UPDATE: Spent the last few hours isolating my max speeds

zylander

Platinum Member
Aug 25, 2002
2,501
0
76
Current system specs;

A64 3000+ s939
DFI nF4 Ultra-D
OCZ Platinum 2gb Dual Channel (stock timing: 2-3-2-5)


This is my first time trying to OC a system. I spent all last night and most of today reading these overclocking guides; UberL33tJarad, Zebo, and this one from DFI-Street.

Ive spent the last few hours trying to find my max numbers and these are what I came up with, tell me what you think;

Max Stable FSB: 256Mhz, multiplier at 7
Max CPU speed: 2.3Ghz, multiplier at 9 and VCore at 1.55V. Temp at full load lingered around 48C, and occasionally hit 49-50C

Max Memory speed: 245Mhz, multiplier at 7, and 2.9V. with a timming of; 2.5-4-3-7


So if I put all that together I should be able to achieve the following;

FSB: 245Mhz
Multiplier: 9 (highest multiplier on this board)
CPU speed: 2.20Ghz

Does this sound right? For some reason I seem to think I should be able to get more out of this setup.
 

robertk2012

Platinum Member
Dec 14, 2004
2,134
0
0
why not go up to the max cpu speed and use a divider. Double check and make sure that is the max cpu speed. lower the htt multiplier to 3 and use a 1:2 divider with cas 3 timings and see how high you can get the clock speed. You should get a much higher "FSB" than 256
 

F1shF4t

Golden Member
Oct 18, 2005
1,583
1
71
Some multiplyers will not work with sertain mem dividers, and some ram sticks will not work with some dividers. For example i can run mem at 120 divider but not the 100 one.
U should be able to go higher on the fsb just try out all the different dividers, also make sure u set the timing manually, some dividers cause mem to use timings that the have no chance at working.
 

Furen

Golden Member
Oct 21, 2004
1,567
0
0
Originally posted by: zylander
Max Stable FSB: 256Mhz, multiplier at 7
Max CPU speed: 2.3Ghz, multiplier at 9 and VCore at 1.55V. Temp at full load lingered around 48C, and occasionally hit 49-50C

Max Memory speed: 245Mhz, multiplier at 7, and 2.9V. with a timming of; 2.5-4-3-7


So if I put all that together I should be able to achieve the following;

FSB: 245Mhz
Multiplier: 9 (highest multiplier on this board)
CPU speed: 2.20Ghz

Does this sound right? For some reason I seem to think I should be able to get more out of this setup.

Looking at your motherboard your max HTT does not sound even close to right. I suppose you could have a very bad motherboard but all the LP Ultra-Ds that I've seen have consistently hit over 300HTT without much trouble. Also, notice how you achieved your "max" CPU speed at your "max" HTT which means that your HTT is actually bottlenecking you. If I had to venture a guess I'd say that your Hypertransport multiplier needs to be lowered so that you can raise your HTT higher.
 

robertk2012

Platinum Member
Dec 14, 2004
2,134
0
0
Originally posted by: Furen
Originally posted by: zylander
Max Stable FSB: 256Mhz, multiplier at 7
Max CPU speed: 2.3Ghz, multiplier at 9 and VCore at 1.55V. Temp at full load lingered around 48C, and occasionally hit 49-50C

Max Memory speed: 245Mhz, multiplier at 7, and 2.9V. with a timming of; 2.5-4-3-7


So if I put all that together I should be able to achieve the following;

FSB: 245Mhz
Multiplier: 9 (highest multiplier on this board)
CPU speed: 2.20Ghz

Does this sound right? For some reason I seem to think I should be able to get more out of this setup.

Looking at your motherboard your max HTT does not sound even close to right. I suppose you could have a very bad motherboard but all the LP Ultra-Ds that I've seen have consistently hit over 300HTT without much trouble. Also, notice how you achieved your "max" CPU speed at your "max" HTT which means that your HTT is actually bottlenecking you. If I had to venture a guess I'd say that your Hypertransport multiplier needs to be lowered so that you can raise your HTT higher.

Thats what I was thinking.
 

zylander

Platinum Member
Aug 25, 2002
2,501
0
76
Few quick questions before I continue on;

HT needs to be less than or equal to 1000, and that is found by LTD * HTT, correct? So to keep it under 1000 I just need to lower the LTD multiplier?

I still dont understand how to use a memory divider, where is this listed in the BIOS?

The 256HTT that I said was my "max" was the highest I could go and still get into windows. I was able to get up to around 310HTT and I got all the way to the windows loading screen before the comp reboot itself.


Originally posted by: robertk2012
why not go up to the max cpu speed and use a divider. Double check and make sure that is the max cpu speed. lower the htt multiplier to 3 and use a 1:2 divider with cas 3 timings and see how high you can get the clock speed. You should get a much higher "FSB" than 256
WHat do you mean my cas 3 timings? Like 3-3-3-7?

If Im trying to find my max clock speed or HTT, wouldnt I just set the mem to 100 so it wouldnt be a factor?
 

robertk2012

Platinum Member
Dec 14, 2004
2,134
0
0
Originally posted by: zylander
Few quick questions before I continue on;

HT needs to be less than or equal to 1000, and that is found by LTD * HTT, correct? So to keep it under 1000 I just need to lower the LTD multiplier?

I still dont understand how to use a memory divider, where is this listed in the BIOS?

The 256HTT that I said was my "max" was the highest I could go and still get into windows. I was able to get up to around 310HTT and I got all the way to the windows loading screen before the comp reboot itself.


Originally posted by: robertk2012
why not go up to the max cpu speed and use a divider. Double check and make sure that is the max cpu speed. lower the htt multiplier to 3 and use a 1:2 divider with cas 3 timings and see how high you can get the clock speed. You should get a much higher "FSB" than 256
WHat do you mean my cas 3 timings? Like 3-3-3-7?

If Im trying to find my max clock speed or HTT, wouldnt I just set the mem to 100 so it wouldnt be a factor?

setting you memory to 100 is the divider I was talking about. Even so I would loosen the timings too.

the multipler should be 3

Then either test the htt by lowering the cpu multiplier or just see what you can get it up to.
 

Billb2

Diamond Member
Mar 25, 2005
3,035
70
86
You're getting close. the dividers are what you have to learn to use next.
When you get that figured out have a look at this puppy. More memory settings than you ever imagined existed! http://avala.yubc.net/~lukija/A64Info.rar
This is how you get those massive memory overclocks or extreemely tight timmings.
Corsair TWINX20483200C2PT DDR400 (Infineon)
http://valid.x86-secret.com/show_oc?id=89624
T
I had to add this for ya, cus you're tryin' so hard.
A64Info runs under windows...yes, you can OC memory without rebooting!
Just change a setting, run SuperPi, write down times or which itration failed, change another, run SuperPi, write down time or iteraton....it saves days of work! It has a CPU-Z like report, and calculates FSB's, multi's, Memspeeds, and Dividers too. Go slowly, most settings can't be changed alot in one jump. Save your settings as you go, and if you get a freeze or reboot you can just load in your last good settings and start right in again.

Now the warning: Is this program dangerous to use?...Hell yes! It can kill your mother and make flames shoot out of your computer. Use at your own risk. If you break something I will not help you fix it. If you even try to contact me I will flame you unmercifully in every forum you post on forever. 'Nuf said?
 

zylander

Platinum Member
Aug 25, 2002
2,501
0
76
Update:

So I seems like the rason all my original numbers were so low was because I was leaving the HT multiplier on auto. I tried everything again and set the HT multiplier manually and got much better numbers;

Max HTT: 335Mhz
Max CPU speed: 2.4Ghz
Max memory speed at 2.9v and timings of 3-4-3-7: 267Mhz

Currently running at;

2.4Ghz
301Mhz HTT
vCore at 1.55v
HTT multiplier of 8
HT multiplier of 3
HT at 901
Memory timming at 3-4-3-7, TREF at 2592, vDimm at 2.9

BUT...I have a little problem. Earlier today once I finished setting everything up, it was running stable. I ran a torture test in Prime95 for about 45 minutes and the CPU temp never went above 50C, only spiked at 50C a few times, mostly lingered around 48-49. I then shut it off and went to work for a few hours. Now I come back and switch it back on, and the second I get into windows, the CPU temp is at 70C! Right when I get into windows I open smartguardian and it says 70C, mbm5 and ntune both read the same.

So I figured that my setting swere just too high for the CPU so I reset everything to deffault, but the CPU is STILL running at 70C. Is it possible that the CPU socket sensor is broken, or all m monitoring programs are reading something else? Will the system even allow the CPU to run that hot? Im using a Thermaltake venus 12 I think it is with arctic silver 3 paste.


edit: When I startup the computer and go into the BIOS, the CPU is stable at arond 32C. Once I exit and get into windows, again it goes up to 70C, I imediately restart and go back into the BIOS and it is back down to around 32C. What could be going on? Could I have burned out my CPU while trying to OC it?
 

Billb2

Diamond Member
Mar 25, 2005
3,035
70
86
That is strange, but I'd believe the BIOS. You could try reinstalling whatever it is you're using to read temps in windows. It could be that the grease got screwed up by a fast temp spike, but unlikely. I don't know what to tell you.
 

robertk2012

Platinum Member
Dec 14, 2004
2,134
0
0
Originally posted by: zylander
Update:

So I seems like the rason all my original numbers were so low was because I was leaving the HT multiplier on auto. I tried everything again and set the HT multiplier manually and got much better numbers;

Max HTT: 335Mhz
Max CPU speed: 2.4Ghz
Max memory speed at 2.9v and timings of 3-4-3-7: 267Mhz

Currently running at;

2.4Ghz
301Mhz HTT
vCore at 1.55v
HTT multiplier of 8
HT multiplier of 3
HT at 901
Memory timming at 3-4-3-7, TREF at 2592, vDimm at 2.9

BUT...I have a little problem. Earlier today once I finished setting everything up, it was running stable. I ran a torture test in Prime95 for about 45 minutes and the CPU temp never went above 50C, only spiked at 50C a few times, mostly lingered around 48-49. I then shut it off and went to work for a few hours. Now I come back and switch it back on, and the second I get into windows, the CPU temp is at 70C! Right when I get into windows I open smartguardian and it says 70C, mbm5 and ntune both read the same.

So I figured that my setting swere just too high for the CPU so I reset everything to deffault, but the CPU is STILL running at 70C. Is it possible that the CPU socket sensor is broken, or all m monitoring programs are reading something else? Will the system even allow the CPU to run that hot? Im using a Thermaltake venus 12 I think it is with arctic silver 3 paste.


edit: When I startup the computer and go into the BIOS, the CPU is stable at arond 32C. Once I exit and get into windows, again it goes up to 70C, I imediately restart and go back into the BIOS and it is back down to around 32C. What could be going on? Could I have burned out my CPU while trying to OC it?


Looks like furen and I were correct about the htt multiplier.

DId you reset your CMOS or try another program to measure the temps. Does it stay at 70 no matter the load?


I would have though you would be able to get the cpu a little higher than that.

What was the settings that you achieved your cpu max at?