Problém je, že nejde bez ďalšich dodatočných znalostí o testovacom softvéri.Sú
1.Lži
2. Veľké lži
3. Obrovské lži
4 .Sľuby politikov
5. Benchmarky ...
1. v článku z roku 2006 (od Portland group, ktorý dnes vlastní nVidia)
AMD vs. Intel The Compiler as Referee
https://www.pgroup.com/lit/presentations/pgisc06... je
porovnanie výkonu Intelu Pentium 4 Woocrest a Opteron K8 v kóde optimálnom pre to na čom bežali, optimálnom pre opačný CPU a na nič neoptimalizovaný kóda napr. na priesvitkách 5 a 7 je,
že
A)na Opteron bol kód optimálny pre Intel o 11% pomalší ako kód optimálny pre AMD, teda optimalizácia na AMD pridala 12,4% výkonuB) na Woocreste je kód optimálny pre AMD o 50% pomalší ako kód optimálny pre Intel, teda optimalizácia na Intel pridala 100% výkonu.
2. Porovnanie optimálneho a neoptimálneho kódu na danom CPU dá ako výsledok hocičo - Ešte dodnes sú Zen-y disriminované
Nov 18th, 2019 10:53 Discuss
MATLAB is a popular math computing environment in use by engineering firms, universities, and other research institutes. Some of its operations can be made to leverage Intel MKL (Math Kernel Library), which is poorly optimized for, and notoriously slow on AMD Ryzen processors. Reddit user Nedflanders1976 devised a way to restore
anywhere between 20 to 300 percent performance on Ryzen and Ryzen Threadripper processors, by forcing MATLAB to use advanced instruction-sets such as AVX2. By default, MKL queries your processor's vendor ID string, and if it sees anything other than "GenuineIntel...," it falls back to SSE, posing a significant performance disadvantage to "AuthenticAMD" Ryzen processors that have a full IA SSE4, AVX, and AVX2 implementation
https://www.techpowerup.com/261241/matlab-mkl-codepat...
>Stemming from Glibc semantics that>>>>>>effectively "cripple AMD">in just checking >or Intel CPUs while AMD CPUs with Glibc are not even taking >advantage of Haswell era CPU features,
https://www.phoronix.com/scan.php...Kód pre AMD v GNU/Linux a väčšine open source pre MS Windows bežali na kóde bez podpory technológií a inštukcií Hasewellu a novších. Aj Zen2 bežal dodnes na kóde pre Sandy Bridge.Haswell
Launched June 4, 2013; 7 years agoPredecessor Sandy Bridge (Tock)
Successor Haswell (Tock/Architecture)
https://en.wikipedia.org/wiki/Haswell_(microarchitecture)Sandy Bridge
Launched January 9, 2011; 9 years ago
https://en.wikipedia.org/wiki/Sandy_Bridge Náhrada s optimalizáciou pre AMD sa len vyvíjalibc-HWCAPS To Help With AMD Zen Optimizations, Other Per-CPU Performance Bits
on 7 July 2020This glibc-hwcaps work sent out under a "request for comments" flag at the end of June was carried out by Red Hat's Florian Weimer. He began this patch series stemming from the work on the bug/issue around providing better optimized AMD Zen support by potentially following Haswell optimizations rather than a more generic target. Prior to this glibc-hwcaps patch series, it was a few months back that AMD was said to be engaged on this matter as well.
https://www.phoronix.com/scan.php
Ale mala by byť k dispozícii v glibc verzie 3.3 začiatkom roku 20212020-08-05: glibc 2.32 released.
2020-02-01: glibc 2.31 released.
2019-08-01: glibc 2.30 released.
2019-01-31: glibc 2.29 released.
https://www.gnu.org/software/libc/