SIMPLY EPYC!

SIMPLY EPYC!

Other urls found in this thread:

phoronix.com/scan.php?page=article&item=ryzen-segv-continues&num=1
reddit.com/r/Amd/comments/6rrbsp/epyc_confirmed_to_suffer_from_the_segfault_issue/
lists.debian.org/debian-devel/2017/06/msg00308.html
intel.com/content/dam/www/public/us/en/documents/specification-updates/xeon-5200-series-spec-update.pdf
intel.com/content/dam/www/public/us/en/documents/specification-updates/xeon-5300-spec-update.pdf
intel.com/content/dam/www/public/us/en/documents/specification-updates/xeon-5400-spec-update.pdf
intel.com/content/dam/www/public/us/en/documents/specification-updates/xeon-5500-specification-update.pdf
intel.com/content/dam/www/public/us/en/documents/specification-updates/xeon-5600-specification-update.pdf
reddit.com/r/Amd/
twitter.com/SFWRedditGifs

wtf did u do user?

>>>reddit

>text based logs
>2017.595890410958
top kek

???

They still haven't fixed it? Lmao

are those the ryzen segfaults under load everyone is talking about ?

what do you want? binary logs by systemd?

phoronix.com/scan.php?page=article&item=ryzen-segv-continues&num=1

here's context since OP is autistic

I want sound based logs voiced by Brian Blessed.

Gentooman btfo.

...

In both screenshots I see 0s and 1s representing tex. Where is the difference?
Protipp: There is none.

>complains about text logs
>offers text logs as an alternative
BUT ITS OKAY IF THE TEXT LOGS ARE IN A GRAPHICAL WINDOW HURRR DUDRRRRRRRR

Every time I see a post by a Windows user I am more and more convinced that Windows users are the niggers of the tech world. Even the mac users on Sup Forums are smarter than them, because they at least do some programming.

well memed

Have you ever used event viewer before? It's obviously more than just text logs

Linux users cant make a graphical chart or output fuck they dont even have a proper image editor ahahahahaha

is dat the ryzen bug?

Soon anons.. Soon

AMD finished&bankroupt

FAKE NEWS

You bought a big stinky turd of a cup. How do you feel about it?

>EPYC confirmed to suffer from the segfault issue. AMD, we *need* an official statement
reddit.com/r/Amd/comments/6rrbsp/epyc_confirmed_to_suffer_from_the_segfault_issue/

>server chip touted as the best for VPS
>can't compile shit
wew

So it turns out to be a RAM issue, as lowering frequency solves this.
This thread can be now closed.

[citation needed]

Get out of here, it never contains any useful information at all
>Muh service crashed
why?
>No info

That's why you always have to make your own logger on top of reporting events

Fucking retard

Not it doesn't, it just gets less frequent.

Fucking fix this shit already, AMD. Epyc can't survive with this crap going on.

>lee better drivers meme
aymd making decent drivers muahahah, NEVER!!!

Working as intended. You're just being entitled.

So there goes the performance..lel
OH JUST BUY 3.6GHZ RAM TO EQUAL INTELS PERFORMANCE

AHHAHHAHAHAHHAHAHAHAHAHAHAHAHAHAHAHHAHAHAHAHAHAHAHAHAHA

wasn't it originally an idea by Sun that Microsoft later copied into Windows NT?

it is a good idea though

Glad I waited for Coffee Lake.

EPYC fail

It would be fixed with a microcode update, retard.

It's Sunday you mongrel

>Linux users cant make a graphical chart or output
Graphical analysis of data output is easy to script with gnuplot. This program has a GUI, because GUIs are for graphical tasks, not text tasks.

go to bed then, ching chong.

who cares?
it's better than text based logs that is for sure.

fucking kill yourself botnet delidshit

[autistic screeching]

Welp I guess it's AMD's turn to have a bunch of autistic chimps screeching about CPU bugs that will eventually be fixed by microcode updates anyway. lists.debian.org/debian-devel/2017/06/msg00308.html

>will eventually be fixed by microcode
AMD didn't even acknowledge it officialy. That's the problem.

Yes they did, and AMD Support ticket is open and active at this very moment.

BUHAHAHA

the game dev culture reached extremly sensitive hardware that can brick your computer.

thats why i will never switch from PowerPC

I was thinking of building a new computer with Ryzen this year, glad I've waited. I'll maybe wait another couple of years or two and hopefully it'll get fixed within that time

of course a FOSSfag would know literally nothing about UI.

To equal intel performance in games you idiot . In other tasks ryzen was already superior without high speed ram

not him, but Intel gains just as much from higher speed DDR retard faggot.
Ryzen is great for the 5% of people who regularly use more than 8 threads. Mediocre for everyone else

EPYC FAULT

So what's the solution? Where is the microcode update?

>So what's the solution?
No one knows yet.

The bug is so elusive that they're having trouble just pinning down what kind of conditions to replicate it under. Some people never have the problem happen to them at all, ever.

Wait for it. AMDfags already have plenty of experience.

It's also yet to happen on anything wangblows.
At all.

>occurs both in Linux and BSD
AYYYMD is literally finished in the enterprise market before enetering it

intel.com/content/dam/www/public/us/en/documents/specification-updates/xeon-5200-series-spec-update.pdf
intel.com/content/dam/www/public/us/en/documents/specification-updates/xeon-5300-spec-update.pdf
intel.com/content/dam/www/public/us/en/documents/specification-updates/xeon-5400-spec-update.pdf
intel.com/content/dam/www/public/us/en/documents/specification-updates/xeon-5500-specification-update.pdf
intel.com/content/dam/www/public/us/en/documents/specification-updates/xeon-5600-specification-update.pdf

>damage control the post

It's sad that Intel employees feel that it's their duty to spam this kind of FUD right before AMD rapes them for a few years. These segfaults are caused by the software and most people can't reproduce the issue.

AYYMD HOUSEFIRES 180W

It happens in both gcc and LLVM/Clang but nice try amdrone.

Yes, in very specific old versions that were released before Ryzen/EPYC support was added. Sad stuff.

>There were also claims of a "fix" in GCC 7. Well, back on the 1800X box I then built GCC 8.0.0 using the 20170730 snapshot for the bleeding-edge GCC compiler code. Nope, when using GCC 8 for the stress-run process, our open-source benchmarking software was still able to trigger 55 segmentation faults per hour.

KYS

Nobody else has been able to reproduce this dude's issue.

Literally a fag with dual socket epyc on plebbit cofirmed it.
KYS

Maybe just fucked up QC on the ryzen processors. Time will tell if Ryzen has any major fuck ups like bulldozer. I wouldn't doubt it.

not true dumb shithead

Jesus Fucking Christ, that damage control on amd subreddit. You didn't see shit like that when the P67 bug occured, now didn't you?

1700x, 1700, C6H and Taichi, only Samsung b-die modules, agesa 1.0.0.6 and all preceding versions, smt on or off, gcc-4.8.5, 6.*.*, 7.*, segfaults occur. Identical system on fx8350: no segfaults.

It's a turd and there most probably won't be fix before b2 stepping or even zenver2. And you-know-who left the AMD before zenver1 release: he won't be doing any bugfixing.

yeah, i'm sure a real person and not some intel-sponsored pajeet shitposter totally confirmed it

This plus you hate change and don't understand Linux at all.

[autistic screeching]

AYYMD IS FINISHED & BANKRUPT

AYYMDPOORFAGS CONFIRMED ON SUICIDE WATCH

>And you-know-who left the AMD before zenver1 release: he won't be doing any bugfixing

Jim Keller wasn't the lead architect on Zen, dumb dumb, Mike Clark was.

Street shitters have done it again. I wouldn't trust those monkeys to make me a decent CPU there literally just shitting out a crappy version of a reverse engineered Intel cpu into the streets of the world

> Identical system on fx8350
Strange, because shit happens on FM2 platform just as much.

It's Barcelona TLB bug all over again and AMD didn't even bother to recall those faulty chips and their "fix" cause large hits in performance

Not disputing any of your points but you can't expect the shit wrecker to hang around to fix bugs

prove it pile of dung

Am3+*

>you hate change
Lennart loves change.
>doesn't understand Linux at all
Lennart doesn't understand Linux at all.

no I was talking about FM2.

8350 isn't socket FM2

wasn't talking about 8350, was I?

No it doesn't. Sabertooth R2.0, FX-8350, zero segfaults in four years. The very same system segfaults on both 1700 and 1700x, on Crosshair and Taichi, but not on FX-8350 (or 1090t for that matter.. or 2500k, 2600k, 3930k).

Recall soon and that will be the end of AYYMD street shitters

AMD can't afford any recall, unlike Intel (see the P67 bug, which is cost billions upon billions for Sintel). Epyc sale might be halted, but that's it.

reddit.com/r/Amd/

The damage control by these AYYMD street shitters is hilarious, trying to save their dying, bankrupt company

>tfw you're still using a B2 stepping P67 board and it still works perfectly

>buy amd server to save a few bucks
>your business goes bankrupt

Why did it suddenly pop up again?

I've known this issue for 1 month now (my original reason for not buying Ryzen and waiting for TR) and been following the AMD support thread closely but no solution yet emerged.
I remain positive that it can be fixed w/ a microcode update but as long as it's uncertain I won't buy either TR nor Ryzen

nah, it's a hardware bug. no point in whitewashing
%rip ends up ~64 bytes off for whatever reason

I am looking forward to reading the errata on this one..

disable opcache, shit works.
Really looks like Linux isn't as fast to react to new hardware as FOSS mentality would let freetads believe

Because people didn't know if it would appear on epyc cpus which are different stepping. AMD remining silent didn't help either.

kys retard
read the thread

>Hardware bug
>**Not actually present on all hardware only some**

I did, nobody here knows or understands what segfault is and why it appears, everyone just shouting "AMD CAN'T INTO HARDWARE" while not realising what and why can present a segfault.

s-stop, you're gonna make milk shoot out my nose!

>all that damage control
holy shit
why do americans take it upon themselves to defend corporations' honor (for free) like that all the time? honestly bizarre

wow okay

Looks like I'll buy a cheap Intel box instead. P4560, here I come