AMD Ryzen 5 3600X Linux Performance

[ryao]

Not very correct then BIOS explicitly refers to legacy case, while the general term for it all is simply "motherboard firmware".

https://en.wikipedia.org/wiki/BIOS

So while some are still calling it BIOS out of habit, it's better to use the general term.

BIOS is a term that describes anything that does the same thing as the original IBM PC BIOS. That includes Compaq’s reimplementation and is the case no matter how many additional features are added. UEFI is just another reimplementation. Calling it a BIOS is not wrong. Different groups have different terminology for the same thing. I am not sure if the term firmware was in use for this at the time the term BIOS was coined. It certainly was not in use for it at IBM when the IBM PC was made. The original meaning of firmware is what is called CPU microcode today.

[f.s.e]

Pretty disturbingly bad: AMD ~66% higher Idle !?

[shmerl]

The original meaning of firmware is what is called CPU microcode today.

It refers to specific device. I.e. microcode is CPU firmware. UEFI is the motherboard one.

[Shevchen]

Michael
Thank you for the test. Great results so far. Could you do an A/B comparison within the next week with tweaked RAM-timings? Windows-users show an impact of ~10% performance increase (in games) with tweaked RAM-settings and I'd like to know the impact with similary tweaked timings on Linux. Esp. on 1900 IF-link-speed (if the chips can do it), it would be interesting to see how Linux reacts to the improved latency.

Little helper tool to tweak it: https://www.techpowerup.com/download...am-calculator/

[bug77]

Probably X CPUs and XT GPUs are for non overclockers or for these who are willing to pay for faster default or just more for more quality silicon or ... you name it

Same story like with RX 5700 vs RX 5700 XT vs 5700 XT 50th Anniversary Edition

That's the theory. But in case of 3600 vs 3600X, they're extremely close without overclocking any of them. It has probably something to do with Zen2 being unable to actually run at advertised turbo clocks.
You do get a better cooler with the 3600X, though.

[Kepsz]

It would have been interesting to include the 8700K in this review to compare 6 core/12 thread processing directly. The 8700k is going for ~$370 nowadays, so it would be interesting to see how much the extra $120 is worth.

Well, Ryzen 5 2600 @ 4.0GHz all core are capable of delivering about 80% of an i7 8700k's performance (i compared them in Cinebench R20 and other sw). Meanwhile you can get two 2600 / 2600X CPU's for the price of a 8700k.
A R5 3600/3600X are simply faster than the 8700k, especially in multi-threaded tasks, while also cheaper. So the i7 8700k are still a bad deal, because of the high price tag. You can compare things in this previous phoronix test: https://www.phoronix.com/scan.php?pa...0x-linux&num=4

[Kimmono]

Pretty disturbingly bad: AMD ~66% higher Idle !?

Yes, what is up with that? I guess that is on Linux only?

[caligula]

BIOS explicitly refers to legacy case, while the general term for it all is simply "motherboard firmware".

So while some are still calling it BIOS out of habit, it's better to use the general term.

Well, the name originates from CP/M in the 1970s. I'd argue that UEFI is still nothing more than a basic input/output system. A modern motherboard could run Linux or whatever, but the UEFI is usually pretty limited. It's very basic.. and deals with I/O equipment. If I had a hybrid GPT/MBR format hard drive and a EFISTUB kernel with both legacy and EFI support, even the boot process would look 99% the same. After booting the kernel, systems with either firmware can read the GPT disk. The main differences would be the EFI functionality (e.g. firmware update capsules) and the different FB console driver. Now if I disable both of these features, there's hardly any difference from my POV.

[caligula]

Pretty disturbingly bad: AMD ~66% higher Idle !?

Maybe they still have some issues with C states? IIRC the first Zen CPUs required some idle=halt tricks or something to avoid crashing when idle. Maybe AMD has disabled the super power savings inside the chips.

[Raka555]

Michael It is not ideal to use 4k resolution for CPU test, the GPU can easily become the bottle neck.
I got much higher fps for the games you tested with my Vega 64 @1920x1080 and 3700x and one cant help to wonder how GPU limited your results were.