AMD Ryzen 7 series and Corona?

[cecofuli]

 EPYC isn't very good CPU for "us". Look at Corona Benchmark.  =)

[maru]

Ryzen APUs for notebooks have just been announced. I am looking forward to benchmark results. :)

[Ryuu]

EPYC isn't very good CPU for "us". Look at Corona Benchmark.  =)

The result in benchmark is not impressive, but I wouldn't make any judgements from just a single sample. There may be something wrong with that guy's HW configuration, etc.

We're planning to investigate Corona on Epyc. Maybe there will be some interesting optimization opportunities.

[Tanakov]

EPYC isn't very good CPU for "us". Look at Corona Benchmark.  =)

Still better performance than i9 in in the lower price range than intel.

30 sec? That sounds good to me.

EPYC isn't very good CPU for "us". Look at Corona Benchmark.  =)

The result in benchmark is not impressive, but I wouldn't make any judgements from just a single sample. There may be something wrong with that guy's HW configuration, etc.

We're planning to investigate Corona on Epyc. Maybe there will be some interesting optimization opportunities.

Well, Im not sure what to say, Thread Ripper 1950~ is 1:30~ so I dont get it... what is wrong with Epyc?

[Ryuu]

Well, so far:

ThreadRipper 1950X scores at 1 minute at the benchmark, while dual Epyc 7551 is at 30 seconds.

That's not that great considering that the dual Epyc has 4x as much cores as the ThreadRipper.

It's even worse when you consider that the ThreadRipper costs around 1k USD while the Epycs cost 2x 3.2k USD (+ increased cost of motherboard, power supply, etc.). 6.5x price is too much premium for twice as much performance for most people.

Granted, I'm comparing Epyc to the best ThreadRipper result which was probably overclocked. But the median TR result is around 1:12 which is not that much different.

But again, that's just from a single Epyc sample and I don't like to draw any general conclusions without having enough data.

[Tanakov]

Well, so far:

ThreadRipper 1950X scores at 1 minute at the benchmark, while dual Epyc 7551 is at 30 seconds.

That's not that great considering that the dual Epyc has 4x as much cores as the ThreadRipper.

It's even worse when you consider that the ThreadRipper costs around 1k USD while the Epycs cost 2x 3.2k USD (+ increased cost of motherboard, power supply, etc.). 6.5x price is too much premium for twice as much performance for most people.

Granted, I'm comparing Epyc to the best ThreadRipper result which was probably overclocked. But the median TR result is around 1:12 which is not that much different.

But again, that's just from a single Epyc sample and I don't like to draw any general conclusions without having enough data.

Thank you for taking time to explain all that.

[3dboomerang]

hey guys;

putting together a renderstation; i have a few questions...

for those amoung you who have free time on your hands, please have a look and tell me where there is room for improvement. Goal is to have CPU power and RAM; I can't be bothered with GPU power since Corona is still CPU-only: see attachment

There is room for GPU expantion in the future perhaps, but not needed at this point.

I'm troubled with the RAM cause I don't really know if it's a necessity to have DDR4 Ram for example, instead of DDR3 which is cheaper. Can someone enlighten me as to where RAM DDR3 or DDR4 comes into play when rendering + the Mhz of it - will I feel a difference when using 2166Mhz instead of 3600Mhz?...

grts