T-Bird 900@1064 Posts, But Windoze Locks. Why?

[PapaRomeo]

OK, My new T-Bird 900 system runs like a clock at 7.5x133 (997.5mHz). (other specs below)

When I set the multiplier to 8.0, it posts normally, but then Windows refuses to launch (it freezes on the "Windows 98" loading screen) and I have to do a hard reboot. This happens consistently. As soon as I set it back to 7.5, it boots up fine. System temps look OK (idling ~ 45C, topping 50 only under sustained stress, and yes, I have a better cooler on order (Alpha 35)

Can any of you give me a clue as to what the limiting factor is that's keeping me from going higher than 7.5 x 133?

Looking forward to any wisdom you can share...



PR

• AMD Thunderbird 900mHz @ 997.5mHz (7.5x133) Mwave OEM HSF
• Inwin IW508-IW mid-tower case, 300W PSU, 2x 80mm (one in, one out)
• Asus A7V133 MoBo
• 256mb PC133 RAM
• 20.5GB IBM Deskstar ATA100 7200RPM
• 3dfx Voodoo5 5500 AGP
• Soundblaster Live! Value
• CL ModemBlaster PCI
• Plextor CD-RW 32x/12x/10x
• Suncom Talon/SFS Throttle/Thrustmaster Elite Rudders (gameport)
• ProView 770 Monitor (17&quot
• Windows 98SE
• DirectX 7.0a

 

[classy]

Have you upped the voltage?

 

[PapaRomeo]

Not yet. I hadn't considered messing with the voltage, since the machine posts normally. Would the voltage make a difference in whether Windows loads or not?

What should I change it to?

I'm new to this OC stuff. Enlighten me...

PR

 

[TonyRic]

Kick up the core voltage in .05v increments until you stabilize...

 

[PapaRomeo]

Thanks, guys. I'll try that tonight and report back...

One question: Should I expect it to run a lot hotter if I up the voltage?



PR

 

[DarkMajiq]

You'll probably run a bit hotter with higher voltage, and it will definitely be hotter at 1064 than 1000 (it's actually 133 1/3 not 133, so you've 1GHz ) regardless of voltage. I would try upping the FSB a bit though, rather than going to 8.0 multiplier. If you can hit, say, 7.5x140, that's 1050 right there, and while in raw MHz it's a tiny bit slower, it will be much faster because your RAM, AGP and PCI will be running faster as well. Even if you only manage something like 1030 with a higher FSB, it will still be faster than just using the multiplier

 

[PapaRomeo]

Thanks for the input, DM. I'll try pushing the FSB first. Will I need to boost the voltage at the same time?

PR

 

[WarCon]

Yeah vcore usually has to go up to stabilize an overclock of any magnitude. Don't worry about it unless you have poor cooling in your case. If I was you though I would invest in a better hsf.

 

[WarCon]

Forgive my previous post about getting a better hsf, just noticed your note about one on order.

 

[PapaRomeo]

Thanks, WarCon:

I have an Alpha Pal6035 with a 27cfm fan (and a tube of Artic Silver ) on order from http://www.coldcpu.com Do you think that will do the trick?

PR

 

[Superdoopercooper]

Not really sure, but I had the problem where my T-bird 850 would post at 1GHz and load into windows, but then it would just randomly reboot. So, I just lowered the multiplier and worked my way up on FSB until I was at a stable 150FSB and 975core... which is stil pretty good.

I think it has to with various things... like a quick, but very small dip on the 3.3V supply, a particular transistor on the chip turning on that pulls enough current from somewhere else to cause a freeze/reboot. Could be a number of reasons... adn that is obviously why your T-bird wasn't sold as a 1050MHz chip... becuase there is obviously some problems up there. Could be cache RAM getting lost? who knows... but I wish you well on the overclock. It's definitly cool to tell your friends "I have a Tbird 600 that is overglocked to 1.21 Gigawatts with a flux capacitor "

 

[PapaRomeo]

<< It's definitly cool to tell your friends &quot;I have a Tbird 600 that is overglocked to 1.21 Gigawatts with a flux capacitor &quot; >>



LOL! Isn't that what this whole forum is all about? Let's face it, it's all about bragging rights!

Thanks for the perspective. I'll try this out tonight and let you guys know how I make out.

Rock on,

PR

 

[pm]

One Comment: raising the voltage reduces the life of the part, and dramatically increases the temperature of the chip (temp is directly proportional to V^2 and only directly proportional to frequency, so a 10% increase in voltage increases temperature much more than a 10% increase in frequency). Even if you have a great heatsink, you are trimming a substantial amount of life off the part if you go higher than 15% over the spec. voltage. There's reasons that AMD isn't selling these things at 1.9V. So, it's worth mentioning the downside of the voltage bump advice. No matter what you do, I highly recommend not exceeding 1.95V... above that and you probably can count the statistical potential life of your part in months instead of years.

 

[Dark4ng3l]

Most people run their cpu's at 1.85 volts(people that o/c i mean)

 

[pm]

I guess my point was to mention that there are limits to how high one should think about going. 1.85 still seems high to me, but at least it's the same as AMD's max spec. (page 26 of their datasheet).