PugetSystem Review on Xeon W 2400 series.
View attachment 79192
View attachment 79193
Intel has released its latest Xeon W-2400 processors, featuring up to 24 cores and quad channel DDR5 memory. Although the W-2400 is not as powerful as Intel's W-3400 line, it is expected to compete well with AMD's lower core count Threadripper Pro processors for several content creation workflows.
www.pugetsystems.com
For General Computing the Threadripper Pro 5965WX with 8 channel it reings supreme.
These are some very interesting results.
The 24 core Intel xeon performs within the margin of error of the 24 core AMD Zen 3 system. All while consuming ~80% the energy of the AMD threadripper in CB.
In blender The 24 core AMD part is 7% faster, while using ~50% more Whr
In Premiere Pro idk why SPR is performing worse than their icelake counterparts. Anyone know why? Idk if this is because of bugs, or because of an architectural difference, or because icelake is consuming >35% more power, because even in what looks like raw MT tests 24 core SPR is losing to the 8+16 RPL too. No idea.
And in Unreal engine, the 24 core AMD part only takes up 75% the time as their SPR competition in compile shaders, and preforms effectively the same in build lighting, but again, all while using >20% more Whr.
A lot of these tests have Intel losses, or near ties, while the Intel part is using marginally to a lot less power. This is not as straight forward as saying the 24 core AMD zen 3 part is better, because if we match up the tests with the power draw results as well, it starts painting a very different picture.
I'm also very interested in what the 2495X vs the w7 3455 is going to look like. Both 24 core parts, one is monolithic while the other one is chiplets based, and also has way more L3 cache. I think the power consumption results are going to be very interesting as well. It's a shame the early preview did not have that, but the website did mention that they will have a review of those CPUs up soon so that's going to be fun.
Just based on the early preview though:
In CB, V-ray, Blender, Unreal Build lighting, Davinci resolve intraframe, the scores are effectively the same
Unreal Compile shaders the monolithic version scored ~10% faster,
In Premiere Pro Raw Result the chiplet version scores nearly ~20% faster
These are from the tests where it wasn't ST based where the 13900k wasn't blowing away everything, and also where the versions of the test are matched together. I might have missed some though. This was a quick glance. Results may vary from the final review, this info is from the early preview after all.