i am using nf7-s........kingston value ram pc3200 with default timings of 3-3-3-8..... the chips on the ram are kingston....
a question for u guys though...
is there a program for checking memory stability other than memtest??? like one that can run within windows? my floppy isn't working so i can't run memtest......so tell me abt any such program..
i am trying to check for cpu stability with "cpu burn in" is it good program for this purpose?? how does it differ from prime95??
Yeah, Prime95's torture test will tax the stability of your overclock, but you won't know for sure what is giving you errors if you are overclocking the proc as well as the fsb speed. If you want to use it to check for memory stability at that 230 fsb, leave the multiplier low to eliminate the proc as a potential problem.
what does an error in prime95 really mean in common application.? like...if the computer gives an error to prime95 does that mean other applications will be equally affected?
Other applications could be. Prime is very sensitive to even minor errors, but if you want your system stable and reliable you should expect it to run prime with no problems.
My guess is that you will need to back off your fsb slightly.
Then once that proves stable you can start slowly increasing your multiplier to get your cpu speed up.
Great work Noman. I know how tough it has been for you to get this system built.
Great work Noman. I know how tough it has been for you to get this system built.
actually i was making it too much of a big thing.. ...it wasn't that tough....heheh...
thanx for support all u guyz...
currently i am trying to find out stable fsb....but ram is really piss poor :shakehead .....although it boots at very high speeds but its not stable .....
i have upped the vdimm to 2.7...trying to check 210...
ohh one more question...
i heard that higher fsb gives better performance compared to higher multi.....but then there was also this mentioned that there are some applications that are more memory dependent and some applications require more cycles or something....
my question is...
which or what type of application utilize more memory?
which or what type of application utilize more clock cycles(or whatever the multi does)??
Usually higher fsb is good, even if you have to relax the memory timings. Now it isn't worth going from cas 3 to 4 if you only get a few MHz increase in bus. But if you can relax the timings and get at least 5MHz then I would do it.
In general machines running faster fsb 'feel' quicker.
i am having a weird kind of problem..
when checking ram stability
when i run prime95 stress test there is an error....but the weird thing is when i up the vdimm the error occurs even more quickly..
for example when i ran 210mhz(all other things constant) at three vdimms this is what happnd
1-ran 210 at 2.6(stock) ....prime error occured after 10 mins.
2-ran 210 at 2.7v ...........................just after 5 mins
3-ran 210 at 2.8v.................................error occured after 1
mins!!!!!!
so ....as i up the vdimm. instead of increased stablity i am having decreased stability....
Kingston frequently doesn't respond well to voltages. You're also using the value line of their memory, which increases the likelihood that A) You won't be able to overclock the memory and B) That it'll respond poorly to voltage hikes.
My suggestion is to run it at 205MHz or less at 2.6v.
i just checked it at 205 & 2.8v...ran prime for 2-3 hrs ok....
what u said is ok.......but isn't it kind of weird???
have u ever heard BEFORE of any ram becoming unstable at higher voltage???bcoz everywhere it is written that any ram will be more stable at high volts...
Yes, I've heard of it. Most value RAM lines are notorious for not accepting higher voltages. I don't precisely know why; it could be many variables.. It could be poorer transistors, an inferior PCB. Something electronic, who knows? Unfortunately, any ram will <i>not</i> be stable at higher voltages -- just <b>most</b> ram.
Most value RAM lines are notorious for not accepting higher voltages.
A lot of the value ram out there don't come with heat spreaders... so more vcore more heat less stability
But with pc 3200 and you're moble I'm sure you could run it at 12.5x200@ 1.775-1.80vcre for 2500mghz. To me that feals faster than runnin in the 2300's with a high fsb.
BTW you could download sandra I beleive it has a torture test for just the memory..
i heard that fsb overclock is considered better than one wtih multi.but i think its my ram....i haven't used memtest yet as my floppy is ****tty.....only using prime's torture blend(one that says it tests a lot of ram)....
i have sandra 2004 but its torture test doesn't run for long i dont know why....but when i benched(cpu tests) different combinations of multi and fsb for same clock speed.....there was a very little difference only 1-200 points ......
i again checked it....ran 1 hr 20 mints with 209mhz at 2.8v
and at 2.6v it was stable for more than 2 hrs.....before i stopped the test.............so now i am sure my ram won't take more volts...
i have this habit of not checking each voltage for stability...i directly go for 1.75 when i am going at 2400 or above....i dont' feel like upping the clock speed and then check stablity and volts everytime
I used to have a stick of kingston value ram (PC3200).. it is decent ram, but voltage increases did not seem to help much in obtaining higher frequencies. I believe the stock timings on the CL3 model is 3-3-3-8, but FYI, it does run 2.5-3-3-7, and u should be able to take it close to 210MHz at stock voltage.
I always felt that a higher FSB gave the PC a zippier feel, but thats all subjective
Thrax: I agree, even with ~2.9V through my ram, it barely gets luke-warm, I have a feeling that with/wo heatspreaders my OC would be pretty similar. More of a bling factor I guess.
Rambus RDRAM needs heat spreaders, like the ones you get with the Thermaltake kits, because the way RDRAM works can concentrate a memory module's entire heat output onto one chip.
PC main memory SDRAM, on the other hand, does not need any sort of heat sinking. It's quite low power, and the heat's spread evenly over all of the chips on each module. They just don't get particularly hot.
The last part says it - the heat is already evenly spread.
I used to have a stick of kingston value ram (PC3200).. it is decent ram, but voltage increases did not seem to help much in obtaining higher frequencies. I believe the stock timings on the CL3 model is 3-3-3-8, but FYI, it does run 2.5-3-3-7, and u should be able to take it close to 210MHz at stock
you are right man!!
its default timings are 3-3-3-8 and i tested it and it doesn't go past 210!...it gives error at 211....
there is one more thing..........it has kingston chips in it....my one friend has kingston valueram with hynix chips....he is stable at 220mhz.....
i have tried various combinations...now i tried running 200x12.5=2500mhz...i put 1.8v vcore..but when running mbm and cpuz show its 1.76l..................the system becomes unstable........ what should i do??
the lowest voltages i got were on 12v rail it showed 11.74 minimumm.......max temp was 45 or so.
.......rest are all ok..
Comments
Good job, Noman. I'm glad to see everything is coming together nicely
a question for u guys though...
is there a program for checking memory stability other than memtest??? like one that can run within windows? my floppy isn't working so i can't run memtest......so tell me abt any such program..
i am trying to check for cpu stability with "cpu burn in" is it good program for this purpose?? how does it differ from prime95??
i have overclocked it at full speeds also...but still doing some fine tuning....
plz tell me abt the question i asked....
and one more thing...
my 12v rail is giving 11.92 or 11.98 at idle and when i start "cpu burn-in"...it drops to 11.86 ....so is it within limits?
and my vcore too..... when i set it to 1.675v ....
mbm shows 1.6v when idle and during burn-in the vcore drops to 1.63v.....
what do u guys have to say abt these volts??
Run FAH, or some type of benchmarking programs. Gaming works as well sometimes.
Your voltages are fine.
so again the same question....is there any OTHER windows based program for checking memory?
http://www.mersenne.org/freesoft.htm
My guess is that you will need to back off your fsb slightly.
Then once that proves stable you can start slowly increasing your multiplier to get your cpu speed up.
Great work Noman. I know how tough it has been for you to get this system built.
actually i was making it too much of a big thing..
thanx for support all u guyz...
currently i am trying to find out stable fsb....but ram is really piss poor :shakehead .....although it boots at very high speeds but its not stable .....
i have upped the vdimm to 2.7...trying to check 210...
i heard that higher fsb gives better performance compared to higher multi.....but then there was also this mentioned that there are some applications that are more memory dependent and some applications require more cycles or something....
my question is...
which or what type of application utilize more memory?
which or what type of application utilize more clock cycles(or whatever the multi does)??
In general machines running faster fsb 'feel' quicker.
so higher fsb is only the faster "feeling"....no other benefits or shortcomings???
and......what abt the benchmarks..
.at a same clock speed..say "X ghz"....if i have
1#higher fsb x lower multi=X
2#higher fsb x higher multi=X
now will the benchmarks be same in both cases..i.e 1# and 2#????
when checking ram stability
when i run prime95 stress test there is an error....but the weird thing is when i up the vdimm the error occurs even more quickly..
for example when i ran 210mhz(all other things constant) at three vdimms this is what happnd
1-ran 210 at 2.6(stock) ....prime error occured after 10 mins.
2-ran 210 at 2.7v ...........................just after 5 mins
3-ran 210 at 2.8v.................................error occured after 1
mins!!!!!!
so ....as i up the vdimm. instead of increased stablity i am having decreased stability....
any thoughts??
My suggestion is to run it at 205MHz or less at 2.6v.
what u said is ok.......but isn't it kind of weird???
have u ever heard BEFORE of any ram becoming unstable at higher voltage???bcoz everywhere it is written that any ram will be more stable at high volts...
A lot of the value ram out there don't come with heat spreaders... so more vcore more heat less stability
But with pc 3200 and you're moble I'm sure you could run it at 12.5x200@ 1.775-1.80vcre for 2500mghz. To me that feals faster than runnin in the 2300's with a high fsb.
BTW you could download sandra I beleive it has a torture test for just the memory..
Congrats on youre new system
Not to go off topic but could you elaberate a little you got me curious.
i heard that fsb overclock is considered better than one wtih multi.but i think its my ram....i haven't used memtest yet as my floppy is ****tty.....only using prime's torture blend(one that says it tests a lot of ram)....
i have sandra 2004 but its torture test doesn't run for long i dont know why....but when i benched(cpu tests) different combinations of multi and fsb for same clock speed.....there was a very little difference only 1-200 points ......
i again checked it....ran 1 hr 20 mints with 209mhz at 2.8v
and at 2.6v it was stable for more than 2 hrs.....before i stopped the test.............so now i am sure my ram won't take more volts...
i have this habit of not checking each voltage for stability...i directly go for 1.75 when i am going at 2400 or above....i dont' feel like upping the clock speed and then check stablity and volts everytime
I always felt that a higher FSB gave the PC a zippier feel, but thats all subjective
Thrax: I agree, even with ~2.9V through my ram, it barely gets luke-warm, I have a feeling that with/wo heatspreaders my OC would be pretty similar. More of a bling factor I guess.
"Great product! Doesn't work!"
The last part says it - the heat is already evenly spread.
you are right man!!
its default timings are 3-3-3-8 and i tested it and it doesn't go past 210!...it gives error at 211....
there is one more thing..........it has kingston chips in it....my one friend has kingston valueram with hynix chips....he is stable at 220mhz.....
i have tried various combinations...now i tried running 200x12.5=2500mhz...i put 1.8v vcore..but when running mbm and cpuz show its 1.76l..................the system becomes unstable........ what should i do??
the lowest voltages i got were on 12v rail it showed 11.74 minimumm.......max temp was 45 or so.
.......rest are all ok..