AMD Ryzen 5 2600X + Ryzen 7 2700X Linux Benchmarks

[Brisse]

I have a Ryzen, and it has never fr...

Did it freeze mid sentence while you were writing ? ^_^

[Kover]

Before replacement I saw few MCE errors, after replecement I haven't seen any , but Im not looking for them in fact ( I have 1700X)

Ah. Perhaps this CPU just is faulty then. Still gotta wait it out just to see if others have a problem with the new chips on the same chipset before RMA'ing it. Haven't experienced any segfaults of system freezes yet after all.

Does it recover after that, or freezes the CPU?

It recovers. I've not experienced any system freezes yet, and were it not for those logs, I'd never know if there was a problem or not.

[shmerl]

It recovers. I've not experienced any system freezes yet, and were it not for those logs, I'd never know if there was a problem or not.

Good to know. I've just ordered 2700X in hope it would actually fix the freezing bug I have with 1700X that required disabling package C6 states to work around it. Disabling even package C6 raises average CPU temperature.

[Anty]

Really impressive results. I'd buy one right away. And that yum-yum L2 cache latency reduction of 34% over the old Ryzen.

Do you know that if you have x370 ASUS board there is an "performance bias" option in BIOS which reduces cache and memory latencies to level close of 2xxx chips for "old" ryzens?
So it is mostly software limitation (AGESA).

[Anty]

After having run the new Ryzen 2700x several hours on a x370 boards with the latest BIOS revision, I sometimes get the following MCE hardware faults:

mce: [Hardware Error]: Machine check events logged
[Hardware Error]: Corrected error, no action required.
[Hardware Error]: CPU:9 (17:8:2) MC3_STATUS[-|CE|MiscV|-|-|-|-|SyndV|-]: 0x9820000000000150
[Hardware Error]: IPID: 0x000300b000000000, Syndrome: 0x000000002a000503
[Hardware Error]: Decode Unit Extended Error Code: 0
[Hardware Error]: Decode Unit Error: uop cache tag parity error.   <------------ this
[Hardware Error]: cache level: RESV, tx: INSN, mem-tx: IRD

This smells like segfault issue. Disabling opcache fixed problem in most cases for old ryzens.
By the chance do you remember batch number? Is it SUT or PGT one?

[Kover]

This smells like segfault issue. Disabling opcache fixed problem in most cases for old ryzens.
By the chance do you remember batch number? Is it SUT or PGT one?

Chip said UA 1803SUS. So neither. Made in China though, not Malaysia

[Dehir]

Pretty supriced that price/dollar 7 1700. So after all i made very good choise :P

[AndyChow]

Do you know that if you have x370 ASUS board there is an "performance bias" option in BIOS which reduces cache and memory latencies to level close of 2xxx chips for "old" ryzens?
So it is mostly software limitation (AGESA).

Really? I thought for the first Rizen series, the 17 cpu clocks for L1 calls was a silicon issue, only fixed in the APU series. I might be wrong.

[leipero]

Interesting that R5 2600x did not beat 8700k at any test on GNU/Linux, but on windows there are few (productivity) tests where that actually happens. For example:

http://www.guru3d.com/articles-pages/amd-ryzen-5-2600x-review,16.html

and one that can be replicated on GNU/Linux (and it would be really fun to see the result IMO):

http://www.guru3d.com/articles_pages/amd_ryzen_5_2600x_review,15.html

If there is a way to include handbrake tests at least in PTS, i think it would contribute more to the "platform comparison".

[Michael]

If there is a way to include handbrake tests at least in PTS, i think it would contribute more to the "platform comparison".

Last I checked, Handbrake couldn't be fully-automated.