FSB(HTT) affected by latencies?

S

silverphoenix

Gawd
Joined
Feb 15, 2006
Messages
780
I'm trying for a best OC and I know my cpu can handle 290HTT, I've ran a divider on a PC3200 ram before. The problem is the ram with a 266 divider ran 2-2-2-5 timings, I've gotten newer PC4200 ram and it runs at 3-4-4-8 @ 266, now I know it can do 3-5-5-10 @ 290. Now nvm that I'm currently clocked at 275 with a 333 divider running 2-3-3-6 timings(lowest ram will go...), now the cpu will chug along fine at this speed, but if I try to run 285mhz I need to loosen timings to 3-4-4-8 and the ram will run without a divider, but when I run mixed test in prime95 for CPU it crashes almost immediately, same thing with 275 if I raise timings just even a little it crashes prime95.

Is my Proc A64 3000 just really picky about timings or is it something else? my ram can handle the high HTT it just loves loose timings, but my cpu seems to only like tight timings for any freq above 275.
 
From what I understand, A64's don't necessarily suffer from timings and command rates as much as the intel CPU's.
 
legendz411 said:
From what I understand, A64's don't necessarily suffer from timings and command rates as much as the intel CPU's.
Click to expand...
I believe it's the opposite actually, A64s are much more affected by timings than Intel CPUs due to the on-die memory controller.
 
Oooo. I thought that the A64's weren't as hit by loose timings due too the on-diecontroller :p

Thanks for clearing that up!
 
My cpu won't prime for any longer than 2 hours with 2-3-3-6 timings at 275mhz, I'm trying to find highest HTT at those timings as those are the lowest these GEIL PC4200 will go... Are most a64's like this? or can most tolerate even those timings at 275ish ?
 
Ive barely noticed ANY speedups between the following timings:

2.5 - 5 - 5 - 7

and

2.5 - 3 - 3 - 5


But going from 200 to 250mhz did net about 1s under SuperPi.
 
I've managed to oc by lowering my multi

currently its running

295x8.5 333divider 2-3-3-6 latency. It is priming now.
 
loosen that shit up a little bit lol, It sounds like either your RAM needs more Voltage or your timings some how are too tight to allow your RAM to achieve those speeds.
 
kill4killin said:
loosen that shit up a little bit lol, It sounds like either your RAM needs more Voltage or your timings some how are too tight to allow your RAM to achieve those speeds.
Click to expand...

its not that the ram can hit those spds but my cpu doesn't like loose timings...
 
it doesnt like changing the CAS timing probably. Try messing with the other ones, my RAM is the same way if thats the case. I can't make the RAM budge from a CAS of 3, if i try it refuses to boot unless i'm at or below stock speeds.
 
kill4killin said:
it doesnt like changing the CAS timing probably. Try messing with the other ones, my RAM is the same way if thats the case. I can't make the RAM budge from a CAS of 3, if i try it refuses to boot unless i'm at or below stock speeds.
Click to expand...


well my memory runs fine as it is with lower latencies and a divider, and the cpu with 8.5x multi will only stay stable at 295mhz, I can raise volts from 1.4875 to 1.55 but the ceiling still remains.
 
It failed prime, and now is failing prime at any combination of speeds I throw at it... is Prime corrupted? should I reinstall it? reinstall windows? It is failing to go over 10min of prime with everything stock.

could it be cuz i'm running 1T
 
Raised it to 2T, now I'm at 8.5x305 with a 333 divider. Priming, can anyone explain what command rate is exactly and how it affects performance.
 
command rate changes the way the memory controller sends commands.. introduces a one cycle delay sorta. it also changes the drive strengths and the way the MC interfaces with the ram a bit, so if you have a somewhat poorly coded bios, i guess i can say that my original statement is wrong, and that there is actually two (or three) timings that can effect htt/fsb:

command rate
max async latency & read preamble (they're somewhat related to each other, the memory ratio, and the round trip time)
 
Ok one last thing for oc, does doing anything else on the machine while prime mix stress testing possibly cause prime errors?
 
Not likely, i have done stuff like web surfing and typing a paper while priming, iv actually even played HL2 once or twice i believe. It didnt cause it to error.

::EDIT::

Also, i noticed that I am also running at 1T, how does one go about changing it to 2T exactly?
 
I saw the tinniest performance bump in 1T for a little bit. but my system ran unstable no matter what spd at 1T
 
You must log in or register to reply here.
Back
Top