FSB Stability with Tbird & XP.

[lhgpoobaa]

After my OS crashed with critical errors due to FSB overclocking, i am forced to conced that my Athlon 1200C AXIA 'Thunderbird' cpu just <font color=purple>DOESN'T</font color=purple> like running at elevated FSB speeds.
Im using an epox 8k3a+ (kt333ce chipset)

and this is removing as many other possible factors as possible.

A. CPU is unlocked so i lower the multiplier to stay WELL within OC bounds.

B. using a KT333 chipset with the 1/5 divider, so 166FSB gives in spec PCI/AGP speeds of 33/66.

C. ram is Corsair XMS3200, easily able to handle 166mhz on hardest mem timings (and ive tested it using memtest86)

D. cooling is more than adequate give the overclock. (MCX-462 with variable speed fan)

So the question arises. Is the tbird naturally bad at High FSB? is any XP cpu better?

I value everyones input here, however i am looking for <font color=green>more</font color=green> than just <font color=red>my cpu does xxxxMhz at xxxfsb with games!</font color=red>

what i want are conclusive results like SuperPI, Prime95 torture tests or AT LEAST 8 hours straight of divx encoding.

see my OS rolled over and BSOD-ed after 8 hours of divx encoding at 1420mhz (8.5x166).
the CPU can go higher, but i get odd results.
for example:
i can do 1466(11x133), but regardless of cooling or voltage applied (up to 2.1v) it BSOD's very fast. same with 1533(12x133)
however it will run at 1500(9x166), on the surface it runs ok, but prime95 torture test dies immediately, and superPI spits out calculation errors after half a minute. so its obvious that cpu calculation issues are occuring.

so what do you have?
can you do prime95 torture test sucessfully for hours at high FSB? or 8 hours of divx encoding?

curious. puzzled.





My Next Performance System!!! - P4 Celleron, 128k cache, SDRAM, Integrated graphics, 5400rpm HDD!

 

[khha4113]

Is the tbird naturally bad at High FSB? is any XP cpu better?

I hadn't run TBird at higher FSB, but on XP I can run it at 177MHz FSB without problem (on Abit KR7A-RAID). However, I couldn't run my XP 1600 at higher FSB on Asus A7V333 as on Abit. It couldn't post at FSB higher than 155MHz with PCI divider at 1/5 (my XP was unlocked to lower its multiplier).

Good or Bad have no meaning at all, depends on what your point of view is.

 

[lhgpoobaa]

okay...
so does prim95 torture test run ok on that XP @ 177fsb?

My Next Performance System!!! - P4 Celleron, 128k cache, SDRAM, Integrated graphics, 5400rpm HDD!

 

[khha4113]

I don't have Prime95 on my system, but I ran Toast for 1/2 hour and <b><A HREF="http://service.madonion.com/compare?2k1=3360740" target="_new"><font color=red>3DMark 2001</A></b></font color=red>.

Good or Bad have no meaning at all, depends on what your point of view is.

 

[lhgpoobaa]

okay... well 3dmark is reasonably FSB sensitive.

the reason why i specified prime95 or superPI is that they have built in error checking that detects computational errors.
see 1420Mhz @ 166fsb SEEMED ok, could do toast for as long as i wanted, but prime95 died almost immediately with rounding errors.

in the long run calculation errors will end up killing your OS, as they did with mine

My Next Performance System!!! - P4 Celleron, 128k cache, SDRAM, Integrated graphics, 5400rpm HDD!

 

[bum_jcrules]

I had some thoughts but I have some questions first.

1. What is the base for the FSB?

2. What is the base clock for the MCH? (Memory Clock)
More importantly...what is the Memory Bus speed?

3. What are the exact speeds that the FSB and the memory
at spec levels? (133.47, 134.21, etc.)

4. Are there half steps fr the Voltage? 8.5, 9.5, 10.5,
etc.?

There could be some issues that these questions might shed some light on.

<b>"Sometimes you can't hear me because I'm talking in parenthesis" - Steven Wright</b> :lol:

 

[lhgpoobaa]

seems the CPU is pretty dead now... allthough i get some degree of stability running lower than 1200mhz (default speed)

and to answer your questions

1. bas FSB was 133.

2. the base multiplier is 9.
memory clock is 133.

3. exact speeds vary a little, but not very much. 7.5x1333=1004, 9x133=1205 10x133=1333 10.5x133=1401 (according to MBM - results vary a little depend on what proggy is used)

4. voltage has 2 stepps...
from 1.40v to 1.90v goes in steps of 0.025.
from 1.90v to 2.2v the steps are only 0.05.

and as ive show with multipliers greater than 11 or 12 it seems to have problems.


:lol: Whats better, <font color=blue>Intel</font color=blue> or <font color=green>AMD?</font color=green> :lol: