Further proof that Windows 10 is causing a perf hit with the Ryzen CPU

[tlink]

1 hour ago, djdwosk97 said:

Watch the New Horizons stream. Lisa Su was going on and on about how amazing ryzen is for gaming (even when she was specifically talking about the R7 lineup).

 

huh well in that case i was wrong.

[LAwLz]

5 hours ago, Valentyn said:

Windows 10 scheduler apparently tries to treat Ryzen SMT threads like actual physical threads, causing performance issues.

It sees Ryzen as 16c/16t in some tasks, as opposed to 8c/16t.

Apparently it'll be fixed via a Windows patch/driver.

This was an issue with Bulldozer as well.

 

Want to know how much of a performance increase the patch which corrected this issue added? 1-2 percent...

 

The reason why I am bringing this up is because it is EXACTLY the same argument people made when Bulldozer was released. "Just wait for the patch and it will be amazing!".

 

Direct quote from AMD regarding the Bulldozer patch (using Archive.org because it seems like AMD has changed their URL structure, so I can't find it anymore):

Quote

Our testing shows that not every application realizes a performance boost. In fact, heavily threaded apps (those designed to use all 8 cores), get little or no uplift from this hotfix – they are already maxing out the processor.  In other cases, the uplift averages out to a 1-2 percent uplift. But heck, it is free performance, and this is the scheduler model that will be used in Windows 8 (along with some further enhancements), so why not add it to your list of downloads?

 

 

 

DO NOT BUY RYZEN EXPECTING A BIG PERFORMANCE INCREASE IN THE FUTURE!

IT MOST LIKELY WILL NOT GAIN ANY SIGNIFICANT PERFORMANCE INCREASE!

[cj09beira]

2 minutes ago, djdwosk97 said:

Well, you could argue that AMD rushed the release and as a result didn't give board manufacturers (or Microsoft) the information they required in a timely manner.

well we don't know how much time was given to them to prepare, i heard the number 3 weeks but i don't know how true that is.

and with the main problem coming from Microshit i don't know if they would have it fixed by then

[79wjd]

@LAwLz in fairness, Zen is a much better architecture than bulldozer. Also was the gaming performance of bulldozer behind what the rest of bulldozer results were pointing to? Because with ryzen, the gaming performance doesn't quite lineup with the other benchmarks (both single and multi-threaded).

 

I expect Zen to have a problem in the long run because it's already behind and Intel is probably just about ready to release Skylale-e (if they decide they need to), and I seriously can't imagine AMD has any improvements to Zen that will be ready to launch in the next two years. With that said, ryzen is different than bulldozer as it is actually pretty competitive (with broadwell-e).

[79wjd]

2 minutes ago, cj09beira said:

well we don't know how much time was given to them to prepare, i heard the number 3 weeks but i don't know how true that is.

and with the main problem coming from Microshit i don't know if they would have it fixed by then

However long it was, it probably wasn't enough considering how many issues have been reported (unless every board manufacturer and Microsoft completely dropped the ball).

[cj09beira]

4 minutes ago, LAwLz said:

This was an issue with Bulldozer as well.

 

Want to know how much of a performance increase the patch which corrected this issue added? 1-2 percent...

 

The reason why I am bringing this up is because it is EXACTLY the same argument people made when Bulldozer was released. "Just wait for the patch and it will be amazing!".

 

Direct quote from AMD regarding the Bulldozer patch (using Archive.org because it seems like AMD has changed their URL structure, so I can't find it anymore):

 

 

 

DO NOT BUY RYZEN EXPECTING A BIG PERFORMANCE INCREASE IN THE FUTURE!

IT MOST LIKELY WILL NOT GAIN ANY SIGNIFICANT PERFORMANCE INCREASE!

at least we will have the same performance as win 7 is having right now, so yes expect performance gains, bulldozer had shared resources which means its not a real 8 core, well ryzen is a true 8 core. 

[cj09beira]

1 minute ago, djdwosk97 said:

However long it was, it probably wasn't enough considering how many issues have been reported (unless every board manufacturer and Microsoft completely dropped the ball).

for some reason they gave less time than they should have given but ms being ms they would probably only have the fix sometime after launch

[JoeCoke]

Wow, so this confirms that Windows 7 will work fine on Ryzen? YESS!!!

[cj09beira]

3 minutes ago, 2Buck said:

Wow, so this confirms that Windows 7 will work fine on Ryzen? YESS!!!

maybe

 

[N3v3r3nding_N3wb]

15 hours ago, Praesi said:

Ya. Give it time till Intel whipe the Floor with it again

Coffee lake isn't supposed to be a huge improvement over Kaby Lake, from what I've read/seen.  It's even staying on 14nm.

[N3v3r3nding_N3wb]

7 minutes ago, 2Buck said:

Wow, so this confirms that Windows 7 will work fine on Ryzen? YESS!!!

AMD said that they tested it with Ryzen on launch and will try to test the chips with W7, but that they can't promise Ryzen will stay compatible with updates (http://www.pcgamer.com/amd-confirms-there-will-be-no-ryzen-drivers-for-windows-7/)

[DildorTheDecent]

kek

[DarkBlade2117]

15 hours ago, nerdslayer1 said:

new chips needs time to settle in, give it 6 months . 

All in all, this isl likely WHY the R5 lineup and R3 lineup are delayed. Most of the market will be looking at the R3-R5 lineup and AMD can afford to take this minor hit witht he R7 lineup, to give develops a bit of time to optimize, Mobo manufacturers to set in BIOs updates and by time it all happens R5/R3 will hit the i3/i5 lineup hard.

[TechGod]

30 minutes ago, LAwLz said:

This was an issue with Bulldozer as well.

 

Want to know how much of a performance increase the patch which corrected this issue added? 1-2 percent...

 

The reason why I am bringing this up is because it is EXACTLY the same argument people made when Bulldozer was released. "Just wait for the patch and it will be amazing!".

 

Direct quote from AMD regarding the Bulldozer patch (using Archive.org because it seems like AMD has changed their URL structure, so I can't find it anymore):

 

 

 

DO NOT BUY RYZEN EXPECTING A BIG PERFORMANCE INCREASE IN THE FUTURE!

IT MOST LIKELY WILL NOT GAIN ANY SIGNIFICANT PERFORMANCE INCREASE!

THERE'S A BIG DIFFERENCE BETWEEN RYZEN AND BULLDOZER. And again. Explain those results then? Why does W7 perform so much better? Because of the scheduler. 

[Doobeedoo]

So Windows 10 scheduler problem along with potential bios and memory quirks. Well patch for Win will come so we'll see difference. Cause some test comparisons are odd.

[DarkBlade2117]

Just now, TechGod said:

THERE'S A BIG DIFFERENCE BETWEEN RYZEN AND BULLDOZER. And again. Explain those results then? Why does W7 perform so much better? Because of the scheduler. 

I'll agree to disagree with Lawlz.

You shouldn't buy Ryzen hoping it will gain a huge boost but people should know that AM4 is completely new, Ryzen is completely new to devs and people are going to have to give it time to iron out. I don't expect a 20FPS jump in 1 update but there are OBVIOUS flaws that can be fixed. Ram, SMT, this issue, the early BIOs problems and so on. I just pray Intel makes the i3 lineup a 4c/4t, i5 4c/8t and i7 6c/12t because if that happens, Devs might actually have motivation to further utilize more cores.

[TechGod]

2 hours ago, DarkBlade2117 said:

I'll agree to disagree with Lawlz.

You shouldn't buy Ryzen hoping it will gain a huge boost but people should know that AM4 is completely new, Ryzen is completely new to devs and people are going to have to give it time to iron out. I don't expect a 20FPS jump in 1 update but there are OBVIOUS flaws that can be fixed. Ram, SMT, this issue, the early BIOs problems and so on. I just pray Intel makes the i3 lineup a 4c/4t, i5 4c/8t and i7 6c/12t because if that happens, Devs might actually have motivation to further utilize more cores.

Oh I agree on that aspect too. Buying something based off future performance is NOT a good idea but as you've said, you can't deny that it's new either. 

[Brooksie359]

4 hours ago, LAwLz said:

This was an issue with Bulldozer as well.

 

Want to know how much of a performance increase the patch which corrected this issue added? 1-2 percent...

 

The reason why I am bringing this up is because it is EXACTLY the same argument people made when Bulldozer was released. "Just wait for the patch and it will be amazing!".

 

Direct quote from AMD regarding the Bulldozer patch (using Archive.org because it seems like AMD has changed their URL structure, so I can't find it anymore):

 

 

 

DO NOT BUY RYZEN EXPECTING A BIG PERFORMANCE INCREASE IN THE FUTURE!

IT MOST LIKELY WILL NOT GAIN ANY SIGNIFICANT PERFORMANCE INCREASE!

Yeah I wouldn't buy a cpu on the possibility that it might get improved performance from optimization and updates. That being said if you aren't in a hurry to upgrade your cpu it might not hurt to wait and see what happens after all the dust settles and the rest of the ryzen lineup comes out. 

[suits]

Those minimum increases. Wow. 

[fragment137]

I can't help but read all these comparisons with Ryzen and wonder where my i7-2700K fits

 

I know that I didn't see much performance difference going from 7 to 10.. but it's a very interesting story to see Ryzen on W7 running faster than W10

[MageTank]

7 hours ago, TechGod said:

THERE'S A BIG DIFFERENCE BETWEEN RYZEN AND BULLDOZER. And again. Explain those results then? Why does W7 perform so much better? Because of the scheduler. 

I didn't dig too deep into things, but did the W10 test happen on a clean install? AMD mentioned issues with using non-clean Installs (updating from a Windows Install from an entirely different platform) on Ryzen. If they didn't start the W10 test with a clean install, it might skew the results in favor of the clean W7 run. 

 

Though, for that drastic of a minimum frame difference, some serious shenanigans would have to be at play. Hopefully this is something that can be addressed by Microsoft, or even a microcode update. As much as it might annoy people for me to harp on it, I still hope there is some sort of fix for the IMC. Still can't see myself touching Zen until that IMC gets a second wind.

[Misanthrope]

22 minutes ago, MageTank said:

I didn't dig too deep into things, but did the W10 test happen on a clean install?

Let me guess, do they also suggest to turn the system off and back on?

[MageTank]

6 minutes ago, Misanthrope said:

Let me guess, do they also suggest to turn the system off and back on?

 

No, but they do suggest High Performance Mode in the power options, lol. Here's where I saw the info:

 

(Got it at the exact time stamp for you)

[TechGod]

31 minutes ago, MageTank said:

I didn't dig too deep into things, but did the W10 test happen on a clean install? AMD mentioned issues with using non-clean Installs (updating from a Windows Install from an entirely different platform) on Ryzen. If they didn't start the W10 test with a clean install, it might skew the results in favor of the clean W7 run. 

 

Though, for that drastic of a minimum frame difference, some serious shenanigans would have to be at play. Hopefully this is something that can be addressed by Microsoft, or even a microcode update. As much as it might annoy people for me to harp on it, I still hope there is some sort of fix for the IMC. Still can't see myself touching Zen until that IMC gets a second wind.

It was tested with a fresh install. He did extensive and thorough tests. 

[alphaproject]

So it performs better on the "unsupported" OS than Win 10 - which was supposed to tbe the best system for gaming. DOH!