PDA

View Full Version : RAM Testers!!!



Punisher!
05-14-2006, 09:03 PM
What would you use to test your RAM and why or, in other words, what do you use to test which dealing with your RAM?

(Why OPB says memtest is not a reliable tester? EVEN if it reports errors.)

I think best is still Prime95 Blend test, in fact I can't still make my UCCC pass it for more than few houres LOL!

Micutzu
05-14-2006, 09:22 PM
HCI Memtest (http://hcidesign.com/memtest/).

mtl_hed
05-14-2006, 09:26 PM
much better: http://www.memtest86.com/

Its a linux live CD, burn it.

boot from it, let it go.

Eldonko
05-14-2006, 09:26 PM
I do dos memtest, 32M, prime blend, 3d in that order.

shadowing
05-14-2006, 09:33 PM
The best way to test memory is simple. DOS Memtest and Memtest for Windows.

Micutzu
05-14-2006, 09:39 PM
Using both is redundant ... Windows Memtest can spot errors when Memtest86 doesn't, but not the other way around.

Zenjirou
05-14-2006, 10:11 PM
dos memetest got 315 no errors

fails at 300 in windows =_=

windows is alot more stress

CapFTP
05-14-2006, 10:53 PM
What would you use to test your RAM and why or, in other words, what do you use to test which dealing with your RAM?

(Why OPB says memtest is not a reliable tester? EVEN if it reports errors.)

I think best is still Prime95 Blend test, in fact I can't still make my UCCC pass it for more than few houres LOL!

Puni....you know..:)

memtest for Windows (DOS version can just give u an approximate idea)...and spi32/3dmark as well.

Mostly 3dmark for 2x1gb kit if equipped with 1st batches of infineon I should say

thunderstruck!
05-15-2006, 11:30 AM
Memtest 86+, Prime blend, 32M, 3d (if infineon). I think using prime blend makes it easier to spot errors. After all is done, I do a final 32M calculation for show.:)

lawrywild
05-15-2006, 11:50 AM
1000% windows memtest for me

ozzimark
05-15-2006, 12:02 PM
Mostly 3dmark for 2x1gb kit if equipped with 1st batches of infineon I should say
any infineon/aeneon CE really.. it affects all the ic's, not just the few couple runs :(

hot_fifty
05-15-2006, 12:19 PM
1000% windows memtest for me

me too m8.. before I used memstes x86 in bios / dos, but I think win memtest is more reliable

ken
05-15-2006, 12:27 PM
1000% windows memtest for me

^

I dont trust the DOS memtest86+ anymore :banana:

Repoman
05-15-2006, 12:53 PM
If my infineon is stable for ~3 runs of 3dmark01 I know it's good, no need for pi 32m

Adam217
05-15-2006, 01:37 PM
Memtest86 is a nice starting point to help from corrupting your OS.

ozzimark
05-15-2006, 02:13 PM
Memtest86 is a good app to find the limits of your RAM in a hurry. I usually find the difference between Windows stable and Memtest86 stable to be about 10 to 15Mhz. Windows can be a real *$&#%@!
depends on the ram really. i've only had that problem with TCCD and bh-style ic's. micron tends to be dead on, no errors in memtest = rock solid in windows.

guess2098
05-15-2006, 02:43 PM
1ooo~5ooo% Memtest Windows Is Good Enough

corvus_corax
05-15-2006, 06:38 PM
Memtest for Windows and RightMark RAMTester. And of course SPi32M and Prime Large

Jonas
05-16-2006, 07:38 AM
looping high detail lobby :)

Dragon69
05-16-2006, 09:47 AM
I also use memtest and Prime Blend and OCCT Torture Test and Goldmemory after this i start with 3D Marks

HousERaT
05-16-2006, 09:58 AM
Windows memtest, superpi 32mb, prime95 (or sp2004) and OCCT. All of these can give you a very good indication about memory stability. Once you think your system is stable run a 3D Benchmark to add further support for system stability.

Right now the order I use when testing memory is superpi 32mb ------> windows memtest ----------> OCCT. If the memory survives all of that then I'll run 3DMark01SE, then 03 then 05 then 06. If I can get through all of that then I'm satisfied.

farksy
05-17-2006, 04:38 AM
dos memetest got 315 no errors

fails at 300 in windows =_=

windows is alot more stress

It's because of your memory controller.

Memtest for me as it's prebuilt in the DFI board, and it gives the true ability of those pair of rams. After that, just play games like NFS:U2 and 3DMark01SE for me, that is the most sensitive ram program.

*Correct me if i'm wrong but i think the performance of the rams is only affected by the chip's memory controller when in windows and not in memtest.

Vincentvega18
05-17-2006, 08:57 AM
It's because of your memory controller.

Memtest for me as it's prebuilt in the DFI board, and it gives the true ability of those pair of rams. After that, just play games like NFS:U2 and 3DMark01SE for me, that is the most sensitive ram program.

*Correct me if i'm wrong but i think the performance of the rams is only affected by the chip's memory controller when in windows and not in memtest.
To some extent this is true i believe, but virtually no memory controller gives windows performance the same as memtest.

JasonDTM
05-17-2006, 08:59 AM
I use 3dMark2001 and CounterStrike:Source.. My testing methods are very odd, but it works for me.

Punisher!
05-17-2006, 12:24 PM
It's because of your memory controller.

Memtest for me as it's prebuilt in the DFI board, and it gives the true ability of those pair of rams. After that, just play games like NFS:U2 and 3DMark01SE for me, that is the most sensitive ram program.

*Correct me if i'm wrong but i think the performance of the rams is only affected by the chip's memory controller when in windows and not in memtest.

Wrong... with memories in slot 3&4 no probs. ... it is the mb that sucks.

thunderstruck!
05-23-2006, 02:00 PM
Can anyone tell me what 3dmark is more stressful for memory? Which gets it hotter, or crashes the most?

RyderOCZ
05-23-2006, 02:51 PM
Can anyone tell me what 3dmark is more stressful for memory? Which gets it hotter, or crashes the most?3D01 is said to be more system stressing not just Vid Card. I have found 3D03 however to crash sooner due to an overclock compared to 01

Sentential
05-23-2006, 03:18 PM
Can anyone tell me what 3dmark is more stressful for memory? Which gets it hotter, or crashes the most?
When I do testing for mushkin I loop lobby high and low + nature for 6 hours straight. If it can run that without crashing then the PC is rock solid, and is always stable enough to qualify RAM.

stealth17
05-23-2006, 04:02 PM
I use 3dMark2001 and CounterStrike:Source.. My testing methods are very odd, but it works for me.

haha I usually fine that CS 1.6 (yes you read that right) is more sensitive to unstable memory. Crazy, I know. Source must not be as memory dependent :shrug:

Vincentvega18
05-23-2006, 04:46 PM
haha I usually fine that CS 1.6 (yes you read that right) is more sensitive to unstable memory. Crazy, I know. Source must not be as memory dependent :shrug:
I would assume thats because the cpu/ram subsystem is less stressed with newer games and emphisis placed on vid cards.

hawkeyefan
05-23-2006, 05:13 PM
S&M test #3, finds 35 errors in 15 seconds on timings that made it to 2000+% HCI memtest

also HCI memtest instantly finds errors that S&M misses

so for short-term stability testing, I use:
memtest86+= 3 passes test #5, 1 pass test #8
boot windows
S&M test #3
HCI memtest 100%

long-term stability imho HCI memtest for 12+ hours as well as 12+ hours stability in Prime95 custom test using 1024-4096k fft lengths and 1800mb memory

stealth17
05-23-2006, 06:12 PM
I would assume thats because the cpu/ram subsystem is less stressed with newer games and emphisis placed on vid cards.
I suppose I can agree. Although BF2 does stress the ram/cpu a lot too. I use that for stability testing sometimes too. Source must not be as cpu orianted I guess.

kakaroto
05-24-2006, 01:13 AM
I always use HCI Memtest (500~1000% coverage) and SP2004 Large FTT for some hours.

But if you want really really stable memory, use QuickPAR and repair a damaged RAR'ed movie. :)

Punisher!
05-24-2006, 01:22 AM
I always use HCI Memtest (500~1000% coverage) and SP2004 Large FTT for some hours.

But if you want really really stable memory, use QuickPAR and repair a damaged RAR'ed movie. :)
Blend is more sensitive to RAM instability than Large, because it uses a VERY lager quantity of RAM (Large uses just 8 mb),

macaddict
05-25-2006, 04:36 AM
But if you want really really stable memory, use QuickPAR and repair a damaged RAR'ed movie. :)
That was always the case for me on all my Athlon XP setups but for some reason it's no longer true with my current A64 rig. I can rebuild a DVD9 PAR set successfully yet fail SP2004 within 2 hours.