Overclocking an Athlon 2400+ XP

[WPKenny]

I've been trying to push a little bit more out of my athlon. I've read that a lot of people can get it running at 166 instead of the plain 133...pretty much as a norm for these chips.

I've got PC2700 Corsair ram so that's not the problem.

I also know that without unlocking the chip I won't get a low multiplier setting below 13 (or is it 12?),

What I'm setting it to is 13x at 166 FSB. I'm upping the voltage a little on the CPU to 1.7v and locking the AGP clock at 66mhz.

It boots fine as a 2700+ at 166 mhz but when it tries to load windows it either blue screens immediately or tells me it's missing system files.

In the BIOS (which are the latest...but I've tried overclocking this chip since I got it...hence the fresh attempt after a BIOS upgrade) it tells me the temp of the chip while overclocked at this 2700+ speed is about 55c so that shouldn't be a problem at all.

I'm at a loss really. If I try and set the chip at anything at all higher than it's 2400+ rating it just bums out on me. I know a 2400+ is what I paid for but it's annoying when I read about others who are able to get higher speeds from their's without a hitch....plus since my last processor was an AXIA Y Athlon 1 gig that overlcocked like a biatch.

Any suggestions? Have I got a duff batch that will only work fine at their set limits or is there a fiddle I can try?

I've tried various FSB's and voltages and god knows what else but it just crashes on anything more than 2400+ standard settings.

 

[Big G]

Isn't this something to do with unlocking it, possibly using the "pencil trick" ?

I have a 2400+ XP too, which runs at 15x and 133 FSB (2000 MHz), but i can fiddle the multiplier (i bought my unlocked from overclockers.co.uk) so that it runs at 16 x 134 (2144 MHz) making it roughly 2700+ speeds. I also run at 1.7v.

Can you try these settings to see if these work for you? I know you want 166 FSB bus, but try it with 133 and upping the multiplier.

I'm trying to get 16.5 x 134 (2211 MHz) to bring it over 2800+ spec, but windows tends to BSOD on me.

G

 

[Embattle]

WP what setting can you reach before the system falls over?

 

[Xavier]

best thing to do is let us know the what stepping of CPU you have and the HSF you're strapping to the top of it.

Don't be afraid of the chip getting hot when you're doing your initial overclock... 75C max won't do any lasting damage and with a pokey fansink you can possibly go 5-10C higher. As long as you aren't using a stock AMD CPU cooler whack the core voltage up a few more notches and see if that leaves your desired overclock any more stable, if so wind the voltage back a notch at a time, rebooting and running an app or two inbetween.

The error about missing files is generally associated with running your IDE controller (PCI BUS) out of spec, is the PCI divider set appropriately for 166FSB?

X4V

 

[SoWat]

I run my 2400+ ok at 13x166, which equates to 2.17Ghz (2700+).

I forget what heatsink and fan I have, but the cpu temp is 45.

Mind you, my motherboard (Epox 8RDA+) has pretty good features for overclocking, such as different memory/fsb timings.

Info from Sandra 2003 Pro:

SiSoftware Sandra

Processor
Model : AMD Athlon(tm) XP 2700+
Co-Processor (FPU) : Built-in
Speed : 2.17GHz
Model Number : 2700 (estimated)
Performance Rating : PR3149 (estimated)
Type : Standard
Package : Socket A PGA
Multiplier : 13/1x
Generation : 7 (7x86)
Name : Duron M8 & Athlon MP/XP (Thoroughbred) 1.5-2.5GHz+ 1.5-1.65V
Revision/Stepping : 8 / 1 (0)
Stepping Mask : B0
Core Voltage Rating : 1.650V
Maximum Physical / Virtual Addressing : 34-bit / 32-bit

Processor Cache(s)
Internal Data Cache : 64kB Synchronous Write-Back (2-way sectored, 64 byte line size)
Internal Instruction Cache : 64kB Synchronous Write-Back (2-way, 64 byte line size)
L2 On-board Cache : 256kB ECC Synchronous Write-Back (16-way, 64 byte line size)
L2 Cache Multiplier : 1/1x  (2172MHz)

Upgradeability
Socket/Slot : Socket A
Upgrade Interface : ZIF Socket
Supported Speed(s) : 2.17GHz+

Environment Monitor 1
Model : Winbond W83627HF ISA
Version : 2.01
Mainboard Specific Support : No

Power Rating(s)
CPU Core Power : 64W (estimated)

Sensors
CPU Temperature : 45.5°C / 113.9°F
CPU Voltage : 1.66V

 

[WPKenny]

The HSF is kind of irrelevant since I said it only reaches 55c when I've overclocked it to 166 fsb and 13x multiplier (2700+ speeds) but since you ask, it's a Coolermaster Heatpipe with a 80mm YS-TECH fan + modifier on it rather than the standard 60mm fan.

The HDD is a 120 gig SATA (on a ASUS A7N8X 1.2 PCB Mobo running 1006 BIOS) but it also did the same thing when it was a plain old IDE HDD.

 

[WPKenny]

Wassamatta? Question too tough for ya?

 

[Xavier]

Cooling in any overclock is never insignificant, it's not just a matter of a particular coolers ability to radiate heat, but also to deal with sudden fluctiations in temperature, so for instance the AVGS sunflowers intel used to send out are excellent for pusing your CPU a bit, but thanks to the primitive heatpipe in the core it can't move heat away from the spreader anywhere near quickly enough if the temperature rises too drastically.

Anyhow, a few things worth looking at:

If you're not able to raise the FSB 5-6Mhz without seeing stability issues then it's a 99% chance that your OC attempt is being fubared by another device objecting to the raise in PCI speeds and is probably writing garbage outside the bounds of its own address space, if you want to get really technical turn off all your integrated peripherals, whip out everything except your graphics card and try to find the limiting component that way by re-installing them one by own at a conservative speed known to cause instability.

With a decent cooler you can probably chuck the core voltage up a bit higher - remember that by increasing your vcore you increase stability by reducing the margins for error, and that most heat is generated by the voltage not the resultant frequency, so if you can get away with 1.85v (raise 0.05v at a time and watch those temps) you've got a greater chance of managing your OC, and if it works you can wind the vcore back down a notch at a time to find the sweet-spot before things get flaky again.

Check the CPU database on overclockers.com to see what chips of the same stepping have been known to do, it's almost certainly not the motherboard so if whacking the vcore up doesn't cut it then it's almost certainly down to a system device.

 

[Dr_Weasel]

And once you think its stable, run download and run the Prime95 torture test for a few hours. If it bails out, then its clocked too high and returning odd results.

It also has the bennifit of testing your CPU to the max thus making it as hot asits ever going to get. Check the temps out and make sure you arent too high when this is running.

 

[Xavier]

don't forget to run tests too with the door on the case, too often people perfect their overclocks with the chassis wide open, slam a door on it, the whole thing rises 5C and bails completely...