Overclocking Need help on setup & Strange heat reports all on prescott

[Meridius]

Hi all how are you.

Well I own a asus p5ad2 deluxe motherboard and p4 prescott 3.6 D0 stepping with Corsair 4200 memory which is 533mhz.

Question 1: when i have my p4 chip set at 3.6 my memory is set at 400 which is 1:1 with the chip which runs at 200bus, in the asus motherboard memory setup i can pick (Auto,400,533,600) i have read that my memory will run up to the 600 is it work doing this ???????? as it will not run 1:! with my chip will i get any inprovment ????

Question 2: I have overclocked my chip to 4.0 (222bus) now the memory options say (Auto,444,592,666) which one should i pick i dont think my memory will go to 666 but if i pick 444 it will run 1:1 with the chip is this write or should i pick 592 but this would not run at 1:1 what would i need to pick ????

so which is better to 1:! to the chip or overcl:banana::banana::banana::banana: it ???

well here is the thing i have compressed a film the cpu reads 100% so its full hog now hers my reading

Stock HeatSink 3.62
Cpu heat = 64c
MB heat = 42c
CPU Fan = 2720rpm
PSU Fan = 1442rpm

Stock HeatSink 4.00
Cpu heat = 66c
MB heat = 43c
CPU Fan = 2830rpm
PSU Fan = 1442rpm

as you can see when i overclocked my cpu at 400mhz over it went up only 2c and the systeam went up 1c and the fan speed went up 110rpm thats is very strange as i thought it would go much higher in heat what do you think ????

also one more question when i ran my bus at 223bus which is 4.02 my computer reset after 5 mins of 100% cpu but at 222bus 4.00 which is only 18 mhz diffrance it never reset would underclocking 18mhz make that muchdiffrance on how it would not reset ?????

i run my cpu at 1.4v to

i hope somone can help me out on these 3 questions so what would you set your memory to on 3.06 and 4.00

oh what is the best program to push your cpu and memory so you know its 100% stable



thanks all

[Meridius]

ok here are some tests

not clocked
200x18=3.6mhz at 800bus memory 4200 (533) at (533mhz 4-4-4-12)

overclocked
200x18=4.0mhz at 889bus memory 4200 (533) at (592mhz 5-5-5-15)
200x18=4.0mhz at 889bus memory 4200 (533) at (444mhz 5-5-5-15)

as you can see my timings go up when i overclcok the memory but what is a suprise is when i under clock the memory it still is 5-5-5-15 timeings i would have thought ti would of bee 4-4-4-12 as my memory runs at 533 at these timings

will the timings make much diffrance fromm 4-4-4-12 to 5-5-5-15 as the memory is running 59mhz faster

so out of the tests what would you say is the best option for the memory when i overclock it should i pick 592mhz at 3:4 or 444mhz at 1:1

what programs do i need to stress test the cpu and memory to see if its stable and to see what speed readigs i can get for best performance

i am compressing a film and surfing the net and playing a movie all at the same time and the cpu is 100% as runs fine no problems at 65c heat

[Romaster]

to test stability.. you can try running sisoft sandra. last time when i overclocked after getting a new cpu.. i was testing my stability.. to do this i just looped mem test and cpu arithmetic test.. both of which push your cpu and ram heavily.. after repeating at least 5 times each nonstop.. you'll know its stable for life.. or close ;)

the way you worded and explained your story is confusing me as hell and I usually know what im reading very clearly :confused:

thats probably why you are not getting much responses (any at all really)


motherboard memory setup i can pick (Auto,400,533,600) Im not sure exactly what you mean.. Im assuming you mean the little selection of default mem speed.. that option is best left on auto or whatever your proc fsb is at default (533,400, or 800 usually)

as for your heat situation of only 2C.. thats odd indeed but there could be many outside factors affecting this such as room temp variation... how long it has been on b4 you checked the temps etc... btw if i recall corrently, asus mobos read 10C too low or something like that.. so effectively your really at like 73-80C which is awful. I suggest you invest in a good heatsink if you plan to overclock. get either the swiftech mcx478V, alpha pal 8952, SP-94 thermalright, or the thermalright xp-90..

next answer: leave the memo thing on auto.. thats my safest answer for that question.. it will keep changing numbers based of your rams mhz... so if at 200mhz.. it will say 400,533,600 for instance.. if at 250mhz it will say 500,somtin,750.. you see the point..


also one more question when i ran my bus at 223bus which is 4.02 my computer reset after 5 mins of 100% cpu but at 222bus 4.00 which is only 18 mhz diffrance it never reset would underclocking 18mhz make that muchdiffrance on how it would not reset

that is an indication that your ram sux... if you loosen your ram timmings using manual options in the bios.. you might be able to do more.. some ram cant do good timings and relatively low fsb like 222..

Also.. many cpus just overheat its an obvious factor and as ive said your asus mb isnt saying the right temps..btw dont go above 1.5v unless you got serious cooling.. in fact dont go above 1.45... some presshots dont like the Vcore being much over default..

next answer: this is for post #2 of yours:

I've never used a 3:4 divider as ive never had a chance to.. I beleive the 3:4 would be better than 1:1 for obvious reasons... then again make sure your ram is very stable on it.. otherwise stick with 1:1 its cool..


not clocked
200x18=3.6mhz at 800bus memory 4200 (533) at (533mhz 4-4-4-12)

overclocked
200x18=4.0mhz at 889bus memory 4200 (533) at (592mhz 5-5-5-15)
200x18=4.0mhz at 889bus memory 4200 (533) at (444mhz 5-5-5-15)

ok first of all.. you worded this completely incorrectly.. which confused me for a second...

the thing where you said "overclocked"
you mean 222x18.. not 200x18.. as 200mhz x 18 multiplier = 3600mhz


as for programs... many programs are good at testing cpu power... sisoft sandra ( a benchmark) 3dmark (a benchmark, 01, or 03 preferably), pcmark 04 ( a benchmark)


this is the longest response ive ever made on any thread.. hope it helps..

[Meridius]

thanks for the help sorry for mistakes as i am new at this

this is what i have tried

3.6 overclocked to
option 1 222x18 = 4.0, Memory 444mhz at 3-3-3-9 1:1
option 2 222x18 = 4.0, Memory 598mhz at 4-5-5-15 3:4
option 3 222x18 = 4.0, Memory 444mhz at 5-5-5-15 1:1

its just i dont know weather to pick between the option 1 or 2 or even 3
ahhhh do you get much performance diffrance anyway or am i making more work out of this than i should.

its just i want to get the best performance out of it do i go for 1:1 or 3:4 i just dont know

[Romaster]

Sry but I just drank some vodka earlier so Itry to make my answer as clearly as i can make,

Im still nto sure exactly what you mean but if im getting it right (and i will by later on) it is clear that either option 1 or option 2 is your best bet.. because option 3 is just option 1 with worse ram timmings, resulting in poorer performance..

My suggestion is you test out both option 1 and option 2 in benchmarking environments and whichever one scores higher is clearly better..

my recommendations are try the following benchmarks:

Pcmark 04 (www.futuremark.com)
SiSoft Sandra - the memory bandwidth test and the cpu arithmetic tests (get it at download.com just search sisoft sandra)
3dmark 03 and 01 -although they wont really be good indicators, especially 03, definetly try 01. try both options twice on 01, just run the video card at stock with the options being the variable in the experiment..

[Meridius]

thanks m8
HI all after alot of sorting got sandra to work at last

here are my results fastest to slowest

Sandra results
CPU: 3.6@4.0, Memory 444mhz, 1:1, 3-3-3-9 = 4857 to 4849
CPU: 3.6@4.0, Memory 592mhz, 3:4, 4-5-5-15 = 4800 to 4800
CPU: 3.6@3.6, Memory 533mhz, 1:1, 4-4-4-12 = 4505 to 4493
CPU: 3.6@4.0, Memory 444mhz, 1:1, 4-5-5-15 = 4432 to 4410

as you can see when i did not have my pc overclocked it came third for memory speed, but as you can see as well there is not much diffrance in memory speed from the top two so i can run my memory in one of them.

so the next thing to do is to try out 3dmark for cpu tests

i noticed sandra had a burn in test for stability but would like to try this prime95 i just git that how do i use it how long will i run it do you just run it ?????

also i have 3dmark05 is this no good should i get the 01 version ????????

[Romaster]

I dont use sandra burn in test.. its good im sure but personally I test stability manually ;)

well based on the memory test its pretty clear the first choice, the 3-3-3-9 is better in terms of bandwidth.. but continue testing and post results..

next to do is the sandra arithmetic test (in sisoft sandra, the one with the pic of the cpu)

dont do 3dmark 03 or 05, just do 01 as it is more cpu/mem etc dependant than the others..

might also want to try pcmark 04.. its a great program which i personally dont use due to lack of interest but it can help in this testing situation.. :)

[Meridius]

mmm can not find 3dmark01 version

the cpu test on sandra where

Sandra results
CPU: 3.6@4.0, Memory 444mhz, 1:1, 3-3-3-9 = 4857 to 4849, CPU 11647
CPU: 3.6@4.0, Memory 592mhz, 3:4, 4-5-5-15 = 4800 to 4800, CPU 11617
CPU: 3.6@3.6, Memory 533mhz, 1:1, 4-4-4-12 = 4505 to 4493, CPU 10489
CPU: 3.6@4.0, Memory 444mhz, 1:1, 4-5-5-15 = 4432 to 4410, CPU 11590

what program you use to find out the stability i herd prime95 is good ??????

[Vryada]

just cpu burn during 20 minutes and its enought to know its rock solid ;D

[Meridius]

HI all after alot of sorting got sandra to work at last

here are my results fastest to slowest

Sandra results
CPU: 3.6@4.0, Memory 444mhz, 1:1, 3-3-3-9 = 4857 to 4849, CPU 11647
CPU: 3.6@4.0, Memory 592mhz, 3:4, 4-5-5-15 = 4800 to 4800, CPU 11617
CPU: 3.6@3.6, Memory 533mhz, 1:1, 4-4-4-12 = 4505 to 4493, CPU 10489
CPU: 3.6@4.0, Memory 444mhz, 1:1, 4-5-5-15 = 4432 to 4410, CPU 11590

as you can see when i did not have my pc overclocked it came third for memory speed, but as you can see as well there is not much diffrance in memory speed from the top two so i can run my memory in one of them.

so the next thing to do is to try out pcmark for cpu tests

pcmark04 results
CPU: 3.6@4.0, Memory 444mhz, 1:1, 3-3-3-9 =5587, CPU=5903 Total=5695
CPU: 3.6@4.0, Memory 592mhz, 3:4, 4-5-5-15=5630, CPU=5900 Total=5692
CPU: 3.6@3.6, Memory 533mhz, 1:1, 4-4-4-12=5174, CPU=5309 Total

as you can see the top two have swaped places for memory tests in pcmark to sandra buts its very littile in it but what is strange is that even though the memory ran a littile slower it got a couple of more points in the cpu tests for total ???? which i would have thought it wouls have been less strange ????

so what would you pick as sandra and pcmark tell a diffrent story is pcmark more for games and sandra more for apps ?????

3dmark05 results
CPU: 3.6@4.0, Memory 444mhz, 1:1, 3-3-3-9 CPU=5012
CPU: 3.6@4.0, Memory 592mhz, 3:4, 4-5-5-15 CPU=5015
CPU: 3.6@3.6, Memory 533mhz, 1:1, 4-4-4-12 CPU=4668

well its the same with 3dmark nothing much in it but dont know what to pick the Memory 444mhz, 1:1, 3-3-3-9 or the Memory 592mhz, 3:4, 4-5-5-15
like i said is pcmark and 3dmark more for games and sandra for apps

now to prime95

i noticed sandra had a burn in test for stability but would like to try this prime95 i just git that how do i use it how long will i run it do you just run it ?????


thans for the help

[Romaster]

as the other guy said 20 mins would be more than enough.. so try that... ;)

btw looking at your scores I'd go with option 1 over option 2..

btw if you can try to run option 2 with lower timmings.. thats the reason it loses otherwise it would win.. see if the comp will boot at 5-5-5-12 or 5-4-4-15 etc... experiment :)

[Meridius]

thanks ran prime95

i use the option under the menu (option, touture test) and ran
the top three options in that menu for 20min is this right
the reason i am asking is because my cpu was only getting used by 54%
and never went up to 100% ????? or am i running the wroung test

oh will this stress test the ram as well as the cpu

just ran prime computer faild 10k test

ok heres the thing ,when i have the voltage set to auto which its default volts reads

CPUID
idel =1.520, load=1.471v now on this reading the load is lower strange ???
ASUS Probe
idel =1.296, load=1.36v

now in the bios i select the volt core and select 1.4v instead of auto

CPUID
idel= 1.550v, load=1.529v now on this reading the load is lower strange ???
ASUS Probe
idel =1.4v, load =1.376v


so heres the question
Q1: how is selecting auto from V1.4 diffrent in the bios as i thought the chip ran at 1.4v selecting auto runs the chip 0.050 less volts form selecting 1.4v in the bios. (when selecting 1.4V in bios my cpu is 100% stable but selecting auto in the bios which runs the volts 0.050 less it is not)

Q2 why is the asus probe and cpuid diffrent volts reading so say if the volts read 1.5V in cpuid it reads 1.328 in pc probe which do i beleave ????

Q3 will uping the volts to this level i know its not much at all do any damage

Q4 so why is uping the voltage from 1.481v to 1.529v making it stable when its not alot of volts more (this is selecting 1.4v in the bios from auto)

Q5 why when the cpu is under load the volts drop a littile ???? thought thay would have went up

god this is crazzy stuff

all i want to know is will it damage the chip runing it at 4.0 selecting 1.4v in the bios and volt results
on cpuid 1.529v or asus prob 1.376v form selecting auto which runs volts at
cpuid 1.481v or asus prob 1.296v

becasue its stable at the 1.4v select in bios

thanks

[Romaster]

hmm alot of questions you have now.. but ill answer them all and then some :D ;)


first off, prime 95 is a very good test so Im sure it has its reasons why its not at 100%.. also since your on a p4 w/ HT i'm not sure it will even say 100%.. im not a big expert on prime 95 so ill answr the questions you posted in order:

A1. not sure.. in theory if your chip is rated on intels referance thing (check the stepping at intels support thing) as 1.4V and you set 1.4v or auto both should logically be the same.. i personally dislike auto and dont trust it whenever i can use manual.

A2. personally i trust cpuid more than asus probe... i dislike the asus probe as its sometimes very inaccurate where certain software is better suited for the job due to programming issues.. theoretically they should both be the same as the cpuid should be reading from the same place the asus probe is reading from.

A3. that all depends.. some presscotts can easily do 1.4-1.6 without a problem well there are a few cases of the "presscott sudden death syndrome" (as its been known to be called) where just a bit of added voltage will hurt the chip heavily maybe even death.. I wouldnt worry about it too much.. as long as your under 1.55v there is more than a 85% you'll not be having a problem (assuming you use a good cooler) recently I ordered an xp-90 for my presscott as it is the second best air cooling out (second to the xp-120)

A4. thats .050 volts.. which really isnt little to be honest.. the .50 increase will definetly increase stability at higher clock speeds...

A5. you would think they'd go up but it varies from mobo to mobo.. some go up about .01-.03 some go down the same ammount.. on my old motherboard it was off by .050 24/7 and went up .02 every so often and then back to off by .050 so it varies.. but its rare to find a motherboard+cpu combination that wont change volts at all.. there is some degree of error while loading and idling etc...

this should answer most of your question, if you have more just post and ill reply nexttime i see it.. i like to help people who are less fortunate in knowledge..makes me feel better about myself and hopefully one day they will do the same. ;)

btw I'm curious.. what are you using to cool this 3.6@4.0?? heatsink? fan mm? how much cfm is the fan (if you know)?

good luck..

[Meridius]

well m8 i am using the stock hs and fan and heres the results

two prime95 programs running 100%cpu results

not overclocked
P4 @ 3.6 voltage set in bios (auto) memory 3:4 533mhz 4-4-4-12
asus probe idle: 1.36 47c CPU fan=2700rpm
asus probe load: 1.296 64c CPU fan=2812rpm

overclocked
P4 @ 4.0 voltage set in bios (1.375) memory 1:1 444mhz 3-3-3-9
asus probe idle: 1.408 51c CPU fan=2636rpm
asus probe load: 1.328c 68c CPU fan=2812rpm

i am going to get the thermalright hs 90 or 120 but what two fans would you recomend for these two heatsinks and a link if poss thanks

i have herd the thermalright heatsinks dont fit the prescotts yet the 775 chipset ?????????

can not find it on there webpage that would fit this

you said (as long as your under 1.55v) is this readings of CPUID or in the bios settings selection i ask this becasue even at normal clock speed and voltage selection CPUID reads 1.520v @ load but when i select 1.375v option in bios it reads 1.550v @ load in CPUID so its like 0.030v more power its puting into the chip at load

at idle its 0.032v more power at idle when selection between auto and 1.375V in the bios

thanks for the help to

[Romaster]

they fit on 478pin presscotts but unless you have a gigabyte ( a mobo makers) retention module, it wont fit on an lga 775.. not yet at least :cool:

tommorow i get my xp-90.. I ordered an enermax 9fab B fan with it.. its a 90mm fan which is good cause the xp-90 takes 80,90 and 92 or just 90 and 92.. it produces 64cfm and only 34 db.. quite leet.. and the fan cost me like $3, well its $6 on newegg (and there is $5 shipping.. so newegg isnt a good option if you want a fan.. $5 on each shipping :( ) heres a link though..

http://www.newegg.com/app/viewProductDesc.asp?description=11-999-011&depa=0

its also nice cause it has manual rpm settings.. not a switch though, just an attachable and unattachable wiring with a resistor on it :rolleyes: but it gets the job done i'd guess..

[Meridius]

do you know what the stock intel 775 fans db sound is and cfm ???