View Single Post
Old 12th June 2017
Originally Posted by msmucr View Post
Have you tried direct comparison of different memory clock speeds at Ryzen? I'd be curious about some close to real-world figures. Say with DAWBench, because its opposite to what Peter Kaine and one other user mentioned in the big Ryzen thread - eg. it's the same as with Intel and memory speed doesn't really affect performance with DAW workloads.

I would be happy to run my tests at various clock speeds to see if that has an impact because I did have a lot of people calling "foul" on my video due to not matching the 3200 Mhz overclock I was able to achieve on the Intel platform.

Let's not forget though that I did clock the memory as high as I could possibly get it on AMD, but that platform is currently experiencing well known limitations in memory speeds. I did see that a new BIOS update just came out for the x370 board I'm using though so I might be able to clock higher now on the Ryzen machine and can run some tests there.

Personally I don't think it would have an impact on max plugin load handling. There have been plenty of reports that on Ryzen going from 2400 Mhz up to an extreme overclock of 4000 Mhz (how they got it that high is beyond me) showed a 9% performance difference in certain benchmarking. As far as I know this has not been tested in audio workloads but if we assume it's true, then it wouldn't even get close to explaining the 150% performance advantage I saw on the 6900K over the 1800x.

This all being said, something I could have stressed more in my video was the price/performance of these chips. Even with a 150% performance lead by Intel, we are talking about a $1050 chip vs a $450 chip, so the AMD still pulls ahead in price/performance. So why did I even bench test these against each other? Because AMD made the explicit claim that the 6900K is their direct competition with the 1800x. And as we saw towards the end of my video, in Cinebench and Blender rendering tests, they really are nearly identical in performance with Intel taking a very small lead. The area we see larger discrepancy is in audio workloads, which could be related to audio plugin developers optimizing for Intel and not AMD since Intel has been the market lead for nearly a decade now. My video is not attempting to state outright that AMD is a poor performer, but that whatever the causative factors are, whether they be hardware related (poor chip performance) or software related (poor optimizations), we do see a large performance discrepancy between these two platforms right now for audio related workloads.