Kernel compilation with no function and no loop alignment.

[_Alex_]

With just one AVX512 vector unit (or 2x256 joining to make 1x512) it's going to be as power-optimized as it gets. Given the process node that is (14nm)......

[_Alex_]

"That code is now in Linux 5.9-rc6 and the performance is back on track with Linux 5.8 while I will have out more benchmark numbers soon on the revised...

[_Alex_]

If one had the resources, it could probably be easy to test thousands of patches over the last few years and find something like a top-10 or top-100 list...

[_Alex_]

https://ark.intel.com/content/www/us...to-3-0ghz.html

I was checking to see if it has AVX512 or not for the big core, but it doesn't even...

[_Alex_]

Remember it's the OS, not the workload itself.

The OS should, theoretically, be as efficient as possible and most of the delay should come...

[_Alex_]

Let me rephrase it another way.

If you are to use 100w in doing AVX2 or underclocked AVX-512, you'll get far more results back with AVX512....

[_Alex_]

I think the real problems of avx-512 is

a) a lot of avx-512 instructions subset fragmentation which makes this thing rediculous. I mean instruction...

[_Alex_]

Speaking of light, maybe "lightweight" or "light" - referring to implementantion should also be removed so that it doesn't hurt the...

[_Alex_]

The OK symbol is similar. Its use as "OK" predates a very small minority like 0.00001% who might have made it as a trolling attempt to be something...

[_Alex_]

That's an open attack vector into anything.

Let's say I hire 100 "white supremacist" trolls that have white supremacy speeches,...

[_Alex_]

If someone doesn't understand that in the context of software, the master/slave notion has no racism extensions, I mean wtf.

Next thing you...

[_Alex_]

The thing is, Intel has been relatively stagnant with x86 for around 8-10 years or so, with around 5 of them being on skylake... with very few performance...

[_Alex_]

In the geometric mean, AMD got more performance (+7.3% vs intel 6.1%): https://openbenchmarking.org/embed.p...ha=432aff7&p=2...

[_Alex_]

Has anyone tried compiling the kernel with -fno-align-functions and -fno-align-loops ?

I tried it on a ryzen 2200G and apparently it reduces...

[_Alex_]

Obviously, but imagine this scenario: You launch the exact same cpu every 3 months and you alternate between 32kb and 64kb l1 cache.

Q1...