Who wants to butt heads with HP alongside me?

LawnMMLawnMM Colorado
edited March 2004 in Hardware
GF's comp was freezing up on her. Its an HP so naturally I asked her why she thought something was wrong to begin with :p

Anyway, long story short...she's got an AXP 2200+ I believe it is, 1.8ghz, being cooled with whats essentially a glorified chipset fan. Locked up after about 15 mins with prime95 running. I put a beefier fan on her heatsink and I'm gonna get a stronger exhaust fan for the duct they have inside her case.

Its not locking up now, but its still failing the prime95 test after 45 mins or so...this is where the question comes in. How abnormal is that? Its an HP so I doubt I'm gonna be able to raise her chipset/ram/cpu voltage...

Should I continue beefing up the cooling in hopes of further stabilizing the machine and getting it to pass prime95 with flying colors, or should I not even bother trying to get an HP to pass prime95 in the first place? She just uses it for surfing and Word/Excel/Outlook so I doubt a minor math calculation error here and there is gonna be an issue...but the tweaker side of me wants to make every effort.

Suggestions...thoughts? Long as it doesn't freeze up on her she'll be happy as a clam and I think that much is already set, I'm just wondering if I have a chance of getting it to pass prime95, who else has dealt with these damn things?

Comments

  • madmatmadmat
    edited March 2004
    Make sure that the duct fan is blowing in and not out as it could be nullifying the cpu fan to an extent.
    If it is blowing out simply reverse it to funnel air down onto the hsf and see if that helps.
  • LawnMMLawnMM Colorado
    edited March 2004
    Not a bad idea, maybe I'll reverse the HSF too...
  • madmatmadmat
    edited March 2004
    Umm...reversing the HSF will just be creating another dead spot unless it's sucking air in the bottom already. You'll end up with the duct blowing air down onto the HSF and the HSF blowing air up into the duct if you see my point.
    The best solution is to have the fan duct blowing onto the HSF and the HSF blowing down through the fins of the sink.
    I've seen HP's that have been configured either way...one will have the duct feeding cool air onto the HSF and the next will have it drawing it out of the case, they are very inconsistent.
  • gtghmgtghm New
    edited March 2004
    LawnMM wrote:
    Should I continue beefing up the cooling in hopes of further stabilizing the machine and getting it to pass prime95 with flying colors, or should I not even bother trying to get an HP to pass prime95 in the first place? She just uses it for surfing and Word/Excel/Outlook so I doubt a minor math calculation error here and there is gonna be an issue...but the tweaker side of me wants to make every effort.

    Suggestions...thoughts? Long as it doesn't freeze up on her she'll be happy as a clam and I think that much is already set, I'm just wondering if I have a chance of getting it to pass prime95, who else has dealt with these damn things?

    I doubt that beefing up the cooling will do much to stop the prime95 errors.
    Usually with prime95 and memtest errors are either hardware related or voltage realeated. Givin its an HP machine I would say that heat is not the issue. Weather or not people beleive it, CPUs and stuff can take quite a bit of heat. Beleive me when I say that 85% of the people that have cookie cutter machines have no idea what their temps are or even care, which is why they are rated to work under pretty high temps.
    As long as you are able to get air to the parts in the same volume levels as it was when it was brand new I would say that the temps would be fine and not causeing the errors.
    I would say that more likely is the case that something is failing. I would consider either the memory or the PSU.

    I know that there are temp zelots that will argue with me... As long as your caes temps are below 45 to 50C and the CPU temps are below 60C you should be fine.

    Check the memory or the PSU...

    "g"
  • LawnMMLawnMM Colorado
    edited March 2004
    Its not the memory. That was my first suspect, but I've swapped it out with a good stick of my own pc2100 and it still froze. I'm certain at this point that its heat related as it became much more stable when I put a bigger fan on the heatsink (you wouldn't believe the little 60mm thing they had on it to start with) and a new exhaust fan. It still freezes up within an hour or so if I start folding and walk away with the case closed. Open the case up and it went something like 8-9+ hrs before it froze.

    What I'm having trouble figuring out is why it froze again if the processor was under load the whole time...why after 9 hours, why not sooner?

    I'm aware that having the fans going in opposite directions is counterproductive. I was saying I was considering reversing them both. The setup was currently up off the heatsink through the duct and out the back via exhaust fan. I removed the duct from the equation while I had the case open so it just shoots the hot air right out of the side of the open case and that seems to be a lot more stable.

    I know its not memory, and while heat seems to be a factor it may not be the smoking gun. I don't really have a way to check the PSU, it appears to be functioning but I'm sure HP uses some silly proprietary PSU and motherboard so I can't plug a regular ATX psu in to see if its a power problem. We all know their bios is crap so I can't up her vcore or memory voltage to try and stabilize it.

    Honestly, I don't really care a whole lot about the prime95 errors. I just want to get it to stop freezing up. It has integrated video and sound so its not a card heating up. I really thought it was the cpu overheating. Her heatsink was noticeably harder than my pal 8045 thats on my 1900+ rig and that thing feels pretty cool even at 45-50C. It ran noticeably longer without a problem once I upped the cooling a bit and left the side of the case off, by almost a factor of 10. I'm just baffled as to why after like 9 hours it decided to freeze then, something changed, I have to figure out what.
  • muddocktormuddocktor
    edited March 2004
    Lawn, did you take the hsf off and clean it? I figure that you did but I just thought I'd ask.

    Do you have access to a digital multimeter by any chance? You can check all the various voltages from the back of the ATX plug with the leads of the DMM and check to see if the pinouts are the same as a standard ATX psu that way also. If you have a fairly recent mobo manual it will have the various pinout volts listed on the ATX plug. Just do it while it's booted up and you should be able to see if any of the readings are way off from what they are supposed to be.
  • LawnMMLawnMM Colorado
    edited March 2004
    Yeah, I cleaned up the heatsink. Its some little chincy aluminum thing. It was dusty but that was about it. Its only rigged for a 60mm fan, I put a stronger one on it, but maybe it needs a real HS and fan to keep it happy. I've never seen somebody try to cool a 1.8ghz amd chip with a frickin 4 dollar heatsink and a fan smaller than stuff I have cooling my chipset.

    I don't have a multimeter so I can't check the pinouts on the PSU.
  • gtghmgtghm New
    edited March 2004
    How about the possibility that the CPU is crapping out?
    Got another AMD CPU around that you could try?

    Just a thought,
    "g"
  • muddocktormuddocktor
    edited March 2004
    A digital multimeter isn't too expensive, Lawn. You can get a DMM for around $20-25 that will be plenty accurate enough for what you need. That's what I have.

    G's idea is also a good one, since that poor Tbred has had such a hot life. It might be getting ready to go tits up totally due to all the heat it's had to live with because of HP's super cooling system.:rolleyes:
  • seversphereseversphere
    edited March 2004
    picture of this heatsink or link to pic of one like it?
  • profdlpprofdlp The Holy City Of Westlake, Ohio
    edited March 2004
    seversphere wrote:
    picture of this heatsink or link to pic of one like it?
    Ghoul! :zombie:
  • LawnMMLawnMM Colorado
    edited March 2004
    I've definitely made some progress but I'm still baffled. I put a better fan on the heatsink, and reversed the fans direction so it blows down into the heatsink rather than away from it as HP had it set up. I also reversed the exhaust fan (don't get started cooling purists!) so that it pulls cool air in and shoots it down the duct right out onto the cpu fan and heatsink. That seems to have all but eliminated heat as a problem.

    Whats still funky, is that for some reason after like 9 hours of running constantly with folding going to keep the cpu hummin' at 100%...it still freezes up. I'm trying to figure out what changes in the 9th hour that causes instability.
  • madmatmadmat
    edited March 2004
    Check task scheduler to see if there's some app firing up to cause it.
  • profdlpprofdlp The Holy City Of Westlake, Ohio
    edited March 2004
    Good idea, madmat!

    You might also check things like AV auto-update stuff, etc. Try killing all extraneous processes and disconnect from the network/Internet. Maybe even leave it on in safe mode overnight.

    Anything in the event log?
  • Geeky1Geeky1 University of the Pacific (Stockton, CA, USA)
    edited March 2004
    As for the cooling, IMO, you should turn the fan in the duct back around so it blows out. The reason the duct fan is like that is to pull the hot air being exhausted out of the heatsink out of the case. Having it blow the other direction may cause problems with it pulling warm air from the PSU in and blowing it across the CPU.

    I'd put the fan back the way it was, take the duct off, and get a halfway decent heatsink for it. They're so cheap now that there isn't really any reason not to.

    For example, for $12, you can't really go wrong with this...
    http://www.nexfan.com/evmocod9uxp3.html
  • Geeky1Geeky1 University of the Pacific (Stockton, CA, USA)
    edited March 2004
    Oh btw, it could just be that the board is on its way out, too. Electronics die for no apparent reason sometimes (especially when I've been working on them... ;D)
  • muddocktormuddocktor
    edited March 2004
    Look at the caps on the mobo and see if any look swollen or are leaking brown stuff.
  • EyesOnlyEyesOnly Sweden New
    edited March 2004
    This thread shows a great example as to why not to use prebuilt comps. At least if you build it yourself you can select quality parts from the beginning and not have to wonder what's in it.
  • Straight_ManStraight_Man Geeky, in my own way Naples, FL Icrontian
    edited March 2004
    LawnMM wrote:
    I've definitely made some progress but I'm still baffled. I put a better fan on the heatsink, and reversed the fans direction so it blows down into the heatsink rather than away from it as HP had it set up. I also reversed the exhaust fan (don't get started cooling purists!) so that it pulls cool air in and shoots it down the duct right out onto the cpu fan and heatsink. That seems to have all but eliminated heat as a problem.

    Whats still funky, is that for some reason after like 9 hours of running constantly with folding going to keep the cpu hummin' at 100%...it still freezes up. I'm trying to figure out what changes in the 9th hour that causes instability.

    In my XP Pro box, the XP plus AV plus etc wanted 2-3% of the P4 CPU for itself. I backed Folding down to 97% CPU use, silly thing is getting .989 effectiveness ratigns anyhow. Try somewhere in the 96-98% range for CPU use, you might settle as I did on 97%. Linux wanted 1% overhead, kept trying to give the client a more increased access to things by giving it amore important priority, when client refused, Linux spawned additional clients that SAT there.... I have had boxes that go unstable at 100% on dialup go REAL stable on 90-93%.... Some modem drivers for less expensive modems can have a huge impact on stability when a client tries to go online. Embedded video on motherboards will consume some overhead, or fixed need for time slices, just to work.

    I would then try backing from 100% to 97%, an dif less unstable but sitll flaky at times, then try 95% or 94%. If STILL unstable, take it clear back to 90% and if still stable a few WUs later, step up 1% every three WUs until you find the minimum OH your box needs all the time. I run my box for stable in use, then if want to crank things up I crank them back down in morning....

    But have pretty much settled for 97%. Note, this the amount of CPU when client is offered slices, not the percentage of slices it will accept in the sense of refusing to accept more 3% of the time on its own cycle of acceptance, it just does not grab ALL the CPU's time when XP lets it run. XP is happier that way, even with lots of services I do not need totally disabled....

    John D.
  • LawnMMLawnMM Colorado
    edited March 2004
    Wow...lets see, updates...

    No leaking caps. Will try reducing folding to 98ish% cpu usage, not exactly sure how to go about that but I'll mess with it and figure it out. I'm not changing the fans back because A - it was a pita to change in the first place, and B - its cooling the cpu better this way. I'm aware of their intent of the original setup, but given what I'm working with its cooling the box better my way.

    I'll try killing all extra apps running and let it fold with nothing else running to see if its a software issue. Thanks for the tips boys, keep em comin.
Sign In or Register to comment.