Gigabyte 965P DS4 and 8GB memory

NathanE

NathanE

NathanE

Soldato
Joined
21 Oct 2002
Posts
18,022
Location
London & Singapore
I upgraded this mobo about a month ago with some 8GB of Crucial Ballistix DDR2-800 memory.

It ran fine for about 2 weeks with no problems. Then I got a BSOD. So I tried lowering the memory multiplier so that they ran at 709MHz instead of 800MHz.

Unfortunately this didn't fix it.

Now, after countless more BSODs. I am running them on 533 (!) MHz and have relaxed the timings significantly - as in 5-5-5-20.

This lasted barely 12 hours until another BSOD. So as of 5 minutes ago I've relaxed the timings to 6-6-6-24. I'm desperate, not expecting this to fix it either - but I will see.

The voltage of the memory is at 2.05v. Rated for 1.8 to 2.0v.

The FSB is overvolted by 0.15v. The northbridge is overvolted by 0.3v as per various recommendations I've read regarding this motherboard.

Everything I try doesn't seem to work. And it is getting worse. It's strange how it was stable for quite some time after I first installed the memory.

Is it possible the memory is being "damaged" by the motherboard and they are becoming less tolerant?
 
Sure enough, 6-6-6-24 just BSOD as well. Barely lasted 20 minutes.

I've taken out 2 of the memory modules now. Running on 4GB, which used to be 100% stable. If this isn't - then looks like there's a serious hardware problem.
 
After doing some more research I've discovered the tRFC timing.

I've put the full 8GB back in and set the tRFC to 64 in the BIOS. So far so good. It was doing Memtest86+ last night for 9 hours, it did a full 6 passes with no errors. And so far Windows has been stable as well all morning. And this is with it running at 800MHz, 5-5-5-18 timings.

Apparently the tRFC value must be raised when running 4 sticks of RAM, especially high densities such as 2GB sticks (8GB total). The SPD on most DDR2 sets the tRFC value too low because manufacturers expect their memory to be used as sold in the kit i.e. not buying two kits for one PC!

Touching wood.
 
OK it was stable for 3 days on the tRFC=64. But now its started crashing again. So I put the tRFC up to a ridiculous 120 and also lowered the FSB to 533. Still Memtest86+ fails within half a minute.

Run out of ideas now :(
 
I did try them in dual channel pairs, last week. And those passed Memtest fine then. I will try them again like this tonight.

It's just strange how I can get it stable then after a few days it becomes catastrophically unstable again (as in Memtest86+ fails within seconds).
 
It sounds like a bad stick of ram, the single stick tests should help to find if this is the case, however it could be a fault in the mainboards slots or the CPU memory controller.... only testing will tell :(
 
You must log in or register to reply here.
Back
Top Bottom