Ga-p35-ds3l no-post and BIOS settings

Nora

Distinguished
Sep 3, 2008
17
0
18,510
Hi, all!


My System:

GA-P35-DS3L v 2.0 (BIOS F6)
Intel(R) Pentium(R) 4 CPU 3.20GHz, 3200 MHz (+Scythe Katana 2)
Kingston 1GB 800MHz DDR2 Non-ECC CL5 DIMM (KVR800D2N5/1G) x1
Seasonic 380W

Gigabyte Radeon X600 Pro (+Passive Cooling)
Silicon Image SiI 0680 Ultra-ATA/133 Medley RAID Controller
Hitachi Deskstar 80 GB, 7200 RPM, Ultra-ATA/133 (HDS728080PLAT20)
Seagate Barracuda 160 GB, 7200 RPM, Ultra-ATA/100 (ST3160021A)
TSSTcorp CD/DVDW TS-H552B

Floppy Drive
USB 2.0 Multi-Card Drive

Windows 2000 Professional SP4


NOT Overclocking
All components are supposed to be compatible
RAM has been mostly on the 3rd slot for space issues.


My problem:

I had this mobo (and setup) working since early september with no problems at all until about a couple of weeks ago.
The first weird thing that happened was my cpu fan not starting. I thought nothing of it at the time. The fan was fairly new, but stuff happens. I replaced it.
Some days later I turn on the PC and - nothing. Fans turned but no boot, no beep, no post, no drives powering on. A few hours later it was working again and then it was dead again.
I tried taking out the battery and clearing the cmos.
It worked for another couple of days, then went back to being dead.
I took everything out of the case and proceed to test it. I thought it was either the mobo or the RAM.
Out of the case it was running again. None of the components seemed to have a problem. I ran memtest for 20 passes and the RAM showed no probs. And that old fan turned out to be working just fine.
I checked if there was any chance of something being shorting out with the case (saw nothing) and even went as far as insulating all the screws with paper!
I put everything back into the case.

It worked fine for another 12 days. And then it died again...

I took it out again, tested everything again (except for running memtest), cleared de cmos with a screwdriver for some 10 seconds (didn't reset the BIOS). When I turned the PC back on it worked again and it's now working on the floor and under the desk as I type. :D


My Question:

So I'm now thinking something in my BIOS settings is probably not getting along with the system.
It occurred to me that for the first 6 months I used the PC for little more than surfing the web and writing, but this last week I went back to working with Photoshop and 3D, and having a lot of stuff running at the same time and putting a stress on the hardware.
I don't usually mess with the BIOS except for basic config, don't know that much about it, and might be missing something here.


Here my MIT settings:

Robust Graphics Booster [Auto]
CPU Frequency 3.20GHz (200x16)
CPU Host Clock Control [Disabled]
CPU Host Frequency (Mhz) (200)
PCI Express Frequency (Mhz) [Auto]
C.I.A.2 [Disabled]
Performance Enhance [Standard]
System Memory Multiplier (SPD) [Auto]
Memory Frequency (Mhz) 800 800

System Voltage Control [Manual]
DDR2 OverVoltage Control [Normal]
PCI-E OverVoltage Control [Normal]
FSB OverVoltage Control [Normal]
(G)MCH OverVoltage Control [Normal]
CPU Voltage Control [Normal]
Normal CPU Vcore 1.36250V


Any thoughts?
(Oh, and sorry for the long post... ^^j)
 

bilbat

Splendid
This sounds to me to be an intermittent power supply problem - does it seem at all to be temperature related?

Oh - and, BTW - don't insulate your standoffs/screws - they are meant to be part of the MOBOs grounding solution - that's why the holes in the MOBO are surrounded by radially oriented solder pads - it's for good contact to case ground...
 

Nora

Distinguished
Sep 3, 2008
17
0
18,510
that's why the holes in the MOBO are surrounded by radially oriented solder pads - it's for good contact to case ground...

Ah! I had wondered about that. :) Thanks for the tip.

The PSU is about a year old, I think, probably less. It's better and has a bit more wattage than the previous one. I don't know... it seems to be working OK.
I'm putting off having to test with another psu because it will mean having to fix the old one.
 

bilbat

Splendid
Here's another trick that may make a difference...

workstation001s.jpg


notice the green wire in the lower left? It's from a screw in a corner of the power supply where the paint's been ground off (an emery board works great) to bare metal in the case floor (Cosmos - all bare metal); I especially need it, as I have my PSU mounted using a silicone rubber noise isolation pad, but, lots of times, the PSU installs in such a way as to supply an 'uncertain' case ground, which can make a (big) difference - especially prone to intermittent wierdness...
 

Nora

Distinguished
Sep 3, 2008
17
0
18,510
Lightning rod style, eh? :)
But I'm a bit confused - does the psu paint provide such good insulation? Whithout any silicone involved, I mean. My PSU is in full contact with the metal on the back of the case and the 2 metal bars that support it (mine is on the top side).
 

bilbat

Splendid
Yeah - most paints make good insulators; but, the reason I use the 'screw & be sure' method is partial insulation: might depend on the temperature, how the screws 'seat' today - crappy grounds are notoriously hard to troubleshoot or diagnose, so old habits tell me just to eliminate 'em whenever possible. There are three grounds (at least - I'd have to think for a while - might find more) involved: electrical ground (established through the third wire of the AC plug [and then, hopefully, the 'premises' grounding of your outlets]); electronic ground (established by the ground (black) wires of your system PSU connectors); and case ground. Grounds are funny - you'd think they just be, well, grounded... But often not so - if grounds are not carefully managed and fully established, it is entirely possible to have circulating currents (known as 'ground loops' - anywhere from a μA up to a few mA) in the grounding system. I do industrial systems where sensors may be blocks remote, and run through standard building wiring conduits - I rely on shielded twisted pairs, with very carefully managed grounding of the shields to carry these signals - a grounding error can lead to, say, me reading pH in a chem tank to within a couple of points (out of fourteen) instead of within a couple of hundreths - a disaster! The problem ground loops cause is that everything your system reads, from disk data to memory access, is essentially read by a threshold comparator referenced to ground, and if that ground is randomly floating and fluctuating, you got all kind of wierd problems - seldom showing symptoms (unless you've been bitten before) one would relate to a missing ground wire...

Cheap and easy insurance!
 

Nora

Distinguished
Sep 3, 2008
17
0
18,510
Only 2 grounds for me. The only grounded sockets in this building are the ones in the kitchens... :sarcastic: A good thing current around here is pretty stable.
These things weren't mandatory 30 years ago, I guess.

I suppose I can try wrapping a clip or bit of wire around the ventilation grid and connect it to the case... But I better wait a few hours for the PSU to discharge, just in case my aiming gets as rotten as usual. :D

Thanks for the tips. :)


Oh, wait - would the lack of ground in the building make that extra wire dangerous?
 

bilbat

Splendid
maybe not dangerous, but possibly a problem - I know where you're coming from, when I first moved in with my ex, we lived in an apartment that, I swear, was wired by ol' Tom Edison himself! The first time I needed to work on the wiring, I told her "we've got to move, as soon as possible... " Then, I met the neighbors in the flat behind us, and discovered that their apartment was a maze of little paths between floor-to-ceiling piles of old (and I mean, 'first world war' old) newspapers - that sealed the date of moving to like, last week!
 

Nora

Distinguished
Sep 3, 2008
17
0
18,510
I just realized it can't be the grounding 'cause the back screws attach to holes with no paint - meaning, it's the same as having the wire.
Mmmm.... Back to the "drawing board"...

Edit - We were typing at the same time :)
Yeah, I had some neighbors like that. Their flat burned. The firemen had to evacuate the building because of the gas sytem.
I guess it might be problem no matter what, then. On the other hand, the PSU does have a bunch of failsafe systems, including power surge protection if I'm not mistaken...
 

bilbat

Splendid
OK - next investigation: what cooling stuff have you got on the CPU/in the case? The one in the picture was designed specifically to move heat, and even it finally dicovered a way to heat-stress itself into rebooting - I started transcoding video in the background, while watching, pausing, and recording TV...
 

Nora

Distinguished
Sep 3, 2008
17
0
18,510
CPU - Katana2 with its scythe fan; Arctic Silver Ceramique (Speedfan currently reads 38ºC. I had 34ºC on the BIOS reading yesterday)

Case - small noisebloker fan on the HD removable block (up front and more or less aligned with the GPU); another scythe fan on the back (aligned with the katana). I don't always have them on, depends on how hot it gets.
It also has 2 grid holes on the side.

HD%20Fan.jpg


PSU - one large fan (ends up being more or less over the space between the katana and the back fan. I don't think there's any vacuum, though).
It's surprisingly cool compared to my previous one (a Chieftec).

GPU - One monster Accelero2 passive sink that benefits from the HD fan (not that it needs it - I kind of overdid it on that one... :D).

naked%20PC.jpg


I'm probably going to reassemble the PC in a couple of hours. I'll get a few more pics then.


 

Nora

Distinguished
Sep 3, 2008
17
0
18,510
OK, current status:

I've had a few more bets on a bad PSU (nothing specific, just random experiences of nasty stuff happening), so, what the hell, I'll try running the old one for a few weeks. Here's the beauty :D :

Old%20PSU.jpg


That's the fan I usually have on the back of the case, btw.


And here´s the pic of the full case:

Full%20PC%201.jpg


There's this electronics repair shop where I sometimes go, I might take the seasonic there to get it checked since I have no PC to test it at the moment.
 

Nora

Distinguished
Sep 3, 2008
17
0
18,510
It's a small shop, not part of any chain or anything. The guy might even do it for nothing.

But those are cute! :) Didn't know about that.
I can only find one on a local store after googling it up and it costs WAY too much. Come to think of it, it costs more than the PSU did. :eek: LOL!
I actually have 3 or 4 meters that belonged to my father but never got around to learn how to use the things... *sigh*
Anyway, I want to see what happens first. It might yet turn out to be a mobo problem.
 

bilbat

Splendid
I like your 'side-saddle' fan there - what is that, the tail rotor off a surplus military helicopter?

Sorry about slack-time getting back - I've been busy - radiator comes tomorrow:
water1002s.jpg

water1006s.jpg
 

Nora

Distinguished
Sep 3, 2008
17
0
18,510
Nah, just the humble twin of the sink fan. The one that broke down on the PSU was a slim profile, couldn't find one of those. :p

Woh! You have been busy What's all that stuff on the floor?
 

bilbat

Splendid
Floor?
Top picture is basement ceiling, directly below my computer; pump and power supply to the left, wooden thing hanging down (temporarily - if I calculated right, radiator will exactly balance level) is radiator support frame...
bottom is wall plate next to comp - fittings to take evil heat downstairs, & dispose of it - I can just about slide a pizza onto my keyboard drawer to bake!
 

bilbat

Splendid
Thought of something to try (from the pictures, I'm guessing you've got cooling covered): have you run memtest on it? It dawns on me that (unfortunately) Kingston and Corsair would seem to share a common problem: they seem to sometimes 'degrade' over time - i.e., all of a sudden, a working config won't run, or will run erratically...
 

Nora

Distinguished
Sep 3, 2008
17
0
18,510
Floor?
Top picture is basement ceiling, directly below my computer;
LOL!
Seen like that it looked like a floor to me. I'm not used to seeing wood on the ceiling, I guess. :p

Thought of something to try (from the pictures, I'm guessing you've got cooling covered): have you run memtest on it?

Yeah, the first time it it happened I ran memtest for 20 passes. It reported no errors...
I find it really annoying not understanding why things happen. Grrr!
 

bilbat

Splendid
Jeeze - you're a hard case - I'm not used to dealing with people here who have already covered the bases; mostly, it's folks who don't know where the bases are, or don't know what a base is, and likely spell it 'bass'!

I find it really annoying not understanding why things happen. Grrr!

I'm with 'ya there, but, doing industrial systems, I've become pretty inured to it - never know 'is the basic chemistry wrong', 'does the wiring to the sensor work', 'do I have a defective board' or [never really a serious consideration] 'is my software not working'? One time, at a brewery, a bird had made a nest in a sensor j-box that the electricians had forgotten to plug all the extra conduit holes in...

Anyway - the BIOS settings you've posted look good to me - but I've really got to go over to TweakTown and review some things about Kingston memory, which may be a weak spot. Basically, the memory hierarchy for GB boards is: 1 (but pricey) mushkin - GB boards seem to 'like' it, & seldom need an adjustment to run up to 8G at stock settings; 2 G.Skill, again, runs great, but might need a tiny tweak; 3 Kingston - often takes some adjusting to get multiple sticks to work; 4 Crucial, which often refuses to run even 2G at stock speed w/o major BIOS surgery, and tends to go to hell over time in more than a couple cases... But still, I've never seen a problem that boiled down to memory timings that MemTest didn't find first, and efficiently - it's usually a great little 'bloodhound'. I'm trying to think of what else 'stress' is actually affecting at the hardware level - temperature, memory bandwidth, MCH, not much left - lemme stew a little and I'll get back with more questions, and a few suggestions for further testing...
 

Nora

Distinguished
Sep 3, 2008
17
0
18,510
Jeeze - you're a hard case - I'm not used to dealing with people here who have already covered the bases; mostly, it's folks who don't know where the bases are, or don't know what a base is, and likely spell it 'bass'!
Hehe! Sorry. :)
Even if we never do figure this one out, I do appreciate you taking the time to go through all this troubleshooting.

I wasn't exactly thrilled at the prospect of getting a Kingston because I had once read some less nice things about their flashcards, I think. But then I did some specific research on the RAM and nobody seemed to be complaining more than usual, so... Besides, most of the RAM on local catalogs is meant for OC geeks. Too much money for stuff I don't need.

I don't know if this means anything but I installed Everest to check some info I needed and I'm getting a warning message on the DMI. It's something like "not being able to garantee DMI's precision"...
 

bilbat

Splendid
De nada... No altruism involved; I learn a lot here, which can never hurt anyone (unless, I guess, you learn too much about the government - then, one fine day - you just 'poof' - disappear!)

Been busy:
radiator001s.jpg


About the DMI error - usually, they're basically SM Bus troubles; either it's not installing, or (and, for some reason, Catalyst seems to often do this) it's been 'hijacked' by an ATI driver, when it's an Intel device... It's supposed to be on the ICH9R - it's on Device Manager under 'System Devices', ID 2930 (I think) - if it shows as 'ATI' SMBus controller, just uninstall it and re-point a new install to an Intel chipset cluster. Just wanted to drop a quick note yet before crashing, will get back AM if it's a problem...
 

Nora

Distinguished
Sep 3, 2008
17
0
18,510
A little update on the situation, for anyone who may one day need the info. :)

Over a month has past and I have had no more problems while using my old Chieftec PSU. I have yet to send the Seasonic PSU for RMA (been busy). It's still within warranty so I decided to take it back to the store and let Seasonic figure it out. However, I was going to run a little experiment today and, after removing the Chieftec, I noticed something I had failed to see before and would have saved me (and other people) a lot of wondering... The CPU's electric plug is fried! 0.o

CPU%20Plug%20.jpg



*Note to self - stop assembling computers at night!*