Discussion AMD Ryzen MegaThread! FAQ and Resources

[jaymc]

Interesting Discussion on Microsoft Forum about the Scheduler:

https://answers.microsoft.com/en-us/windows/forum/windows_10-hardware/fyi-smt-configuration-error-affecting-amd-ryzen/6f911994-1c17-4886-98ab-93c55852285a

 

[juanrga]

As said before, beware of using average, because you can't put into the same calculation the games that are not affected by the issues. For example, if 9 games are single threaded and one is multi threaded, average will be 1/10th of the increase in that game.

And, as I posted earlier, performance can increase by as much as 15% both in average and minimum FPS, which is substantial.

There is no problem with averages. If a "patch" improves performance by 10% on a given game, but the improvement is 0% in other nine games then the average is 1%. The same happened with the Bulldozer patches. One game got 10% improvement a pair of games got 4% improvement and almost everything else performance remained unchanged. That is why AMD mentioned that the Bulldozer patches increased performance by 1--2%:

AMD was pretty honest when it described the performance gains FX owners can expect to see from this update. In its own blog post on the topic AMD tells users to expect a 1 - 2% gain on average across most applications.

However, now AMD itself claims that the W10 scheduler is working fine for RyZen and that there is not need for any update:

We have investigated reports alleging incorrect thread scheduling on the AMD Ryzen™ processor. Based on our findings, AMD believes that the Windows® 10 thread scheduler is operating properly for “Zen,” and we do not presently believe there is an issue with the scheduler adversely utilizing the logical and physical configurations of the architecture

No patch is coming because there is not anything to fix. What is more, as part of the investigation, AMD has determined that an outdated version of Sysinternals Coreinfo utility was responsible for originating the incorrect topology data that has been widely reported in the media.

If there's nothing wrong with how Win10 handles Ryzen why did Microsoft say they were releasing an update for it in their batch Windows Updates due for release in April ?

The update is addressing other issues, power management issues:

In the near term, we recommend that games and other high-performance applications are complemented by the High Performance plan. By the first week of April, AMD intends to provide an update for AMD Ryzen™ processors that optimizes the power policy parameters of the Balanced plan to favor performance more consistent with the typical usage models of a desktop PC.

 

[juanrga]

Interesting Discussion on Microsoft Forum about the Scheduler:

https://answers.microsoft.com/en-us/windows/forum/windows_10-hardware/fyi-smt-configuration-error-affecting-amd-ryzen/6f911994-1c17-4886-98ab-93c55852285a

I don't find any interesting in that link just the same people repeating same false points ever and ever and ever. The OP in that link mentions AT forums. I checked the AT thread yesterday and people is now inventing conspiracy theories after being proven wrong. They claim now that the scheduler issue is real (because they say so) but that Wintel (sic) is forcing AMD to deny the existence of the issue in public. Those guys are hilarious.

When PcPer analyzed the issue and demonstrated that the W10 scheduler is working correctly a bunch of angry people went to the PcPer site to insult them and attack them saying that are being paid by Intel. LOL

After AMD official statements PcPER wrote a sarcastic article in revenge for all the nasty attacks

http://web.archive.org/web/20170313223822/https://www.pcper.com/news/General-Tech/AMD-Running-out-Intel-Sheckels-Renews-Contract-Defame-Own-Products

Check the comment section. The same people continue rejecting facts and attacking PcPER. LOL

 

[dgothi]

I just build my new PC with Ryzen 7.
My PC Specs:
Ryzen 7 1700 (no OCing for now), Corsair Hydro Series H60 cool for Ryzen 7 CPU temp at average 36c, Asus x370 Pro motherboard, Corsair Vengeance 2X8GB DDR4-3000 with XMP enable, seasonic power supply 750 and EVGA Geforce 1070 SC. I did test benchmark two games - for honor and The Division.

For honor @ 1440p Benchmark resulted: https://goo.gl/photos/j4EF4qHJVcHo1mC57
The Division @ 1440p Benchmark resulted: https://goo.gl/photos/aE7tLKkHFF8RbEV19

I think this performancece is very well with Ryzen 7. also, I notice The Division's benchmark result show CPU average is only 38% and GPU average is 96%. GPU is loading a lot more than CPU. I assume no bottleneck for CPU because of multi-threading.

For my opinion, I think either Intel and AMD's new Ryzen both are very well. Nothing wrong either one of those you'd buy.

I am trying to find benchmark tools for some games. Do you know any good benchmark tools software? I'd test with TitanFall 2.

-DG

Hi,
I've got the same build as you but instead of the 1070 i've got the RX480.
I think i'll have some good room to play and do fun stuff.

That is so nice, Hope you enjoy it Let me know how is it.

-DG

 

[Redneck5439]

I think the most important thing to be considering at this point in time is that Ryzen is a very good processor, and will only get better. Ryzen is as big of a leap, if not bigger than Sandy Bridge was for Intel. Sandy Bridge didn't come out of the gate perfect with no issues and no room for improvement. Intel has been refining the advances they made with Sandy Bridge for several "generations" of processors and AMD will do the same with Ryzen. I think its important to look at the fact that Ryzen is very good as is and in some cases and some applications can best Intel processors costing twice as much. That in itself is huge and being largely overlooked because Intel's latest and best processors have up to 20% better FPS in some (but by no means not all) games. The simple fact of the matter is no matter what game you play on a Ryzen 7 platform your going to have a butter smooth great gaming experience. Yes there are Intel processors that are better in some games, but Ryzen keeps up very good. It comes down to what you need and what you expect.

If all you use your computer for is a glorified console system and outside of gaming your computer is a very large paperweight then Intel is definitely the way to go. If you need to game on a budget AMD may be a better solution when they release their R5 processors.

If you use your computer for more than just gaming then AMD Ryzen is definitely the way to go. The R7 line can take it right to the $1000 i7 6900K in workstation application and at half the cost. For when you just want to relax and play a game the R7 line is more than capable of giving a butter smooth gaming experience giving up only a few FPS to their Intel counterparts in most games. Ryzen is also very new and will be optimized so the gaming experience that is already very good will only get better.

For enthusiast gaming with a healthy budget then Intel i7 processors are great and still ahead of AMD at this point. If you use your computer for anything more than just gaming the R7s are workstation powerhouses that are also more than capable of delivering enthusiast level gaming. All the reviews I have seen have basically said the same thing in their 'conclusion' sections.

 

[8350rocks]

We can put the Windows Scheduling issue to rest:
https://community.amd.com/community/gaming/blog/2017/03/13/amd-ryzen-community-update?sf62107357=1

So that means performance is final? Then yeah if so even when games use 8 cores Ryzen loses to a 4 core Kabylake and in those same games the 8 core broadwell part beats a 4 core kaby lake.

Yeah the R7 Ryzen should not be considered gamer first CPUs

R3 quad cores should be decent for budget gamers however.

Not quite...

AMD has dropped the idea of using a NUMA approach for Ryzen, it seems they did not actually incorporate some specifically NUMA bits into the design...(I am assuming IF is the biggest contributor here...).

Having said that...Software optimization will still decrease thread associativity to a degree. This means that software optimized for Ryzen will not thread jump as much to different cores, meaning that there are still gains to be had. Might be ~5% in some cases at 1080p. There is a BF1 video somewhere showing the dips in FPS compared to intel processors occurring when the scheduler routes threads from 1 CCX to the other. Software developers can make threads more locked into place in some circumstances that will increase performance, but likely not as much as a full push to NUMA would.

 

[8350rocks]

Interesting Discussion on Microsoft Forum about the Scheduler:

https://answers.microsoft.com/en-us/windows/forum/windows_10-hardware/fyi-smt-configuration-error-affecting-amd-ryzen/6f911994-1c17-4886-98ab-93c55852285a

I don't find any interesting in that link just the same people repeating same false points ever and ever and ever. The OP in that link mentions AT forums. I checked the AT thread yesterday and people is now inventing conspiracy theories after being proven wrong. They claim now that the scheduler issue is real (because they say so) but that Wintel (sic) is forcing AMD to deny the existence of the issue in public. Those guys are hilarious.

When PcPer analyzed the issue and demonstrated that the W10 scheduler is working correctly a bunch of angry people went to the PcPer site to insult them and attack them saying that are being paid by Intel. LOL

After AMD official statements PcPER wrote a sarcastic article in revenge for all the nasty attacks

http://web.archive.org/web/20170313223822/https://www.pcper.com/news/General-Tech/AMD-Running-out-Intel-Sheckels-Renews-Contract-Defame-Own-Products

Check the comment section. The same people continue rejecting facts and attacking PcPER. LOL

PCPer tested, claimed it worked fine, and then watched windows scheduler push one thread to the other CCX randomly. While I think the windows scheduler is generally working well, I do think that core affinity for threads should be a bit stricter in this regard. There is no reason to push a thread from a core doing the work to start a new process when there are available cores.

I also tend to think this may relate to core parking in many ways. I believe Ryzen would benefit greatly from disabling core parking on the CPU to prevent windows from parking cores and then pushing threads later when you do not want that thread moved.

 

[gamerk316]

Software developers can make threads more locked into place in some circumstances that will increase performance, but likely not as much as a full push to NUMA would.

I'll note this approach to threading has significant downsides. Take this for example: A kernel thread bumps one of the threads you locked to a specific core. Now, because you locked it to a core, you have a thread that is ready to execute that instead has to wait for a period of time before it can run again.

This is the exact type of thing I've been complaining about the Linux scheduler for ages now, because this approach sucks for gaming.

 

[8350rocks]

https://translate.google.com/translate?sl=auto&tl=en&js=y&prev=_t&hl=en&ie=UTF-8&u=https%3A%2F%2Fwww.computerbase.de%2F2017-03%2Fryzen-windows-7-benchmark-core-parking%2F&edit-text=

Core parking issue in Win10 confirmed!

Win10 parks all but 2 cores, where as windows 7 only parks half the cores (the logical ones), so when windows 10 pushes a thread to a new core, it must wake the core, load the cache, etc.

Need to find someone to run a test with core parking disabled to confirm if there is any significant improvements (i.e. >=5%), but it *does* explain the gap between win7 and win10 nicely.

 

[gamerk316]

https://translate.google.com/translate?sl=auto&tl=en&js=y&prev=_t&hl=en&ie=UTF-8&u=https%3A%2F%2Fwww.computerbase.de%2F2017-03%2Fryzen-windows-7-benchmark-core-parking%2F&edit-text=

Core parking issue in Win10 confirmed!

Win10 parks all but 2 cores, where as windows 7 only parks half the cores (the logical ones), so when windows 10 pushes a thread to a new core, it must wake the core, load the cache, etc.

Need to find someone to run a test with core parking disabled to confirm if there is any significant improvements (i.e. >=5%), but it *does* explain the gap between win7 and win10 nicely.

Wouldn't that imply that Intel CPUs with HTT are also being negatively affected?

Hence why I view this with a decent amount of skepticism.

 

[jdwii]

We should do a poll over what people would rather have a 144hz 1440P monitor or a 4K 60hz panel for gaming.

I know what i prefer but i can say i don't want a 1080P panel again unless its small like my second 21.5 inch monitor

Maybe gamers do want 4K 60FPS or something i'm not sure also VR demands 90fps or more at all times. I just can't recommend getting less performance for the same amount of money for a gaming rig.

 

[juanrga]

https://translate.google.com/translate?sl=auto&tl=en&js=y&prev=_t&hl=en&ie=UTF-8&u=https%3A%2F%2Fwww.computerbase.de%2F2017-03%2Fryzen-windows-7-benchmark-core-parking%2F&edit-text=

Core parking issue in Win10 confirmed!

Win10 parks all but 2 cores, where as windows 7 only parks half the cores (the logical ones), so when windows 10 pushes a thread to a new core, it must wake the core, load the cache, etc.

Need to find someone to run a test with core parking disabled to confirm if there is any significant improvements (i.e. >=5%), but it *does* explain the gap between win7 and win10 nicely.

It says exactly the contrary:

On average over the 13 games tested by ComputerBase, there is a difference of two percent, but in favor of Windows 10.

If AMD's statement on the scheduling at the word is taken, the speed gains listed in this article are the only ones that can be expected by Windows 10 on the operating system level.

 

[-Fran-]

It says exactly the contrary:

On average over the 13 games tested by ComputerBase, there is a difference of two percent, but in favor of Windows 10.

If AMD's statement on the scheduling at the word is taken, the speed gains listed in this article are the only ones that can be expected by Windows 10 on the operating system level.

Leave Ashes out of the average and take another look. Ashes is behaving as a heavy outlier.

Cheers!

 

[AlexTheRoot]

Could owner of 1700/1700X/1800X test CPU using Indigo Bench? Here is link:
https://www.indigorenderer.com/indigobench
Unfortunately uploading isn't working right now, please post result here.

I can do that today. I have AMD Ryzen 7 1700 Will run it then let you know.

Thanks! Waiting impatiently!

 

[rgd1101]

ryzen building tip
https://community.amd.com/community/gaming/blog/2017/03/14/tips-for-building-a-better-amd-ryzen-system

 

[8350rocks]

https://translate.google.com/translate?sl=auto&tl=en&js=y&prev=_t&hl=en&ie=UTF-8&u=https%3A%2F%2Fwww.computerbase.de%2F2017-03%2Fryzen-windows-7-benchmark-core-parking%2F&edit-text=

Core parking issue in Win10 confirmed!

Win10 parks all but 2 cores, where as windows 7 only parks half the cores (the logical ones), so when windows 10 pushes a thread to a new core, it must wake the core, load the cache, etc.

Need to find someone to run a test with core parking disabled to confirm if there is any significant improvements (i.e. >=5%), but it *does* explain the gap between win7 and win10 nicely.

Wouldn't that imply that Intel CPUs with HTT are also being negatively affected?

Hence why I view this with a decent amount of skepticism.

Except that MS has been optimizing for their implementation of SMT for over a decade.

 

[8350rocks]

It says exactly the contrary:

On average over the 13 games tested by ComputerBase, there is a difference of two percent, but in favor of Windows 10.

If AMD's statement on the scheduling at the word is taken, the speed gains listed in this article are the only ones that can be expected by Windows 10 on the operating system level.

Leave Ashes out of the average and take another look. Ashes is behaving as a heavy outlier.

Cheers!

Precisely.

 

[8350rocks]

We should do a poll over what people would rather have a 144hz 1440P monitor or a 4K 60hz panel for gaming.

I know what i prefer but i can say i don't want a 1080P panel again unless its small like my second 21.5 inch monitor

Maybe gamers do want 4K 60FPS or something i'm not sure also VR demands 90fps or more at all times. I just can't recommend getting less performance for the same amount of money for a gaming rig.

4K @ 60 Hz for me this time around.

 

[cdrkf]

I'm on 3x 1080p 60hz panels in eyefinity here

 

[8350rocks]

I'm on 3x 1080p 60hz panels in eyefinity here

If Vega reaches my expectations, I might go CF and run 2x 4K panels...

 

[Nope 1151]

Take a look, link is down again:
http://i.imgur.com/UTHepzS.png
Original link:
http://www.guru3d.com/news-story/amd-announces-ryzen-series-5-with-six-and-four-core-processors-available-april-11.html
(Returns 404)

Looks like Juan's guess on the clockspeeds is accurate.

 

[BudgetBoy]

If that link is accurate, I wonder how the 1600X will stack up against the i7 7700k in terms of gaming value?

1600X - 6C/12T 3.6/4.0 = $250
i7-7700K - 4C/8T 4.2/4.5 = $350 + cooler

 

[juanrga]

Leave Ashes out of the average and take another look. Ashes is behaving as a heavy outlier.

Precisely.

Not true. The review says clearly that removing Ashes of the Singularity doesn't change the conclusion:

On average over the 13 games tested by ComputerBase, there is a difference of two percent, but in favor of Windows 10. This statement does not change if Ashes of the Singularity is deleted from the final result, which is significantly faster under Windows 10 than under Windows 7 due to DirectX 12.

 

[juanrga]

If that link is accurate, I wonder how the 1600X will stack up against the i7 7700k in terms of gaming value?

1600X - 6C/12T 3.6/4.0 = $250
i7-7700K - 4C/8T 4.2/4.5 = $350 + cooler

I guess the 1600X would be between the i3 and the 1700X

 

[dgothi]

Could owner of 1700/1700X/1800X test CPU using Indigo Bench? Here is link:
https://www.indigorenderer.com/indigobench
Unfortunately uploading isn't working right now, please post result here.

I can do that today. I have AMD Ryzen 7 1700 Will run it then let you know.

Thanks! Waiting impatiently!

here is result after benchmark on my PC.
https://goo.gl/photos/DaKwc35Dot8vHJQi7

-DG