Everyone's finished, back to the drawing board

File: 1488596291374.jpg (177KB, 885x1132px) Image search: [Google]

177KB, 885x1132px

>>59227826
Everyone must have headaches having to test the same chip 20 times.

Looking forward to these improved power figures, Zen's perf/watt is already insane.

>>59227826
No one's interested?

>>59227826
So are the Gigabyte boards the only "good" ones at the moment?

>>59228076
Unknown. Only safe assumption is that there is more performance left on the table across the board.

>>59228076
Apparently, everyone's pushing BIOSes out but the newest ones from ASUS and MSI still have performance differences, some 5-7% for MSI compared to the newest ASUS.

There's still obviously work to be done, especailly in SMT land.

>>59228076
Bump

File: file.png (133KB, 306x306px) Image search: [Google]

133KB, 306x306px

Jesus Christ AMD, get your act together. It's because of shit like this, people avoid buying your products.

>>59228094
And yet there's still bugs with MSI boards.

 * Superpi 1M

- each loop varies a lot in time taken, you can visably see it as it's going through the loops.. resulting in inconsistent final times.
-Setting affinity to one core results in a 56s or so score.. yet the core does appear to be out of idle, (3.2+ghz) when monitoring with HWmonitor.

Fritz Chess:

Is only detecting 8 logical cores.
is only using 1 CCX .. the other is sitting idle.

This is quite interesting
https://forums.anandtech.com/threads/official-amd-ryzen-benchmarks-reviews-prices-and-discussion.2499879/page-128#post-38774366

each zen thread is being registered as an individual core with its own L2 and L3 cache. i.e. totaling 136 MB cache!!. this is using Windows Sysinternals. This explains the SMT troubles in the event that a thread bounced to a HT thinking its the real deal.


This could explain part of the issue with SMT scaling under Windows.

>>59227826
Lmao this is pathetic

"ALL THE BENCHMARKS WERE FAKE"

Jesus christ AMDtards are so fucking stupid.

>>59228123
shit, apple does it, everyone continues to suck their dick

AMD can fuck up and be fine too, not like anyone is expecting anything different

>>59228198
windows 7 fags btfo

>>59228217
Why do you shitposting little kids even bother?
https://www.computerbase.de/2017-03/amd-ryzen-1800x-1700x-1700-test/2/

Für die von AMD als Muster verteilten Mainboards Asus Crosshair VI Hero und MSI X370 Xpower Gaming Titanium empfiehlt der Hersteller mindestens den Einsatz der BIOS-Versionen 5704 respektive 117 und in der Tat gab es auf beiden Platinen damit einen deutlichen Zugewinn an Leistung – teilweise zweistellig bis zu 25 Prozent.


For the AMD Asus Crosshair VI Hero, and MSI X370 Xpower Gaming Titanium, the manufacturer recommends at least the use of the BIOS versions 5704 and 117, respectively, and in fact there was a significant gain in performance on the two boards with a double-digit figure up to 25 percent.

>>59227875
I knew the 1700 will be the processor to watch out for this is fantastic.

File: file.png (82KB, 225x224px) Image search: [Google]

82KB, 225x224px

>>59228220
>Apple does it so why can't AMD
AMD doesn't have anywhere near the market share, nor the same market competition that Apple does.
If they want their 30% of the current marketshare to drop even further, then go right ahead and release unfinished products.

>>59228276
This. The 1700 is the go to chip.

File: the real curse.jpg (231KB, 869x943px) Image search: [Google]

231KB, 869x943px

One of these mornings, I'm going to wake up and find out AMD updated, and fixed everything, making Ryzen the best CPU.
One of these mornings, the past will fade, and it will all be like a bad dream.

>>59228369
You won't, it won't be the best gaming CPU, and that's final.
It doesn't need to, it delivers what the 6900k is for half the price or less, and that's even better, because the market for a lot of efficient cores is a way bigger market for AMD than the gaming one.

>>59227826
>Want to compete
>underdog
>don't give your products the best chance possible

I don't understand. Why did they not make sure these issues were solved before launching. They just look incompetent now. Especially with all teh fuckign hype for the past 6 months.

File: intel.png (148KB, 763x930px) Image search: [Google]

148KB, 763x930px

>>59228431
>Implying intel didn't have a hand in this

ASUS MSI etc make most of their money selling intel compatible products.

>>59228431
ASUS sat on the BIOS updates too long and people tested them before they were up.

>>59227826


Left to do
>fix SMT (Microsoft?)
>fix CCX utilization
>RAM latency and timings
>Overvolting for yields, again(think 480)
>Ryzen Master temp and voltage readings
>Create simple branches in game engines to optimize Ryzens SMT

You left some things off.

>>59227826
>ASROCK and Gigabyte are the only ones who didn't fuck up the launch

Who would have thought.

>muh anus muh msi

>>59228559
The issues with SMT are apparently with the games themselves and not a problem with windows. It looks like the way AMD implemented SMT is different in some way than the way it was implemented by Intel so games have to be patched to support it which AMD is working with devs to do.

The other stuff I don't know

File: file.png (495KB, 1280x720px) Image search: [Google]

495KB, 1280x720px

>>59228559
They should have postponed the release by a week to fix those issues.

>>59228901
no then all the funposting on /g/ would've never happen

>>59228628
It's AMD's fault for only giving the manufacturers 3 weeks

>>59228901
Hype was already hitting a critical point. They should have announced a much later date earlier on.

>>59229045
To be fair ASUS did have a BIOS ready a week ago and sat on it for whatever reason.

And it just turned out that AMD review samples were mostly ASUS boards.

>>59228652
See >>59228198
AMD had code in Linux for Bulldozer that prevented the kernel from treating adjacent threads as SMT (because Bulldozer did that stupid fucking module crap), and they had to patch it for Zen so that SMT would work right. It's possible Windows had similar code and they haven't fixed it yet (unlikely considering how fucking stupid that's be of them but who the hell knows these days) or more likely MS hasn't pushed the update.

>>59229073
>>59229120
Some conspiracy theories, but I find it hard to believe Intel isn't involved in these delays at all.

>>59228901

They needed another month at least.

>>59229132

Or maybe AMD rushed the launch, and having BIOS updates at the 11th hour was a by product of AMD not having the man power to hit their street date and making sure everything was ready to go on day one instead of botching yet another critical launch of a product?

>>59227826
There are reports of Windows power profiles effecting performance too now. Supposedly setting the power profile to high performance mode makes a significant difference.

Its shit

>>59229193
I wouldn't be surprised, Zen clearly has some secret sauce involved with its power states, it's using half the power at idle.

I never expected intel to bribe motherboard manufacturers.

so AMD is requesting everyone to work around their slow chips?

>>59229212
>we've only begun to unlock Zen's power levels

>>59229215
>bribed dell $4 billion to not use AMD CPUs

Did you really?

>>59228123
Sandy Bridge literally fried it's own SATA ports from overvoltage on the original chipset. Kaby Lake runs at a million degrees because of the TIM and had issues with high speed memory. Teething issues aren't unexpected.

>>59228243
What, are we supposed to use Windows as a Service 10? Are you having a fucking giggle? When 7 becomes unusable I'm going Linux with a 10 VM for the two programs I need Windows for.

>>59229232
>Slow chips
You're a little slow yourself, so I'll explain in simple terms.

Ryzen as a chip is really strong. The design is good. But it's having serious driver issues on Windows, because Windows was written for Intel Hyperthreading instead of AMD's somewhat different implementation.

>>59227875
how is power consumption when some cores are deactivated?

>>59227826
Ryzen was right all along

>>59227826
Anon you're finished. Joker admitted since then that he had no fucking idea WTF he is doing and reran those benchmarks without a GPU bottleneck showcasing the same result of Ryzen being 15-20% behind 7700k and on par with 7600k.

File: full retard.jpg (43KB, 560x463px) Image search: [Google]

43KB, 560x463px

>>59227826
Did people seriously not expect Day 1 problems with a brand-new architecture?

Unless your CPU is dying, there's no reason in the current no-gains CPU market to rush out and buy Ryzen until things have settled and been fixed.

>>59231554
FOR THE LAST FUCKING TIME, THE R7 (WHICH HAS 8 CORES/16 THREADS) IS NOT SUPPOSED TO GO AGAINST THE 7700K 4C/8T

THE R5 SERIES IS THE ONE

FUCK OFF AND KILL YOURSELF YOU CYNICAL ASSHOLE