PDA

View Full Version : SETI & Overclocking



SteveC
2nd June 2000, 03:35
Hi Guys,

Finally got round to puttung SETI on my home machine last night, and found that ti kept crashing out after a few mins (Under win2k). I then put my CPU back down to 500 (from 620MHz) and it seemed to run fine. How comes though that I don't have any other problems when at 620, only SETI crashes.... http://forums.murc.ws/ubb/frown.gif

------------------
Cheers,
Steve

"Life is what we make of it, yet most of us just fake"

cjolley
2nd June 2000, 08:10
Steve,
I had the same problem.
My Celery 333@500 would almost never make it through a whole WU unless I clocked it back.
The P3 650@910 that I have now never has a problem.
Both machines used the same two sticks of PC100 ram & the same MB.
Go figure!?!?!?
chuck


------------------
ABit BF6, P3-650@910, 128mb gh@140cas3, 10gb IBM@7200, SB Live Value@3.0, Pioneer 104s DVD, Mitsumi CDRW@2x2x8, Acatel 1000 ADSL@1.5mb/sec, Linksys EtherFast NIC, LG 995e, USB IntelliEye,Matrox G400 MAX!!!! :D

Assimilator1
2nd June 2000, 10:21
Steve C

1 thing that SETI does that many programs/games dont is too run the cpu at full load continiously.It could be a heat build up problem.
BTW you say your cpu temp is 30C ,how is that measured?
If it is an external sensor next to the cpu slot then these don't really give a very good idea of cpu temps.
What m/brd do you have?

BTW2 I find that Unreal(& maybe UT too) is an excellent program to weed out any flaky overclocked systems.
I've found that whilst I can run SETI endlessly on my Celeron 366 @ 578MHz @2.1v ,if I run Unreal it will bomb out of the game after about 1/2hr.It needs 2.15v to be totally stable.

What cpu have you got?
What cpu voltages have you tried?

Of course it could be your RAM http://forums.murc.ws/ubb/smile.gif

cjolley

Looks like your 333 was too close to its limits to be totally stable.What voltage did you run it at?

------------------
I guess I won't recruit for my team here ;) TA

cjolley
2nd June 2000, 10:35
Voltage is an interesting point too. http://forums.murc.ws/ubb/wink.gif
The seti-unstable celery was @ +.2v and 52c
The seti-stable P3 is @ stock 1.65v and 48c
(both at full load)

I'm afraid to take the P3 apart to put on a better hs/fan.
Untill some good i820 mbs come out I won't be able to get it significantly faster anyway.

chuck


------------------
ABit BF6, P3-650@910, 256mb@140cas3, 10gb IBM@7200, SB Live Value@3.0, Pioneer 104s DVD, Mitsumi CDRW@2x2x8, Acatel 1000 ADSL@1.5mb/sec, Linksys EtherFast NIC, LG 995e, USB IntelliEye,Matrox G400 MAX!!!! :D

CharlesWA
2nd June 2000, 11:37
SteveC: Are you sure the 30C is your "CPU temp" and not your "motherboard temp"? I've got one heck of a heatsink on my CPU, and it runs at about 70-75C when set to it's defaults. Though it is Slot 1, not Socket 370... (more details below)

cjolley: Given the disaster with the i820 chipset and the MTH translator, I don't know if we'll ever see a "good" board using it. Basicially i820 and SDRAM are a bad idea. I don't think that Rambus is all Intel and Rambus want us to think it is either. I've seen enough tests showing that PC133 SDRAM is basicially better...

All: Anyway, back to the real topic. I recently bumped my PIII 450 up to 600 (ASUS P3V4X, Win 98 SE) and it seemed to be running fine, but sometime over night it froze up (while running SETI and nothing else besides "normal" background stuff). I had it at 2.0 volts (standard for the CPU) and at 600 (133 FSB) it was running between 85 and 95C. When I dropped it back to 450 (100 FSB) it dropped down to 70-75C. I've got a big old Global Win VOS32 (or something to that effect) dual fan cooler on it (currently "stock", though I have plans to try to tweak it in an Alpha cooler sort of way). I'm guessing that the ASUS software must be reading a temperature sensor built into the CPU. I'm not sure though. In both cases, it also reported that the motherboard temperature was 25-30C. Of course this is with the sides and top of the case off (still in the upgrade/tweak phase), so maybe things will get a little better once I put the cover on and let the intake and exhaust fans do some good.

I could bump my voltage up to 2.05V (or higher, I forget what they are exactly), but I'd like to try to get the temperature of the CPU down to 80C or less (when at 600/133). Maybe my idea to "duct" the cooler will do it. If not, then I guess a peltier might be the only way.

I'm not totally ruling out my ram (PC100, CL=3), but it's running at 100 thanks to the ASUS P3V4X, even though the CPU is using a FSB of 133. Though I'm planning on getting some good PC133 CL=2 ram soon.

I'm not a computer "newbie" at all, but haven't yet dealt with a peltier, so if anyone has any good sites with hints, tricks, etc. please let me know...

Thanks,

Charles

cjolley
2nd June 2000, 13:03
Pehaps I should have said i820esque. http://forums.murc.ws/ubb/wink.gif

70-75c is too hot.
Something is wrong.
I think thats about what intel has as the max operating temp for your cpu.

Check that your hs goop hasn't dried out
and/or the hs is cattywampus and not making full contact with the cpu.

chuck

CharlesWA
2nd June 2000, 13:27
I'm using one of these Global Win coolers on my CPU:
http://2cooltek.safeshopper.com/2/106.htm?764

I cleaned up the CPU after I took the "Retail" heatsink off, then applied some of the included thermal compound (very thin coating) and then installed this guy. I used the retention mechanism as the instructions indicated. I'm not sure where I could have gone wrong!

I can double check the readings in the BIOS monitoring just in case Windows 98 is getting (ASUS PC Probe 2.11 I think) bad info or something.

HELP?!?!?

Charles

cjolley
2nd June 2000, 13:56
Are you sure Probe isn't set to show degrees F ?
That would be more like it.
70-75 just above room temp, 95 just warm.
chuck


[This message has been edited by cjolley (edited 02 June 2000).]

CharlesWA
2nd June 2000, 14:41
It was showing both C and F degrees, and the lower of the two was in the 70-75 degrees C range.

All I can figure is that I need to yank it all out and clean off all the thermal goop and put some other stuff on it. I've got somethat came from an electronics store that I'll try...

Charles

Kruse
2nd June 2000, 14:58
I have a P!!!450 @ 558 with a normal size heatsink/fan that performs slightly better than the original Intel one (I know - I tested both of them).

My operating temperatures at 558 MHz are 60 C for the CPU and 30 C for the MB. If I stop SETI the CPU temp. drops to about 42 C.

I accidentally ran the CPU without proper thermal paste connection once when I changed the heatsink/fan. That made it run at about 90 C without any load at all (other than starting up Win98).

CharlesWA
2nd June 2000, 15:31
Kruse: What brand of motherboard is that? I'm wondering if my new ASUS P3V4X has screwed up hardware monitoring or something. Given the type of cooler I'm using, I should be running as cool as you if not cooler, especially at 450!

Maybe the thermal paste stuff that came with the cooler is total crap. I'm tempted to run by the electronics store on the way home to see if they have a thermal probe for my Fluke DVM in stock...but they may close before I can get there. That would clear up the issue of faulty hardware monitoring. At the minimum, I'll try some other thermal paste.

How thick or think of a layer for thermal past should I be applying? Maybe I just didn't put enough on. Can anybody point me to a web site that has pictures of the proper application?

Thanks for the ongoing help.

Charles

Maggi
2nd June 2000, 16:33
Hi Steve,

maybe the instructions that SETI polls from your CPU are located in a certain area of the die that is not as good cooled as the rest of it ?

Have a look at your temps and try CPU Idle with settings of highest priority when reaching a certain temp.

Cheers,
Maggi

SteveC
2nd June 2000, 16:39
Hi Maggeroni http://forums.murc.ws/ubb/wink.gif

My CPU is a flip-chip PIII and from what testing I've done, is being cooled quite well - hovers around 30 degC. Suppose it could also be my memory is being flakey at over 100Mhz... time to get PC133 ram methinks.

------------------
Cheers,
Steve

"Life is what we make of it, yet most of us just fake"

Maggi
2nd June 2000, 16:50
oh ... ok, could be your mem.

CU,
Maggs

Assimilator1
2nd June 2000, 19:44
CharlesWA

I recently was involved in a thread all about the Asus P3V4X cpu temp measuring.
Alot of people have had it happen to them to ,but for different reasons ,check the thread out ,should give you some ideas http://forums.murc.ws/ubb/smile.gif.
http://anandforums.gisystech.com/messageview.cfm?catid=28&threadid=175559

>>>>Can anybody point me to a web site that has pictures of the proper application?<<<<

Yep! http://forums.murc.ws/ubb/smile.gif ,here it is.....
http://www.agnhardware.com/hardwaresection/default.cfm?revid=253

Hope this helps



------------------
I guess I won't recruit for my team here ;) TA

CharlesWA
2nd June 2000, 21:02
Thanks Assim. I think I may have gone a little too light on the thermal paste. I'll take things apart in a while and try again.

According to Intel my CPU's "Thermal Spec" is 90 degrees C, so I doubut I did any damage to it, even when it ran up to 90 or a bit above.

That's assuming it actually went that high. I skimmed that thread you guys have going on the ASUS board. I'm going to check out the bios and PC Probe versions. I'm pretty sure my board has 1003 out of the box, but maybe the new revision is out of beta by now (or soon, hopefully).

If everything checks out, then I'm back to wondering why I can't run at 133 FSB/600 Core. There's always a chance that between clocking the memory at 100 Mhz and the FSB at 133 Mhz the chipset may be unstable. Maybe not under "normal" use, but under the constant pounding of SETI. I'm just going to have to order that PC133 RAM sooner than later...

Charles

Assimilator1
3rd June 2000, 05:17
CharlesWA

You need to try higher cpu voltages ,you'll be OK up to 2.2v.But increase the voltage by the smallest increments & check for stabilty.
If it is 'nearly' stable at 2v then theirs a good chance it'll be ok at 2.05v(that's what Intel did BTW for 600MHz on a Katami).

Of course you'll want to confirm that your cooling is working OK.! http://forums.murc.ws/ubb/smile.gif

BTW have you tried Mother Board Montior to see what cpu temps you get with that?

Here's another thread that'll interest you
http://anandforums.gisystech.com/messageview.cfm?catid=28&threadid=177645

Oh ,nearly forgot ,I stumbled across another forum with a section dedicated to the P3V4X!
Their should be loads of info their for you http://forums.murc.ws/ubb/smile.gif
http://www.nightmachines.net/cgi-bin/oc/ubb/forumdisplay.cgi?action=topics&forum=Asus+P3V4X&number=7&DaysPrune=30&LastLogin=


------------------
I guess I won't recruit for my team here http://forums.murc.ws/ubb/wink.gif TA



[This message has been edited by Assimilator1 (edited 03 June 2000).]

CharlesWA
3rd June 2000, 10:03
Update to the story... The retention mechanism that ships wit the Global Win cooler that I've got seems to be at least partially to blame. I took the CPU and cooler apart, cleaned it, and noticed that there seemed to be a little "wear spot" on part of the cooler right where one corner of the CPU contacts it. The corresponding part of the CPU is worn slightly too. Upon closer inspection, it appeared that only that corner of the CPU was touching the cooler!

I did some bending and tweaking of the mechanism and now it seems better, though I still think it's too hot. With the CPU at 600/133 FSB the temp likes to hover in the low 60's but I've seen it jump up to almost 70 C. This is much better than the 90+ from a previous trial. In fact, it ran fine this way all night. It's still at 2.00 V for the core. A brief test at 450/100 showed a temperature of about 10 degrees cooler (50-55). I've read other places that this board likes to report 5+ degrees warmer than the CPU actually is, so that'd put it at more like 45-50 C for the standard clock/voltage settings. Does that seem reasonable? Or is my CPU cooler still not running at maximum efficiency?

Incidently, ASUS has a "release" version of BIOS 1004 posted for this board, but that BIOS is JUNK! Let's just say that the multiplier settings don't work right. I've seen other posts saying the same thing.

More later or tomorrow after more testing/tweaking.

Thanks all,

Charles

Assimilator1
4th June 2000, 11:30
45-50 C for internal cpu temp is fine http://forums.murc.ws/ubb/smile.gif

I'm still surprised that it hits 70C though, are you ABSOLUTLEY sure that the heatsink is sitting flush onto the cpu this time?

------------------
I guess I won't recruit for my team here ;) TA

Kruse
5th June 2000, 14:34
A somewhat late reply, but better late than never right...?

CharlesWA:

My MB is an Asus P3B-F. In my humble opinion the best BX board ever made. I have no reason to believe the measurements are off... the heatsink feels quite warm to the touch, so I'd say 60 C for the CPU is very likely.

Your new temperatures sound quite reasonable to me too.

One extra thing: I recently cut a big hole in my cabinet (just the metal, not the plastic) where the intake fan is (front bottom). Before that there were just a lot of small holes. Now my operating temps are 55-58 C for the CPU and 26 C stable for the cabinet while running SETI. Besides the lower temperature it also makes less noise now http://forums.murc.ws/ubb/icons/icon7.gif

jms
5th June 2000, 22:17
Kruse,
the lower temperature must have made your machnine crunch WUs a hell of a lot faster. http://forums.murc.ws/ubb/smile.gif You'll probably pass me in a couple of days. Oh well, for the best of the team and so on... http://forums.murc.ws/ubb/smile.gif
Jan

CharlesWA
5th June 2000, 23:51
Kruse: I actually haven't had the case cover back on yet. The temperature may or may not go up when I do that. I've got one intake fan (front/bottom area) and two exhaust fans plus the power supply fan. One is right behind the CPU and directly below the power supply. The other is up higher in the back (full tower case). My case has lots of little holes in front of all the fans. You've got me thinking about maybe opening up at least the front one to increase air flow.

All: So far so good. The machine has been running pretty much continuously with SETI (and occasionally me messing with it) since Saturday. No lock-ups, no restarts, etc. The temp has been in the 60-70C range the whole time. I'm feeling ok with that since Intel says I'm ok up to 90C. I still may try an Alpha cooler since I've read a few places that they do fit ok.

Charles

Kruse
6th June 2000, 01:41
jms:

If only that was so... I took advantage of an extended weekend (a few national holidays) to add three more machines from work. They're down again though, so I'm back to around 10 WUs/day.

CharlesWA:

That's a lot of fans! If I were you I'd certainly make a bigger hole at the intake fan. All those exhaust fans need some air flow.