A RAM triggered query about PSU Wattage

[goosee]

Hi Guys

Not really sure where this post belongs, but I'm opting for here as RAM seems to have be playing a role in my problem.

My set up is pretty much that which I detail in my sig, below. Before I got my 6800GT, I was running a 9800Pro and all ran very sweetly off my 430W Antec PSU. However, since I added the 6800GT I started to have random lockups on boot and also crashes to BSOD from Windows. This would happen sporadically, and I could never recreate the crash. I tried all the classic GFX Card tests/fixes/theories, but to no avail. The crashes were very random and often difficult to pin on the GFX Card.

I recently swapped out the OCZ RAM (which is faultless in Memtest and worked fine with the 9800Pro) and replaced with some old 1GB Corsair 3700. The system is now (almost) completely stable.

Based on the fact that the 9800Pro worked fine with the OCZ and the 6800GT works fine with the Corsair, I theorise that the crashes are not related to hardware faults, but to a lack of juice getting to the componants from time to time.

As well as that stated in my sig, I run a lot of USB kit and a pro soundcard and suspect that the 430W is simply not enough for the system.

So, good diagnosis, easily fixed by 550W (which PSU would you recommend by the way)?

Or should I be picking another tree to bark up?

Thanks for your time.

[h20-ski]

Actually that is a good hypothysis. The best way to validate this, is to unplug, unhook all of the non essintials and see if there is some stability gained. If there is stability from unplugging extra devices, then the PSU might not be up to the task.

[twajetmech]

You are correct, the new 6800gt is a bit more power hungry than your old 9800. That 430W psu was living on a thin line as it was and now with the added load from the 6800gt it can no longer meet the demands. The HD's you have are also quite power hungry. Chances are your memory issues were due to power flucuations from the over stressed psu. The mobo you have is an excellent one, but tends to have poor voltage regulation therefore I suggest you invest in a good quality unit from PC Power & Cooling or similar with very tight regulation and strong rails to support your system (look at Enermax, Fortron and Sparkle as less expensive alternatives) I would look for units of 500W or better. I also reccomend, if you have not done so to update the drivers of the 6800 to the 81.89 release, which I have found to be quite stable. Also remember to adjust your psu requirements by factoring that any usb device you plug into if it is to be powered by the usb port is an additional 15W load on the psu. And, finally, Corsair memory usually require a slight bump up in the ddr ref voltage from what they advert as stock voltage, I would expect to run at least 2.7-2.8 volts to get them runing as they should

[goosee]

Thanks for the replies guys. I'll go and do some shopping!

[specmike]

Since you have concluded that the video card is likely the culprit and since you are PSU shopping, I'd recommend looking for a PSU with very solid 12v rails. I'd suggest 24a minimum and 30a+ to be sure. This can be a single big 12v rail or the total of 2+ 12v rails. This is more important than overall wattage.

PC P&C is reputed to be the best PSU on the planet. I have personally had good luck with Enermax. You can't go wrong with either.

Mike.

[goosee]

Hi

Thought I'd bump this rather than start a new thread as it's all related. As you'll see from my earlier posts I was having various random problems that I couldn't quite narrow down, but was suspecting a graphics card/PSU issue. A few months down the line, after further tesing (and holding back on a new PSU - due time issues) I am now convinced it is linked io the RAM.

BUT I'm not sure whether it's the RAM itself or simply the power getting to the RAM, due to the fact problems only occur from a cold boot. See below for my updated explanation of the problem, which I have also posted at Bleeding Edge

I've had this problem for a few of months now and in the process of trying to solve it myself have encoutered many similar problems (including some on this forum) but none exactly the same.

My problem is that (almost always, though not every single day) on its first boot of the day (ie after at least 8+ hours down time) my system will either lock-up on the Windows boot screen (black screen, XP logo, blue progress bar) without any BSOD or make it into Windows and BSOD within about 2 minutes. The strange thing is that, upon reboot, everything becomes fine and 100% stable for the rest of the 'on-time'.

When this problem first started, the BSOD errors would be very random. error codes would claim to relate to driver issues (according to the MS KB) or point to PAGE_FAULT_IN_NON_PAGED_AREA and other seemingly unrelated stuff. More recently however, the BSODs do not report anything specific and only returrn bug codes that MS KB reports to be a RAM error.

Straight-up faulty RAM I could accept, but I am bemused as to why this only happens off a cold boot. In general, I can play demanding games, run benchmarks and stress tests all without issue.

Interestingly, Memtest follows a similar pattern to Windows. If run after the PC has been going a while, the tests sail through. I've regularly left it to run through the night from this stage and it reports zero errors. If run from cold, however, memtest often reports errors from the off.

I'm almost at a loss. I've been on the verge of investing in a new 600W PSU in the hope that it was a power issue (some of my research had pointed at this), but now I'm not so sure.

Anyway, any advice or suggestions for further testing would be most appreciated.

Many thanks

(PS As far as I can tell, my BIOS is set-up correctly and all drivers are up to date)

UPDATE.....

I ran Memtest from cold this morning and, as has been the trend, it reported errors after about 20 seconds. I left it to continue for 2 minutes, then restarted the testing process (via the key-press option). Surprise, surprise, the rerun test had not reported one error after 20 minutes.

Any thoughts on this guys?

[3 of 7]

I'm going to go out on a limb with a guess here and suggest reseating the video card yet again (wouldn't hurt to do the ram while your in there) ..... Hot parts are physically bigger and can mask a poor connection, and your problems began when you disturbed these connections.

Are your Mobo mounting screws nice and tight?

[Gorganzola]

During this testing did you use the Corsair or OCZ memory?

In either case, have you tried bumping the voltage in the BIOS that is available to the memory? Try setting it to 2.7V or 2.8V as previously suggested.

Beyond that, just because memory test reports memory errors doesn't necessarily mean it is the memory. Low voltage to the memory can cause it. Not enough power to the CPU can cause it. Bad CPU can cause it. Issue with the motherboard can cause it. The third planet being in the wrong house can cause it

When does the instability occur? Sitting at the destop? Surfing the web? Playing a game?

Originally Posted by goosee

Hi

Thought I'd bump this rather than start a new thread as it's all related. As you'll see from my earlier posts I was having various random problems that I couldn't quite narrow down, but was suspecting a graphics card/PSU issue. A few months down the line, after further tesing (and holding back on a new PSU - due time issues) I am now convinced it is linked io the RAM.

BUT I'm not sure whether it's the RAM itself or simply the power getting to the RAM, due to the fact problems only occur from a cold boot. See below for my updated explanation of the problem, which I have also posted at Bleeding Edge



UPDATE.....

I ran Memtest from cold this morning and, as has been the trend, it reported errors after about 20 seconds. I left it to continue for 2 minutes, then restarted the testing process (via the key-press option). Surprise, surprise, the rerun test had not reported one error after 20 minutes.

Any thoughts on this guys?

[Protoplasm72]

So you are still using the same psu? I'd bet my donkey that it is a psu problem. On my old system I upgraded my video card and a week or 2 later I started having trouble cold booting. It would take 3-4 restarts before it would finally load without giving my a blue screen. It took a while to register in my head but I finally realized it was a power issue and bought a 530 watt ps (old one was 380 watts). It worked perfectly after that.

[twajetmech]

As much as I love my P4C800-E Del mobo....it is notorious for voltage fluctuations and supplying slightly lower voltages than what its set too, (long has been Asus' weak point for their vrm's) so bumping both the cpu volts up a bit or setting it to auto would help, the memory on the other hand would certainly benefit from a slightly higher voltage....don't worry the mobo can't supply enough volts to do any damage to your memory as its limited to 2.85V max. While I do believe the voltage bumps will help you, since you are only having this issue on cold start, I susspect that your psu is unable to provide enough power during startup, and thus giving you the instability you speak of. Since Corsair mem almost always requires a bit more volts than they advertise and its running at pc3700 (is your ocz the same speed ?) it will require more than the average 2.65V to run with tiight latancies and at those speeds. I would also make sure that the ras to cas timing of the memory is set to 3....the intel 875 chipset can be quite tempermental with a ras to cas of 2. You could try timings of 2-3-2-6 (3 being the ras to cas) or 3-4-4-8 to take the memory timings out of the picture and see if you still have the instability. I will also say that when I installed my 6800gt, I found that since it didn't exhaust the hot air out of the case, but instead blows fwd, it caused my memory temps to skyrocket....thus until I had figured that out and solved it I too had sudden crashes as a result of overtemp'd memory (granted I'm oc'ed to 3.8 with a ocz booster....mem is @ 3.0V 24/7) Hope this helps you out.

[goosee]

Hi guys

Thanks for all your responses. Ok here's the deal:

All my recent testing has been with the OCZ 2GB Plats at 2.85v. The Corsair mem is simply some old stuff I've got, which has never given me trouble and still doesn't. I simply used it as a 'control'. With the Corsair mem installed, I can boot from cold every time.

Ignorant question, but are two 1GB sticks of OCZ at 2-3-2-5 gonna suck more juice that two 512MB sticks of corsair at 3-4-4-8 (with mobo RAM voltage at 2.85v)? If the answer is a definite 'yes', then I'm pretty much convinced it's my PSU. If not, I find it hard to accept that the OCZ is faulty, as it runs so well from 2nd boot onwards. :confuse2:

I'm looking at the Seasonic 600W now!

EDIT: Couple of things I missed.....

If it makes it through the boot to Windows, the BSOD either occurs whilst services are finishing loading (i.e. straight away) or when I go about my business (eg open Outlook, I.E., etc). There doesn;t seem to be any particular pattern other than it's within about a minute of my desktop appearing.

When I boot tomorrow, I'll try the OCZ at 3-4-4-8 as twajetmech suggests to rule out a timings issue.

Cheers guys

[FightingChance]

Bollocks this is a power problem. I run just about the same setup on my Antec 380W TruePower I and it's fine. Maybe the quality of power, but not the quantity.

[twajetmech]

2 1Gb memory modules are going to use more power than 2 512Mb modules, but in all honesty the difference should not make or break the stability of the system, however trying to address 2Gb of memory does in fact put a lot more stress on the memory controller which can indeed cause instability...one it would need a bit more power to address that much more memory, which as a result would also generate more heat....this would also cause instability.....to rule that out run the computer with the side of the case open and with a desk fan blowing into the case...if its no longer giving the bsod's then you have a heat issue, if it still bsod's then your probably in need of the psu. I would also suggest that you try one memory module out at a time to rule out if you have a bad module of memory. Make sure that in the bios Turbo is disabled, spread spectrum is disabled and legacy usb is disabled (all well known causes of instability)

[specmike]

Originally Posted by goosee

Ignorant question, but are two 1GB sticks of OCZ at 2-3-2-5 gonna suck more juice that two 512MB sticks of corsair at 3-4-4-8 (with mobo RAM voltage at 2.85v)?

Cheers guys

Is 2,3,2,5 the EXACT order you are listing for your timings? If so, try 2,2,3,5 on that specific MOBO as the RAS to CAS delay needs to be 3, not 2. The AMI bios lists them in a different order than than the older Phoenix, Award, etc.

Mike .

[h20-ski]

Voltage fluctuations can cause memory instability. I have tested with my volt modded PC-DL. If I pull the trimmer off for a second, and then put it back on, it will cause errors in Memtest. Also back when my main rig was folding QMD's the OCZ Powerstream voltage LED's would flash. If I pushed the system up to the maximum overclock, the 3.3V LED would flash like a Christmas tree, but while at stock the LED would flash only extreme, extreme memory transfers.