- Jun 30, 2004
- 16,340
- 1,891
- 126
I've worked with all sorts of memory modules going back to "EDO" and back further to hand-socketing Toshiba RAM chips into a Tall-Tree Systems JRAM-3 ISA card (1985).
I've never had the troubles I'm about to describe.
The system I call "Chrome Lightning" has:
C2Q Q6600 B3 stepping
Striker Extreme motherboard BIOS v 1303
2GB 2x1GB Crucial Ballistix DDR2-1000
BFG nVidia 8800 GTS 640mb
3Ware RAID5 9560SE PCI-E x4 controller plus four Seagate 7200.10 320GB drives (and 9/10 of a terabyte)
The PSU is Seasonic 700W M12
The Crucials have run beautifully. I can tighten the latencies down to pre-DDR2 "high-performance" levels while running the modules 1:1 above 720 Mhz (2 x 360, and the processor 2.4 @ 3.2+ Ghz).
Now the story begins. I had trouble with the Media Center extension in XP MCE 2005. The ehrecvr.exe program that manages the TV tuner-capture was grabbing almost 1GB of memory on its own. But the problem wasn't the ehrecvr.exe -- it was the tuner-capture driver itself. Everything was fine after I reinstalled the driver, but not before I decided that I needed more RAM and ordered a 1GB 2x512 kit of the same Crucial Ballistix RAM. I should have decided right away that nearly 1GB for that exe file was too much and not normal, but . . . . what was I to expect or to know?
This was going to give me 3GB of RAM. I wasn't in a hurry to add the extra gig, though, because now I could see I didn't need it. But finally, and since I couldn't just send the kit back to the reseller and ask for a refund, I decided to go ahead with it.
Although I should've done so, I didn't CLR-CMOS. Instead, I just changed all the settings back to default in the BIOS.
After adding the memory, the system will not complete posting. Several resets and I was able to enter BIOS, but I couldn't get it to complete system post.
I tried swapping the 512mb modules between the channel A2 and channel B2 slots. No cigar.
During this time, I noticed that the module locked into place for slot B2 felt "wobbly." No matter what I did, it seemed looser than the rest. I swapped the 512mb modules between slot A2 and B2 -- and slot B2 memory still felt "wobbly." And I couldn't get the system to post. The latches for the wobbly slot seemed completely closed and locked into place.
I decided to get an RMA# for the modules and replace them. But after I got the RMA#, I also decided to test them in another identical motherboard, and MEMTEST86 ran 26 iterations and another 15 iterations of test #5 -- with no errors. So I cancelled the RMA.
Finally, I decided to try them again. This time, I was careful to CLR-CMOS, and this time -- with the 512mb modules in slots A2 and B2 again -- I was able to get the system to post, but it seemed on reboot that it was a 50-50 random proposition -- sometimes it would post, sometimes not. I played with the modules some more. I got the system to post with the MEMTEST86 disk ready to boot.
MEMTEST86 would come up, run test 0 for about 60 seconds, and then lock up with "funny-character/symbol" corruption on the screen.
Finally, I decided to put the 512mb modules in slots A1 and B1 with the 1GB modules in slots A2 and B2. I notice that the 1GB module in slot B2 gives that comfortably reassuring and gritty "interference fit" feeling as it locks in. The 512mb modules also feel snug in slots A1 and B1.
there would have been no damage to the B2 slot --- no cracks in the plastic or damaged latches -- because those slots were never used before.
So how can a module feel loose in slot B2 and snug in slot A2, and a tightly-fitting 1GB module that was snug in B1 now fits snugly in B2?
Also -- and this may have something to do with the 3GB of memory, since MEMTEST86 website suggests that "the program sometimes has trouble 'finding' all the memory" -- on tests 0 and 1, the tests seem to complete maybe a third of their run before going on to the subsequent test, but those tests show no errors. So far, MEMTEST86 is showing completion of 7 iterations of all 9 or 10 tests without error.
Any opinions or insights out there in Anandtech-geek-land? I don't want to miss something on these MEMTEST86 runs and then find that I've got data corruption or memory malfunction somewhere down the road.
I've never had the troubles I'm about to describe.
The system I call "Chrome Lightning" has:
C2Q Q6600 B3 stepping
Striker Extreme motherboard BIOS v 1303
2GB 2x1GB Crucial Ballistix DDR2-1000
BFG nVidia 8800 GTS 640mb
3Ware RAID5 9560SE PCI-E x4 controller plus four Seagate 7200.10 320GB drives (and 9/10 of a terabyte)
The PSU is Seasonic 700W M12
The Crucials have run beautifully. I can tighten the latencies down to pre-DDR2 "high-performance" levels while running the modules 1:1 above 720 Mhz (2 x 360, and the processor 2.4 @ 3.2+ Ghz).
Now the story begins. I had trouble with the Media Center extension in XP MCE 2005. The ehrecvr.exe program that manages the TV tuner-capture was grabbing almost 1GB of memory on its own. But the problem wasn't the ehrecvr.exe -- it was the tuner-capture driver itself. Everything was fine after I reinstalled the driver, but not before I decided that I needed more RAM and ordered a 1GB 2x512 kit of the same Crucial Ballistix RAM. I should have decided right away that nearly 1GB for that exe file was too much and not normal, but . . . . what was I to expect or to know?
This was going to give me 3GB of RAM. I wasn't in a hurry to add the extra gig, though, because now I could see I didn't need it. But finally, and since I couldn't just send the kit back to the reseller and ask for a refund, I decided to go ahead with it.
Although I should've done so, I didn't CLR-CMOS. Instead, I just changed all the settings back to default in the BIOS.
After adding the memory, the system will not complete posting. Several resets and I was able to enter BIOS, but I couldn't get it to complete system post.
I tried swapping the 512mb modules between the channel A2 and channel B2 slots. No cigar.
During this time, I noticed that the module locked into place for slot B2 felt "wobbly." No matter what I did, it seemed looser than the rest. I swapped the 512mb modules between slot A2 and B2 -- and slot B2 memory still felt "wobbly." And I couldn't get the system to post. The latches for the wobbly slot seemed completely closed and locked into place.
I decided to get an RMA# for the modules and replace them. But after I got the RMA#, I also decided to test them in another identical motherboard, and MEMTEST86 ran 26 iterations and another 15 iterations of test #5 -- with no errors. So I cancelled the RMA.
Finally, I decided to try them again. This time, I was careful to CLR-CMOS, and this time -- with the 512mb modules in slots A2 and B2 again -- I was able to get the system to post, but it seemed on reboot that it was a 50-50 random proposition -- sometimes it would post, sometimes not. I played with the modules some more. I got the system to post with the MEMTEST86 disk ready to boot.
MEMTEST86 would come up, run test 0 for about 60 seconds, and then lock up with "funny-character/symbol" corruption on the screen.
Finally, I decided to put the 512mb modules in slots A1 and B1 with the 1GB modules in slots A2 and B2. I notice that the 1GB module in slot B2 gives that comfortably reassuring and gritty "interference fit" feeling as it locks in. The 512mb modules also feel snug in slots A1 and B1.
there would have been no damage to the B2 slot --- no cracks in the plastic or damaged latches -- because those slots were never used before.
So how can a module feel loose in slot B2 and snug in slot A2, and a tightly-fitting 1GB module that was snug in B1 now fits snugly in B2?
Also -- and this may have something to do with the 3GB of memory, since MEMTEST86 website suggests that "the program sometimes has trouble 'finding' all the memory" -- on tests 0 and 1, the tests seem to complete maybe a third of their run before going on to the subsequent test, but those tests show no errors. So far, MEMTEST86 is showing completion of 7 iterations of all 9 or 10 tests without error.
Any opinions or insights out there in Anandtech-geek-land? I don't want to miss something on these MEMTEST86 runs and then find that I've got data corruption or memory malfunction somewhere down the road.
Facebook
Twitter