doveman

So perhaps it might be a bad idea to apply these network related hotfixes to a normal, non-Enterprise desktop PC and it would be better just to apply the individual boot-related hotfixes? Have any of the individual hotfixes been included/superseded by Windows Update yet as they're quite old and I'm wary of applying them in case they undo something good that a newer hotfix has done (i.e. make things worse)?

xbootmgr -trace boot -traceFlags BASE+CSWITCH+DRIVERS+POWER -resultPath C:\TEMP did cause my PC to BSOD on boot and rebooting it gave me the options of Startup Repair or Boot Normally and the latter caused it to show the Windows boot animation and then reboot again. The next time, I pressed F8 and selected Last Known Good Configuration and it booted OK. I understand Windows creates this backup on each successful boot, so it would be identical to the previous working boot and won't result in any loss of settings, so this is probably a better, quicker option than using System Restore. When using the xbootmgr command, where does it set it and is it wiped on the next successful boot, so that it will no longer be in effect if I reboot now? Otherwise I'd better remove it manually. Using the Last Known Good boot, xbootmgr did still run once Windows had loaded but then gave an error once the countdown had finished, no doubt because whatever it had set was not included in the Last Known Good configuration.

I've installed Defraggler, which replaces the Windows defrag tool (it doesn't have to but I selected that option at install and it shows under Advanced, so I guess it can be disabled after install) and set the Windows defrag service to disabled, to avoid any clashes. Do you know if this method will still work, using Defraggler instead of the MS tool, or will I need to disable Defraggler and restore the MS tool first?

Don't know what that means... but I reiterate that (probably) even a RAMdisk is paged (it IS in RAM, you know) or the Applications themselves COULD be facilitating paging via the API, in either case you'll be PAGE'd "somewhere". Heck, putting a VHD in a RAMdisk and using it instead (double-disks?) may do the same.It means that it allows you to to put the pagefile on the RAMDisk I think, basically that it loads early enough in the boot process to allow this That would only make things worse. The VM, unless running from a RAMDisk, would be loading from HDD and so it would be no better than loading from HDD under Win 7 x64 (probably slower in fact) and the emulated graphics in a VM wouldn't be able to cope with the games anyway.

jaclaz No but I wasn't sure I should start creating user permissions where they don't already exist, in case there's some security risk from giving Everyone Full Control to keys (I presume there's a reason why the entire registry doesn't just already give Everyone Full Control).

Yep sorry, typo there. The CU key exists up to Common. The LM key ends at Office, with an Outlook subkey/folder. Checking permissions on the latter (right-clicking on Office), it shows: Creator Owner - Special SYSTEM - Full Control + Read Administrators - Full Control + Read Users - Read The user I'm operating with is in the Administrators group. Checking permissions on the CU key (right-clicking on Common): RESTRICTED - Read SYSTEM - Full Control + Read Ste - Full Control + Read Administrators - Full Control + Read Ste is the administrator user I'm operating with.

With the ArmA2OA 8GB RAMdisk loaded and ArmA2 running, Resource Monitor shows 11851MB In Use 875MB Modified 1720MB Standby 1884MB Free Available 3602MB Cached 2592MB Total 16348MB Installed 16384MB and Process Hacker shows that the pagefile usage is 1.84GB. So clearly Windows is paging something out unnecessarily. It may not be ArmA, but instead some other stuff that Windows think is less important but nonetheless it doesn't appear to be needed. Shortly after typing the above though, I alt-tabbed back to Arma2OA and it took ages to unpause so it seems likely it had paged some of the game files out. Shortly after that, I got a Windows warning saying that I was low on memory and asking me to close Arma2OA. Once I closed it, pagefile usage dropped back down to 564MB.

As I say, Windows Resource Monitor under Disk shows which processes (Image) are using the pagefile (currently only System). ProcessHacker is useful for showing the total usage (currently 422MB and increasing), thanks.

Even now, whilst I'm not playing any games and have about 12GB available, I can see in Resource Monitor that's paging some of my open tabs in Iron as well as System, Jitsi, Game Booster Tray, itype (some of those are going in and out of the pagefile as I watch) so it does seem to be paging unnecessarily. It's obviously frustrating when I have plenty of RAM and it still takes a while when I change tab in Iron because it's decided to page it out (at least I'm assuming that's what's causing the delay, rather than Iron itself paging out the non-current tab contents and having to load them in again when that tab is switched to).

OK. As I said though, the highest spike was 2639us. I was defragging at the time and obviously don't care about latency whilst doing that but if HDD activity can cause such spikes, I imagine they could occur whilst a game is loading files from the HDD as well, which would be a problem.

I've recently done a fresh install of Windows 7 x64 Ultimate and am going through installing the software. I've got to MS Word Viewer and it's throwing up error 1406 repeatedly, saying it can't write to Software\Microsoft\11.0\Common\ That key does exist under Current_User but the only thing under it is a key "PersonalMenu", none of the things that the installer is trying to write (LanguageResources, General, etc). I found this http://support.microsoft.com/kb/838687t but I don't even have a C:\Users\All Users\Microsoft\Office let alone anything below it.

Yeah, maybe. ArmA and DCS are both pigs to play on any system. I've got a Phenom II X4 955@3.5Ghz, 16GB RAM, HD6950 2GB, 2TB Samsung and 500GB Samsung, so by no means a super-computer and it seems ArmA at least much prefers Intel CPUs. Nonetheless, it's normally just about playable (varies from 10fps up to 50fps depending on the amount of AI in use) and the stuttering I noticed when using the RAMDisk was something new. As the author of SoftPerfect has told me that it somehow prevents the RAMDisk itself being paged out, I assume Windows has some mechanisms to facilitate that. If so, it's a shame that games can't use the same mechanism but I guess they have to allow for it, for systems where there might not be enough physical RAM free (although as ArmA can only use 2GB and I've never seen DCS use much more than that, despite being 64-bit, I'd imagine most systems these days have that spare). I need to use my dual-boot to setup a clean copy of Windows just for ArmA and DCS. Mind you, I think I've done that before and tend to end up installing everything else on top as well, so I need to try and plan a bit better next time

You are seemingly not familiar with "directional hmmmming" , it was aimed @submix8c . jaclaz Hmmm (that was a general contemplative hmmm not aimed at anyone in particular), I'll blame my PC's dodgy positional audio for that one

Hmmm. jaclaz Which part of that 6 month old thread are you Hmmming about?

I'm using SoftPerfect RAMDisk, which testing showed to be somewhat faster than IMDisk. I've checked and the author assures me that it has measures to prevent the RAMDisk itself being paged out but of course once the game files load from the RAMDisk into RAM then I guess it can't prevent Windows swapping them out if it feels like it. It does seem that Windows gets very upset if you disable the pagefile completely, whether it should strictly be needed or not.

I'm pretty sure it is and it only seems to happen when I've been using the RAMDisk and the way the game starts stuttering and takes ages to close, followed by HDD thrashing clearly points to the game files being swapped out to the pagefile. I've actually got the debugging information set to None, which I presume can only reduce memory usage. I've set the pagefile to a fixed 2GB now though, so I'll see if it being smaller reduces how much Windows tries to use it.

I'm running Win7 x64 Ultimate and have 16GB RAM. Left as System Managed, Windows creates a 16GB pagefile (not a problem in itself). However, it seems to page stuff out when there's plenty of free RAM, which obviously slows things down as it has to page it back into RAM to use it. I've been trying to use a RAMDisk to run some games from (ArmA, DCS World). I used a 8GB RAMDisk and had about 13.5GB free before mounting and so about 5.5GB after. ArmA itself can only use a max of 2GB, so there's still 3GB+ free when it's running but even so, it seems to page stuff out, even ArmA itself perhaps, as it's starts stuttering after a while and takes a long while to close and causes HDD thrashing for a while after, so it's obviously using the pagefile. I tried disabling the pagefile completely but then I got error popups when playing about being low on RAM, even though there was over 2GB free. I've got DisablePagingExecutive set to 1, LargeSystemCache set to 0 and NtfsMemoryUsage is set to 0. Is there anything else I can do to force Windows to use the actual RAM rather than the pagefile when at all possible?

the DPC issues come from the sptd.sys driver (Demon Tools) Thanks. I'll uninstall the sptd driver and use Alcohol 52% instead. I note the driver tab of LatencyMon doesn't show sptd.sys very high up the list, so I guess that's not the useful for identifying the problem. Regarding the Explorer usage, do you know what was triggering the SearchFolder function as I wasn't actively searching in Explorer at the time? EDIT: Also, is it normal that latency is high when the HDD is busy, such as when defragging? I'm doing that at the moment and whereas normally it hovers around 19-30us but does spike to 100-200, whilst defragging it's minimum is about 160us going up to about 350us (the highest spike in the 3minutes I've been measuring was 2639us though).

I seem to have "acquired" a DPC latency issue again. I'm also hearing a lot of random HDD activity for extended periods so maybe that's a clue. At the moment, it seems to be explorer.exe causing it, as that's using 10-15% CPU whilst everything else is idle. Defrag isn't scheduled and nor are there any other scheduled tasks I can see that would cause this disk activity. We previously fixed the DPC problem by plugging my mouse into the motherboard and not a hub but it's still plugged into the motherboard so shouldn't be causing any issues. https://docs.google.com/file/d/0B1fDI89phEESeTRDZ3hyektqTWM/edit?usp=sharing

Strangely, after installing the Net Framework 4.5 and the Performance Toolkit, it didn't automatically add the path so I had to add "C:\Program Files (x86)\Windows Kits\8.0\Windows Performance Toolkit" manually so that I could run xperf without having to go into that directory each time. Then I got this C:\Windows>xperf -on latency -stackwalk profile -buffersize 1024 -MaxFile 256 -FileMode Circular && timeout -1 && xperf -d DPC_Interrupt.etl xperf: warning: This system is not fully configured for x64 stack tracing. Please modify the registry under: HKLM\System\CurrentControlSet\Control\Session Manager\Memory Management and set the value: DisablePagingExecutive (REG_DWORD) = 1 Then reboot before retrying tracing. Note: Tracing has been enabled, this is just a warning. xperf: error: NT Kernel Logger: Cannot create a file when that file already exists. (0xb7). I don't recall either of these problems before but maybe I already had DisablePagingExecutive set then. I fixed the "cannot create file" with "xperf -stop" although I'm not sure why it happened when this was the first time I'd run xperf since installing it. I got it to start after that but then when I pressed a key to make the file I got Press any key to continue ... xperf: error: NT Kernel Logger: The instance name passed was not recognized as valid by a WMI data provider. (0x1069). but then I tried again and it's worked on several runs now so it looks like that was just a first time glitch.

Hmm, what about this then? The latency's still under 1000 but isn't the high DPC execution time of storport.sys a problem?

OK cool, I thought I recalled reading somewhere that it should ideally stay around 20-30 but nevermind

remove the driver with the installer: http://www.duplexsecure.com/downloads/ Thanks, hopefully I don't need it for anything The nusb3xhc.sys (Reneas USB3 Host Controller Driver) also seems to be a problem, causing latency up to 500us and execution time of 0.5 but I think that might only happen when actually transferring files to/from a USB3 stick plugged in, so I can live with that if so. EDIT: Whilst it hovers around 20-40 a lot of the time now, it still spikes to 100 and even 400-500 quite regularly. Note that storport.sys only appeared at the top of the list after I launched Iron, which has about 30 tabs open in it.

I thought I had this licked but monitoring today I got this: Clearly sptd.sys is the problem. Is this a known issue and is there a fix for it?

I can't think why that would have changed though, just because he's moved house The only thing I can imagine is that perhaps he had the monitor connected to the onboard graphics (which is disabled) rather than the card when he first tried and therefore Windows couldn't see a monitor and reset the refresh rate to something that his monitor doesn't suppor,t although I don't know if it does that or if it should automatically go back to the refresh rate that was being used before once the monitor is reconnected. The proper resolution for his monitor is 1024x768@60hz.