Microsoft acknowledges slow NVMe SSD write speeds in Windows 11 - working on a fix.

by Hilbert Hagedoorn on: 12/11/2021 12:33 PM | source: support.microsoft.com | 23 comment(s)

Windows 11 has had a bug for a long time that slows down the write speed of hard drives and SSDs. Microsoft is releasing a fix for the problem under patch KB5007262.

Recent tests by Reddit user PleasedPen25317 showed that the random write speeds of the Samsung 980 Pro SSD went down a lot when Windows 11 was installed on any partition that had the SSD. It must have been more widespread than previously thought because the problem affects all drives, not just NVMe SSDs, which are very, very fast.

November 22, 2021—KB5007262 (OS Build 22000.348) Preview -

Addresses an issue that affects the performance of all disks (NVMe, SSD, hardisk) on Windows 11 by performing unnecessary actions each time a write operation occurs. This issue occurs only when the NTFS USN journal is enabled. Note, the USN journal is always enabled on the C: disk.

Read more about it here, a whole lot more has or will be fixed.

Related Stories

Microsoft adds H265 and H264 video encoders to the DirectX12 API for graphics cards. - 12/09/2021 02:27 PM
Microsoft has announced new DirectX12 API compatibility for Windows which will offer a new way for apps to efficiently encode video using the GPU....

Windows 11 Build 22000.318: Microsoft rolls out AMD fix with KB5007215 to all users - 11/11/2021 10:08 AM
Update KB5007215 to build 22000.318 is being released as part of Patchday November, and the patch for the L3 cache of Ryzen CPUs is being made available to all users - this is the first time that cust...

Microsoft halts Onedrive support on Windows 7, 8 and 8.1 - 11/08/2021 10:30 AM
March 1st, 2022 will be the date that the desktop application for the cloud storage service Microsoft OneDrive for Windows 7, 8, and 8.1 will no longer be supported by Microsoft....

Windows 11 Performance Issues on Ryzen confirmed Fixed with Updates from Microsoft and AMD - 10/22/2021 03:35 PM
Microsoft and AMD delivered software upgrades on Thursday to fix two performance issues with Windows 11. Problems with L3 cache latency and the "Preferred Cores" mechanism. The issues redu...

Microsoft promises free Flight Simulator GOTY Edition Upgrade - 10/21/2021 09:19 AM
Microsoft has announced that the Game of the Year Edition of Flight Simulator will be released on November 18 as a free upgrade for existing gamers, according to a blog post by the firm. On the same d...

5 pages 1 2 3 4 5

PrMinisterGR
Senior Member

Posts: 8058
Joined: 2014-09-27

#5972626 Posted on: 12/13/2021 12:48 PM

What point?
This one.
Resources do not mean that you can just bring a mass of people to solve any problem and call it a day. If the issue is complex, it doesn't matter that much if there is 100 versus 5 people investigating it, because at least one person needs to fully understand the whole structure of the issue first, since we are talking about code here. Organizing even few people to research the issue at hand effectively is the best solution and in fact the only working one instead of throwing money into the air. Heck the problem could even be that they have too much staff doing the same thing and switching places so nobody knows exactly what is going on, but this is just me thinking, I really don't know. Resources can solve problems quickly, but not nearly everything.

EDIT. added some

They can easily afford to hire hundreds of experts in several areas if needed. They just don't really care that much. They are so competent that they manage to invent new bugs where there haven't been before.

Microsoft has more than 100,000 programmers employed. The problem is that at this point the source code of Windows is so large that they had to modify Git to be able to even work on it.

https://arstechnica.com/information-technology/2017/02/microsoft-hosts-the-windows-source-in-a-monstrous-300gb-git-repository/

shadex
Senior Member

Posts: 673
Joined: 2008-08-19

#5972837 Posted on: 12/14/2021 07:04 AM

Sigh.. this is bugging me:

MP33 M.2 PCIe SSD using NVMe 1.3 protocol:

https://www.teamgroupinc.com/en/upload/product-note-pic/c5b68811632bdcc2e516ca197fe7bf1d.jpg

Read and Write should be 1,800 MBP/s.

I am getting 800 MBPS both read and write. :\. See my benchmark:

https://anonfiles.com/Hd58S20cv7/pix-jpg

MSI B350 Tomahawk (AMD). See CPUz ID for more info:

https://anonfiles.com/3aG1S508v2/pix2-jpg

PCI-e bandwidth chart (for reference):

https://anonfiles.com/b4HcSa0bv4/pix3-jpg

Thanks.

I may have a working theory behind this. Currently, I am using PCI-e adapter for NVM-e until I get some spart parts (screws) for my NVMe on my motherboard. A NVMe slot is a dedicated lane to the CPU unlike PCI-e x8/x16 slots? This might explains why I am only seeing 800 MBP/s.

shadex
Senior Member

Posts: 673
Joined: 2008-08-19

#5974939 Posted on: 12/21/2021 07:46 AM

Update. After doing some research, I found out that my secondary PCIe slot is operating 2.0 x2 bandwidth which is 800 MBP/s because one of 2.0 x4 slot was used (PCIe x1). So I took it off and my speed was at (R): 1,400 MBP/s and (W): 500 MBP/s. I find that very odd because 1) with PCIe x1 in use, it operates both R/W at 800 MBP/s. By making PCIe x1 slot available, write speed seems to be a lot slower, but the read speed doubles. I wondered why.

It wouldn't really matter because PCIe slots are on shared lanes anyways. M.2 is on dedicated lane, correct me if I am wrong. I will find out on the 29th once the correct screws for M.2 NVMe board arrives from MSI.

5 pages 1 2 3 4 5

Post New Comment

Click here to post a comment for this news story on the message forum.