Support NeoGAF

[Risk Breaker]

Cool you buy AMD stocks... nice.

What anything of this has to do with the thread lol

Dude stahp, you just got rekt'd. Let the thread go on.

Edit: ah jeez, top of the page. In any case, I also wanted to comment on those 2 very different Hitman benchmarks. Choosing to spread one as gospel but ignoring the other one is quite BS.

I think we should all wait to draw final conclusions when the dust, and the bugs, have settled.

 

[ethomaz]

Dude stahp, you just got rekt'd. Let the thread go on.

That is what you get?

Just go back and looks at my comments in this thread and compare with his... you will be surprised.

Seeing the results for ourselves as you recommended shows benchmarks results all over the place so the wise suggestion would be to keep a level head and give the platform time before adopting a posture and acting on it online. It just released one weak ago and look at yourself already, pushing panic.

Lets wait.

Panic? lol lol lol

 

[opticalmace]

Ugh, ok, Win 10 then.

As much as everyone has helped me in this thread and answered questions/engaged in discussion, the lack of objectivity has made it really tough to make decisions and has honestly made things harder for a first time build.

How I see it...

if it's a work machine -- look up benchmarks of the exact software you need to use, decide based on those. Depending on the nature of the work, could be either ryzen, z270 or x99.

if it's mostly for gaming/general use -- just go intel z270.

My gaming rig is a 6700k, work rig is ryzen 1700.

 

[spyshagg]

Are you from AMD?

Serious question because the only think you did in this thread since beginning is marketing Ryzen in games against all solid evidences that it is indeed delivery less being more expensive in games.

What you objective lying in a forum where people is instructed enough to see the results for themselves? I really don't understand what you gain with that and yes I already saw most guys are already avoiding to reply to you.

Seeing the results for ourselves as you recommended shows benchmarks results all over the place so the wise suggestion would be to keep a level head and give the platform time before adopting a posture and acting on it online. It just released one weak ago and look at yourself already, pushing panic. Internet remembers forever, so present yourself the way it suits you but dont complain when it bites you in the arse.

Lets wait.

 

[Coldfriction]

Ugh, ok, Win 10 then.

As much as everyone has helped me in this thread and answered questions/engaged in discussion, the lack of objectivity has made it really tough to make decisions and has honestly made things harder for a first time build.

It's not that hard.

http://www.anandtech.com/show/11182/how-to-get-ryzen-working-on-windows-7-x64

 

[Vipu]

Thank for these links, I'll dig through! I honestly prefer Win 7, so this is good(ish?) news for now. Can you get a legit copy of Win 7 anymore?

FAKE EDIT - I'm a bit concerned though, 'cos don't games generally run better on Windows 10?

https://www.youtube.com/watch?v=XzSSidHU54g

 

[spyshagg]

Any solid evidence that Ryzen performs better on Win 7 rather than 10? Or to the contrary?

https://ibb.co/jEZXDv

Some difference. Still early days.

If you are a gamer with hard-earned money, waiting is your best friend right now.

 

[spyshagg]

Panic? lol lol lol

Thats how you sound like. Not my fault.

Its a cpu that launched a weak ago. Loosen up.

 

[ethomaz]

Thats how you sound like. Not my fault.

Its a cpu that launched a weak ago. Loosen up.

Panic for what? That even makes sense?

OT.

AMD finally updated their review... it is full now... just not include games but there 5-6 pages with info about Ryzen's Arch... they give even 3 graphs of Perf/$.

http://www.anandtech.com/show/11170...-7-review-a-deep-dive-on-1800x-1700x-and-1700

 

[Datschge]

Any solid evidence that Ryzen performs better on Win 7 rather than 10? Or to the contrary?

Windows 7 most certainly is less broken than Windows 10 even though neither contain official support so far and Win7 can be tricky to install. Anyway some posts from the superb "Ryzen - Strictly technical" at anandtech forums (the OP is a strongly recommended all encompassing technical read): https://forums.anandtech.com/threads/ryzen-strictly-technical.2500572/

Regarding performance Win7 vs Win10:

Draw call benchmark https://forums.anandtech.com/threads/ryzen-strictly-technical.2500572/page-5#post-38773988

Same driver version, same everything else.
17.8% faster than Win 10

Total War: Warhammer (game with the biggest SMT regression in Win10) https://forums.anandtech.com/threads/ryzen-strictly-technical.2500572/page-8#post-38775732

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)

Regarding installing Windows 7:
Slipstreaming drivers: https://forums.anandtech.com/threads/ryzen-strictly-technical.2500572/page-10#post-38776685
Using DISM: https://forums.anandtech.com/threads/ryzen-strictly-technical.2500572/page-11#post-38776813
Without early drivers using PS/2 mouse/keyboard: https://forums.anandtech.com/threads/ryzen-strictly-technical.2500572/page-11#post-38776848

 

[Datschge]

LOL, it looks like we can throw out a lot of Win8/8.1/10 Ryzen benchmarks (essentially all overclocking ones) for being inaccurate due to a technical glitch?

Ryzen Platform Affected by RTC Bias; W8/8.1/10 Not Allowed on Select Benchmarks

In a statement issued by the Head of Moderation Christian Ney, we have confirmed that the AM4 platform is affected by the Windows 8/8.1/10 RTC bias. The bias occurs when adjusting the reference clock at run-time and will affect the Windows timer, causing benchmarks to perceive time slower (or faster) than it really is. This results in benchmark scores affected in such a way that the benchmark scores reported do not reflect real performance.
(...)
Timer is skewed when changing REFCLK in Windows 8+. Additionally the default systimer has issues with OS ratio changes unless HPET is enabled. To summarize, always enable HPET on this platform.
(...)
When downclocking the system under Windows8, the Windows RTC is affected as well. The biggest difference between Windows7 and Windows8 is that now all benchmarks (no exception) are affected.

 

[theultimo]

LOL, it looks like we can throw out a lot of Win8/8.1/10 Ryzen benchmarks (essentially all overclocking ones) for being inaccurate due to a technical glitch?

Ryzen Platform Affected by RTC Bias; W8/8.1/10 Not Allowed on Select Benchmarks

There is a "sleep bug" possibly too in windows 10 in an reporting oc incorrectly issue.

https://youtu.be/S93asfXAu9M

 

[Datschge]

There is a "sleep bug" possibly too in windows 10 in an reporting oc incorrectly issue.

https://youtu.be/S93asfXAu9M

Yeah, this video has been posted before.

 

[ethomaz]

LOL, it looks like we can throw out a lot of Win8/8.1/10 Ryzen benchmarks (essentially all overclocking ones) for being inaccurate due to a technical glitch?

Ryzen Platform Affected by RTC Bias; W8/8.1/10 Not Allowed on Select Benchmarks

Actually they removed all Benchmarks with Win8/8.1/10 with overclocked that are not approved apps (3DMark suite, GPUPI, HWBOT X265 Benchmark, Y-Cruncher, Realbench and CPU-Z).

Default clock has no issue.

This bias exists in several Intel platforms too and now in AM4... to overclock you need to use their guides to avoid the bias and the benchmark be accepted.

SkyLake same issue: http://hwbot.org/newsflash/3182_imp...kylake__clarification_on_engineering_hardware
List of Intel Platform with the same issue: http://hwbot.org/news/9824_breaking_windows_8_benchmark_results_no_longer_accepted_at_hwbot/

There is a "sleep bug" possibly too in windows 10 in an reporting oc incorrectly issue.

https://youtu.be/S93asfXAu9M

In the article says they are investigating this case yet.

http://forum.hwbot.org/showthread.php?t=167702

 

[Übermatik]

OKAY. Thank-you for the replies everyone, I'm going to address them one by one...

I'm sure it varies by game. Regardless, if you want to take advantage of DX12, Windows 10 is your only choice.

I likely will, as the RX 480 I'm getting takes advantage of DX12

I think its a strange situation where a "winner" so to speak isnt really clear cut. Actual competition is good for this, and there really isnt a downside in the long run on either case.

There are positives in both archs, and honestly it all comes down to "what do you do want to do with it, and are aware of certain restrictions"

Its an exciting time for PC development.

Primarily, this is a 'workstation', a PC build for productivity

Honestly, with the way things are in such flux right now and a total lack of reviews on motherboards, it would still be hard to give you advice.

That said, I just found out that Newegg is selling Ryzen CPU + mobo combos for Gigabyte, at the very least, so once things settle down, things might be easier for you.

The trials of being a (fairly) early adopter it seems! Newegg don't have such deals here in the UK, but not to worry too much. I'll keep shopping around for best prices - the price alerts on PCPartPicker I've set up should help with that.

How I see it...

if it's a work machine -- look up benchmarks of the exact software you need to use, decide based on those. Depending on the nature of the work, could be either ryzen, z270 or x99.

if it's mostly for gaming/general use -- just go intel z270.

My gaming rig is a 6700k, work rig is ryzen 1700.

It is indeed a work machine, and having just looked up some benchmarks for the programs I'll be using, such as Maya, I'm more confident in my choice now. The lower end 1700 I'll be buying dukes it out with even the 7900K, at a fraction of the price, so it seems only logical to pull the trigger on that CPU:

It's not that hard.

http://www.anandtech.com/show/11182/how-to-get-ryzen-working-on-windows-7-x64

Noted, thank-you!

https://www.youtube.com/watch?v=XzSSidHU54g

I saw this! Not worth the faff I don't think, though...

https://ibb.co/jEZXDv

Some difference. Still early days.

If you are a gamer with hard-earned money, waiting is your best friend right now.

Can't afford to wait, really. I need the PC now (laptop is dying) and the sooner I get up and running the better. If I wait forever, I won't get anything!

Windows 7 most certainly is less broken than Windows 10 even though neither contain official support so far and Win7 can be tricky to install. Anyway some posts from the superb "Ryzen - Strictly technical" at anandtech forums (the OP is a strongly recommended all encompassing technical read): https://forums.anandtech.com/threads/ryzen-strictly-technical.2500572/

Regarding performance Win7 vs Win10:

Draw call benchmark https://forums.anandtech.com/threads/ryzen-strictly-technical.2500572/page-5#post-38773988


Total War: Warhammer (game with the biggest SMT regression in Win10) https://forums.anandtech.com/threads/ryzen-strictly-technical.2500572/page-8#post-38775732


Regarding installing Windows 7:
Slipstreaming drivers: https://forums.anandtech.com/threads/ryzen-strictly-technical.2500572/page-10#post-38776685
Using DISM: https://forums.anandtech.com/threads/ryzen-strictly-technical.2500572/page-11#post-38776813
Without early drivers using PS/2 mouse/keyboard: https://forums.anandtech.com/threads/ryzen-strictly-technical.2500572/page-11#post-38776848

Great resources, thank-you! I'll trawl through these and see what info I can extrapolate.

As mentioned, what I'm looking at now is:

RX 480 8GB
Ryzen 1700
Gigabyte/MSI/Asrock Mobo
Win 10, for ease

 

[ethomaz]

RX 480 8GB
Ryzen 1700
Gigabyte/MSI/Asrock Mobo
Win 10, for ease

I guess you choose that pages ago

Win 10 is preferable even if with bugs... there is nothing worst to work in a old Windows and for some reason the app you use to work crash because the OS didn't support the new hardware.

Unless the software you will use is too old that requires a old OS then go with Win 10... even so I prefer use VM on Win10 to run old softwares.

About the mobo I love Gygabyte... so I'm biased but from these only MSI is in the same quality... Asrock for me is the low quality of Asus... Asus has probably the best quality mobo in the market but they are the most expensives.

BTW the best place to get help with your config: http://www.neogaf.com/forum/showthread.php?t=1222740

I build my PC last year only using this thread for all my doubts... it is active and the guys has knowledge to help.

Give it a try.

 

[Übermatik]

I guess you choose that pages ago

Win 10 is preferable even if with bugs... there is nothing worst to work in a old Windows and for some reason the app you use to work crash because the OS didn't support the new hardware.

Unless the software you will use is too old that requires a old OS then go with Win 10... even so I prefer use VM on Win10 to run old softwares.

About the mobo I love Gygabyte... so I'm biased but from these only MSI is in the same quality... Asrock for me is the low quality of Asus... Asus has probably the best quality mobo in the market but they are the most expensives.

BTW the best place to get help with your config: http://www.neogaf.com/forum/showthread.php?t=1222740

I build my PC last year only using this thread for all my doubts... it is active and the guys has knowledge to help.

Give it a try.

Windows 10 it is! Again, mostly to keep things simple haha
As for the mobo... well, I'm somewhat (rightly or wrongly) swayed by the colour palette of the board itself - I'm getting the Silver Enthoo Evolv ATX, my GPU is white and the other components are monochrome, so far. The thing that turns me off Gigabyte, despite the Dual Bios, is the flashes of red all over the place... MSI's is far subtler, and Asrock's PRO is even better, being grey and all.

Now we mention Asrock, I saw a whoooole bunch of recommendations for that particular B350 mobo on the AMD subreddit. Why is that? Do they know something I don't? I hadn't heard of Asrock before now, so I was a little puzzled.

And yes, I've spammed the PC build thread too, don't you worry! I think things are finally coming together though...

 

[dr_rus]

Core by Core, MHz by MHz with AMD RYZEN 7 1800X vs Intel Core i7-7700K [English]

 

[Louis Cyphre]

Seeing the results for ourselves as you recommended shows benchmarks results all over the place so the wise suggestion would be to keep a level head and give the platform time before adopting a posture and acting on it online. It just released one weak ago and look at yourself already, pushing panic. Internet remembers forever, so present yourself the way it suits you but dont complain when it bites you in the arse.

Lets wait.

This is very rational, I also tend to tune out those who keep posting over and over again as though they have some sort of agenda.

Ryzen is a brand new product and I'm willing to see what transpires over the next few months. Gamers should actually be looking at the Ryzen 5 because those will be cheaper and more comparable since they have less cores like the Intel chips. No I don't think they will outperform but value wise they ought to. Ryzen 7 seems like a great overall product that does everything well just not the high end gaming where people want to argue over 120 fps versus 100 fps.

Honestly I'd rather sacrifice a bit in gaming performance for a long term product then to have Intel continue it's dominance.

 

[Übermatik]

Honestly I'd rather sacrifice a bit in gaming performance for a long term product then to have Intel continue it's dominance.

Basically what I'm settling on.

As it goes, is the 1700X worth the extra £30? My understanding is that 1700X clock speeds can be achieved with the 1700 in overclock. But then £30 isn't a lot...

AMD Ryzen 7 1700X: £368.99
http://amzn.eu/hloBpoF

AMD Ryzen 7 1700: £332.60
http://amzn.eu/fIcLD2h

 

[Datschge]

As it goes, is the 1700X worth the extra £30? My understanding is that 1700X clock speeds can be achieved with the 1700 in overclock. But then £30 isn't a lot...

Are you conscious of power consumption or do you plan to overclock all the time? If the former I'd pick 1700 as it's closer to the sweet spot in the performance vs consumption ratio, for the latter 1700X may be the better deal depending on binning.

 

[Sectorseven]

Basically what I'm settling on.

As it goes, is the 1700X worth the extra £30? My understanding is that 1700X clock speeds can be achieved with the 1700 in overclock. But then £30 isn't a lot...

AMD Ryzen 7 1700X: £368.99
http://amzn.eu/hloBpoF

AMD Ryzen 7 1700: £332.60
http://amzn.eu/fIcLD2h

The 1700X doesn't include a cooler, so you'll have to add that cost as well.

 

[Übermatik]

Are you conscious of power consumption or do you plan to overclock all the time? If the former I'd pick 1700 as it's closer to the sweet spot in the performance vs consumption ratio, for the latter 1700X may be the better deal depending on binning.

Yeah, I'd rather not buy a more expensive PSU than I have to. Overclocking would be a niche treat for me.

The 1700X doesn't include a cooler, so you'll have to add that cost as well.

Aw shit, good spot. This seals it then.

 

[Steel]

Yeah, I'd rather not buy a more expensive PSU than I have to. Overclocking would be a niche treat for me.



Aw shit, good spot. This seals it then.

To be clear, a 4.0 Ghz overclocked 1700 will use more energy than the 1700x at base clocks(They'd be similar at the same speed, however). You'll have to buy the same PSU regardless.

 

[ethomaz]

Core by Core, MHz by MHz with AMD RYZEN 7 1800X vs Intel Core i7-7700K [English]

Interesting... about 10-15% behind Kaby Lake IPC... that means around Haswell IPC.

I think AMD has a great arch in hands right now that suffer from the low clocks... for the test they had to overclock to 4.0Ghz (max they get stable) while downclocked 7700k to 4.0Ghz... that is 500Mhz difference from the Ryzen over to 7700k base.

 

[Locuza]

Interesting... about 10-15% behind Kaby Lake IPC... that means around Haswell IPC.

I think AMD has a great arch in hands right now that suffer from the low clocks... for the test they had to overclock to 4.0Ghz (max they get stable) while downclocked 7700k to 4.0Ghz... that is 500Mhz difference from the Ryzen over to 7700k base.

1080p results with ultra settings showing mostly average FPS are not a IPC comparison.

Here is a table from PCGH versus BDW-E @ 3,2 Ghz:

http://www.pcgameshardware.de/Ryzen-7-1800X-CPU-265804/Tests/Test-Review-1222033/

 

[Renekton]

1080p results with ultra settings showing mostly average FPS are not a IPC comparison.

Here is a table from PCGH versus BDW-E @ 3,2 Ghz:
https://abload.de/img/broadwellryzenxfjgr.jpg
http://www.pcgameshardware.de/Ryzen-7-1800X-CPU-265804/Tests/Test-Review-1222033/

Ouch, that's why AMD loves cinebench

 

[Locuza]

Fortunately PCGH used the same settings for the SMT on/off comparison, I included the SMT off results on the right:

The most noticeable result is from Far Cry 4, where SMT off can bring 24%-39% more performance.
There is also AC: Syndicate which runs ~8% faster without SMT but the rest reacts pretty little.

On the other hand applications show more differences and can lose much more performance if SMT is turned off.

 

[Steel]

Ouch, that's why AMD loves cinebench

Well, that is a comparison between a $1k chip and a $500 chip, in fairness.

 

[ethomaz]

1080p results with ultra settings showing mostly average FPS are not a IPC comparison.

Here is a table from PCGH versus BDW-E @ 3,2 Ghz:

http://www.pcgameshardware.de/Ryzen-7-1800X-CPU-265804/Tests/Test-Review-1222033/

What do you mean? Their IPC comparison are wrong? It is bigger the difference?

 

[Locuza]

What do you mean? Their IPC comparison are wrong? It is bigger the difference?

I painted this extreme case:

If you use the second test case you could make the wrong statement that the IPC of every processor is the same because you get equal FPS across the board.

Their results are smoothed out by the higher resolution and make it impossible to produce precise judgement about the IPC difference.
The difference is definielty higher.

 

[LelouchZero]

I painted this extreme case:

If you use the second test case you could make the wrong statement that the IPC of every processor is the same because you get the equal FPS across the board.

Their results are smoothed out by the higher resolution and make it impossible to produce precise judgement about the IPC difference.
The difference is definielty higher.

This is a good visual representation of CPU and GPU limitations, I like this!

 

[Coldfriction]

This is a good visual representation of CPU and GPU limitations, I like this!

It's a good simple visualization. One would assume any of the three CPU's are overkill for gaming under a GPU limited situation. Looks like more money should have been spent on a better GPU and paired with a cheaper CPU. However, GPUs are usually replaced at a far higher rate than a CPU, so maybe more money should be spent on the CPU knowing the future will change. This is a primary argument for not getting a Ryzen chip, and one of the most solid arguments out there. Gaming at 144hz is no joke if you want to be competitive at the highest levels. I would say that my biggest concern about Ryzen is whether it starts to limit a high refresh rate system at say 1080p. I can tell the difference between 120hz and 144hz personally. I'm not sure it even matters though when the server tick rates are something like 30-60 hz. The monitor isn't showing what the server says the game is doing at that point; it's maddening to be hit when you swear you had cover and miss when you are certain you shot someone. I'm super stoked Quake Champions announced a 120hz tick rate on all servers.

I'm of the opinion that whatever a person builds with somewhat high end parts, the GPU will probably always limit the CPU through the usable lifespan of any new system. People have to go to extreme settings (720p) to find a situation not GPU limited. For a visual learner, these sorts of figures are invaluable in explaining someone's wall of text; there are probably a lot of neogaf readers that are new to PCs for the first time and need this sort of information presented in this fashion.

"Simplicity on the far side of complexity is powerful, simplicity on the near side of complexity is dangerous." -One of the best things I've heard in my life.

 

[LegitMaan]

It's a good simple visualization, but what is the takeaway message? Any of the three CPU's are overkill for gaming under a GPU limited situation. Looks like money should have been spent on a better GPU and none of those CPUs.

CPU 2 will hold you back when you upgrade your GPU in the future.

 

[theultimo]

It's a good simple visualization, but what is the takeaway message? Any of the three CPU's are overkill for gaming under a GPU limited situation. Looks like money should have been spent on a better GPU and none of those CPUs.

It has nothing to do with being gpu limited.

It just shows actual cpu differences, in which higher resolution would not show reliability bevause the gpu would limit any data.

 

[Coldfriction]

CPU 2 will hold you back when you upgrade your GPU in the future.

That's an assumption. How could you know that?

 

[Coldfriction]

It has nothing to do with being gpu limited.

It just shows actual cpu differences, in which higher resolution would not show reliability bevause the gpu would limit any data.

The real world value of any of the CPUs in the figure for gaming is equal across the board at the resolution people play at (I don't know anyone playing at 720p on a PC today). It tells me someone has more CPU than they need assuming all they do is play the game in the bar graph represents, assuming it represents a game. IF the bar graph represents benchmarks for an average of many games, it might be more representative of typical expected peformance, but the chart doesn't say that does it? There is a whole bunch of stuff that chart doesn't tell you that you might believe it does. You can't pull meaning out of something that doesn't contain it, don't extrapolate incorrectly from a limited data set.

The limitation at 720p means nothing. The higher resolution isn't hiding anything; the lower resolution exposes nothing of value except potential future limitations assuming a future GPU will no longer be the bottleneck. Also an assumption, yet an honest concern. You guys are taking the simplicity on the near side of complexity approach to the figure.

 

[·feist·]

For a range of reasons from AGESA to availability, decent AM4 motherboard reviews have been difficult to come by. I've linked a few earlier, and will try to compile a section for X, and B series reviews. Perhaps A series as well if there's enough interest or reason to do so.




PC Perspective's take on Windows 10 scheduling:

AMD Ryzen and the Windows 10 Scheduler - No Silver Bullet
https://www.pcper.com/reviews/Processors/AMD-Ryzen-and-Windows-10-Scheduler-No-Silver-Bullet

Alien: Covenant | Meet Walter | Official HD | 2017
https://www.youtube.com/watch?v=Ckst-p0c4ZU

https://twitter.com/20thcenturyfox/status/839883776914173952

I might try to get an extra 200 MHz but I don't intend to do any hardcore overclocking.

I went with the Noctua NH-U12S SE-AM4:


Noctua has 3 "special edition" AM4 coolers, this is the mid-range one. Will post some temperature #s later if anybody is interested. (Just got Win10 installed this morning before I had to leave for work, can't wait to get home and play with my new toy!)

Nice choice. I hope to be able to finally get a hold of my Noctua AM4 kit soon.

 

[Renekton]

https://www.pcper.com/reviews/Processors/AMD-Ryzen-and-Windows-10-Scheduler-No-Silver-Bullet

This suggests any scheduler fix is not a magic bullet

In this way at least, the CCX design of 8-core Ryzen CPUs appears to more closely emulate a 2-socket system. With that, it is possible for Windows to logically split the CCX modules via the Non-Uniform Memory Access (NUMA), but that would force everything not specifically coded to span NUMA nodes (all games, some media encoders, etc) to use only half of Ryzen.

 

[Paragon]

New article from PC Perspective: AMD Ryzen and the Windows 10 Scheduler - No Silver Bullet

Closing Thoughts
What began as a simple internal discussion about the validity of claims that Windows 10 scheduling might be to blame for some of Ryzen's performance oddities, and that an update from Microsoft and AMD might magically save us all, has turned into a full day with many people chipping in to help put together a great story. The team at PC Perspective believes strongly that the Windows 10 scheduler is not improperly assigning workloads to Ryzen processors because of a lack of architecture knowledge on the structure of the CPU.

In fact, though we are waiting for official comments we can attribute from AMD on the matter, I have been told from high knowledge individuals inside the company that even AMD does not believe the Windows 10 scheduler has anything at all to do with the problems they are investigating on gaming performance.

In the process, we did find a new source of information in our latency testing tool that clearly shows differentiation between Intel's architecture and AMD's Zen architecture for core to core communications. In this way at least, the CCX design of 8-core Ryzen CPUs appears to more closely emulate a 2-socket system. With that, it is possible for Windows to logically split the CCX modules via the Non-Uniform Memory Access (NUMA), but that would force everything not specifically coded to span NUMA nodes (all games, some media encoders, etc) to use only half of Ryzen.

Doesn't look good for people holding out for a scheduler fix.

I have an AMD stock position

I'm shocked.

 

[saunderez]

https://www.pcper.com/reviews/Processors/AMD-Ryzen-and-Windows-10-Scheduler-No-Silver-Bullet

This suggests any scheduler fix is not a magic bullet

Doesn't explain why performance is better under Windows 7 though. Clearly something is wrong with Windows 10 if it's not the scheduler it's something else.

 

[·feist·]

Did the two of you happen to look one post above yours while hitting submit?



Level1Linux — Ryzen: PCIe Passthrough Working, But...*
https://www.youtube.com/watch?v=ep_2dDaqBU4

 

[Renekton]

Doesn't explain why performance is better under Windows 7 though. Clearly something is wrong with Windows 10 if it's not the scheduler it's something else.

Yeah the PCPer guy came on Reddit to agree with this, W7 is still a mystery for now.

Linux on the other hand probably treats the system like a NUMA already.

Did the two of you happen to look one post above yours while hitting submit?

Sorry got distracted by Fassbender

 

[Unknown Soldier]

https://www.pcper.com/reviews/Processors/AMD-Ryzen-and-Windows-10-Scheduler-No-Silver-Bullet

This suggests any scheduler fix is not a magic bullet

Ryzen is apparently a 2x4 core design but whatever method is used to glue the core logic it's not what you would expect from an MCM. I mean Intel was doing 2x2 MCM as far back as Core 2 Quad without causing problems like this. I don't get what AMD did with Ryzen but it's obviously not kosher as far as Windows is concerned.

 

[DonMigs85]

Well at least the 4-core chips should be decent i3/i5 alternatives

 

[Coulomb_Barrier]

New article from PC Perspective: AMD Ryzen and the Windows 10 Scheduler - No Silver Bullet

Doesn't look good for people holding out for a scheduler fix.

You couldn't wait to post that could you.

 

[jediyoshi]

LOL, it looks like we can throw out a lot of Win8/8.1/10 Ryzen benchmarks (essentially all overclocking ones) for being inaccurate due to a technical glitch?

Ryzen Platform Affected by RTC Bias; W8/8.1/10 Not Allowed on Select Benchmarks

How would this affect benchmarking where results are done by frame analysis?

 

[Renekton]

Ryzen is apparently a 2x4 core design but whatever method is used to glue the core logic it's not what you would expect from an MCM. I mean Intel was doing 2x2 MCM as far back as Core 2 Quad without causing problems like this. I don't get what AMD did with Ryzen but it's obviously not kosher as far as Windows is concerned.

No idea lol. I didn't even know it is a MCM.

 

[Übermatik]

Is the report that the issue doesn't rely on Scheduler bad news because that would have been 'easy' (or easier) to fix? If this is a more complex issue, can we ever expect a fix at all?

 

[Locuza]

No idea lol. I didn't even know it is a MCM.

Summit-Ridge (Ryzen Desktop Products) is not a Multi Chip Module.
Intel used two chips and connected them with the (slow) Front-Side-Bus:

Summit-Ridge is just one chip but with basically the same "problem".
Two major parts are slower interconnected than the communication inside of them.