The fuck, Ryzen is actually faster on Windows 7

>>59239925
ayyyyyy lmao how much did Wintel pay for the Ryzen downgrade on win10!

>>59239938
I'd say it's more likely them trying to be clever in the scheduler/power control and the idea backfiring because Ryzen is different to everything before it.

But still, fuck that.

I think the guy said in that thread that W7 is very slightly, but measurably faster than W10 overall, even in non-gaming applications, but I can't find it now.

It seems the difference is only big in these game-related tests, at least.

File: Capture.png (8KB, 291x197px) Image search: [Google]

8KB, 291x197px

The SMT actually works BETTER on Windows 7 according to this. This is how you'd expect it to work.

Maybe it's still not completely correct though because I wouldn't expect Windows 7 to have any specific optimizations for Ryzen, maybe it is misidentifying it as a Bulldozer chip or something.

>>59239925
M$ is gimping Ryzen confirmed!

>Currysoft products past Wintendo 7

On the plus side, if there is so much of non-optimal behavior in BIOSes as well as OS and individual games, the performance of the Zen uarch has some nice margin for growth now, as software matures.

Also from reading that thread: he says W7 works perfectly. It just lacks the driver for USB 3.0 controller out of the box and thus fails to install from USB, unless you modify installer.

Installing from SATA optical drive is working perfectly. After install you just install the regular drivers for everything.

How does this keep happening to AMD? It's always a software problem

>>59240137
Yes this is what plebs dont understand, they expect perfect software laucnh, which means that they should have "launched" it for developers over a year ago.

>>59239925
Because NT is known to be a shit-tier kernel. Even .net core runs faster on linux than on windows.

>>59239925
DELETE THIS

>>59240109
> Clinging to Windows 7 and blocking Get Windows 10 updates
> tfw someday I may be forced to use 10 and can never go back

>>59240585
GNU/Linux and virtualization is the best option I think. I'm planning on doing this some point after Windows 7 gets dropped. Ryzen does actually support IOMMU so you can virtualize graphics cards.

>>59240018
I actually take a very slight (~10-15 CB out of 500 CB) performance hit going from Windows 10 to Windows 7 on my i3.

>inb4 you call bullshit on an i3 hitting 500CB multithreaded
It's an overclocked 6100

>>59240585
When Win7 support is finally dropped I'll change to Linux plus a Windows 10 VM (hopefully by then passthrough will be good enough for muh gaems)

>>59240730
I'll probably just dual boot. I gotta get my hands on a ltsd2015 release.

File: ryzen SMT.png (22KB, 904x294px) Image search: [Google]

22KB, 904x294px

So the original assumption that Windows was somehow at least partially responsible for these unusual scores was actually closer than AMD themselves anticipated

>>59240441
AMD doesn't control the 3rd party software that uses their hardware, they only control their hardware.

They can make something that gives 150% but if the software is shit and only uses half of that power, there's not much they can do.

>>59240624
>Ryzen does actually support IOMMU so you can virtualize graphics cards.
I thought you were talking about passthrough for a moment there. Isn't that shit reserved for the enterprise? Also, which kernel supports GPU virtualization?

>>59240811
IOMMU is required for GPU pass-through. It also can be used to virtualize other hardware but for /g/ the most common use case is using a graphics card in a VM.

I think you can basically use any Linux distro that has KVM(I think all of them), and then you just have to find a guide for setting it up.

Sweet vindication for people like me who cling to Win7.
Seriously though I found windows 8 and 10 unusable. The schizo interface really triggered my shit and I'm not willing to look past/get used to bullshit at such a fundamental level.

should I get asrock with 9+2 phases or gigabyte 7+1? I heard that asrock uses cheap ass vrms.

This thread is going to turn to shit real quick. But thanks for posting this op, theres some interesting reading in that thread.

>>59240963
Gigabyte tends to have better hardware yeah, but check what exact type of chip those VRMs are. The decision depends on the exact mobo models, also I wouldn't be really asking here but on Overclock.net or such place with more versed people.

>>59241003
Yeah, that analysis is great/interesting, Stilt is quite the information well.

>>59240911
I actually kinda like W10 (OP), so I hope the MS folk fixes this. Bleh.

New Ryzen build for me fukken confirmed.

File: 1488184395192.png (148KB, 763x930px) Image search: [Google]

148KB, 763x930px

>>59240767
Same tricks, different year.
Should star petitioning for embargo's on Intel to the EU market until they pay their fine.

i don't know why i dont just leave this piece of shit board and just lurk anandtech

>>59240093
>not realizing that microshit was crippling performance of opengl applications in favor of their directshit crapwork since vista
come on lad, they do it on everything

>>59242630
Probably because you like to shitpost too much like everyone else.

>>59242706
fuck off jani i only post the highest quality memes and catchphrases here

File: 1488678336902.png (408KB, 882x1418px) Image search: [Google]

408KB, 882x1418px

>>59239925
NOOOOOOO DELET

>>59240963
Careful with the vrm counts on all these. (assuming you are talking about the matx boards) The asrock has 6 phases for vcore, 3 for soc_voltage and 2 for ram, the gigabyte looks like to be 4+3+1.

File: 1487077561341.png (305KB, 799x634px) Image search: [Google]

305KB, 799x634px

>>59239925
interesting

File: 1b70c8cd-73d1-4cab-bc44-903e46d0351a[1].png (105KB, 255x345px) Image search: [Google]

105KB, 255x345px

>>59239925
>Because he's the hero /g/ deserves, but not the one it needs right now. So, we'll troll him, because he can take it. Because he's not our hero. He's a silent guardian. A watchful protector. A Mastermind Demigod.

>>59239925
Makes me feel less bad about my Ryzen pre-order. Hopefully, this gets fixed.

>>59239925
Windows 10 - 1080 Ultra DX11:

8C/16T - 49.39fps (Min), 72.36fps (Avg)
8C/8T - 57.16fps (Min), 72.46fps (Avg)

Windows 7 - 1080 Ultra DX11:

8C/16T - 62.33fps (Min), 78.18fps (Avg)
8C/8T - 62.00fps (Min), 73.22fps (Avg)

>Win10
>more threads, lower min framerate, same average framerate
>Win10
>more threads, same min-framerate, higher average framerate

Ar these niggas for real?
Is there just one thing in Win10 which they haven't fucked up?

>>59245611
they will fix the scheduling soon and there will be game mod also
by the time Ryzen 5 is released the platform will work as it should and /g/ will be flooded in intelbois tears

Meanwhile, in Linux with up-to-date kernel:
https://www.phoronix.com/scan.php?page=article&item=amd-ryzen-gaming&num=2
https://www.phoronix.com/scan.php?page=article&item=amd-ryzen-cores&num=2
It seems that gaming performance in Linux does not benefit well with loads of threads all open. Sometimes disabling some of the threads can help a bit or just makes no difference, in some case (more correctly "unoptimised(?)" Vulkan games) it can help a lot.
In synthetic/productivity benchmarks in Linux Ryzens perform normally though.

>>59239925
I like how the Stilt's numbers seem to indicate the cache problem (if it can be resolved) would net Ryzen 9.8% FPS improvement across the board

They also indicate outside of gaming SMT it actually about 40% better then hyperthreading.

>>59245802
doesn't it just show that dota2 is shit for this test?

>4+0 slower than 2+0 slower than 3+0

wat

metro shows a more reasonable performance-curve.

File: dontbeasucker.png (235KB, 637x367px) Image search: [Google]

235KB, 637x367px

>>59245640
>All Mad Dinguses are THIS delusional that they THINK their stupid shit niggerzen 5/3 will save them

LMAO the only thing ryzen 5/3 will save is me money on the i5 and i3 intel chips haha

>>59245611
makes me so fucking glad im sticking to 7 now.

>>59240070
bulldozer was newest amd on win10 released, so it is highly likely.

gpu drivers are also likely using code optimized for bulldozer, which is less than ideal.

File: Intel_Inside_logo_(2013).png (109KB, 900x980px) Image search: [Google]

109KB, 900x980px

>>59246207
This post has been sponsored by Intel™ The Way It's Meant To Be Played™

File: btfo.gif (2MB, 500x281px) Image search: [Google]

2MB, 500x281px

>>59246207
Reminder: Being underage is against the rules here.

>>59242687

They did this on XP as well.

File: Top Tier.jpg (278KB, 1059x827px) Image search: [Google]

278KB, 1059x827px

>>59248867
>mfw I didn't know there was a second season so I get to experience new RM all over again

>>59242687
That was just with the default drivers though, wasn't it?

>>59245586
I'd say he's the hero we need, but we don't deserve.