Exhausted memory
 

I am posting this in the hardware forum because I suspect a hardware problem, but I need feedback. Didn't see any topics with the words "exhausted memory" in them either.

Ran the torture test two times. I've set the prime95's memory parameters to be 550MB out of 768MB on one run and 500MB out of 768MB in another. Both times it stops (AMD XP 1700, Win98SE, 768MB) when it hits the 2048K FFT and says "exhausted memory"
The next line says the test is being stopped with "0 errors and 0 warnings"

What's up with that? Do I have a hardware problem?
Thanks -=- john

I don't have much experience with this sort of thing on x86, but I've seen similar happen using Mlucas in self-test mode on a few platforms. What seems to happen is that as each suseqent job allocates the memory it needs, the memory freed by the previous run is not being recycled, so if (for example) you run around 100 trials at an average FFT length of 1024K, each trial needing 8-10MB of RAM, you wind up exhausting a 1GB virtual memory space.

But I don't know if that's what's happening on your system - George would be the person to answer that.

Argg! I turned off virtual memory pagefile and forgot to turn it back on. I feel stupid!

Since it turned out to be a software problem, I moved this topic to Software.

Sometimes it's hard to tell 'till you FIND the problem.

8-)

Could Windows98 be using that much memory for an exhaust :shock: ?! 768-550=218. I would think that Windows98 could contain itself on that :? . I feel like im still in the memory stone ages :D .

I realize you found a solution ( turning the virtual memory paging system back on), but if after the first run you exited Prime95 then that memory should have been released ( how it takes to release is an issue ).

98se has some sort of OS memory issue if you have more than 512 MB, edit System.ini in the vcache section set the max amount of memory for caching ( MinFileCache and ChunkSize too if you want).
[vcache]
MaxFileCache=524288

or less if you want (what I have done, to give programs more, cache less)
[vcache]
MinFileCache=4096
MaxFileCache=24576
ChunkSize=512

Only the first 512MB can be cached.

dsouza123.
The first thing I did after loading Sin98, but before turning off the pagefile, was to set the vcache parameters (MaxFileSize=16384, I like it small). I probably should have mentioned that earlier. The vcache bug is a design flaw, but I'm stuck with this OS since I have apps that won't run on XP or linux. Regardless, Microshaft gives the lowdown [url=http://support.microsoft.com/default.aspx?scid=kb;en-us;253912]here[/url].

Actually, I didn't find a solution yet. I turned the pagefile back on, got impatient, and I reported back to this thread success before verifying it (that'll teach me). :surprised:ops:
Now I get
[code:1]Out of memory
0 errors, 0 warning.[/code:1]
So now I am out of memory, which is baloney, as opposed to "exhausted memory" (probably a malloc() status due to the pagefile missing).

I have my suspecions about the mobo being DOA (because of other weirdness) but can't prove it yet. I'll double check for typos on the vcache settings. For now, I've taken out one of my two sticks of memory and torturing again. Maybe hardware. Maybe software. I haven't encountered this combination before. Don't know where this will end up.
Suggestions are welcome.

[quote="E_tron"]Could Windows98 be using that much memory for an exhaust :shock: ?! 768-550=218. I would think that Windows98 could contain itself on that :? . I feel like im still in the memory stone ages :D .[/quote]
I love operating systems internals and the subject of memory management fascinates me (as does process/thread scheduling). In short, the OS seemed to take so much RAM when all along that was a side effect of turning off the pagefile. Turning on the pagefile keeps part of the OS and its processes, DLL's, etc. in RAM and part in the pagefile (and other places, but no need to go into that here). PM me if you want to know more.

-=- john

Have you shut down the PC since getting the messages ?

Some OS stuff doesn't get restored/fully restored until a full shutdown and startup.

Maybe if the memory is still allocated before the pagefile was reactivated it is working with a reduced amount of memory, the allocated memory from the runs is now locked and unavailable for use.

I notice after a program crashes the system resources aren't returned until a full shutdown and restart. Then they are completely restored.
Maybe the allocated memory from Prime95 remains allocated and only after a shutdown will virtual memory work correctly and the memory be freed.

I realize it is obvious that a shutdown will free all memory and resources because Windows starts fresh with each restart, but am mentioning it just to be thorough.

I'm curious what program you run needs W98... XP has a real nice compatibility mode that does surprisingly well with older programs... If you want I could test your program with it to see if it works...

[quote="Xyzzy"]I'm curious what program you run needs W98... XP has a real nice compatibility mode that does surprisingly well with older programs... If you want I could test your program with it to see if it works...[/quote]
Recording studio stuff. There's a device driver for the 128 channel recorder/mixer that controls a proprietary sound card driving the DAC/ADC's. A year or two ago there were rumors going around that some of the users themselves started writing an XP driver . . . I'll have to track down the user group.
Thanks for the offer. I may take you up on it if the search is positive.

Still says "no memory" when it hits the 4096K FFT; the 3584K is fine. Apparently, Win98 needs a huge pagefile for FFT's that large.
So I'm going to declare the CPU and memory just fine. Got other problems though . . . :?