The ultimate CPU guide (overclocking and cooling) - Page 4

MrSeanKon · 29th May 2005, 10:06 AM

Quote:

Originally Posted by !JD!

i have a P4 2.0ghz im not quite sure but i thik it has a FBS of 100 so the multiplyer should be x20?? right?

Yeah you are right.
CPU speed = FSB (Front Side Bus) * Multiplier.

Quote:

Originally Posted by !JD!

so how would i change my multi??

Forget it!

All Intel's (else of ES=Engineering Samples) are locked=you cannot change multiplier.
Of course you must have great relations to find an ES Intel

mazda_626 · 29th May 2005, 8:02 PM

my P4 3.4E can have either a 17 or 14 multipleir and i think the 3.2 can do the same.
the down side is that the multiplier can't go up so it's pretty pointless unless you are only going for more fsb, but the cpu runs slower so is it really worth it?
I've personaly found it not to be worth it unless your trouble shooting and want to see if it's ram or cpu holding you back.

MrSeanKon · 30th May 2005, 7:58 AM

Quote:

Originally Posted by mazda_626

my P4 3.4E can have either a 17 or 14 multipleir

I did not know it..
Anyway...

Quote:

Originally Posted by mazda_626

the down side is that the multiplier can't go up so it's pretty pointless unless you are only going for more fsb, but the cpu runs slower so is it really worth it?

No.
Assume 3.4GHz = 17 * 200 so 14 * 243 = 3.4GHz too.
Same CPU and FPU performance.
But higher FSB increases RAM performance.

SteakTheMooCow · 30th May 2005, 8:06 AM

Quote:

Originally Posted by MrSeanKon

Same CPU and FPU performance.
But higher FSB increases RAM performance.

erm, unless your ram can't take it in which case you suddenly get an amazing decrease, and in some cases, smoke...

MrSeanKon · 31st May 2005, 3:21 AM

I agree but if memory timings and divider is the same (of course RAM must be 100% stable always) you gain performance as FSB increases.

mazda_626 · 31st May 2005, 5:30 AM

from what i've noticed with my system, 17x236 compared to 14 x 250, fsb of 236Mhz still has better performanceand has a ram bandwidth of about 5,900 while 250Mhz is only about 6,100.
real world performance, higher multiplier and best fsb is what i found to be the best, still working on getting to 247 stable (ie crashes)

RodneyJM · 1st June 2005, 3:13 AM

I have a mobo that claims to support DDR2 600 mhz natively, I tested it at that bandwidth with Sp2004's test for stressing RAM on priority value of 10, for 12 hours and had no errors or crashes of any kind. This is with 2:3 ratio. CPU at default clock speed and FSB at default value.
However when I O/C my CPU more than 8% and boost the voltage it crashes within minutes of starting Sp2004 @ priority 10. This is with the test for stressing ram only. Boosting the CPU voltage to the next increment in the scale does'nt solve the problem and boosting it again to the next increment after that still does not solve the problem. If I step up higher again in the CPU voltage scale it crashes sooner than a few minutes have passed from the first increase in CPU voltage test with ram stress.

I can O/C the FSB by 33% and its rock stable under 100% CPU load for 24 hours, this is with the CPU core only O/C 4% @ "Auto volts" in BIOS.
I don't understand how a CPU can have its FSB overclocked to 33% and be stable yet with only a small 8% O/C on the core it can't handle it - even with several increments in voltage.
I've disabled energy saving features in BIOS such as "C1" and "TM" which the BIOS describes as reducing voltage to save power when idle.
Could I have a flaky MoBo? maybe one of the IC's is a bit wonky?
I have good quality power supply in the form of "Antec 550W True power."

Annoyed and Puzzled.

Edit: Apparently the problem is either with the 3-phase CPU power regulation circuit or the 20-pin to 24-pin MoBo power connector. In the 955X chipset mobo, Asus decided to put a 4-phase CPU power circuit in place... why did'nt they do that before... go figure.

debsy · 21st June 2005, 12:38 AM

Hey guys just wondering if someone could explain the difference between the AMD 3200+ 64bit chips that are rated at a definite 1.4/1.5V and those rated as "variable"

ndt · 23rd July 2005, 11:10 PM

Could a laptop CPU be overclocked?
woul the lack of room for cooling be a problem?

mazda_626 · 24th July 2005, 1:39 PM

as far as i know, laptops are just as overclockable as desktop pc's, it just depends on whether the bios supports it

I.E., some desktop mobo's dont have any overclocking setings and i would guess most laptop mobo's wont either.

why would you want to oc' a laptop? cos you can?

there may also be some hardware hacks that you could do but I personally would keep away from those unless you really know what your doing

BuZeR · 2nd September 2005, 2:45 PM

My BIOS only has CPU Fequency (native 200), CPU Multiliper (3.0 - 9.0, native 9.0), some PCI-e one (native 100), and voltage options.. Does this mean my "CPU Fequency" is my CPU FSB?

If so, to find out the max FSB of my ram i keep turning up the CPU Fequency (and turning down my CPU Multiplier so its equal or below my native CPU speed - CPU Feq * CPU Multiplier = CPU Speed) untill i crash or hang, then i need to up my RAM voltage and see if it still hangs, if not i return to raising my CPU Feq and vlotage when nessasary untill i hit "the wall", and cant raise either anymore w/o a crash.. I once i hit "the wall" i put my CPU Feq back 3MHZ, and if it boots i do the RAM test and see if its faulty or not, if it is i keep turning it down till i find a stable speed. Then thats my max RAM FSB?

All the different names of things in my BIOS is starting to confuse me, so i just wanna make sure i dont blow something up by changing the wrong thing :P

Deltoid · 19th January 2006, 4:14 PM

Sometime on from when this thread started.

Any new ways for stress testing? Or is running prime for a few hours still the best way to stress test a cpu.

I'm mainly after testing the cpu here (I've primed the rest of the system just need to test the cpu speed).

Priming for a few hours can be a pain so if there is a quicker way let me know please.

Utetopia · 19th January 2006, 5:13 PM

The CPU tests in 3DMark '03 & '05 are a good alternative. Run them in a loop for a few hours.

SuperPi as well.

gregzeng · 24th February 2006, 4:10 PM

Pulled apart my HP Pavilion Notebook (Intel Pentium 3 ghz) myself, & quickly cleaned the air flows. Not much of an inprovement. Then had the repair shop overhaul it properly. Great improvement - the CPU FAN never seems to be needed now.

Also running latest version of Cpuidle. CPU temperature drops at least 10 degrees centigrade from previous operations. Haven't run Benchtests with CPUIDLE YET. One day ...

Greg.

nitestick · 18th April 2006, 1:24 PM

Quote:

Hey guys just wondering if someone could explain the difference between the AMD 3200+ 64bit chips that are rated at a definite 1.4/1.5V and those rated as "variable"

different core revisions. the 1.4v/1.5v are revision D and the variable voltage ones are revision E. revision E is rated for 1.3v-1.4v

29th May 2005, 8:02 PM	#47
mazda_626 Member Join Date: Jun 2004 Location: New Zealand Posts: 139	my P4 3.4E can have either a 17 or 14 multipleir and i think the 3.2 can do the same. the down side is that the multiplier can't go up so it's pretty pointless unless you are only going for more fsb, but the cpu runs slower so is it really worth it? I've personaly found it not to be worth it unless your trouble shooting and want to see if it's ram or cpu holding you back. __________________ (\___/) This is Bunny. Copy and paste bunny into your signature to help him gain world domination. (='.'=) ASUS maximus Formula, Q6600 (3.4Ghz), 6Gb G.skill ram, ATI 4890, (")_(")2x 250Gb RE3 raid 0, silverstone 850w PSU, DIY wooden case V2.0, Swiftech water cooling Server case is DIY wooden case V1.8

31st May 2005, 5:30 AM	#51
mazda_626 Member Join Date: Jun 2004 Location: New Zealand Posts: 139	from what i've noticed with my system, 17x236 compared to 14 x 250, fsb of 236Mhz still has better performanceand has a ram bandwidth of about 5,900 while 250Mhz is only about 6,100. real world performance, higher multiplier and best fsb is what i found to be the best, still working on getting to 247 stable (ie crashes) __________________ (\___/) This is Bunny. Copy and paste bunny into your signature to help him gain world domination. (='.'=) ASUS maximus Formula, Q6600 (3.4Ghz), 6Gb G.skill ram, ATI 4890, (")_(")2x 250Gb RE3 raid 0, silverstone 850w PSU, DIY wooden case V2.0, Swiftech water cooling Server case is DIY wooden case V1.8

1st June 2005, 3:13 AM	#52
RodneyJM (Banned or Deleted) Join Date: Mar 2005 Location: Engaged in Silicon Wars ! Posts: 1,916	I have a mobo that claims to support DDR2 600 mhz natively, I tested it at that bandwidth with Sp2004's test for stressing RAM on priority value of 10, for 12 hours and had no errors or crashes of any kind. This is with 2:3 ratio. CPU at default clock speed and FSB at default value. However when I O/C my CPU more than 8% and boost the voltage it crashes within minutes of starting Sp2004 @ priority 10. This is with the test for stressing ram only. Boosting the CPU voltage to the next increment in the scale does'nt solve the problem and boosting it again to the next increment after that still does not solve the problem. If I step up higher again in the CPU voltage scale it crashes sooner than a few minutes have passed from the first increase in CPU voltage test with ram stress. I can O/C the FSB by 33% and its rock stable under 100% CPU load for 24 hours, this is with the CPU core only O/C 4% @ "Auto volts" in BIOS. I don't understand how a CPU can have its FSB overclocked to 33% and be stable yet with only a small 8% O/C on the core it can't handle it - even with several increments in voltage. I've disabled energy saving features in BIOS such as "C1" and "TM" which the BIOS describes as reducing voltage to save power when idle. Could I have a flaky MoBo? maybe one of the IC's is a bit wonky? I have good quality power supply in the form of "Antec 550W True power." Annoyed and Puzzled. Edit: Apparently the problem is either with the 3-phase CPU power regulation circuit or the 20-pin to 24-pin MoBo power connector. In the 955X chipset mobo, Asus decided to put a 4-phase CPU power circuit in place... why did'nt they do that before... go figure. Last edited by RodneyJM; 21st June 2005 at 3:18 AM.

21st June 2005, 12:38 AM	#53
debsy Member Join Date: Jun 2005 Posts: 3	AMD 64bit voltage ratings Hey guys just wondering if someone could explain the difference between the AMD 3200+ 64bit chips that are rated at a definite 1.4/1.5V and those rated as "variable"

24th July 2005, 1:39 PM	#55
mazda_626 Member Join Date: Jun 2004 Location: New Zealand Posts: 139	as far as i know, laptops are just as overclockable as desktop pc's, it just depends on whether the bios supports it I.E., some desktop mobo's dont have any overclocking setings and i would guess most laptop mobo's wont either. why would you want to oc' a laptop? cos you can? there may also be some hardware hacks that you could do but I personally would keep away from those unless you really know what your doing __________________ (\___/) This is Bunny. Copy and paste bunny into your signature to help him gain world domination. (='.'=) ASUS maximus Formula, Q6600 (3.4Ghz), 6Gb G.skill ram, ATI 4890, (")_(")2x 250Gb RE3 raid 0, silverstone 850w PSU, DIY wooden case V2.0, Swiftech water cooling Server case is DIY wooden case V1.8

31st May 2005, 3:21 AM	#50
MrSeanKon Member Join Date: May 2005 Location: Athens in love with Anna! Posts: 506	I agree but if memory timings and divider is the same (of course RAM must be 100% stable always) you gain performance as FSB increases.

23rd July 2005, 11:10 PM	#54
ndt Member Join Date: Sep 2004 Location: Perth Posts: 1,179	Could a laptop CPU be overclocked? woul the lack of room for cooling be a problem?

2nd September 2005, 2:45 PM	#56
BuZeR Member Join Date: Apr 2005 Location: Townsville, Qld Posts: 571	My BIOS only has CPU Fequency (native 200), CPU Multiliper (3.0 - 9.0, native 9.0), some PCI-e one (native 100), and voltage options.. Does this mean my "CPU Fequency" is my CPU FSB? If so, to find out the max FSB of my ram i keep turning up the CPU Fequency (and turning down my CPU Multiplier so its equal or below my native CPU speed - CPU Feq * CPU Multiplier = CPU Speed) untill i crash or hang, then i need to up my RAM voltage and see if it still hangs, if not i return to raising my CPU Feq and vlotage when nessasary untill i hit "the wall", and cant raise either anymore w/o a crash.. I once i hit "the wall" i put my CPU Feq back 3MHZ, and if it boots i do the RAM test and see if its faulty or not, if it is i keep turning it down till i find a stable speed. Then thats my max RAM FSB? All the different names of things in my BIOS is starting to confuse me, so i just wanna make sure i dont blow something up by changing the wrong thing :P __________________ \|\|Intel Core i7 2600 \|\| Gigabyte P67A-UD7-B3 \|\| Kingston HyperX 16gig \|\| Gigabyte GeForce GTX 570 1280MB \|\| G.Skill Phoenix Pro Series 120GB SSD\|\|Sound Card: Auzentech X-Fi Forte\|\| Last edited by BuZeR; 2nd September 2005 at 3:07 PM.

19th January 2006, 4:14 PM	#57
Deltoid Member Join Date: May 2003 Location: Brisbane Posts: 9,463	Sometime on from when this thread started. Any new ways for stress testing? Or is running prime for a few hours still the best way to stress test a cpu. I'm mainly after testing the cpu here (I've primed the rest of the system just need to test the cpu speed). Priming for a few hours can be a pain so if there is a quicker way let me know please. __________________ PSN:Deltoid-

19th January 2006, 5:13 PM	#58
Utetopia Member Join Date: Jun 2004 Location: BNE/TWMBA Posts: 2,517	The CPU tests in 3DMark '03 & '05 are a good alternative. Run them in a loop for a few hours. SuperPi as well.

24th February 2006, 4:10 PM	#59
gregzeng Member Join Date: Apr 2002 Location: ACT Posts: 1,045	CPUIDLE, & clean the cooling air flows ... Pulled apart my HP Pavilion Notebook (Intel Pentium 3 ghz) myself, & quickly cleaned the air flows. Not much of an inprovement. Then had the repair shop overhaul it properly. Great improvement - the CPU FAN never seems to be needed now. Also running latest version of Cpuidle. CPU temperature drops at least 10 degrees centigrade from previous operations. Haven't run Benchtests with CPUIDLE YET. One day ... Greg. __________________ TBI=>Extreme Emotional Disinhibition. Sorry! Last edited by gregzeng; 24th February 2006 at 4:19 PM.