Forum Home
Press F1
Thread ID: 79557	2007-05-24 02:37:00	memory Testsers and Reliability	pctek (84)	Press F1

Post ID	Timestamp	Content	User
552610	2007-05-24 02:37:00	. firingsquad . com/hardware/windows_vista_x64_troubleshooting/page8 . asp" target="_blank">www . firingsquad . com I tried reinstalling Vista x64 and I was now getting blue screen of death errors on this MSI board! At this time, the only thing I hadn’t tested was the memory . I ran Memtest 86 and Microsoft’s memory tester . It was perfect, there were no errors . We returned all of the AMD products and motherboards to Fry’s . I returned home and built the system . But on the second day, the BSOD’s returned . Once, there was a weird STAGE ZERO BSOD . We re-ran Memtest 86 and Microsoft’s Memory Tester . Again no issues . So, we replaced the HDD, replaced the DVD burner, and even replaced the physical Windows Vista disc just in case . Vista x64 did install, but we were back to square one with intermittent and seemingly unpredictable system crashes . We tried the system with only one module installed . It worked . We tried the other module, it crashed . We repeated this several times to confirm our findings . All along, our memory had been bad . I tried another Memtest 86 and Microsoft Memory Test . It still passed .	pctek (84)
552611	2007-05-24 04:34:00	Unfourtunatley, i think you've answered your own question. Vista x64:horrified	wratterus (105)
552612	2007-05-24 05:05:00	been a while, but last time I used a memory tester for a fault ( memtest I think ) there was something like "press X for extended tests" on the screen. My RAM passed the standard 6 tests fine, no matter how long I ran it, but failed half way through test 7 in the extended tests. Never rely on the standard tests, if you suspect faulty RAM always use the extended test and run through several times, some people leave it overnight.	dugimodo (138)
552613	2007-05-24 05:23:00	Memory testing is incredibly difficult . All a test can do is say: "I haven't found any", or "I have found errors" . You can usually assume that when a memory test finds errors that it has actually found bad memery . When it can't find errors, there might still be faulty memory under certain extremely rare bit patterns . When HP's main business was test equipment they made some minicomputers . Thne HP2000 range were produced using the latest technology, including 4kx1 dynamic memory chips which they used on the memory boards . They discovered that these chips could read bits wrongly wihen there were particular bit patterns, as well as when cosmic rays struck . They finished up by designing the boards to have error detection and correction, using 21 bits for each 16 bit word . They also logged every error . When the maintenance engineer came round, he would replace any chips which were having "too many" errors . All the chips were socketed, and the sensible argument was that the boards would become more reliable over time . One of the computers I worked on had a 4kx12 core memory board out of use for quite a while waiting for replacement driver chips to arrive . When they came, they were installed and the mainenance guy ran the memory diagnostics for a few hours . No problems . Except that when I tried to run the "real" work programmes, they continued to crash when they used that memory area . There was some odd timing difference which would stop it working . Those were small memory devices . . . imagine the scope for pattern sensitivity in chips with megabits . The hope is that memory tests will cover enough cases to catch frequently (and consistently) occuring errors . Usually that's enough . But no test can prove perfection; it can only report faults it can find . In a case like yours, I would be inclined to test the memory when its temperature is higher than normal . It's quite reasonable to expect errors to increase (or start occuring) when the chips get hotter . I'd expect Vista to work the CPU harder than a memory tester, so the whole box will be warmer . Perhaps the memory tests should include a test which warms up the CPU, then test the memory .	Graham L (2)
552614	2007-05-24 11:29:00	well it sounds like vista found a faulty ram stick, but have you tried the stick in another machine yet? or tried running xp to see if it isn't just vista. it could be the motherboard having compatibility troubles, check the manurfacturer's website if you haven't already	motorbyclist (188)
552615	2007-05-24 22:05:00	It wasn't me. It was the guys on that forum. I have used memtest but only on RAM that appeared to be fine. (Which it was). I've found if you do have faulty RAM you don't need to run memtest because the PC is either not booting at all, beeping or giving BSODs. I just thought that info might be of interest. Whether or not its because of Vista I don't know.	pctek (84)
552616	2007-05-24 22:15:00	Maybe it's a 64 bit thing. The Memtest86 site has the following as a planned project "Native 64 bit support with 64 patterns. With 64 CPUs becoming common a native 64 version of Memtest86 is needed. Test algorithms need to be updated to use 64 and 128 bit patterns. A simple flat memory model can now be used that will eliminate a great deal of complexity and add efficiency." Memetest86+ has something similar.	PaulD (232)
1