D'oh. Been running RAM with wrong settings for months.

Soldato
Joined
15 Jan 2006
Posts
7,811
Location
Derbyshire
Silly boy...

In the last couple of weeks had a few random BSODs pop up for the first time in years. Figured my CPU overclock probably wasn't stable in the warmer weather. Just went into BIOS to drop it down a bit and have realised that I've been running my RAM at CAS3 since upgrading from DDR400 in February. It's rated as CAS5. :rolleyes:

Put it at CAS5 and DDR2800 speed and just doing a spot of stability testing. Seems OK so far. Will throw in a few benchmarks later. Who knows what difference it'll make? I have no idea what speed the RAM's been running at. D'oh!

I suppose it helps to pay attention when you're on an odd upgrade path. Wonder if my X2 5000+ BE will go any higher than the 3 GHz I've been running?...

Edit: 3GHz it is. I seem to be stable again though.
 
Last edited:
how did you go from DDR to DDR2 on the same board?

<edit> Ah, you're using one of those dual-whatsit motherboards. *shudder*

Shudder you may. I don't think these are the dual-whatsit's your thinking of.

This boards based on ULi m1695: one of the best and most versatile budget chipsets ever. It's so good that nVIDIA have re-released it under their own name rather than kill it off and Phenom supporting AM2 boards based on it have come out recently.

I'm convinced that nVIDIA bought ULi because they were scared rigid of them. Budget motherboards were staring to be released allowing dual x16 SLI based on m1695 at half the price of the NForce4 SLI X16. There were rumours of Dual x16 crossfire on the same board with a driver update (alledgedly demonstrated behind closed doors). They ran much cooler than nVIDIA chipsets too. ULi could have blown nVIDIA out of the enthusiast motherboard market for months, if not years. Unfortunately the SLI implementation relied on drivers and nVIDIA have, unsurprisingly, not updated them for the Geforce 8 series onwards.

There's a review of one of the SLi x16 boards here: http://anandtech.com/mb/showdoc.aspx?i=2706

Edit: In other news - no more BSODs yet. Passed a few runs of S&M, now been running prime95 for a few hours. Seems like all's well.
 
Last edited:
Oh sorry. It's not the AM2CPU expansion board that supports Phenoms, which is what I guess you were asking about. I was talking about newer M1695 based motherboards.

This is the most recent one: http://www.asrock.com/mb/overview.asp?Model=ALiveDual-eSATA2&s=n

Theoretically it might be possible for the 939Dual series to support Phenoms with an AM2CPU upgrade board but there's a problem. Phenoms apparrently need at least a 4MBit (preferably bigger) BIOS chip. The 939Dual-SATA2 has a 2MBit BIOS chip so it's very unlikely that it'll ever be supported. Unless some genius makes it happen with a BIOS flash (unlikley) the only chance is a BIOS chip replacement (which hasn't been done yet AFAIK).

The 939Dual-VSTA has a 4MBit BIOS and ASROCK are gradually getting around to updating their older boards for Phenom support so it's not out of the question that Phenom support will turn up. In fact, I'm certain that if ASROCK don't do it, a modder will.

Edit: The maximum a 939dual-SATA2 will currently support with an AM2CPU board is the Athlon X2 6400+.
 
Last edited:
Don't know anything about Gigabyte dual motherboards. I think it was only ASROCK thad did the whole 939 -> AM2 upgrade path thing, using their own design.

There have been quite a few other designs that have done both AGP and PCI-E, most of them with a nasty bottleneck on one of them. Quite a few 775 motherboards offer both DDR and DDR2. I wasn't aware that they were that bad.
 
Oh - should you be interested in moving to AM2 and getting a faster Athlon X2, I have a spare AM2CPU board on the MM.
 
Uh oh. Just had a BSOD when doing nothing more complicated than browsing a web page with CPU and RAM at stock. Had an instant reset earlier when just listening to a CD in media player. This one's got me confused.

It's generally prime stable, S&M stable and hadn't had anything more than a CTD or two in one game since the last new hardware I installed in February. I didn't install any new hardware, change any settings and I'm pretty sure the crashing isn't heat related. I can game for hours with no problem but it will just randomly reset or pull a BSOD with no CPU or GPU load.

I hate instability. I've been running stable on this rig for years through several hardware changes. The only thing I've changed during the last couple of weeks is install the GRID demo and Deus Ex: Invisible War. Will run memtest later. Any other suggestions?

Edit: I have just spotted one unusual thing: I have strange low temperature spikes being misreported in speedfan. CPU temps are generally 35 idle, 50 load (say prime) 56 during SnM FPU test. Every so often I get the odd reading of between -50 and 0. That never happened before. Something fishy is going on.
 
Last edited:
Running 2T. 1T won't behave at anything like stock timings.

Edit: Just stuck the RAM voltage on high, which is about 1.95v. It would have been running at 1.8v before, which is within my RAM's operating range but on the low side.
 
Last edited:
Oh well. Thought things were looking up after the RAM voltage change. No problems yesterday. This evening it's back with a vengeance. Two BSODs so far.

Prime95 and S&M are still generating no errors. Just dug out the Memtest disc and will be running that later. This could be a pain to diagnose.

Edit: Could be anything. I'm not liking this. Only recently installed software is AVG8. Just Googled it and there are a few people compaining of BSODs with it.
 
Last edited:
Well I'm baffled. The BSODs cleared up with no obvious action on my part. Nothing for nearly two weeks. Maybe we got local micro-powercuts or something?
 
I never knew what that 4/8 beats thing did. I haven't changed it. Left it on 8, I think.

Edit: This led me to a RTFM incident. Just looked it up in my motherboard's manual and it says set to 4 beats for "64 Bit Dq". I have no idea what that means. A quick google suggests it might have something to do with single channel memory but I didn't turn up anything conclusive.
 
Last edited:
I told a lie. Was using 4 not 8.

No problems for a month but I've started getting reboots and BSODs again. The only thing that I can think of that's changed is the ambient temperature. It was warm when I last had problems and it's warm again.

Heavy load doesn't seem to do much so what do you all reckon? RAM chips getting warm? North Bridge (it's passively cooled)?

Have just put everything stock and increased the speed of the case fans a notch. We'll see what happens next.

Temps in speedfan after increasing the case fan speed look fine. Temp 1 36º, Temp 2 (CPU) 37º, temp 3 30°, Hard Disk 32º. Maybe I'd decrease the speed later and see if there's much difference.

Edit: Those temps are all idle.
 
OK - I'm fairly certain I pinned down these crashing problems. It seems to be sorted now (assuming it's not just the colder weather).

In short nVIDIA drivers - probably originally the ones I installed for Assassin's Creed (which I think were beta). Installed some WHQL ones a few weeks ago and I've gone from at least one (usually several) BSODs a day to none at all.

I'm rather annoyed I went through so much grief while doing my MSc dissertation, all for the sake of a game (which I didn't have time to play much because I was busy - GRR).
 
Was you able to see what the error was on the BSOD's, as it might have lead you to that conclusion much sooner?

Yep. 0x00000124 variants. People seem to get them for lots of reasons though so it was hard to pin down.
 
Wasn't the nVIDIA drivers. The BSODs came back after a while. I changed to an HD4850 and that didn't sort things.

Current prime candidate for the problem cause is a fairly big bit of fluff that was wedged in the socket under one of my RAM sticks. Just blew it out with compressed air last night.

Don't want to count my chickens because I've thought the problem was solved on several occasions. We shall see.
 
Back
Top Bottom