- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi,
dwm.exe (Desktop-Window Manager) uses high memory with Intel Intel HD Graphics 630. This problem has been reported before:
Here's a screenshot of my dwm ram usage after 2 hours (1.6 GB):
Please, this bug is already known for 3-4 months, and I do not want to reinstall a 2019 driver...
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
It seems that everyone has posted positive results so far about beta driver 101.1121 resolving this dwm leak issue. I'm going ahead and marking this thread as resolved! Please note that we'll specifically call out this fix in the release notes of our next production driver release to the Download Center which will be found here once available.
Special thanks to @HKlum, @TurboTacho, @Ansuel, and the MANY others that dedicated their time testing and providing various dump files. I really appreciate you all sticking by our side after all this time to make sure this long-standing problem gets resolved. You folks are what make this Community forum so great. We couldn't have done it without you! 😊
For those of you that continue to exhibit issues: Please create new individual threads so we can assess if it is a separate issue or not for your particular case.
Thanks again everyone!!
Link Copied
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
I've been experiencing this bug intermittently and I've finally managed to get a dump file of two different conditions - moderate DWM usage (150ish MB RAM) and high usage (7000+ MB). I hope these can be useful in diagnosing this problem - its endurance and the response to it are not inspiring so far.
Moderate:
https://drive.google.com/file/d/1ObM8q7K6RIJyOYsVDGgY-e2p-hrps59J/view?usp=sharing
High:
https://drive.google.com/file/d/1Hj2OHCnhRX1-ziGk03fihIoTWk76AAIw/view?usp=sharing
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Thanks for joining us on desesperately trying to make things move with our ram dumps. We are clearly all experiencing the same thing with proofs, there is already a lot of dumps here. I hope it will be enough for Intel to do something.....
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
hi all
i'm newbie member
nice to meet you all
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi @Bryce_Intel and @RonaldM_Intel , as well as everyone here.
I would like to add the 10 dumps I've taken from my PC to the ones already uploaded to help to find the issue.
https://1drv.ms/u/s!AhNGmCtr4NeUpKAfbIiKziUjkcFBrA?e=lAjFa5
In this Onedrive folder, there is 10 dumps and they are named with the approximate RAM DWM.exe was taking when I created the file (in Mb).
I tried taking them when the issue was the most critical (at 3.3gb for example). But it was difficult to know if it was really the same issue or not. Because when the RAM usage is at the highest, my PC completely freeze and the only way is to force power it off. Which of course makes me impossible to create a dump file. So I tried my best to provide files where I think the issue was present, but I can't be 100% sure. I hope it will still help to solve the issue, which is apparently the same for everyone here and really present. It happens around 1 time each day.
My PC is a Lenovo LEGION Y530 with an Intel UHD Graphics 630 + an Nvidia GTX1060 mobile. I am on latest Windows 10 Pro 20H2.
I hope this will be useful.
Have a great day everyone.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
I would like to add the 10 dumps I've taken from my PC to the ones already uploaded to help to find the issue.
https://1drv.ms/u/s!AhNGmCtr4NeUpKAfbIiKziUjkcFBrA?e=lAjFa5
In this Onedrive folder, there is 10 dumps and they are named with the approximate RAM DWM.exe was taking when I created the file (in Mb).
I tried taking them when the issue was the most critical (at 3.3gb for example). But it was difficult to know if it was really the same issue or not. Because when the RAM usage is at the highest, my PC completely freeze and the only way is to force power it off. Which of course makes me impossible to create a dump file. So I tried my best to provide files where I think the issue was present, but I can't be 100% sure. I hope it will still help to solve the issue, which is apparently the same for everyone here and really present. It happens around 1 time each day.
My PC is a Lenovo LEGION Y530 with an Intel UHD Graphics 630 + an Nvidia GTX1060 mobile. I am on latest Windows 10 Pro 20H2.
I hope this will be useful.
Have a great day everyone.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi @Bryce_Intel and @RonaldM_Intel ,
My coworkers (Hi Ismael, Neil and Kevin!) managed to capture a memory dump that shows the actual allocation stack of the blocks leaking (gflags.exe +ust).
Driver version is 27.20.100.8935 (I know, not the latest), and running on an Intel UDH Graphics 630.
They disabled the Windows LowFragmentationHeap to be able to browse the heap.
The memory dump is located at : https://genetec365-my.sharepoint.com/:u:/g/personal/jvary_genetec_com/EZ5WWmIqDipPqHTFL81Rv4QBauSFTfaL5XASVSEzKtu9_Q?e=kja0JZ
Results:
Flags enabled:
0:000> !gflag
Current NtGlobalFlag contents: 0x00001810
htc - Enable heap tail checking
htg - Enable heap tagging
ust - Create user mode stack trace database
Heaps:
0:000> !heap -s
LFH Key : 0x76d6e0b4f51df676
Termination on corruption : ENABLED
Heap Flags Reserv Commit Virt Free List UCR Virt Lock Fast
(k) (k) (k) (k) length blocks cont. heap
-------------------------------------------------------------------------------------
00000229ed6c0000 0c000022 388688 375972 388688 1465 14922 28 2 2f #whole process is usually ~80MB, that heap itself is ~390MB
00000229ed9b0000 0c000022 1080 448 1080 440 2 2 0 0
…
Stat for that big heap:
0:000> !heap -stat -h 00000229ed6c0000
heap @ 00000229ed6c0000
group-by: TOTSIZE max-display: 20
size #blocks total ( %) (percent of total busy bytes)
1d0 438ef - 7a73130 (49.32) # ~50% of that heap is block size 0x01D0 (~128MB)
68 87734 - 3706d20 (22.16)
30 6660c - 1332240 (7.73)
1100 11da - 12f7a00 (7.64)
...
Listing a bunch of blocks of size 0x01D0:
0:000> !heap -flt s 1d0
_HEAP @ 229ed6c0000
HEAP_ENTRY Size Prev Flags UserPtr UserSize - state
00000229ed6c1710 0022 0000 [00] 00000229ed6c1740 001d0 - (busy)
00000229ed72b3a0 0022 0022 [00] 00000229ed72b3d0 001d0 - (busy)
...
00000229fd57aab0 0022 0022 [00] 00000229fd57aae0 001d0 - (busy)
00000229fd57aed0 0022 0022 [00] 00000229fd57af00 001d0 - (busy)
...
Actual allocation stack of a leaking block
(can’t exactly tell what is going on within igd10iumd64.dll, we don’t have the .pdb’s in https://software.intel.com/sites/downloads/symbols/ ?)
It is always the same stack that pops-up.
0:000> !heap -p -a 2299ba2cf20
address 000002299ba2cf20 found in
_HEAP @ 229ed6c0000
HEAP_ENTRY Size Prev Flags UserPtr UserSize - state
000002299ba2cf20 0022 0000 [00] 000002299ba2cf50 001d0 - (busy)
7ffe91ff21e3 ntdll!RtlpCallInterceptRoutine+0x000000000000003f
7ffe91f0fd00 ntdll!RtlpAllocateHeapInternal+0x00000000000009e0
7ffe8774dfc0 igd10iumd64!OpenAdapter10_2+0x00000000001ec040
7ffe875baa10 igd10iumd64!OpenAdapter10_2+0x0000000000058a90
7ffe876fdbef igd10iumd64!OpenAdapter10_2+0x000000000019bc6f
7ffe876fe06d igd10iumd64!OpenAdapter10_2+0x000000000019c0ed
7ffe8770035a igd10iumd64!OpenAdapter10_2+0x000000000019e3da
7ffe876fd355 igd10iumd64!OpenAdapter10_2+0x000000000019b3d5
7ffe874ae8af igd10iumd64+0x00000000000be8af
7ffe874fe51f igd10iumd64+0x000000000010e51f
7ffe8b57143a d3d11!CContext::TID3D11DeviceContext_DrawIndexed_<1>+0x000000000000003a
7ffe8b991fc3 dwmcore!CD3DBatchExecutionContext::SetBlendMode+0x00000000000000f3
7ffe8b990a77 dwmcore!CHWDrawListEntry::ApplyRenderState+0x0000000000000027
7ffe8b96da71 dwmcore!CDrawListEntryBatch::Render+0x0000000000000131
7ffe8a8edf9e d2d1!CD3DDeviceLevel1::ProcessDeferredOperations+0x00000000000002ae
7ffe8a8ecc84 d2d1!CHwSurfaceRenderTarget::FlushQueuedOperations+0x0000000000000394
7ffe8a8da944 d2d1!CHwSurfaceRenderTarget::ProcessBatch+0x00000000000000c4
7ffe8a8d427c d2d1!CBatchSerializer::FlushInternal+0x00000000000000cc
7ffe8a8da593 d2d1!DrawingContext::Flush+0x0000000000000083
7ffe8a8da22e d2d1!D2DDeviceContextBase<ID2D1DeviceContext6,ID2D1DeviceContext6,null_type>::EndDraw+0x000000000000011e
7ffe8b942f2c dwmcore!CD2DContext::EndDraw+0x0000000000000044
7ffe8b942db7 dwmcore!CD2DContext::PopTarget+0x00000000000000e7
7ffe8b95033a dwmcore!CDrawingContext::EndFrame+0x00000000000001de
7ffe8b94dc2f dwmcore!CHwndRenderTarget::Render+0x00000000000004bf
7ffe8b97147a dwmcore!CComposition::ProcessComposition+0x0000000000000b5a
7ffe8b96e72a dwmcore!CPartitionVerticalBlankScheduler::ProcessFrame+0x000000000000049a
7ffe8b96b9e2 dwmcore!CPartitionVerticalBlankScheduler::ScheduleAndProcessFrame+0x0000000000000162
7ffe8b9f04f5 dwmcore!CConnection::RunCompositionThread+0x0000000000000169
7ffe904c7974 kernel32!BaseThreadInitThunk+0x0000000000000014
7ffe91f5a2f1 ntdll!RtlUserThreadStart+0x0000000000000021
I hope this might help to speed-up the issue troubleshooting on your side.
Best regards,
Julien
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Isnt it "interesting" how everyone can reproduce this except anyone at Intel? And how every single machine with a UHD Graphics 630 is affected by this it seems, but Intel is speaking of a "very small percentage of users"?
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi,
I've added 2 new dumps to my collection, still accessible here : https://1drv.ms/u/s!AhNGmCtr4NeUpKAfbIiKziUjkcFBrA?e=lAjFa5
These two (the most recent ones) were taken when the issue was clearly present, with DWM taking 2 to 3gb of my RAM.
I hope it will be enough to finally work on this issue, after all these months.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
@Bryce_Intel @RonaldM_Intel can you please have a look at this post:
your "moderator colleges" seem kinda confused about it and dont seem to realize the context of it. Would be nice if you could open a bug ticket for it, it might be related to the memory leak of DWM, and if not, it's another rather concerning bug.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi @HKlum
Done! thanks for bringing this to our attention.
I have added the information provided in that thread to the bug ID 22012524606 to check if it helps our devs investigating this issue. If it turns out to be a separate bug then we will handle it as such.
Best Regards,
Ronald M.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Just wanted to add the following: I noticed yesterday, that modern standby was, as usually, again broken on my Dell XPS 15 9570, no deep sleep C10 states of the SoC and Windows 10 just mentioned a "NoHwDrips.INTERNAL_BLOCKER" as the cause, so no much help, how could it be different, as what was the cause.
I had installed 27.20.100.8476 btw some days ago and could confirm, that it also didnt cause any memory leaks with dwm.
I installed latest Intel drivers 27.20.100.9466 and with those, modern standby and C10 states were working again.
So I have to warn people about that 27.20.100.8476 will break modern standby and C10 states at least for me.
Just amazing. You have the choice between cholera or pest.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Does anyone know if other integrated Intel HD Graphic also have this issue? Or is it just HD Graphics 630?
I'm buying a new laptop, so I'm worried whether their newer lines are also having the same problem or not. Do they use the same driver?
Also, anyone have experience with Ryzen laptop? How is their integrated GPU?
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
The new laptop with the card as they currently mark in laptops as only UHD has a memory leak on new Intel graphics drivers.
So there is a risk that new laptops in a similar series unfortunately have the same leak.
e.g. Lenovo Legion 5 (production 2021) Intel UHD Graphics
4095MB NVIDIA GeForce GTX 1650
Intel Core i5-10300H CPU @ 2.50GHz has a leak. So there is a risk that a new laptop with an i5 series processor and integrated UHD graphics may leak.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
The same issue also occurred on my Intel NUC NUC8i7BEH.
OS: Windows 10 Professional 20H2 (Build 19042.964)
CPU: Intel Core i7-8559U
Graphic: Intel Iris Plus Graphics 655
Driver: 27.20.100.9466
The driver version 27.20.100.8681 provided by the support page of NUC8i7BEH also triggers this issue, so I decided
This issue also occurred in the driver version 27.20.100.8681 provided by the support page of NUC8i7BEH, so I decided to find an up-to-date driver and installed 27.20.100.9466; however, the issue couldn't be solved.
Here is my dump file of dwm.exe: https://drive.google.com/file/d/1aFyW49R47q-G0B1Mo-5nagYWaYlpX6-X/view?usp=sharing
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hello Everybody,
I've been struggling with this issue for as long as I can remember now, using my Asus X556UAK laptop (i5-7200U CPU with HD 620 GPU), running Windows 10.
Unfortunately, I haven't figured out what might be the trigger for the issue - though so far I believe it may be either sleep or hibernation -, however, today I've made an observation once the issue had already been triggered: when I'm just idling on my desktop, dwm.exe is more or less stable, but when I opened an application using DirectX, the RAM usage kept growing. I tried this with two applications (Guild Wars 2 game launcher and OBS Studio) and both produced the same behavior: when minimized to the desktop the RAM usage didn't change; otherwise it kept growing by about 1MB/s.
I hope this helps the investigation in any way.
Edit: Driver version is 27.20.100.9168, Windows 10 version is Version 10.0.19042 Build 19042 (20H2, 19042.928).
Best regards,
Alex
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
I'm going to post a bolded text of this mild "workaround" that I've discovered thanks to a comment by someone on one of the many threads (microsoft website, reddit, here I can't remember exactly where) about this issue. I've tweaked the method they posted slightly but it'll be the same main premise.
I hope that it'll help but it is obviously not really a proper fix to whatever the issue actually is. It's a way to "deal" with the issue and keep it under control to some degree, I use my computer mainly for gaming and after discovering the DWM issue (I had a BSOD or two before I discovered it. Unfortunately my computer doesn't have a lot of ram at the moment, only 8gb Windows uses around half by itself. So, having DWM gobbling 1-2gb is very bad)
Workaround:
The poster said the following, he noticed that the memory leak only happens when you shut down and then turn the pc on (people have said stuff about sleep, hibernate, etc) and that the issue seems to go away or be much less if you restart the machine.
So my methods have been this:
> If you've shut your computer down when you first turn it on.
> Login and restart.
> Use computer as normal
Second part of workaround:
I've noticed that after a bit it still leaks but it's less aggressive and after a couple days of shutting down at night, restarting after startup and login that the computer will begin to slow down and get clunky. So I tried the following and it's worked well because it seems to de-gunk the memory clog as it were.
Workaround part 2:
> Do the above steps as normal for your day, if you notice it's "laggier" than normal do the following.
> Reboot computer
> In the Windows login screen, click shut down. (It gives you the warning about people losing saved work before you shut it down, that's why I tried it)
> Start up computer
> Login
> Restart again.
Windows will seem much zippier and faster after this for a few days.
I hope that the issue gets solved soon, I don't know how to revert the driver so for now the above steps are what I've been doing to cope. I've tried downloading the old version, uninstalling the driver (rollback wasn't available for me because the driver update came with the 20H2 updates) and it just wouldn't let me install the old version. It said it found the appropriate driver in windows (some windows default driver) and wouldn't accept the old-but-new Intel one. I couldn't find an adequate walkthrough to install the older drivers so these are the steps I've been taking to make the memory leak somewhat manageable.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
It sure is interesting workaround you've provided. As for my rather unique case, I actually did a simple full shutdown (Start > Shift + Shut down) and that effectively resolved my DWM bloating issue. Even if I put the laptop to sleep, the DWM issue would come back, and Intel Graphics Command Center would crash (requires restarting the app to get it running again).
In the meantime, I'd send my hello to @Bryce_Intel @RonaldM_Intel , just a friendly poke to check on the status of the issue (I hope you guys were able to upload the huge dump files into the troubleshooting tool). Still keeping my fingers crossed for any great news.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
@Serene ... .... ..... (-: I dont want to be rude but....... You wrote that long post with several """workarounds""", and basically all you just said is: "restart windows." if it starts to leak again do the other workaround "restart windows several times". Erm... right. Yes, obviously that will clear the memory. But that is not a workaround.
Only workaround so far is to use <= 27.20.100.8476.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
I said a mild workaround with quotation marks for one, babe and for two it's more specific steps than just "restart kekw it'll be better then" because part 2 mentions that eventually it gets sloggy even after "restarting windows several times"
But I guess if you actually read the post you'd understand that. I'm not going to apologize for wanting to share what I'm personally doing because I know other people are having issues, waiting patiently for a fix.
Imagine forum flaming in this day and age?
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
"I dont want to be rude but." So you say you don't want to be rude but you then decide to try **bleep** on someone for giving a mild workaround. Also, you do know what workaround means, right? You can't be such low IQ that you say "REEE NONSENSE DELETE REEE". I see your issue, you use intel. I know it's hard being a user of such a bad CPU chip, but for a very low price of £588 you could get yourself a Ryzen 9 5900x, which far out performs Intel CPU's.
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page