Message boards :
News :
VM applications
Message board moderation
Previous · 1 · 2
Author | Message |
---|---|
Send message Joined: 14 Jan 10 Posts: 1158 Credit: 7,118,800 RAC: 1,948 ![]() ![]() ![]() |
You can't make the used memory by a VM visible with task manager or process explorer. With the sysinternals util RamMap it's able to show the total amount of "Driver Locked Memory". The VM's are using that part of reserved memory. |
Send message Joined: 27 Sep 08 Posts: 745 Credit: 558,647,676 RAC: 334,134 ![]() ![]() ![]() |
|
![]() Send message Joined: 5 Nov 15 Posts: 144 Credit: 6,301,268 RAC: 0 ![]() ![]() |
Process Hacker shows Working set, Private bytes and Virtual size. (I gave up on Process Explorer after finding Hacker had a greater set of variables to display, better column sorting and real time graphing. I get update versions of both through Portable Apps but Hacker is also available here: http://processhacker.sourceforge.net/.) vLHC has over 880MB claimed after summing up all supporting processes (Wrapper, Headless, etc) under the Virtual size. All the VM WU's (ATLAS, CMS, LHC, LHCb) seem to require an overhead of another 250MB for wrapper, headless manager and such but this is more than physical RAM. The processes are laying claim to swap. From experience working on a Windows Vista machine with only 2GB I can get 2 vLHC working. Process Hacker shows 1.9GB of Physical+Swap being used with no swap thrashing. The machine was able to run other things and started thrashing when the Virtual size broke 2.4GB on the Commit charge graph from Process Hacker. I'd replaced Vista's explorer.exe shell with Process Hacker and just launched with run command and stored run history. @Toby Broom Can we get the requirements pinned in the Q&A message board, please? The RAM requirements you list seem less than what I've seen empirically. The Virtual Box headless management adds overhead to what the 630M required by the VM internally. Reserving 880MB when planning how many vLHC WU to run at one time is how I avoided swap thrashing on my machines. BTW, one of the reasons I posted on this particular thread was that ALICE isn't listed in the FAQ located here: http://lhcathome.web.cern.ch/what-specification-each-application which Laurence linked for us in message 27919 ALICE jobs are a complete unknown to me. |
Send message Joined: 27 Sep 08 Posts: 745 Credit: 558,647,676 RAC: 334,134 ![]() ![]() ![]() |
Where would you like in Q&A? I only get ALICE on the dev project not here yet. If I see one come up I can check it. I checked Process Hacker, the number are the same, I didn't include all the other memory types so it was lower number. |
![]() Send message Joined: 5 Nov 15 Posts: 144 Credit: 6,301,268 RAC: 0 ![]() ![]() |
Where would you like in Q&A? I posted your answer in Questions and Answers : Windows Message 28682 thread. Ideally, the forum location to pin client requirements should be Questions and Answers : Getting started, but in most all the BOINC forums I've visited, it seems like majority of users go straight to the Message boards : Number crunching to find their technical answers. So either of those two boards, both for best exposure to incoming eyes? I only get ALICE on the dev project not here yet. If I see one come up I can check it. That'll be nice, thanks.
So after adding up the first headless.exe, the console.exe, the 2nd headless.exe and the 3rd headless.exe in the WU process tree, did you get about 880mb committed? |
Send message Joined: 27 Sep 08 Posts: 745 Credit: 558,647,676 RAC: 334,134 ![]() ![]() ![]() |
I get about 880MB, in the Virtual Size, each one is slightly different. If I add the private bytes and working set, this adds a further 181MB bringing total to 1061MB I would say that 900±100MB |
©2023 CERN