I am having issues with a memory leak that is eventually consuming all my RAM, even if I don't use my computer after a while. This is utterly annoying, mainly because I have to reboot my system after a while.
I have tried to trace back the cause of the leak with poolmon.exe, and apparently, it has to do with the Event Tracing for Windows (ETW). I have tried to disable some of the logs in computer management->performance->Data Collector Sets and tried to trace precisely which one is the one causing the issue with logman/xperf.
Any ideas?
Thanks everyone!
Details
System
Razer Blade 15 Advanced Model - QHD 240Hz - GeForce RTX 3070 - Black (Reference here)
RAM 64GB
Windows 10 Pro 21H2 19044.1889
Here is a screenshot of the task manager showing the memory performance in a session when this memory leak happens.
Image:
Task Manager Paged/Non-paged pool consumption
As you can see, it is consuming 63.4 GB of memory as a paged pool resource!
Image:
poolmon.exe debugging
According to Microsoft, these are the values for these tags:
EtwR - nt!etw - Etw Registration
EtwD - nt!etw - Etw DataBlock
Related
I have had created VM instances using google cloud platform (using console). The VM is a based on WINDOWS SERVER 2019! I have been successful in making one but unable to get some virtual ram in the instances. It (VRAM) shows zero. Does adding GPU not increase the vram? If not then what increases them? I am looking to increase the same for gaming purposes and using software like ADOBE AND AUTODESK too...
Instances created with additional GPU's (Like Tesla K80 and other) have all specified amount of GPU memory (VRAM).
You can find list of all GPU's in the documentation.
Every GPU has an amount of memory specified in the table.
If you create a VM with one K80 GPU it will kave 16GB of DDR6 memory available (nothing to do with the type of the machine or actual RAM assigned).
You can find how much of VRAM a GPU has in the Device manager; find "Display adapters" and expand it and find your card; it's all in the "general" tab.
And regarding any Adobe or Autodesk software I can't really tell if having Tesla will be of advantage..
Every working day, all unity developers have to launch the play mode in a Unity more than once. And if you think about it, it may take you a long time to wait for the launch of this mode.
My question is advise how I can speed up the launch mode and scripts update?
I have already taken the following actions:
Created a RAM disk and installed Unity on it.
Transferred all project-related files to the SSD (I am afraid to transfer the disc to RAM because of the possibility of losing everything in case of failure)
EDIT:
it running on:
i7 3770, ram ddr3 16gb(12 ram + 4 ram disk),Slim S55 240GB,gtx670
I've been having a lot of trouble over the past couple of days. I set up a Win7 work environment in Virtualbox and everything has been running smoothly until I tried to use Remote Desktop. I can log in to the remote machine normally and everything works for a while but then all of a sudden, my guest Win7 machine restarts due to a BSOD. I thought it might be a memory or space issue so I upped the VM's RAM to 4GB (Host system is 8GB) and the HDD to 45 GB. I also enabled I/O APIC and PAE/NX which improved performance on the guest machine but running RDP for a few minutes still triggers a BSOD. I also increased video memory from 128MB to 256MB. I don't know what else to do and I don't know how to analyze dump files. I was able to see that the last 4 error check strings were the following:
KERNEL_MODE_EXCEPTION_NOT_HANDLED
IRQL_NOT_LESS_OR_EQUAL
MEMORY_MANAGEMENT
KERNEL_MODE_EXCEPTION_NOT_HANDLED
I'm attaching the last 4 minidumps in here:
https://drive.google.com/file/d/0B-hWzWXaH29hZXVVbjVmTkV4Vkk/edit?usp=sharing
Any help would be immensely helpful. Thank you all in advance.
Cheers,
Luis
The problem was that virtualization was disabled on the host machine's BIOS which prevented me from accessing the Acceleration tab in the VM's settings. Once I changed the host machine's BIOS settings to enable virtualization, I was able to enable Hardware virtualization in the VirtualBox settings and the problem went away. I also added extra disk space for good measure.
I'm the Accurev administrator at work. I have one of my users on Mac OSX that is having really bad performance - it takes about 45 seconds between launching the app and displaying the login screen. It takes 8 seconds to run "accurev info". Our server is version 5.3a and the client is version 5.3b. We had the same issue on version 4.7. We opened a support case a month ago and have had a lot of back and forth.
Here's the hardware information
Model Name: iMac
Model Identifier: iMac11,3
Processor Name: Intel Core i7
Processor Speed: 2.93 GHz
Number Of Processors: 1
Total Number Of Cores: 4
L2 Cache (per core): 256 KB
L3 Cache: 8 MB
Memory: 8 GB
Processor Interconnect Speed: 4.8 GT/s
Boot ROM Version: IM112.0057.B00
SMC Version (system): 1.59f2
Here's the output from "accurev diag"
Basic CPU : 108233.76
Host name resolution : 286.69
Memory : 582.40
Guaranteed disk write : 135.52 MBytes/sec
Network read : 11.20 MBytes/sec, 11472.14 KBytes/sec
Available network bandwidth is equivalent to typical 100Mbit LAN
Network write : 10.72 MBytes/sec, 10973.83 KBytes/sec
Available network bandwidth is equivalent to typical 100Mbit LAN
All other application are really quick on this machine. I have a Macbook Pro with an i7 and 4GB Ram and accurev works great on it. That said, I'm not a Mac expert.
I'm at a loss and this user is really becoming frustrated. Do any of you have suggestions?
Thanks in advance for reading this question.
Could you post the support case ID so I could look into the history? My first inclination would be to isolate if it's the accurev client in general or limited to the Java GUI. How long does it take run commands like "accurev show -fx depots" from a command shell, compared to opening the list of depots in the GUI, for example?
If it turns out to be the AccuRev client in general, is it possible that there is some kind of duplexing going on from the client machine over the network? I've seen this happen via VoIP phones. The "accurev diag" output makes this seem less likely though. Have you tried having them delete their "preferences.xml" file that lives in the user's .accurev directory under their home directory?
Just some thoughts...
~James
If you use GlobalMemoryStatusEx to get statistics, one is dwMemoryLoad which is the approximate percentage of physical memory in use. Is memory that SuperFetch has consumed added to dwMemoryLoad? I'm working on software which uses this statistic to manage its own caching, flushing cache when the percentage goes too high. I'm worried that SuperFetch causes false results.
Reading about SuperFetch, games users on Vista often say turning it off improves performance. That confirms my hypothesis that SuperFetch will cause an application to falsely believe more physical RAM is in use than is actually used by the apps being run.
Mark Russinovitch's Vista Kernel posting has a nice explanation of SuperFetch.
In an interesting example of where the original Task Manager is more informative than Sysinternals Process Explorer, Jeff Atwood has pictures of the Physical Memory usage with Superfetch in Task Manager.
Note the 6MB free Physical Memory!
I'm not just being lazy - all our 32bit test machines are running XP and I only have access to 64 bit Vista or Windows 7 machines, so I'd still like to hear from people as to how it affects 32 bit systems.
Tentatively, the answer is no it doesn't get included - the cached amount is accumulated separately.
I turned on SuperFetch on my Vista x64 machine (interesting to note it was off for some reason) and built the sample program in the GlobalMemoryStatusEx docs. At a time with little happening on my machine, the sample was reporting 50% of memory in use but the Windows task manager showed Physical Memory stats of:
Total 4093
Cached 2285
Free 92