CPU test using DAWbench - New results

There are new results from testing the latest current generation of CPUs using the latest version of DAWbench. You can read up on it in these places:

DAWBench DSP / VI Universal - Cross Platform DAW Benchmarks : - Page 14 - Gearspace.com



(the second link doesn’t work for me).

For those who don’t know how the test results differ and don’t want to read up on it the short version is that the DSP test focuses on running plugins whereas the “VI” test focuses on the number of voices played back when using virtual instruments. There’s an argument to be made then that if you’re mostly mixing and not playing back instruments in realtime the DSP test might be a better indicator of what performance you’ll get, whereas if you play virtual instruments in realtime the “VI” test is more appropriate.

Please note: this test is performed using Reaper which if I understand correctly scales much better across CPU cores than Cubase/Nuendo does. So the test shows the relative performance between the CPUs, not an absolute number that you can achieve in Cubase/Nuendo.


Thank you I will consider that .
Much appreciated mate​:grinning::metal::+1:

This is a very interesting test, thank you.

So if one is considering having two systems, one for audio, and one for virtual instruments/samplers, it could be a good idea to use one of each CPU?

Well, it depends, right? Curiously ever since Ryzen was introduced a few years back people who have generally been very favorable of Intel have pointed to Intel’s lead in the VI benchmark and the argument was at least partially that if you wanted low-latency work then Intel was the chip to have - and you wanted that because you didn’t want latency while playing a VI. This time around there was a line in Pete’s (Scan) article if I remember correctly that basically said people increased latency when working with virtual instruments, or at least didn’t use the extremely low ones at 32 or 64 samples. That seems backwards to me.

So anyway, if you look at the chart it seems that indeed at 128 samples and up Intel indeed does better. But at 64 samples the 5950x by AMD does best. The question then is what size buffer you would use. People were complaining about Ryzen chips not being able to perform as well as Intel at those very lowest buffer settings so… they should presumably be happy about this.

Regardless it’s hard to go wrong. Add to that new Intel chips are coming out soon and I would guess they’ll be competitive. And then AMD has updated chips next year. I think both are promising roughly the same amount of improvement over previous generations, maybe a bit more by Intel which would put them on par. I guess we’ll see…