That's certainly one possible cause, though that could lead me to suspect a driver bug. But it's also quite possible for the system to get overcommitted in other ways through misbehaving applications or just something with a legitimately enormous active working set. In the Win 7 resource manager you can try sorting by Working Set and by Private Bytes to see if something stands out. You may find that nothing does, however, because Resource Monitor doesn't show everything. For example, you could have a misbehaving Service: services are listed on the CPU tab, but not on the memory tab, so it's difficult to monitor their memory usage. To really track down the problem (assuming it continues to happen) you'll need to get something like Process Explorer (Resource Monitor is really a prettified and simplified version of it).just brew it! wrote:Sounds to me like you just got "had" by a workload that caused Windows to chew up lots of physical RAM for I/O buffers, indirectly resulting in a bit of pagefile thrashing on the processes it stole those pages from. Windows generally does a pretty good job of managing its physical RAM, but it isn't perfect.
I.S.T. wrote:Hard Faults is what the Resource Manager on Vista was telling me. *shrug*
P5-133XL wrote:I.S.T wrote:Every time my system went wacky, it was downloading.
Why look for the complex, when you supplied a likely simple answer.
Well, it makes sense: since Windows isn't flushing pages from memory right away (unless the system is already overcommitted) and in fact is pre-fetching pages in many cases, it's not uncommon for the kernel to discover that the page it needs to fault into memory is in fact already in memory, just not mapped appropriately. So that's a "soft" fault and it can just do a few fix-ups in the page tables to make everything work without hitting the disk at all; faults where it does actually have to go to the disk thus become "hard" faults.just brew it! wrote:Sorry for the confusion, I mostly stopped using Windows around the time Vista came out; apparently that's what Microsoft calls page faults these days.
Yes, but the BIOS has to be set to remap the memory at the top of the 4GB space that would otherwise be shadowed by PCI devices, which is what I was asking: some motherboards of that vintage were set to do it automatically while others were not (and some didn't even offer the option).I.S.T. wrote:Yes, I see all the RAM. I am running a x64 OS(I chose it on purpose to future proof my ass)
UberGerbil wrote:Yes, but the BIOS has to be set to remap the memory at the top of the 4GB space that would otherwise be shadowed by PCI devices, which is what I was asking: some motherboards of that vintage were set to do it automatically while others were not (and some didn't even offer the option).I.S.T. wrote:Yes, I see all the RAM. I am running a x64 OS(I chose it on purpose to future proof my ass)
Well, I'm not sure about that but clearly there's something amiss. I suspect that the combination of x64 and 4GB is kind of a worst case situation, since 4GB is pretty much the minimal size for anyone who actually installs x64. And 64bit enables (indeed encourages) taking larger bites of the apple, which is a problem if you're dealing with a limited fruit. I wouldn't be surprised if Steam stabilizes at some working set that doesn't cause a problem on a larger system, though clearly this is a configuration that should be in their test environment (and accounted for in their design).I.S.T. wrote:Valve needs to fire some people.
I.S.T. wrote:So, Steam updated tonight, and one of the notes was lowered disc writing when downloading games using the new content system.
Users browsing this forum: No registered users and 1 guest