Nope, I'd do a memtest86+ scan. I've seen that IRQL error with bad memory before.

Will do. Anyway Superfetch constant disk thrashing can be quite annoying.

DosFreak wrote:

Nope, I'd do a memtest86+ scan. I've seen that IRQL error with bad memory before.

I would recommend the same thing, this could be just a coincidence that it fxied the issue and my guess is it will return of indeed you have a faulty stick of RAM 😉

Sometimes it can be one or more sticks have come loose in the DIMM slot, just by removing the sticks and reseating them can help too.

Or maybe Superfetch is trigging the AMD TLB bug on my old Phenom. I am running with the erratum fix disabled. Anyway, I will post the memtest results when the test is done.

I have run Memtest86+ for 42 minutes and it reported no errors. I have just reenabled Superfetch and Prefetch. I will keep testing. Maybe it was coincidence.

Be careful, it's possible that the capacitors of the graphics card and/or the mainboard are starting to fail.

Failing capacitors are the most incredibly annoying things on earth. They are totally random and give numerous (but misleading) blue screen errors.

They drive people crazy!

I have run Memtest86+ for 42 minutes

Not nearly enough. Minimum is 24h, best to keep it running for two or more days if you have more than 2GB RAM. Yes, I'm serious. This is how you use the tool properly.

I am not wasting 24h for testing purposes. I will keep using the computer keeping an eye on the prefetch/superfetch thing.

ADDiCT wrote:

I have run Memtest86+ for 42 minutes

Not nearly enough. Minimum is 24h, best to keep it running for two or more days if you have more than 2GB RAM. Yes, I'm serious. This is how you use the tool properly.

I always stick with doing at least 3 complete runs, but I agree that the longer you run, the better it is. I once had a problem with RAM which wouldn't start appearing after the system had been running for about 2 hours. The RAM would start erroring in a most peculiar way but seemed to run fine at first.

Heat apparently can have an influence on RAM that's not quite dead, but is of the annoying type.

I find it easier to simply exchange the RAM and start using the system again. If the problem is gone, it's probably the RAM.

Well, I wouldn't want to do _any_ work on a machine I'm suspecting to have faulty memory. Bad RAM can lead to all kind of "interesting" effects, of which crashes are the least problematic. I've had clients with hundreds of corrupted business documents as a result of a bad RAM stick, for example. Anyway, it's your call, I'm not going to waste more time trying to help a know-it-all.

I usually RMAed the memory with the new stick when I encountered the IRQL_Not_Less_Or_Equal BSOD.

I have done this two times on two different systems, and the problem gone.

I'm debating going with ECC memory on my next home builds. Still haven't decided.

My home file server definetly should be using it. Hell I'm using ZFS with the filesystem may as well go with ECC while I'm at it.

ECC is pretty cheap nowadays it seems.

Same here DosFreak, my next "Home Server" will definitely have ECC memory. It's difficult to find a reasonably priced and energy efficient Mobo/CPU combo with ECC support, though. Sandy Bridge, for example, only supports ECC with expensive Xeon CPU's. I'm going with a Asus AM3+ Mobo this time, my first AMD machine since the KT133/A debacle (; .

24h uptime + large amounts of RAM = need ECC.

I really doubt my ram sticks to be bad, because I changed the two of them not so long ago. And besides, getting BSODs on my system is quite uncommon. Another thing: if ram was faulty there is a great possibility than anything else would be corrupted too. After deleting the prefetch folder everything seems fine so far. Anyway I will keep testing. I am pretty sure it was the TLB bug that was trigged for some reason (I am running without the erratum fix as I said before), so it's pretty safe to assume the bug will appear from time to time... Maybe it's about time for me to update. After working with AMD all these years, I am thinking about in getting an Intel system this time, probably an i7 or whatever thing is available.

I am really thinking into going the Superfetch disabled path, although I agree it speed up things a little.

PS - Do you know what chipsets support ECC?. How about prices? Is really that expensive?

Did you get the complete stop error? In my experience the IRQL Not less or equal error is likely a driver issue. If you still have the dump file, you can use the BlueScreenView to get more information about it.

No. I do not have it, but today I have find out that Superfetch does indeed weird things in my setup. I had just reenabled it yesterday and today my scanner driver refused to load. Disabled it and voila, working perfectly fine.

So, I will left Superfetch disabled for good. Not a great loss anyway...

I've forgotten about superfetch, prefetch, defrag and all that crap since I bought an SSD drive.

Seriously, this is the most noticeable upgrade I've done to my computer in MY ENTIRE LIFE.

I was extremely skeptical at first because they were so damn expensive, but now I can tell you that the hype is really true. The difference is like night and day.

Leolo wrote:

I've forgotten about superfetch, prefetch, defrag and all that crap since I bought an SSD drive.

Seriously, this is the most noticeable upgrade I've done to my computer in MY ENTIRE LIFE.

I was extremely skeptical at first because they were so damn expensive, but now I can tell you that the hype is really true. The difference is like night and day.

The price is what bugs me 🤣

Tetrium wrote:

The price is what bugs me 🤣

Totally...

RAM is sooo cheap right now. But SSD still commands a huge premium...