Radeon Series RX 6700 XT preview & analysis
Corsair MM700 & Corsair Katar Pro XT Review
Guru3D Rig of the Month - February 2021
ASUS GeForce RTX 3060 STRIX Gaming OC review
EVGA GeForce RTX 3060 XC Gaming review
MSI GeForce RTX 3060 Gaming X TRIO review
PALIT GeForce RTX 3060 DUAL OC review
ZOTAC GeForce RTX 3060 AMP WHITE review
Fractal Design Meshify 2 Compact chassis review
Sabrent Rocket 4 PLUS 2TB NVMe SSD review
Microsoft Windows Bug Is Holding Back AMD Ryzen
We've been addressing the topic widely in our reviews, it makes little sense for Ryzen processors that are that powerful to not perform as well in games in CPU bound situations. Microsoft now kind of confirms the issue that Windows 10 does not detect the simulated SMT threads properly, and the simulated SMT threads are weaker compared to real CPU cores. This could hinder the game performance.
In nothing more then a tweet Microsoft seems to have confirmed the existence of this issue. So basically each core gets an extra simulated thread, that is SMT (simultaneous multi-threading) or what Intel refers to as hyper-threading. A real core obviously is stronger compared to a virtual threaded one. Aka we are back to thread-scheduler bugs. Things remain a bit trivial we feel, when we disabled SMT in the BIOS (and just ran 8-threads) performance only increased marginally on some titles, with an exception for some. But who knows, titles like Rise of the Tomb Raider (which is hit the most) might benefit from properly threaded processors big-time.
There's more going on with the scheduler though, a weird fact exposed itself that the output of the diagnostic tools for the Windows 10 scheduler that distributes these computing tasks across cores, estimated an overly optimistic cache memory per core for Ryzen. A Ryzen 7 CPU seems to be recognized as a processor with a 136 MB combined cache, where it is in reality is obviously is 20 MB for L2 and L3 caches.
Meanwhile Windows 10 RTC seems to be disliking 0.25x multipliers, causing timing issues and Microsoft’s scheduler might not understand the different CCX core-to-core latencies either. So yes, most paths now lead back to the thread scheduler.
In short, the Windows thread-scheduler does not line up with the AMD Ryzen architecture.
Back in the days when Intel introduced Hyper-threading many similar issues have been addressed. From here on-wards it will be a waiting game as we'll have to wait and see how Microsoft will address this matter and see what effect these two factors will have on the slightly disappointing game performance results. But sure, it is terrific news that Microsoft found an issue and is correcting it, let us hope that Microsoft sorts out the scheduler and cache issues as soon as possible. Maybe the upcoming Patch Tuesday update scheduled for March 14 will bring this coveted patch.
AMD Teams up with Microsoft to Advance Open Source Cloud Hardware - 03/09/2017 08:12 AM
AMD announced their collaboration with Microsoft to incorporate the cloud delivery features of AMD's next-generation "Naples" processor with Microsoft's Project Olympus -- Microsoft's n...
Microsoft Stops OneDrive unlimited storage plans - 03/07/2017 08:23 AM
Announced last year and starting today Microsoft no longer offers unlimited storage to OneDrive users. Subscribers of Office 365 could previously store an unlimited amount of data on Microsoft&rs...
Microsoft Adds Ability to Block Non-Windows Store Apps in Win10 - 02/28/2017 07:22 AM
Is Windows 10 on full device-style lockdown, requiring you to use apps from the Windows Store? Beta News reports, this is only an option which can be enabled should you so desire, and while it is in t...
Microsoft will release Windows 10 Cloud - 02/06/2017 05:31 PM
Buty Windows 10 Cloud has actually very little to do with the cloud, think of it as a simplified Windows RT. Earlier on it was rumored already that Microsoft is planning to launch another version of ...
Microsoft to add Refresh on Demand to Windows Defender - 01/11/2017 08:32 AM
Microsoft will be adding a new featture towards Windows Defender called Refresh on Demand. The function will allow you tor ecover from a slow or crashing setup by reinstalling the base Windows and re...
Neo Cyrus
Senior Member
Posts: 9695
Joined: 2006-02-14
Senior Member
Posts: 9695
Joined: 2006-02-14
#5404737 Posted on: 03/10/2017 10:52 AM
I just realized there was a mistake in my post, and it's too late to edit it again. A stress test in fact doesn't make full use of a core, that's why SMT in such an artificial benchmark will yield much better results.
I'm super tired, what was going through my brain as I was typing that, is that a stress test (some, not all) is one of the few things that can saturate a core properly for both threads. That's what I was thinking of when I said it can make full use of a core. In my half asleep state that somehow translated into the wrong statement. Sorry.
I just realized there was a mistake in my post, and it's too late to edit it again. A stress test in fact doesn't make full use of a core, that's why SMT in such an artificial benchmark will yield much better results.
I'm super tired, what was going through my brain as I was typing that, is that a stress test (some, not all) is one of the few things that can saturate a core properly for both threads. That's what I was thinking of when I said it can make full use of a core. In my half asleep state that somehow translated into the wrong statement. Sorry.
nevcairiel
Senior Member
Posts: 744
Joined: 2015-05-19
Senior Member
Posts: 744
Joined: 2015-05-19
#5404740 Posted on: 03/10/2017 11:03 AM
One would think AMD would be smart enough to work with Microsoft to have this sorted out in Windows 10 for the processor launch. Especially after their last processor architecture which needed Windows scheduling changes as well, and happened only after the launch. But apparently not.
One would think AMD would be smart enough to work with Microsoft to have this sorted out in Windows 10 for the processor launch. Especially after their last processor architecture which needed Windows scheduling changes as well, and happened only after the launch. But apparently not.
SirDremor
Senior Member
Posts: 586
Joined: 2008-06-20
Senior Member
Posts: 586
Joined: 2008-06-20
#5404745 Posted on: 03/10/2017 11:13 AM
If this is an issue in Windows 10, why does this affect only Ryzen?
Or is Intel also somehow affected by this?
If this is an issue in Windows 10, why does this affect only Ryzen?
Or is Intel also somehow affected by this?
Neo Cyrus
Senior Member
Posts: 9695
Joined: 2006-02-14
Senior Member
Posts: 9695
Joined: 2006-02-14
#5404748 Posted on: 03/10/2017 11:17 AM
If this is an issue in Windows 10, why does this affect only Ryzen?
Or is Intel also somehow affected by this?
Read my earlier posts in this thread and moeppel's post as well.
The TL;DR is that it only affects Ryzen because their implementation of SMT isn't detected properly by Windows. If you want more details read the posts.
If this is an issue in Windows 10, why does this affect only Ryzen?
Or is Intel also somehow affected by this?
Read my earlier posts in this thread and moeppel's post as well.
The TL;DR is that it only affects Ryzen because their implementation of SMT isn't detected properly by Windows. If you want more details read the posts.
Click here to post a comment for this news story on the message forum.
Senior Member
Posts: 153
Joined: 2015-06-30
Let me explain it from what I understand, someone correct me if I'm wrong.
The SMT thread uses leftover or unused resources of a core to get maximum use out of the core.
A core is made of many components, and even if it's used to "100%" as Windows's task manager (but not others) may show, it's not common for all the parts of the core to actually be in use at the same time. A program is pretty much never (aside from stress tests) going to use exactly all the resources a core has, it needs certain amounts of certain things in a certain ratio, that means there will be something left unused generally and can't make use of that remainder because there isn't enough of other resources (the ratio I talked about). SMT is a clever way of scheduling to make use of those available resources which adds up to quite a bit when you have multiple cores.
So this means the SMT thread is nowhere near as powerful as the main thread of the core, it's not simply 2 threads attached to one core. One is a standard thread, the other is cleverly using leftover resources. Both threads have to be in use at the same time to saturate a core.
Though I thought I read in the past that Intel (at the time, this was a long time ago) did include a minor amount of extra physical components in their SMT CPUs. This was so long ago it may have been with a Netburst chip or derivative with their first implementation of Hyper Threading. I have no idea if that still holds true today, but I doubt it. I've asked about this at least once before but no one had an answer.
Futher Ryzen consists of CCXs (Core Complexes, in R7's case 2 CCXs with 4 Cores each) - Due to architecure and the nature of Ryzen it's further helpful to have threads and resources reside in the same CCX insofar the threads are dependent on one another.
Interal CCX communication is a lot faster than cross CCX communication. Having threads which depend on one another run in a different CCX increases latency unnecessarily.
That said, I've absolutely no idea how much a proper scheduling could get out of Ryzen. I'd expect anywhere from 5~10% but that may just be completely wishful already.
Time will tell.