Every once in a while my computer BSOD’s probably due to memory issues. It has 12 gigs and I thought I’d just replace the RAM since it is so cheap - just $75 to eliminate the possibility of a defective stick.
I was surprised at the low cost of RAM and thought I’d order another 12 gigs to bring the system up to its max of 24 gig. Yeah, I know, I may never need that much but what the heck; it is cheap and having a 24-gig box has bragging rights and will surely compensate for my small penis. However, would getting excessive RAM increase the chance of crashes/BSOD’s due to exposing the system to more silicon that might have a defect…kinda like RAID 0 increases the risk since either drive’s failure could bring down the system?? What say ye??
Of course any time you increase the number of components in a system you increase the chance of a defect occurring. But in this case I think the increase is negligible.
But seriously, 24 GB of RAM? I work in a research institution where we need to process huge amounts of data, and even our systems don’t need anywhere near that much memory. For normal desktop use you are not going to see any speed or other performance increases beyond a certain amount of RAM, and that certain amount is well below 24 GB! You are just throwing your money away and wasting resources.
it’s possible, yes. first, as you and psychonaut said, when you add more parts you increase the chances of one of them being defective.
the other thing is that RAM modules are installed in parallel with each other on their channel (“multi-drop” bus.) as you add more modules, you are increasing the electrical load on that channel and if something is non-compliant with specification (or on the edge) then the increased load can cause problems. It’s safe to assume that if your board is from a reputable manufacturer, it was validated with 24 GiB of RAM. However, it was probably not validated with all possible 24 GiB configurations, so a dodgy DIMM can cause problems.
I look at it this way, if you bung all 24 GiB in there, and it passes Memtest86+ run at least overnight, things are probably just fine.
I’m not sure if it is relevant to your situation, but having different brands or even different series in the same brand can wreak absolute havoc with most motherboards out there.
On every machine I build, I look up what the motherboard manufacturer says will work with the board, and stick with matched sets of that. Strangely, I’ve always had issues with Corsair, even though it is generally a premiere brand. Currently, I’ve had no problems with G.Skill sticks.
Boards occasionally have issues with a certain brand or style of ram. We had a very nice system build on a gygabyte board we did a few years back, that for whatever reason would not post with Crucial DDR2-800, DDR2-533 or 667, no problem. Corsair DDR2-800 worked perfectly. It actually was some kind of BIOS issue that was later fixed with an update.
I’ve read, on this board IIRC, that having that much RAM will actually slow down your computer because the processor has to use cycles just to maintain unused RAM. Also, just because your board supports 24 gigs doesn’t mean that it was tested or certified for any actual real-world use.
For the first point, I don’t think that’s true since the system memory controller will provide the refresh to DRAM as a matter of course. it shouldn’t take CPU time at all.
as for the second point, if the board manufacturer says it can take 24 GiB but never tested it with such, then stop buying boards from that manufacturer.
On top of that, certain versions of Windows have memory limits. You’ll need Windows 7 Ultimate, Enterprise, or Professional to see more than 16 GB. See here.
I’ve got an Asus Sabertooth X58 motherboard (using safe defaults) running Win7 Pro. The board has an LED that indicates problematic or improperly-seated memory modules. I swapped the memory modules and slots one at a time to try to discover a defective module or slot. There was no perfect correlation but 1 module tended to fail a lot.
I updated the system BIOS tonight with a version that includes fixes to improve system stability and memory compatibity. The BIOS version that my system had was not available in the history of BIOS’s available on the Asus website so I wonder if that version had problems.
Tomorrow, I will get the G Skill modules to replace the (also Asus-approved) Adata modules. I’ll then run mem tests. Hopefully, the replacement RAM and new BIOS will take care of the problem. I’ll probably just install 1 module (4 gigs) for awhile and, if things go well, increase it to 12 gig while keeping the other three 4-gig sticks as backups or future expansion if needed.
Also, more memory means you can run more concurrent processes. Bad programs or unusual combinations in memory might contribute to instability. It gives the processor more to do managing the memory subsystem and might highlight flaws elsewhere [processor / chipset / hard drive] driving the virtual memory space.
no it does not. Virtual memory abstracts away the actual amount of physical RAM in the system. as far as any given process believes, it’s running on a system with 4 GiB of RAM even if there’s only 512 MiB physical installed. you can run as many processes as the OS will allow regardless of the amount of RAM; the OS will probably sit there thrashing stuff in and out of the pagefile/swap but they’ll run.
I performed the BIOS update and re-ran the tests with the old memory sticks. This time the tests were very conclusive. No memory module failed except for one that failed 5/6 times. The slots appear to be fine.
I went ahead and replaced all 3 memory modules with the new RAM. They all passed in testing.
Bottom line is that their was a bad memory module but it took a new BIOS for the motherboard to conclusively identify the problem.