chipset and clock generator cooling?

hi
i have a bx-133 raid. i can post but it only boot halfway into windows at 182fsb with my 800eb/1.1ghz, i know the processer isnt the limiting factor here. i can only get it to run stable at 155fsb but memory performance is poor with all setting at low.
if i added a large heatsink and fan to the chipset will this make my memory overclock higher or able me to reach a higher fsb? i am using cas3 pc133 micron at the maxium voltage 3.9v, seems the only way to get memory to go faster?
as for the clock generator, why do people put heatsinks on those? i touched it with my finger and it wasnt very hot at all. how does the clock generator get under any stress since it just sets the speed for the mobo. is there any benfit in cooling this chip? will that make a higher fsb more stable or is it just a myth.

its probably not letting you go any higher because your throwing

your PCI\AGP bus out of whack

whats your divider set too?

and your memory may have found a stooping point also

any ideas?

yo
not many out there clued up on chipset cooling? where do i go to get in the know?
thanks

Re: any ideas?

mircofire said:

yo
not many out there clued up on chipset cooling? where do i go to get in the know?
thanks

Click to expand...

Most boards have holes so you can put a HS on your chipset. That is it. As for hitting 180+ fsb with a agp of 2/3 that is plain crazy Most likely that is what is holding you up as also your pci bus too. You need a turbo pLL if you want to go higher with that board!

Northbridge cooling gets very important the higher you go up in FSB. However, even though this will often help overclock the system higher, I doubt adding chipset cooling will help the memory much. You will need better and faster RAM for that.

Evidence does suggests that clock generator cooling sometimes does help a little at very high FSB, but certainly, northbridge cooling is far more important.

Trey is right, even using 1/4 divider, the PCI bus is running at 40 MHz at a FSB of 160 MHz and that's about the limit for most controllers and PCI cards.

I'm running 172FSB with my p3 1g but ive got a huge HSF on my nortrhbridge, get a medium sized HSF for a small CPU [larger then a 486] half the size or bigger then your current CPU HSF you can attach it with thermal compound in the center and 4 drops of super glue in the corners. This may or may not help you as at high FSB settings you have a number of things that can hold you back. I cant believe you havnt fried your memory at 3.9 we wish you luck!

better now

hi
put large aftermarket p3 heatsink and fan on bx chipset, and also a heatsink on clock generator.
i attached it with metal bolts so it rock tight on the mainboard with artic silver paste.
the result is better but i have not fully tested it yet but it seems good for 162fsb at this stage before i could only get 155fsb, thats like stable type speed.
yeah my memory seems to get hot when i touch it under full load, not sure what you mean about memory havent fried it i have tested with a few different modules and all are still ok. 3.9 is max on the board so i dont think its that unsafe dont you think? anyhow it doesnt wanna go much higher fsb unless i do it. i notice that you are running 2.1v on your cpu i thats high enough for me to say the same but i wont cause i run that on my 700e and its fine, it becomes unstable at 2.2v. i think that the one that fry are just bad luck cause theres a whole world out there and some haveta go wrong.
any other ideas on issues and tips are welcomed

At 2.1 ive been gambling for a while but i think VIO settings turned up is gambling even more VIO will fry things faster [a little less forgiving] anything over 3.6 is gambling at 3.9 you may expect problems or a very short lifespan for MB components, ive seen many people in here who have fried parts,MB's ect from VIO but very few Intel CPU's fried from high CV. I think you have said a key word about cooling or adding HS's or HSF's to chips and that is if its hot to the touch add a HS if its cool theres no need. I would make a HS for your ram if you have 1 stick then add HS's to both sides and this may help to keep your ram alive longer, but you are still pumping allot of V to other components so i would check for other hot spots.

Testing completed

yo hi

yeah been testing over this weekend and i have now finalised my stable speeds right down to the last usable mhz.
the input/output voltage had to be reduced to make it stable, i am now using 3.7 it really likes this voltage, any less no boot, any more unstable after a few mintues of max2001.
i have 955mhz@159fsb and get a memory bandwidth of 430/490@cas333
or i can have 935mhz@155fsb and get a memory bandwidth of 445/505@cas332 , 2 being percharge if i got the numbers right, better bandwidth but less cpu mhz, tested with qauke 3 timedemo and its about .6 fps slower so i decided to use the 955mhz/159fsb one.
the memory i am using is a nova 256mb cas333 stick (dont trust a computer salesman they can often rip ya off, he told me that it was cas232 or as he refer to it as cas2.5) anyhow can only use the one stick at these speeds.
not sure whats holding me back from going higher, i know its not the cpu cause i can post it at 1.1ghz and its an old cB0 stepping (pretty good i thought), it could be the chipset maxing out even with my large fan and heatsink on it but i think its most likely the ram at the end of its usable limits. its hard to get good ram here in new zealand so i not sure if its worth the extra cost of sending overseas to get it at a high exchange rate plus freight. ice pc150 might be the go if so. single 256mb or two 256mb? i not sure dont want it to be unstable for a higher cost. ummm?

your thoughts on this topic will be well accepted. thanks

lee.

I bet a dollar the thing holding back from going any
higher is your PCI bus speed.

Unless you got a motherboard with a 1/5 divisor on it,
you are getting close to the limits.

My hard drive started flaking out last time I tried to go
any higher than 162. Its one of those IBM 60GXP 40g
drives. Got the PC150 Kingmax memory.
(I had turned the ram speed down on my last few tests
so I know it wasnt a factor in that.)

You might want to run with the 155fsb so you arent
stressing the rest of your system quite as much.

Good luck !

you may have to pay your betting money to me

when i was testing, i was testing without any pci cards in there the only thing leaft would be the on board raid chip and a 1xcdrw/1xdvd but you may have seen many running much higher with this board. one more mhz (160mhz) unstable after few mintues, but at 159 i ran for 9 hours looping 3d max 2001. yeah that 155 could be a good idea but if theres no problems at 159 then having 155 percharge 2 maybe more unstable (have not tested it yet for 9 hours+) if it not crashing then there must not be that much of a stress. i have a 30gb ibm 7200 deskstar and today i just added my old flakey 13gb quatium fireball 5400 (useto not work but leaft it sitting for a year and now it seem ok, well for now anyhow).
are you sure that it would not be the memory since its way over spec 26mhz, pc133 isnt really a great overclocker compared to pc100 was, then again i managed to almost make it into windows with 182fsb, so the memory much be sort of ok? or maybe it was 3.9v pushing it sort of the way there?
so many ? i think i have tapped this systems limits.
are reacon that ice pc150 is not worth the trouble since there are much fast things out now.
i guess i get caught up in the challange of trying to get as much as possible out of it instead of using the bloody thing.
i have got what i call it "CHA" short of 'computer hardware addiction'
i wont be happy untill smoke starts pumping out of the back of the pc unit? hmmmm.

forgot something

i also decreased my core voltage from 1.90v to 1.70v today cause 1.90 was just for testing. this old stepping chip much be one of the good ones, i wouldnt of guess so cause its old stepping and a malay sl3y2 800eb. funny thing is i can run it at default 1.65v but i thought that 1.7v would be better since these abit bx-raid boards drop upto -0.05v under full load, at default voltage that affectlly gives me a range of 1.60-1.65v so i choose 1.70v so it gives me a more constant default of 1.65-1.70v. i dont know if it means anything but i have stability in mind. have i done the right thing? i know at 1.60v it crashed into windows, that most likely cause at that moment the core goes to 1.55v underload of booting.