Dying ram?

[infrared]

After an overclocking session a few nights back that i thought went very well, i've realised that my ram is performing no-where near as well as it should be.

I've got 4x vdata 512mb CL4 533mhz ddr2. Rated for 1.8v

Can normally run at 600mhz 4-4-4-10 @ 2v
Now it can only run at 580mhz @ 1.95-2.15v

And the memory bandwidth score in sisoft sandra has gone right down to 4700mb/s. From 6500mb/s.

I think i might have half killed them with too much voltage, at too high frequency for too long, what do you think?

I'll try each dimm on it's own, and find their maximum clocks, then pair them up in dual channel and see what the bandwidth's like. Does anyone else have any ideas?

Looks like i'll have to make some money and buy a pair of 1gb ocz pc2-8000 dimms (I wish)

 

[Jimmy 2004]

Sounds quite messed up. Only thing I would do to check is memtest, but if that shows no errors than it's strange!

 

[infrared]

Ok, i've tested each dimm individually, and they all seem fine, and overclock up to 650mhz @ 4-4-4-10, and overclock well in pairs in dual channel, but if there's 4 stick in, the maximum clock drops to 550mhz, when i can get 640mhz with 2 sticks.

Strangely, the issue with lower bandwidth has dissapeared! 6800mb/s again.

I have a horrible feeling that the problem lies with my NB. I gave it quite a large overvolt a few days back to stabilise an overclock, and i think it may have done some damage. Either that or the power supply circuits to the memory is failing. I might be able to rma the board, but i hate rma'ing stuff... takes forever for replacements to come.

Edit, just been running prime95 for 10 mins, and it seems even 550mhz is too fast.

 

[Jimmy 2004]

Guess it's time to test each individual stick with Prime95... fun task for you! Sounds like it could be the mobo to me. Did you ever check your OC to 650MHz with Prime95 before your problems? It might have even been too far then.

 

[infrared]

Yep, and i tested each stick on it's own, and for some reason i only experience the problem when all 4 are installed.

I tested each one with prime 95, and they all managed 650mhz no problem.

Also, the overvolt i was using shouldn't have caused any damage, because i had a fast 40mm fan on the NB at the time, and 1.65v isn't that much. Some people run it at 1.6v 24/7.

I've contacted asus, and am going to try to rma the board direct to them. Since the board will still boot into windows, the distributor will plug it all in, turn it on and see that it's working, charge me £10, and send it back

Thanks for your help jimmy, it's appreciated!

 

[Sasqui]

May seem like a stupid question, but did you try a CMOS reset? I encountered a situation with my Abit board that only a reset would fix - seems that some multipler in the NB got stuck and I could no longer get to the FSB I could before... I was testing the PCIx frequency. As soon as I reset the CMOS, I was back in businiess, even though there were no settings changed when all said and done. I actually RMA'd my original board not knowing this!!!

 

[infrared]

yeah, i tried re-flashing the bios as well to make sure that wasn't corrupt. I've definatly found the problem though. I can get it fairly stable now, if i overvolt the NorthBridge to 1.65v (maximum), but as soon as i lower it down, i get an almost instant crash. I'm hoping running at 1.65v will finish it off soon, so i will be able to rma it with the distributor. At the moment, under load @ 1.65v, I can't touch the heatsink it's that hot, so it's gotta be around 70c upwards. Shouldn't last too long.

Thanks

 

[POGE]

yeah, i tried re-flashing the bios as well to make sure that wasn't corrupt. I've definatly found the problem though. I can get it fairly stable now, if i overvolt the NorthBridge to 1.65v (maximum), but as soon as i lower it down, i get an almost instant crash. I'm hoping running at 1.65v will finish it off soon, so i will be able to rma it with the distributor. At the moment, under load @ 1.65v, I can't touch the heatsink it's that hot, so it's gotta be around 70c upwards. Shouldn't last too long.

Thanks

You could just kill it with static if you wanted to speed the process up. Just rub a sock on your head and then rub the northbridge (with heatsink off).

 

[infrared]

hehe, i'll see if we've got a static generator at college, i'll zap it several times with that

 

[Sasqui]

hehe, i'll see if we've got a static generator at college, i'll zap it several times with that

Does it look anything like this?

Built a 3ft tall Tesla coil in highschool, darned physics teacher only gave me a B+ . Probably becuse I almost set the class on fire. I love the smell of ozone in the morning!

 

[infrared]

Wow, awsome!

I may hafta have a go at making one! How is it powered, and how many watts does it draw?

 

[Frogger]

infrared; you mabe on the right trak about the volatage try unpluging the psu for a couple hours and see if will work at stocks again F

 

[infrared]

nope, won't even post at stock volts anymore

 

[Sasqui]

Wow, awsome!

I may hafta have a go at making one! How is it powered, and how many watts does it draw?

It's a labor of love. The cheapest way is to get an old furnace starter - they output roughly 50kv, but the amperage is tiny. I'd assume going that route would draw less than 1000 watts - I can't remember exactly. The Tesla coil bumps the voltages into the Megavolt range, at very high frequencies. The hardest part of it is building the secondary coil - took me many, many hours. 1000's of tight copper windings arouns the mast - 3 or 4 inch PVC pipe in my case - it was 2 decades ago. I think my parents still have the coil in thier attic . Plenty of info on the web to build one.

Wow - talk about going OT!

 

[infrared]

sweet

 

[Sasqui]

sweet

Too bad you can't zap the MB to life... watch the irony, you'll RMA the MB and find it was something else.

 

[infrared]

Hmm, i've had the northbridge on this board at it's maximum of 1.65v for a couple of days now... no death, so i tried packing the heatsink with toilet paper and playing games... not death, so i tried carefully swiping a blowtorch across the HS, which got it way over 100c, and it still live on

How should i kill this stupid thing?!?!?! It'l probably be bullet proof too i suspect!

I'm thinking i might get a conductive pen, and bridge a few legs on some chips inconspicuously.

 

[POGE]

How bout just take of the heatsink and run prime?

 

[infrared]

hehe, yeah ok, i'll give that a try

 

[Solaris17]

that would fuck it up but if he rma'd they may notice he took it off


hmmm how to kill an operational computer................ground the mobo?

 

[POGE]

If he tries to kill his computer and not just the northbridge he could also damage other components. Usually they dont check under the HS, and if he still has the origional thermal tape on there he should be fine.

 

[infrared]

ah, nice idea. i was gonna break a few connections, but i was worried it might overvolt any of the other components by sheer bad luck.

I'll try swiping a grounded wire around the contacts around the clock generator. or get two wires of a small 12v battery, and burn out a few chips with the comp powered down, this seems the safest. I could remove all the components, burn out a chip or two, re-assemble it on the bench, and make sure it's dead.

I feel really bad doing this, I love this motherboard

 

[POGE]

ah, nice idea. i was gonna break a few connections, but i was worried it might overvolt any of the other components by sheer bad luck.

I'll try swiping a grounded wire around the contacts around the clock generator. or get two wires of a small 12v battery, and burn out a few chips with the comp powered down, this seems the safest. I could remove all the components, burn out a chip or two, re-assemble it on the bench, and make sure it's dead.

DONT. Plugging parts into a dead motherboard can kill parts. Just a warning. Back in the days of Socket A I had a hole rig taken down by a faulty motherboard. :shadedshu Todays motherboards have a lot of protection to prevent this from happening, but better safe than sorry...

 

[infrared]

hmm, i'm thinking i could blow up an insignificant chip like the realtek sound, or marvel gigabit lan

 

[POGE]

Just take a razor blade and swipe in under the chipset. Make it disconnect one of the soldier points.