memtest86+

nomorewindows · 2011-11-03 22:16:07

I was running the current version of memtest86+ (4.2) on my hardware that seems to be dragging itself around. I decided to run memtest figuring that it *MIGHT* be a problem with memory. However, when I start memtest86+ after using it before the L1 and L2 cache speeds are way lower than I remember in the past. Running memtest86+ on another machine in the office with an identical motherboard confirms that I wasn't dreaming this up. It should be 5 digits long for both the L1 and L2, but is only 3. I put back in the original memory and get no improvement in speed, making me think that something has gone weird on the motherboard.

fukawi2 · 2011-11-03 22:21:51

Is there a question here or are you just sharing experience...?

nomorewindows · 2011-11-03 22:49:57

I'm trying to figure out whether the sluggishness I seemed to be experiencing is related to the values is confirmed from something like memtest86+. Memtest states that it reads the values as they are even on an overclocked processor (which this one doesn't have a way to do that). If it really is as slow as it says it is, when I know it used to read higher would seem to confirm that something odd has happened but I'm not sure I have an explanation. I'm thinking that some piece of hardware has degraded in performance and trying to confirm it using diagnostics.

R00KIE · 2011-11-04 11:26:58

Check the cpu heatsink for dust and proper seating.

nomorewindows · 2011-11-04 13:28:17

While I was at it, I did blow out some dust and testing various memory combinations afterwards didn't seem to make the numbers come back to expectations. I put the big 2GB DDR2 stick in (limited by 32/36 bit address boundary and manufacturer even though it is 64-bit processor). I tried the original doubles that came with the computer, and it seemed to speed back up, but got even lower values. One of the matching sticks in one slot would give a different value than putting it in the other slot by itself (2 slots). And then the higher of the two would result from putting the doubles in. The highest value resulted with the single 2GB. It just started getting sluggish the other day with a noticeable screen smear commonly associated with being overburdened. Some applications just refused to respond.

lagagnon · 2011-11-04 16:30:11

I doubt very much that faulty memory, or non-dual channel memory problems would cause "noticeable" sluggishness. Faulty memory is normally associated with complete freezes or reboots. My suggestion is that when you are experiencing your so called "sluggishness" that you do a bit of detective work using the output from the following commands, to see if there is an errant process consuming CPU cycles or RAM: top, ps aux. Also check your CPU temperature when you notice the slowdowns. You might also want to install "mprime" and do a full test to see whether the CPU and motherboard are producing the correct answers.

Last edited by lagagnon (2011-11-04 20:24:17)

nomorewindows · 2011-11-05 00:46:19

I tried an older version of memtest 1.2 and it told me that my cache was operating at the same speed as reported with 4.2.
There was a day where I posted about weird video and a crash that happened, could still be memory, but it took me 8 hours instead of 30 minutes to an hour to test my memory on the first pass with memtest. I think my hardware has bit the dust, but L1 cache is on the processor, I could possibly try swapping processors even though that seems a little trivial. My processes are pretty clean, and every thing is sluggish, regardless of operating system. I'll try the mprime after (looks like a boinc similar program?) a while. The cache speeds that I've been getting on Pentium I's are faster than what my machine is doing right now. The only thing I can figure is plugging into the ethernet port the other day could've cooked my board somehow? It was working before I did that. I do check my temperatures quite reguarly.

nomorewindows · 2011-11-05 03:46:17

I got ahold of another motherboard that is just exactly the same and switched processors between boards and found that my board is toast and the other board worked with either processor. Any reason why it might be the processor (a possible placebo maybe if run long enough might have an ability to ruin a good one?) I noticed after putting the other board back together it's numbers slipped ever so slightly, but not near the crawl that mine is. Looks as if my mobo has lost its' mojo.

Last edited by nomorewindows (2011-11-05 03:56:02)

zenlord · 2011-11-05 09:26:39

Last on the list to try is the bios settings. Maybe try resetting it to stock settings?

nomorewindows · 2011-11-05 12:19:11

It's one of these modern BIOS; there's practically nothing to change. Looks like it'll be shade tree mechanic will be changing out mobos. From what I've seen here memtest was an accurate tool for finding/confirming that something was indeed wrong, even though it couldn't tell me exactly what the problem was. The next thing I might be trying is to make sure something like the power supply is putting out the right voltage and not prematurely burning something out.

Arch Linux

#1 2011-11-03 22:16:07

memtest86+

#2 2011-11-03 22:21:51

Re: memtest86+

#3 2011-11-03 22:49:57

Re: memtest86+

#4 2011-11-04 11:26:58

Re: memtest86+

#5 2011-11-04 13:28:17

Re: memtest86+

#6 2011-11-04 16:30:11

Re: memtest86+

#7 2011-11-05 00:46:19

Re: memtest86+

#8 2011-11-05 03:46:17

Re: memtest86+

#9 2011-11-05 09:26:39

Re: memtest86+

#10 2011-11-05 12:19:11

Re: memtest86+

Board footer