMD and Intel that allow users to render the same scene on the same hardware twice as fast as previously.

Also, there is no such thing as AVX3 (yet), there's AVX2 which is often used interchangeably with FMA3 although strictly not same thing. If it used this I'd expect Intel CPUs to be way ahead of AMD, given that's not the case maybe it isn't used at all or not in a significant manner.

 

[Kenrou]

Maybe he meant AVX-512 ?

 

[mackerel]

Maybe he meant AVX-512 ?

Doesn't look like it is using that, look in 2nd post. Consumer Intel beat it.

 

[trents]

CBR20 takes maybe twice as long to render (guessing) but "needs 8x the computational power of R15"? I see evidence of a little more power draw but not that much.

Edit: CBR15 pegs the memory usage out at 26% (according to Task Mangager) and CBR20 about 36%. "4x the memory"?

 

[Kenrou]

Well TPU and TechReport say it uses the open source Embree rendering engine so right off the bat we have SSE3+AVX, and SUPPOSEDLY it supports AVX2 and AVX-512. Maybe they will make different modules for what's missing in the future ?

 

[EarthDog]

CBR20 takes maybe twice as long to render (guessing) but "needs 8x the computational power of R15"? I see evidence of a little more power draw but not that much.

Edit: CBR15 pegs the memory usage out at 26% (according to Task Mangager) and CBR20 about 36%. "4x the memory"?

well... computational power is not power used...

Second.. that is system memory use, not application use, right?

I also think it said it supports instructions allowing it to run twice as fast.

 

[trents]

well... computational power is not power used... But it actually states it needs that 8x computational power. If it needs, it is seems logical to me that it would use it.

Second.. that is system memory use, not application use, right? Correct, but then that tells me that CB uses a pitiful small amount of memory to begin with if there is only a 10% increase in total system use when CBR 20 is run.

I also think it said it supports instructions allowing it to run twice as fast.

 

[EarthDog]

But computational power is not correlated with power use... at least it isn't close to a linear relationship. So, it may use have 8x the ability, but that doesn't translate to 8x power use.

Not sure what a pitiful amount means to you... that said, R15 peaked at ~660 MB used (opened around 150 MB, shot up to 660 when running) while R20 peaked at 2400 MB (started around 130 MB)... so, in this case, 4x is about right. Try it and see for yourself... look at the actual cinebench process in task manager instead of grouping the system memory together so you have an accurate representation of the memory increase from the application itself.

 

[UltraTaco]

In my case, because it takes so long to complete, I can easily use it as a quick stress tester to check if my particular overclock is relatively stable or not. Who needs prime95! It's so 1996
FTW!! (for the win? Right?)

 

[RJARRRPCGP]

https://www.maxon.net/en/products/cinebench-r20-overview/

https://www.anandtech.com/show/14053/maxon-releases-cinebench-r20-benchmark

Official page at top, and Anandtech writeup of it below. Haven't run it yet, it requires you to use MS Store to download it, and I've blocked all MS servers from my home network so I'll have to temporarily get around that. Time to get benching!

And at the same time, Maxon's web site says that it supports Windows 7! I smell bull plop!

 

[Johan45]

And at the same time, Maxon's web site says that it supports Windows 7! I smell bull plop!

I tested it in Win7 it works https://www.overclockers.com/forums...scores-here!?p=8108970&viewfull=1#post8108970

 

[Johan45]

Looks like Maxxon has changed their mind and are now offering a free download from their website that doesn't require the MS store. http://http.maxon.net/pub/cinebench/CinebenchR20.zip

 

[habbajabba]

Too late too slow