Jump to content


Photo
- - - - -

Errors w/S&M "Night" test (DOS floppy version), system s


  • Please log in to reply
9 replies to this topic

#1 anthonys123

anthonys123

    New Member

  • Members
  • 11 posts

Posted 16 February 2006 - 09:24 AM

Hi Everyone,

S&M is turning into a blessing and curse for me. :) When I run the S&M "Night" test (DOS floppy version) on all of my rigs that I thought were stable, it inevitably results in literally millions of errors by test's end. S&M "Normal", "Long", and "Slow" tests do not reveal any errors, but I haven't tested with the "Fast" and "Short" tests - my thoughts are that the Normal and longer-term tests are the most stressful ones. My rigs test out successfully with the other apps I've thrown at them:

- OCT Stability and 12-14 hour Torture test

- OCT running concurrently with Norton AntiVirus scan and Microsoft AntiSpyware scans

- Prime95 small, medium, large FFT -> all run for at least 17 hours

- Prime95 running concurrently with Norton AntiVirus scan and Microsoft AntiSpyware scans

- OCT and Prime95 running concurrently -> 12-15 hours

- MemTest86 Test #5 -> 17 hours

- MemTest86 Test #8 -> 17 hours

- MemTest86 All Tests -> 17 hours


And yet, the S&M Night test reveals errors - what are your experiences with this particular test? I've resigned myself to stopping all overclocking/over-volting on one of the systems and running the test just to make sure no errors surface, but could there be a problem (dare I say bug?) with S&M itself? I'm also going to try the test on several plain-jane Dell and HP systems to see if they show any stability.

- Anthony

#2 medianOCer

medianOCer

    Geek

  • Members
  • PipPipPip
  • 714 posts

Posted 16 February 2006 - 01:56 PM

I have yet to mess much at all with the S&M FDD versions.
I just mainly use the Windows version of it.
I'll take a look at the newer revision of the FDD version and
see what gives with it.
I briefly messed around with the FDD program awhile back,
but only for a very brief time, sort of just a quick glance at it.

Have you tried just running the S&M.exe within Windows and
using the program like that? You'll also need to copy the io.dll
and paste it to the Windows/System32 folder before it'll run
properly; just in case you were unaware of that.

EDIT/Update:
I made an S&M FDD and took a look at it. I was able to successfully
run it in a DOS environment, but I used the quick and fast modes and
ran them a few times on each DIMM that I have with my rig currently.
I passed all the tests without errors.

Have you run less lengthy modes of the testing on each DIMM that
you're using, and without getting any errors?

Also, each RAM stick needs to be configured properly within the FDD
program and according to the layout/config of each RAM stick. If you
don't have a RAM stick in one or more slots, then it looks like you need
to set the program to have none being indicated in the said slot(s).

You can also manually indicate which slot is being tested at a time, as
well as how much memory is being tested.

I started to run the night mode, but it looks like that would take awhile
to run. Were you testing one RAM stick at a time, or what config did you
have setup to run in night mode. Also, how long did it take the test to
run? Kind of curious as to how long it took you to run it and how much
RAM was tested when you used that mode.

#3 anthonys123

anthonys123

    New Member

  • Members
  • 11 posts

Posted 16 February 2006 - 07:46 PM

Thanks for the info regarding the Windows version of S&M - I had no idea how to run it in Windows and didn't realize that the executable could run either in a DOS or Windows environment. I will definitely try that tonight.

I haven't run the "Fast" or "Quick" tests on the systems - I figured that "Normal" and the slower tests would stress the systems more? The Normal test needs something like 7-8 minutes to run and has proven useful for doing quick spot-checks of stability. The Slow and Long tests have been useful too and thank goodness all of my rigs pass these three tests.

I agree - the UI leaves a bit to the imagination in terms of understanding. ;) It appears to auto-sense the stick in slot 1 but will not auto-sense any other sticks. I've simply booted from the floppy, exited the "Volkov Commander" by hitting F10, and running S&M manually from the DOS prompt. When the app starts, I tab over to the right-hand menu and select which test type I want and hit Enter to start the test. I think in this mode I'm only testing the first stick, though I haven't figured out how to make it test both sticks in the same test.

As an update, I disabled all overclocking/over-volting on one of my rigs this morning (Sempron 2500+, 166MHz memory bus, 1.6V core, 2.6V DIMM) and ran "Night" mode. Low and behold I get home tonight and there are even MORE errors than when it was overclocked! :)

- Anthony

#4 medianOCer

medianOCer

    Geek

  • Members
  • PipPipPip
  • 714 posts

Posted 16 February 2006 - 08:41 PM

Np on using S&M in Windows. Thanks for the
appreciation. ;)

I'll need to look at the night mode a bit more.
I'm wondering if some of the code for the night
mode needs to be cleaned up a bit more, and if
there might be something wrong with the prog
itself? Maybe not, just a thought.

Although, if it is quite a bit more stressing, then
it might be possible that it could run across errors
which might not be detected with the shorter run
of testing.

There's quite a bit of configurability with the DOS
version, and a lot of options/menus/modes you
toggle between. There might be something you're
missing that you need to set manually for it to run
properly. Unsure on that though.

Personally, I'll probably end up using it for testing
along with Memtest, and with the normal and/or
quicker modes of testing. After that, I'll move into
Windows for further testing of settings and etc.

#5 anthonys123

anthonys123

    New Member

  • Members
  • 11 posts

Posted 18 February 2006 - 09:28 PM

medianOCer - I tried running the Windows version of S&M on one of my Win 2K Server rigs and the test never starts. The application starts, I can run the Wizard to set options appropriately or set them manually, but when I click on the "Start" button, the test never starts (0% CPU). S&M works fine on my Windows XP Pro rig - is there an incompatibility you're aware of with Windows 2000?

- Anthony

#6 medianOCer

medianOCer

    Geek

  • Members
  • PipPipPip
  • 714 posts

Posted 19 February 2006 - 01:47 AM

I'm unaware of any prog incompatabilities with Win2K, but I don't have a copy of it
to test S&M out on. Wish I had a copy of Win2K Pro though.

I would think you ought to be able to run S&M on it. Wouldn't it just be a matter of
copying the io.dll file to the system32 folder, just like with WinXP? Is there possibility
that you need to change some type of configuration within the program itself, and
which is keeping you from running it successfully?

#7 NEOAethyr

NEOAethyr

    Certified Geek

  • Members
  • PipPipPipPip
  • 1239 posts

Posted 19 February 2006 - 05:38 AM

Did you mod the heck out of your 2k?

I modded the heck out of my 2k pro sp4, not a sr ver though btw(been to lazy).
Works fine too.

It does not however run on my 2k3 pro, the io.dll thingy.
Ver 1.7.3 does however run fine on it.
ver 1.7.6 is the one that does'nt.
I don't have any ver's between this.

If you need a copy of 1.7.3 just holler and I'll upload it for ya.

You can make a windows pe build if need be, and run 1.7.3 from there, ie where my 2k3 pro comes from, it's the pe ver of 2k3 server.


I have 2k server adv, vlm myself.
I have'nt used it in ages though.

S&M works fine on a modded 2kpro.
Also works fine on a modded 2k3.
Should work on xp pro as well lol.
You don't need to mod anythign to get it working either, I just said that because my os'es are modded, and they work fine.

The thing is though, 1st run on a fresh os, it says io.dll cannot be connected to but is present or something.
Run it again and it works fine from then on.

I never had to copy it to my system/system32 folder.
Dll's in a folder with the program that links to them should work fine, even with directx files and such.

#8 anthonys123

anthonys123

    New Member

  • Members
  • 11 posts

Posted 19 February 2006 - 08:09 AM

Thanks NEOAethyr - yeah, if you could upload the 1.7.3 version, that would be great! Where is the location it will be stored at? I'm new to the forum, so I'm still learning my way around.

#9 anthonys123

anthonys123

    New Member

  • Members
  • 11 posts

Posted 19 February 2006 - 08:52 AM

- What is "Disable Safe Mode" in CPU Test and Memory Settings? It's un-checked by default. Does checking these options make the tests more stressful? Does it disable temperature monitoring?

- What is the "Variable pattern" setting in the Memory Setting? Does checking it make the memory test more stressful?

#10 NEOAethyr

NEOAethyr

    Certified Geek

  • Members
  • PipPipPipPip
  • 1239 posts

Posted 22 February 2006 - 12:43 AM

I'm sorry about the delay, I've been working my butt off on a custom display driver.

Major overhall...
I mean major lol, just need a voodoo card now though which I'll have in the begining of next month....

Anyways I can't upload to savefile right now apperenlty, says fileservers are full :.

If you pm me an email addy and I can send that way if you want ;).


As for the options...

Disable safe mode, safe mode count.
Safe mode means it stops the test as soon as it notcies an error, normally within an error or 2.
I usually leave safemode disabled so it runs through the full set of tests, however I do use safe mode on the cpu tests.


Varible pattern.
I just use it, piece of mind.

Single thread, disabled on the newer ver's.
Use single thread mode on a signle cpu system, use mutlible thread mode on a smp system.
It's faster to use single thread on a single cpu system ;), very little diff but still.


The test lengths.
I use normal length, sometimes called avg length in the older ver's.
For cpu wise, fpu, I use short and loop and long.
Short to test, long to make sure these days, and loop to break in.

Make sure you're running the load% at 100% when you need to test stability.
Lower it for breakin if need be, if you need to, run another program in the background to take away cpu time.
It'll allow it to test without errors, cpu wise, and therefor continuing breakin, errors are bad too.
Meaning they reverese breakin, errors = bad ;).

Anyways drop me a msg and I'll get you that older s&m.
I'll keep trying to upload it otherwise, hoping that savefile fixes there servers...