What's up with the newest SMP Memtest 3.5?

[Rubycon]

I'm not the only one after doing some googling either...

No matter what settings I use it freezes shortly.

I found on overclockers forums that Memtestx86+ v.2.11 is the program to use and let it run for as long as I could today.

I guess it's safe to say it's stable...

Any other ways to check?

I'm pretty sure memory is OK at this point.

I think my X3 1,6kW PP is dying! 🙁

 

[taltamir]

occpt?

 

[Rubycon]

I don't trust the new Occt honestly. Linpack is broken in Occt 3.0. There is a memory tester that runs in Windows but you have to run lots of instances of it on a 12GB system.

 

[n7]

I always use the + version personally.

And you should run HCI Memtest.

It's much harder on things than Memtest86/86+ is.
Set up 8 instances to utilize all of your 12 GB, & run that for as long as you can...you'll likely need to do it for a day to really test all 12 GB.
At least overnight.

 

[Rubycon]

Originally posted by: n7
I always use the + version personally.

And you should run HCI Memtest.

It's much harder on things than Memtest86/86+ is.
Set up 8 instances to utilize all of your 12 GB, & run that for as long as you can...you'll likely need to do it for a day to really test all 12 GB.
At least overnight.

I've never gotten that program to find errors when I had them in the past from too tight timings, etc. Now if memtest86 started scrolling red on test 4 - thousands of errors (thank you crucial ballastix tracer junk!) - then sure but Windows often had the stability of a canoe in a tropical cyclone. :laugh:

 

[n7]

It might be something to do with me having 8 GB; not sure.

But Memtest86+ will pass fine @ say DDR2-900 5-5-4-12, yet HCI will fail (need to do 5-5-5-12).

Or i can do DDR2-1050+ 5-5-5 just fine in Memtest86+, but nowhere close in HCI...have to back down to a lower speed.

I will admit...even when HCI fails i've usually found everything i use the PC for to be rock stable...but i'm kinda picky so i prefer it to pass.

 

[VirtualLarry]

Originally posted by: Rubycon
I don't trust the new Occt honestly. Linpack is broken in Occt 3.0. There is a memory tester that runs in Windows but you have to run lots of instances of it on a 12GB system.

Why do you say that Linpack is broken in OCCT 3.0? Have you tested 3.0.1?

 

[Rubycon]

Originally posted by: n7
It might be something to do with me having 8 GB; not sure.

But Memtest86+ will pass fine @ say DDR2-900 5-5-4-12, yet HCI will fail (need to do 5-5-5-12).

Or i can do DDR2-1050+ 5-5-5 just fine in Memtest86+, but nowhere close in HCI...have to back down to a lower speed.

I will admit...even when HCI fails i've usually found everything i use the PC for to be rock stable...but i'm kinda picky so i prefer it to pass.

HCI results off the grill err Dominator Heatsinks. 😉

Originally posted by: VirtualLarry

Why do you say that Linpack is broken in OCCT 3.0? Have you tested 3.0.1?

It does not load the CPU nearly as much as the real thing. The GPU (mem) test is confirmed broken in 3.0.0 as well. Poor release and NOT recommended! :thumbsdown:

I have 3.0.1 now and will put through paces when time permits.

 

[n7]

Beautiful stuff Rubycon 🙂

Looks like memory isn't unstable, that's for certain.

As for OCCT, i see no reason to use it, as i've never liked the custom P95 test it uses (i prefer P95), & LinX works great for LinPack usage.