Thunderbird Adventures

[paulcs]

Running everything at spec, my newly built T-bird 700 MHz/Abit KA7-100 system has been as solid as a rock. If I up the FSB by a relatively small amount, depending on how much I overvolt the CPU, a SETI commandline client will eventually lock the system up hard.

I considered heat as a potential issue, but the more I overvolt the CPU, the longer the SETI client will run before the lockup. I've manipulated the "CPU Strength" settings as well, but no luck.

(Just as an aside, SETI runs really slowly with this configuration. I believe Mega had similar findings with his tests.)

I just broke down, cracked the T-bird open and added a GFD. It's now overclocked to 750 MHz. (This was not fun or pretty.) So far, so good, although it once took a few hours before SETI locked the system up with a 107 MHz FSB overclock.

The KA7-100, supposively, has a reputation for good FSB overclocking results. Possibly, my problems could be a T-Bird thing. Given most of my problems have been with SETI, it could be a SETI thing.

I just touched the heatsink and the top of the heatplate. Neither was hot.

Any scientific theories and wild speculations would be appreciated.

Paul
paulcs@flashcom.net

[This message has been edited by paulcs (edited 13 July 2000).]

[paulcs]

Some specs:

700 MHz Athlon Thunderbird
128 MB Crucial CL2-rated PC133 RAM
15 GB IBM 75GXP Deskstar (UDMA/66)
Matrox G400, GeForce 256 DDR, and GeForce2 GTS
Creative Soundblaster Live MP3
PC Power & Cooling 400 Watt Power Supply

The ATA-100 controller is disabled.

Paul
paulcs@flashcom.net

[This message has been edited by paulcs (edited 13 July 2000).]

[LAMFDTK]

hehe

[Nuno]

Maybe AMD was right when they stated the only chipsets supporting the Thunderbird were AMD 750 and KT133?
Another chance is that you got a DOG cpu that really won´t overclock. It has a 700 core, right?

And seti runs slower on a thunderbird then on a classic athlon. As it also runs slower in a coppermine vs a katmai. Seti really depends on a 512Kb L2.

[paulcs]

The "dog factor" occured to me, although changing the multiplier seems to be working (knock on wood). I haven't exposed the core, as that involves removing the heatplate as well as the plastic housing. It looks as if getting the clamps back on would be *extremely* hard. I haven't heard of anyone online with a 700 MHz T-bird getting stuck with a 700 MHz core, but I wouldn't be surprized if I were the first.

As for SETI, on similarly configured machines, with matching clock speeds and procesing the exact same WU, I have a Coppermine beating the T-bird by close to two hours. I've never seen a disparity like that between a Coppermine and a Katmai.

The T-bird is faster in every other test I've thrown at it: Q3, Q2 Crusher, Content Creation Winstone, Business Winstone, the Sandra CPU and Memory Benchmarks, etc.

I'm not complaining, by the way. This is first and foremost a test machine, and if the thing catches fire or dances naked through the streets, my job it to report my findings. I just want to make sure I haven't missed something. I'm writing a review about two products from companies with some very passionate avocates (AMD and Abit). I don't want to miss something important, write up a review, and get flamed for no good reason.

I'm just polling the MURC "General Hardware Think Tank" for ideas.

Paul
paulcs@flashcom.net

[UncleJoe]

There have been a couple guys reporting succes with T-bird 700 slot-A and KA-7. They have used ry-bios (beta ?) and T-birds are week 20 origin. (reported in finnish hardwaresite called muropaketti www.muropaketti.com) Moderator of that site have allso tested overclocking T-bird on MSI k7pro and got up to 900 with air cooling, 1000+ with water and water&peltier results will be available in near future ...

[paulcs]

I'm using the RAID capable version of the RY BIOS. I don't seem to be having a problem overclocking by manipulating the multiplier. I've been stressing the CPU for about 16 hours now at 750 MHz. No lockups so far (again, we knock on wood).

If all goes well, I'll jack it up to 800 MHz tomorrow.

Paul
paulcs@flashcom.net