Memtest86+ V2.00 (08/02/2008)

NoobyDoo

Senior member
Nov 13, 2006
463
0
71
Link

*** Enhancements in v2.00 : ***

New Features
Major Architectures changes
Modulo test now use random pattern for better accuracy
Added Advanced DMI Errors Reporting Mode
Added support for bus ratio changes on Intel Core CPU
Added support for non-integer bus ratio on latest Intel CPU
Added SPD Data Display for all Intel Chipsets (more to come)
Added serial support as a linux boot parameter (Thanks to Michal S.)
Added preliminary support for VIA CN Isaiah CPU
Added preliminary support for Intel Nehalem
Added support for VIA C7/C7-D/C7-M/Eden on Esther Core
Added support for AMD K10 (Phenom) CPU w/ timings detection
Added support for Intel Pentium E w/ 1 MB L2 Cache
Added support for Intel Core 2 45nm (Penryn)
Added support for FSB1333/FSB1600 Intel CPU
Added support for Intel 5400A/5400B w/ timings detection
Added support for Intel Q35/P35/G33/Q33 w/ timings detection
Added support for Intel X38/X48 w/ timings detection
Added preliminary support for Intel 5000P/V/Z
Removed on-fly memory timings change (unstable)


Bug Fixes
Numerous (really) bug fixes

*** Enhancements in v2.01 : ***

New Features
Added support for 45 nm Mobile Core 2 w/ 3 Mo L2
Added support for i945GM/PM/GME & i946PL/GZ
Added support for iGM965/iGL960/iPM965/iGME965/iGLE960
Added detection for SiS 649/656/671/672
Added detection for i430MX/i430TX
Added an optional beep mode (pass completed w/o error)
Pass duration 20% reduced
Removed the blinking cursor
Reverted Test #0 to cached

Bug Fixes
Solved a major bug in Memory Address Errors Reporting
Patched for Intel-Powered Mac
Corrected Intel 3-Series (P35/X38) chipset init
Corrected a bug with SPD Display and ESB6300
Correct a detection bug on P965/G965 C-Stepping
Solved a incoherency with pass progress indicator
Patched Makefile to compile on x86_64
Bootable Memtest86+ ISO more compatible
 

M1A

Golden Member
May 27, 2003
1,214
0
0
I downloaded this and am having problems running it with a floppy or usb pen drive?
Both seem to have made the disk ok but when rebooting it will just go past the floppy boot or usb boot right to Vista 32? As is says on the readme it doesn't show up anything in windows? Anyone else have this?
Also running the Vista 32 memory check you can go in in the vista start window I get errors when 4-1G Ballistic sticks are in but with only 2 sticks in I am fine. Don't matter which 2 are in they check out fine, just when I use 4 it gives errors. I am overclocked as you can see in my signature so tonight I will go back to stock and check again. May just be pushing the memory too much. Just kind of strange with 2 sticks its ok?
Thoughts?
 

VirtualLarry

No Lifer
Aug 25, 2001
56,587
10,225
126
Let me guess, it still lacks 64-bit support, so it can't test more than 3.5GB worth of RAM, right?
 

sblake

Member
Jun 28, 2001
163
0
76
When I run it from the floppy, the words are cut off on the left side of the screen i.e. I see "el" instead of "Intel". Anyone else having that problem? When using v1.70, all is well.
 

Zenoth

Diamond Member
Jan 29, 2005
5,202
216
106
Alright... hummm... so I downloaded from Guru3D, and I had two archives, one contained an .ISO, and the other contained an .EXE named mt200. Now I want to remove mt200.exe but it says that it is being used. I tried to remove it in Safe Mode and it refuses to go.
 

Zenoth

Diamond Member
Jan 29, 2005
5,202
216
106
Anyone knows how I can remove mt200.exe? It's always being used, even in Safe Mode. I shouldn't have touched it.
 

LOUISSSSS

Diamond Member
Dec 5, 2005
8,771
58
91
Originally posted by: VirtualLarry
Let me guess, it still lacks 64-bit support, so it can't test more than 3.5GB worth of RAM, right?

why 3.5gb? isn't that a windows problem? doesn't memtest86+ not relate to windows at all and boots off a CD?
 

VirtualLarry

No Lifer
Aug 25, 2001
56,587
10,225
126
No, 3.5GB (approx) is the limit of the 32-bit address space. It has to fit hardware I/O in there somewhere, so that leaves less than 4GB for RAM. So you need 64-bit mode to access mode. Hence my question.
 

bharatwaja

Senior member
Dec 20, 2007
431
0
0
good heavens, if memtest lacks 64 bit support then do we have to test each module separately when using 6 Gb of RAM?
 

n7

Elite Member
Jan 4, 2004
21,281
4
81
Originally posted by: bharatwaja
good heavens, if memtest lacks 64 bit support then do we have to test each module separately when using 6 Gb of RAM?

No, it will test all 4+ GB, but it has to do it in portions, instead of simultaneously.

For this reason, i am finding Memtest86+ less of a viable stresstesting tool.

I have found that HCI Memtest, which is run in Windows, finds errors or instability much faster usually, since you can run as many instances as needed to simultaneously stress ALL your RAM.

It also tends to show NB/MCH instabilty because of that as well, since if you're running 4 dimms, it puts tremendous stress on that.

http://hcidesign.com/memtest/
 

NoobyDoo

Senior member
Nov 13, 2006
463
0
71
V 2.01 released.

*** Enhancements in v2.01 : ***

New Features
Added support for 45 nm Mobile Core 2 w/ 3 Mo L2
Added support for i945GM/PM/GME & i946PL/GZ
Added support for iGM965/iGL960/iPM965/iGME965/iGLE960
Added detection for SiS 649/656/671/672
Added detection for i430MX/i430TX
Added an optional beep mode (pass completed w/o error)
Pass duration 20% reduced
Removed the blinking cursor
Reverted Test #0 to cached

Bug Fixes
Solved a major bug in Memory Address Errors Reporting
Patched for Intel-Powered Mac
Corrected Intel 3-Series (P35/X38) chipset init
Corrected a bug with SPD Display and ESB6300
Correct a detection bug on P965/G965 C-Stepping
Solved a incoherency with pass progress indicator
Patched Makefile to compile on x86_64
Bootable Memtest86+ ISO more compatible
 

Rubycon

Madame President
Aug 10, 2005
17,768
485
126
Originally posted by: n7
I have found that HCI Memtest, which is run in Windows, finds errors or instability much faster usually, since you can run as many instances as needed to simultaneously stress ALL your RAM.

It also tends to show NB/MCH instabilty because of that as well, since if you're running 4 dimms, it puts tremendous stress on that.

http://hcidesign.com/memtest/

Because of this, I don't find this very good. Running four instances will only test 4GB.

If you want to test memory for errors, that is all you should be stressing! Verify that the MEMORY ONLY is good, then go on to other testing. Sometimes a bad module that will not run at stock speed and stock voltage could be holding you back and you would not know it.

 

n7

Elite Member
Jan 4, 2004
21,281
4
81
Originally posted by: Rubycon
Originally posted by: n7
I have found that HCI Memtest, which is run in Windows, finds errors or instability much faster usually, since you can run as many instances as needed to simultaneously stress ALL your RAM.

It also tends to show NB/MCH instabilty because of that as well, since if you're running 4 dimms, it puts tremendous stress on that.

http://hcidesign.com/memtest/

Because of this, I don't find this very good. Running four instances will only test 4GB.

If you want to test memory for errors, that is all you should be stressing! Verify that the MEMORY ONLY is good, then go on to other testing. Sometimes a bad module that will not run at stock speed and stock voltage could be holding you back and you would not know it.

Run more than 4 instances.

I usually run 7 or 8 instances when stressing my 8 GB.

I'm not saying HCI is perfect; it's certainly not.

But it errors with what Memtest thinks is stable for me sometimes...
 

Rubycon

Madame President
Aug 10, 2005
17,768
485
126
Originally posted by: n7

Run more than 4 instances.

I usually run 7 or 8 instances when stressing my 8 GB.

I'm not saying HCI is perfect; it's certainly not.

But it errors with what Memtest thinks is stable for me sometimes...

Is it testing independent rows for each instance?

 

Zenoth

Diamond Member
Jan 29, 2005
5,202
216
106
I concur, HCI is great, I've used it a lot to troubleshoot with my previous 2GB kit.

Now that I've got two sticks of 2GB each I've noticed that Memtest86+ will do each tests two times instead of one, for example, with my previous kit it would do each tests in chronological order, but only do them one time until the last test, and then loop all of them from the start again, and so on. Now with 4GB the loop works, but each tests are done twice, thus taking twice the time to complete a single pass.

With HCI and 4GB I can manually adjust the desired amount of memory I want to test, in fact I can test so much that it some times also test the actual paging file (if I put a large number going above my physical RAM), which stresses the whole system instead of just the RAM. That way other issues will be detected in the process such as n7 said. It's a test under the Windows environment, so any NB/FSB-related issues will pop up to HCI and it'll in turn give you errors fast enough. Basically if an HCI test passes without errors it is a very reliable result, if not better than Memtest.
 

Mondoman

Senior member
Jan 4, 2008
356
0
0
n7, memtest86+ is NOT designed to stress RAM, just test it thoroughly under fairly normal conditions. Thus, as you say, it's more of a floor on which to add stress testing.