paul
Granpa Paul
Premium,MVM,ExMod 2006
join:2001-07-14
clubs:  
| memtest86 conditioning
After ramping up a new 8RDA+ with some XMS pc3200c2, I hit a wall pretty quick. Never really had a problem before, with older boards I would have some peripheral hold me back anyway.... with pci and agp lock, nothing but the ram and chipset holds me back, so I decided to do some semi-serious burnin with memtest86 on the stick.
After getting these tips from libertate, I started some serious burnin... (edited slightly for clarification) The following is a little guide to loop test 5 in all memory ranges.
said by libertate  :
when memtest86 boots up,
hit C for configuration..
then 2
then 5 for test selection
then 5 for test 5
then hit enter and it'll say somewhere on the screen "Test 5"
once the specific test is being looped, we can tell it to cover the entire range of memory. (we're still in the configuration screen)
3 for address range or memory used
select 3 for entire address range
hit enter and it'll come back to the test
Just leave it running through test 5 (it automatically restarts the test after it's finished for as long as you can spare the time).
Thought I'd start this thread for anyone who'd like to bookmark it for future use. Basically once memtest86 starts, hit c,2,5,5,enter,3,3,enter, and you'll be looping test 5 in all the memory range.
If you are getting errors, back off the front side bus until it shows no errors for the first 10 loops or so. Run it for a few hours, restart, bump the fsb up 1, run the burn in again. Keep going until it you hit a wall, you can usually get 10 mhz fsb in a day, but once you start getting errors, you may have to leave it burnin overnight at that setting before moving up to the next step.
Anyone with other thoughts or procedures, please post so we'll have a good memory burnin thread for reference.
Paul
--
looking for little green men....
Team Starfire Queue VI
[text was edited by author 2003-05-06 21:06:10] |
|
tencz51
Kats The Man
Premium
join:2000-07-06
Ocala, FL
|  well done  maybe this script for memtest86 should go in the overclockers guide/help section ??
--
Hail 2 the Chief ! support our Troops ,right or wrong thats the call !! |
|
noe1
The Future Was Not What It Use To Be
Premium
join:2002-09-10
Charleston, WV
| reply to paul
Q: Hypothetical. Okay, is it more effective to run memtest86 at highest fsb default multi or highest fsb stable at lower multi?
Like with a 2100+, run it at 195x12.5 or 220x11? Or, put another way, run it at max clockspeed or at max Sandrs mem bandwidth? Assuming both are stable in windoz.
--
The Wino and I Know!  |
|
TOTC
I only exist to make you cry
Premium
join:2001-05-16
Van Nuys, CA
 ·Pacific Bell - SBC
| reply to paul
Ohh man I need floppies 
I have a couple good BIOS's sitting on some floppies.
Now I need a perm memtest86 floppy.
--
My AV only expresses "The way nature intended." *Disclaimer* Not my fault for your imagination going somewhere else |
|
sjohnson
join:2001-02-28
Sioux Falls, SD | reply to noe1
Both... |
|
jsimmons
Premium,MVM
join:2000-04-24
Falls Church, VA
| reply to TOTC
I used the memtest86 iso and burned a bootable memtest CD. Man it boots into memtest so much faster.
--
"Everything should be made as simple as possible, but not one bit simpler."- Albert Einstein |
|
Exit
Premium,ExMod 2002
join:2001-04-10
Canada
·TekSavvy Solutions..
| reply to paul
Quick question. Why only loop test 5? I don't get any errors on test 5 mine all start on tehst 4 or 6. I ususally loop memtest while I sleep every night.
I run mine at max fsb with a low multi just because I like to keep the strain down on my cpu. That being no high clocks or volts when burningin my ram
--
If we are what we eat, then i am fast, cheap and easy!!! |
|
sjohnson
join:2001-02-28
Sioux Falls, SD
·Golden West
| reply to paul
I think Paul's recommending test 5 because it's like Sandra mem benching - it helps to condition memory for higher bandwidth.
You really need to read up on the test descriptions on the memtest web site, then use that knowledge to select the test or tests that help you achieve what you want.
--
Don't confuse me with the facts... |
|
TOTC
I only exist to make you cry
Premium
join:2001-05-16
Van Nuys, CA
·Pacific Bell - SBC
| reply to paul
I usually only get the errors on Test #5 so I will follow the instructions indicated
My Corsair loves to get errors @ 225+ in 255MB~256MB sections of the RAM.
--
My AV only expresses "The way nature intended." *Disclaimer* Not my fault for your imagination going somewhere else |
|
Exit
Premium,ExMod 2002
join:2001-04-10
Canada | reply to paul
Use what ever works for you I guess. My old stick used to get errors only in test 5 but the new stick gets other ones
Go figure
Steve
--
If we are what we eat, then i am fast, cheap and easy!!! |
|
noe1
The Future Was Not What It Use To Be
Premium
join:2002-09-10
Charleston, WV
| reply to sjohnson
said by sjohnson :
....You really need to read up on the test descriptions on the memtest web site, then use that knowledge to select the test or tests that help you achieve what you want.
When I have time to read the durn website when you guys got me runnin' memtest all the doggone time??  :D
--
The Wino and I Know! |
|
The WeaseL
Premium
join:2001-12-03
Minnesota | reply to paul
I run all the tests...Seems to work well, gets a good blend in...although running just test 5 now and again would be good variety.
--
How lucky am I to have known someone who is so hard to say good-bye to. |
|
NutCase4
Mad Overclocker
Premium
join:2001-12-10
Livermore, CA
| reply to paul
Test 5 is a good test to run for Burnin and conditioning the system as it moves large chunks of Data around. It also uses the Cpu a lot so it can condition it also.
But this test is useless for finding a problem. It also has a high tendency to fail on AMD athlon's even though everything is good.
Tests 0-4 are better at finding a problem in the memory. |
|
noe1
The Future Was Not What It Use To Be
Premium
join:2002-09-10
Charleston, WV
| reply to paul
Okay master guru know-it-alls of this forum , I've run Prime96 and f@h here for several hours with no problem or error showin up. I run test 4 and 5 in memtest at same settings and get multiple errors in each test. What say ye?
2100+ @ 1.95 vcore, vdimm = 3.0
213 fsb x 10.5
[text was edited by author 2003-05-07 19:06:19] |
|
paul
Granpa Paul
Premium,MVM,ExMod 2006
join:2001-07-14
clubs:
| All I can relay is my own experiences. In an AMD setup, the box ran fine, but had errors like you have. I backed off the fsb a few mhz until it ran memtest86 no errors, then looped test 5 in all memory ranges for an hour or so, bumped it up a notch, repeat. Errors went away by conditioning....
Paul
--
looking for little green men.... Team Starfire Queue VI |
|
noe1
The Future Was Not What It Use To Be
Premium
join:2002-09-10
Charleston, WV | Thans Paul, that's good enough for me.
--
The Wino and I Know! |
|
iamthehunter
Premium
join:2002-03-01
Pelham, AL | reply to paul
timely information....nf7-s, xp2100 tbredB, corsair twinx1024-3200LL, slk900u, shipped today from newegg and excaliberpc. will be burning in and oc'ing this weekend:)
--
damn im old..... |
|
Exit
Premium,ExMod 2002
join:2001-04-10
Canada
·TekSavvy Solutions..
| reply to noe1
said by noe1 :
Okay master guru know-it-alls of this forum , I've run Prime96 and f@h here for several hours with no problem or error showin up. I run test 4 and 5 in memtest at same settings and get multiple errors in each test. What say ye?
2100+ @ 1.95 vcore, vdimm = 3.0
213 fsb x 10.5
[text was edited by author 2003-05-07 19:06:19]
I found that if prime 95 is stable then so is your ram. Since it does math calculations it is a really good way to check for errors
Steve
--
If we are what we eat, then i am fast, cheap and easy!!! |
|
drew
Reformation
Premium
join:2002-07-10
Port Orchard, WA
clubs: | reply to paul
Nice job man.. I'll be using that guide when it comes time.
Submit the link with a description to the FAQ! I'm sure FE will add it. |
|
TOTC
I only exist to make you cry
Premium
join:2001-05-16
Van Nuys, CA
·Pacific Bell - SBC
| reply to paul
Nothing would help my RAM get out of the efficiency hole.
But this is a good guide for anyone making sure there RAM is all worked out.
--
My AV only expresses "The way nature intended." *Disclaimer* Not my fault for your imagination going somewhere else |
|
