Overclocking an atholn 64 3200+ on a 939

[MadMike]

i just got done (for now) oc-ing to find a stable setup for my system.

athlon 64 3200+ on an asus a8v deluxe 939 chipset kt800 PRO

200mhz fsb X 10 cpu multi 2000mhz, 1.4v core default 1000mhz ht

heres what im at

261mhz fsb X 9 cpu multi 2349mhz, 1.5v core current ht 1000mhz

ht speed, is this a multi? options are 200 400 600 800 1000mhz (is this 1-5X)?

this is all 32bit windows xpsp2, i havent done anything with 64bit os yet.

also does stable on 32 bit windows mean it will be stable under win xp 64?

cpu temp has not registered over 49c even with toast (stock hs and fan)

seems to idle at 40c (lowest ive seen with these settings)

i have a pci\agp lock at i think 33.33/66.66, theres a higher setting, would that help anything? mb can take up to 8xagp(old) and my current card is only 4x, but in cpu-z it says transfer rate 1x max supported 4x

its already at 4x in bios

could it help to bump up agp voltage (1.5 to 1.6)? as screen seems slow

prime 95 failed a ways into testing at 267mhz after passing superpi 32m so i backed it off to 261mhz and i havent dialed it in closer, i ran the prime 95 tests for about 45 minutes each with no trouble

memory is at 213mhz and ddr427 and this is a cheap kingston 2700 ddr333

i guess this might be harder to do with more than one module

mem timings 2.5-3-3-7 ive run it 1T and 2T, both pass memtest86 with a slightly faster time with 1T. as well as a better little time in superpi 32m.

34:58 1T 36:34 2T, any benefits to 2T?

any reason to bump volts, on auto, i guess 2.5v options, auto 2.6 2.7 2.8

other options for memory (i think this is called whats called the divider) my bios calls it limiter

if i remember right options are auto or limit,

limit options are 1:1, 4:3, 3:2, 5:3, 2:1 1:1 is 200ddr 2:1 is 400ddr, i have it on auto and it registers as ddr427, and passes memtest

these memory options are new to me (as is most of this) so please someone let me know if i could benefit from limiting the memory, (or if my memory is actually working as fast a i think it is,(a ddr333 module firing 213mhz 427ddr)

also the test-pass meter goes up a little over 100% before its done ?????

v-core voltage from asus probe reads usually 1.438-1.454 with 1.486 spikes

v-core voltage from cpu-z usually stays between 1.488-1.504 highest spike i saw was 1.536 and once it was as low as 1.456

im pretty happy with this setup, as long as im not overlooking any dangerous situations i might of created :nonoi hope not)

but i like to tinker around, plese let me know if anyone has any ideas, espessilly those of you who might of owned a simmilar setup, i should mention that my cpu multiplier goes in half steps

(10X, 9.5X, 9X, 8.5X, 8X, etc.)

i put screen shots of cpu-z if anyone's interested, they look a little goofey cause i saved them in 16 color mode.

thanks to any who take the time to read this, all comments and suggestions (besides "go buy a better system") welcome and appreicated.:wavey:

 

[Rastalovich]

Gah ur pics aint opening for me. But squinting really hard under the micron-microscope (actually I just looked at the figures u put at the top of the thread) - those values are pretty much what I can remember getting on a stock cooler, regular memory, and possibly even the same mobo.

2.4ghz seemed to be the stability ceiling under standard cooling conditions, no secret modding or skyscraper coolers involved. 49 degrees is pretty darn good, I can recall hitting the 60s.

I *think* with a better air cooler u can push 2.7 stable. 2.6xx something.

Some1 would have to confirm that tho.

I know u said not to mention it, but upgrading cooling and stuff would really be financially ineffective for what amounts to 0.2xxghz. But if u could acquire one cheaply, that's another story.

 

[Mul.]

I would say that the CPU is close to it's limits if you're already applying between 1.45-1.5V. I generally found that a lot of the 90nm Socket 939 AMD K8's hit a brick wall about 100-200MHz from when such voltages are required. Aside a few batches like the initial ones off the line and the "0448 SPAW" batches, the Winchester 90nm's were fairly naff overclockers compared to the later Venice and San Diego cores. The Via K8T800 Pro generally doesn't like much over 250 on the base HTT either. Your overall HTT is also above 1000MHz, which is known to cause instability on most AMD K8 boards.

For a start, I would increase your CPU's Multiplier back to the default 10x to take some strain off the chipset and lower the LDT Multi (Hypertransport Multi) to 3x (600MHz). Start with a base HTT of 240 with the current Voltage to give you 2.4GHz and see how that goes for you. If it seems alright, keep raising it in 5-10MHz increments and see where it takes you

 

[MadMike]

thank you ill try that later on tonight, maybe ill try the 9.5x setting too, i havent really played with it for a few days but ill probally try that tonight and ill let you know how it goes

as for a cooler, i was thinking about getting one of those if i want to push it hard,

ill mess around with it like you were saying with 10xcpu and test and let you know how it goes

thanks for replying

 

[MadMike]

heres the pics i put before in jpeg

pics from cpu-z

 

[MadMike]

highest stable ocs i can get with this board at stock voltage and cpu 1.5

(1.4v default)

237X10 2370

250X9.5 2375

261X9 2349

in any of these cases the temp never gets over 50c running toast

would it be safe to attempt to raise the bus up to like 270 280 with a lower divider?

and with a stock voltage of 1.4 about whats a general (safe) temperature/voltage range?

 

[MadMike]

safe cpu voltage

my processor is default 1.4v can i safely pump it up past 1.55 if its under 50c

just for the hell of it it put the fsb to 240 and put everything else on auto, it booted nice, then i got into windows and cpu-z reading on cpu voltage was bouncing around 1.67-1.73v! could this be safe?!?! i used the reset button.

i got it to pass superpi 32m 240X10 at 1.55v, why would the bios pump it up that high, that kinda scared me but i think everythings ok