Gizmos Freeware Reviews  

Go Back   Gizmo's Freeware Forum > Freeware Forum > General Computer Support

Reply
 
Thread Tools Display Modes
Old 25. Sep 2009, 08:39 AM   #1 (permalink)
dre
Senior Member
 
dre's Avatar
 
Join Date: Jun 2008
Posts: 143
Default Memtest

Good day

Have anybody tested with this utility before to see if you have defective RAM but the test come out clean, after longs tests even. Use the RAM again and gives you still the same intermittent problems that you had before. Then you replace the RAM and problem magically is gone! (same module)

The point is can we always trust all the utility's that is out out there for testing and so on. Answer is clearly no i suppose in my opinion.
__________________
Windows 3.0, IBM PS/2 i486/DX2, 256K RAM, 50MB HDD, Integrated onboard gfx...
dre is offline   Reply With Quote
Old 25. Sep 2009, 09:28 AM   #2 (permalink)
Member
 
Join Date: Sep 2009
Location: Denmark
Posts: 44
Default

Yes I have seen clean Memtest86+ tests hide the fact something was not working too well. But typically only with overclocking, ram tweaking or other strange things - like using 4 modules on a chipset which dont work too well with 4 unless speed is lowered 1 notch.

Sometimes moving module to other slot can help - moving process fixed problem because it was not properly attached at first

You are talking about Memtest86+ right? There are other "memtest" tools. You can trust 86+ one, industry does. If you go to memory maker and ask for help they will suggest you run just that. Problems is considered proof module is bad and so will be replaced. Think "bad sector" in hd terms. Must test properly of course, with right timings/voltage and I think a few other powers to consider. Check Corsair forum, they have guide on how to run Memtest86+ correctly.

To be sure also run "blend test" in Prime95. Memtest86+ test if module is working/healthy, Prime95 if module is working under real life conditions. Actually extreme conditions but if success you know nothing will break no matter what you run. This also test cpu, motherboard, power supply. Parts are connected. Im not into overclocking no more but you can probably find some questioning of Prime95 like you just did 86+. Those tools are also updated over and over so possible bugs etc. I would say both are needed to run. Not wise to ignore any error.
js960 is offline   Reply With Quote
Old 25. Sep 2009, 09:52 AM   #3 (permalink)
dre
Senior Member
 
dre's Avatar
 
Join Date: Jun 2008
Posts: 143
Default

Quote:
Originally Posted by js960 View Post
Yes I have seen clean Memtest86+ tests hide the fact something was not working too well. But typically only with overclocking, ram tweaking or other strange things - like using 4 modules on a chipset which dont work too well with 4 unless speed is lowered 1 notch.

Sometimes moving module to other slot can help - moving process fixed problem because it was not properly attached at first

You are talking about Memtest86+ right? There are other "memtest" tools. You can trust 86+ one, industry does. If you go to memory maker and ask for help they will suggest you run just that. Problems is considered proof module is bad and so will be replaced. Think "bad sector" in hd terms. Must test properly of course, with right timings/voltage and I think a few other powers to consider. Check Corsair forum, they have guide on how to run Memtest86+ correctly.

To be sure also run "blend test" in Prime95. Memtest86+ test if module is working/healthy, Prime95 if module is working under real life conditions. Actually extreme conditions but if success you know nothing will break no matter what you run. This also test cpu, motherboard, power supply. Parts are connected. Im not into overclocking no more but you can probably find some questioning of Prime95 like you just did 86+. Those tools are also updated over and over so possible bugs etc. I would say both are needed to run. Not wise to ignore any error.
Fully aware bout what you are saying. I run a overclocked rig of 4 Gig on my PC... That was not what i asked. I said if you no your RAM is defective, and you change the module with exactly the same one and your intermittent problem is gone. I know there is a other utility called Goldmem. Have used it before...
__________________
Windows 3.0, IBM PS/2 i486/DX2, 256K RAM, 50MB HDD, Integrated onboard gfx...
dre is offline   Reply With Quote
Old 25. Sep 2009, 10:06 AM   #4 (permalink)
Member
 
Join Date: Sep 2009
Location: Denmark
Posts: 44
Default

Im trying to count modules. You think of 1 or 2? - or just moving 1 module about and then suddenly problems magically go away? First option is of course, second possible. Until you have run P95 does it matter? probably not
js960 is offline   Reply With Quote
Old 25. Sep 2009, 10:30 AM   #5 (permalink)
dre
Senior Member
 
dre's Avatar
 
Join Date: Jun 2008
Posts: 143
Default

Quote:
Originally Posted by js960 View Post
Im trying to count modules. You think of 1 or 2? - or just moving 1 module about and then suddenly problems magically go away? First option is of course, second possible. Until you have run P95 does it matter? probably not
Prime 95 is used for overall system stability test yes, or shall i rather say it's actually a torture test on your system to verify your system stability...

one module i'm talking about. then replacing the faulty with another one
__________________
Windows 3.0, IBM PS/2 i486/DX2, 256K RAM, 50MB HDD, Integrated onboard gfx...

Last edited by dre; 25. Sep 2009 at 10:33 AM.
dre is offline   Reply With Quote
Old 25. Sep 2009, 11:00 AM   #6 (permalink)
Member
 
Join Date: Sep 2009
Location: Denmark
Posts: 44
Default

Well a big yes then. Guess why Microsoft build Windows Memory Diagnostics in to Vista. Too common to ignore.

As an overclocker main use of P95 torture is to check if setup hold water, cpu/ram overclocking/tweaking. "Overall" test is side effect since important parts of computer is running full speed. P95 needs a little hd-test, video-test, usb-test, just "more" to be a real system test. I think P95 author also mentions this in readme.txt - add some heavy games to list of what to check. If games crash then 12 hours of perfect P95 testing dont really help. May be a This-is-really-fast preset in bios also tweaked pci-e settings in an unfortunate way - or worse auto/software-oc did. You can be "fully aware" about your hardware, chipset/bios settings but oc/tweaking on the edge is more like witchcraft. Oc-forums are typically very busy for this reason. Must be humble and obey properly executed test-results, facts of life, but most also want the last 100mhz, a weeny bit faster ram timings, and who wants to run a stupid test for more than 30min! It is "stable", game on - what makes it fun...
js960 is offline   Reply With Quote
Old 25. Sep 2009, 11:07 AM   #7 (permalink)
dre
Senior Member
 
dre's Avatar
 
Join Date: Jun 2008
Posts: 143
Default

Quote:
Originally Posted by js960 View Post
Well a big yes then. Guess why Microsoft build Windows Memory Diagnostics in to Vista. Too common to ignore.

As an overclocker main use of P95 torture is to check if setup hold water, cpu/ram overclocking/tweaking. "Overall" test is side effect since important parts of computer is running full speed. P95 needs a little hd-test, video-test, usb-test, just "more" to be a real system test. I think P95 author also mentions this in readme.txt - add some heavy games to list of what to check. If games crash then 12 hours of perfect P95 testing dont really help. May be a This-is-really-fast preset in bios also tweaked pci-e settings in an unfortunate way - or worse auto/software-oc did. You can be "fully aware" about your hardware, chipset/bios settings but oc/tweaking on the edge is more like witchcraft. Oc-forums are typically very busy for this reason. Must be humble and obey properly executed test-results, facts of life, but most also want the last 100mhz, a weeny bit faster ram timings, and who wants to run a stupid test for more than 30min! It is "stable", game on - what makes it fun...
Pointless running 12hour test, Cause then you know your system has a stable OC regardless if it fails then. Your components will never be stressed like that to the extreme gaming or whatever.
__________________
Windows 3.0, IBM PS/2 i486/DX2, 256K RAM, 50MB HDD, Integrated onboard gfx...
dre is offline   Reply With Quote
Old 25. Sep 2009, 12:07 PM   #8 (permalink)
Member
 
Join Date: Sep 2009
Location: Denmark
Posts: 44
Default

P95 dude say min. 6 hours I think but 12, 24 or more is also a challenge/task for some oc-people. Like benchmarking is for others, 3D, Superpi, anything that spit out numbers will do. Like finding "screenshot-stable" setups interesting. All part of the same tinikering really. I was a 24+ oc´er Actually more of a forever oc´er since I dont really understand the concept of fixed hours when thinking about it - but to maintain sanity I settled for 24 hours. To begin with more like few hours if even that but eventually result of planned longer tests would be adjusted to. If crazy about benchmarking there is a tendency to slack if not skip testing, some dont care as long as games work. I never understood that, and perhaps also noticed a connection between those who researched and tested properly, and those who just pushed buttons/moved sliders I bet same goes today. So if stability is a big issue you risk getting even more determined on rigid testing than make sense. Due to the many horror stories and unbelievable user bugs available. A PC does not become ready for mission critical work just because P95 has run for a week. Still cheap components but understand hunt for P95 approval is about the same as 3Dmarks or similar.
js960 is offline   Reply With Quote
Old 25. Sep 2009, 01:19 PM   #9 (permalink)
dre
Senior Member
 
dre's Avatar
 
Join Date: Jun 2008
Posts: 143
Default

Quote:
Originally Posted by js960 View Post
P95 dude say min. 6 hours I think but 12, 24 or more is also a challenge/task for some oc-people. Like benchmarking is for others, 3D, Superpi, anything that spit out numbers will do. Like finding "screenshot-stable" setups interesting. All part of the same tinikering really. I was a 24+ oc´er Actually more of a forever oc´er since I dont really understand the concept of fixed hours when thinking about it - but to maintain sanity I settled for 24 hours. To begin with more like few hours if even that but eventually result of planned longer tests would be adjusted to. If crazy about benchmarking there is a tendency to slack if not skip testing, some dont care as long as games work. I never understood that, and perhaps also noticed a connection between those who researched and tested properly, and those who just pushed buttons/moved sliders I bet same goes today. So if stability is a big issue you risk getting even more determined on rigid testing than make sense. Due to the many horror stories and unbelievable user bugs available. A PC does not become ready for mission critical work just because P95 has run for a week. Still cheap components but understand hunt for P95 approval is about the same as 3Dmarks or similar.
30min and i'm happy with my OC. Like i said there is no way any other application will take the temperature to such a extreme as Prime95 does. 3Dmark is a synthetic benchmark. Prime95 is a torture test.
__________________
Windows 3.0, IBM PS/2 i486/DX2, 256K RAM, 50MB HDD, Integrated onboard gfx...
dre is offline   Reply With Quote
Old 25. Sep 2009, 02:24 PM   #10 (permalink)
Member
 
Join Date: Sep 2009
Location: Denmark
Posts: 44
Default

You can just use a cpu-burner if max. temp is only target for testing. P95 test relationship between certain hardware components and the way you have set them up - by crunching gazillion of calculations. Test much more than heat. A slight mismatch and it fails = setup is wrong. Fix and repeat, fix and repeat... Temperature does go up but just another parameter to count with. Does not matter if parts heat up from "synthetic" test, game or a hot summer day. Heat is heat. P95 also test cooling but temp can change dramatically when big vcard is running full speed, previous P95 test could be redundant Should not be a problem but depends on how far there is to critical state. P95 is not just about making sure computer run right now, but to make sure it runs under all conditions, in theory 24/7 365 days per year. To keep healthy distance to critical state is the big trick. Longer distance equals lower output from benchmarking - see the conflict? Well, P95 has limitations regarding heat. Im pretty sure a gaming session can tax cooling setup more than P95 on its own. You also get usb devices working, sound, may be networking. Ever heard of strange usb events at xxx fsb? All connected in mysterious ways = mysterious bugs and errors. Chipset can have specific "features" one must know/research to make the most of options. Like a big puzzle. No chain is stronger than the weakest link so testing is quite a job if really really "stable" is goal

I do know school of 30min-is-enough, 1 level above "screenshot-stable" but good luck With quality components may be you dont need much but I was never on that wagon for sure, mainly because luck should not play any role with tinkering. Do it right or not at all. Whole idea of testing, and I would add a bit of research and time spend understanding how things work, is to neutralize luck. Luck cant be trusted and bsods, other errors are not an option. Settings will work on hottest summer day as well as during winter. Dont matter which game/program, what is plugged in or active - all perfect. Anything else is unacceptable. How I looked at it anyway.

Oc lost me when makers stop warning against it, now an advertised feature. Progress but boring. Can also be costly, seems like a waste of money now.
js960 is offline   Reply With Quote
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off



All times are GMT +1. The time now is 02:42 PM.


Powered by vBulletin® Version 3.8.7
Copyright ©2000 - 2019, vBulletin Solutions, Inc.