overclocked P4 stability problems..

[touser]

hi all my problem is i am trying to overclock my P4 1.7ghz to 2ghz i set the fsb all the way up to 1.85v (as high as my motherboard would go) because any lower and it wasnt very stable at all..and at that it got through all of the 3dmarks2k1 benchmark but when i try and run a game like tribes2 it freezes up in less than a minute at that..my full load temp was only around 40C load at 2ghz so i doubt it is a heat issue, i am running 256mb of samsung rdram and an abit th7-raid mobo with a 450w enermax power supply..can anyone help me with this problem? it would be greatly appreciated!. thanks in advance.

[Burning Phoenix]

There are a couple of other guys here overclocking their p4 1.7 over 2.0 and they'll probably jump into this thread. I have one but i'm waiting for a dangerden waterblock and peltier first. I think it still maybe the temps. It seems to me that as you overclock to higher MHZ the underload temps shouldn't be compared to what they used to be at stock settings. So as you get towards 2.0 i think the temps should be far lower underload to be stable.
I'm curious as to what your 3dmark2001 scores were at that speed. Also to those others who may post to this thread who also overclocked their P4 I would like to hear about their scores.

[touser]

thank you Burning Phoenix for the reply, i just dont think i can get much lower temps with my watercooling rig withought going the pelt route and i really dont want to do that just yet.. :( hopefully one of the guys with the 1.7@2gig can help me out :)

[Burning Phoenix]

peltiers aren't so bad as long as your completely insulate from the peltier down from condensation. I had my last computer a P3 700 @1022 running temps at idle 8 F (-13 C) and 35 F (3 C) under load. I was using only a 72 watt peltier which i'm going to use 1st on new waterblock when it arrives. I have 2 of them but one my be burnt.

[JetMech]

Originally posted by touser
hi all my problem is i am trying to overclock my P4 1.7ghz to 2ghz i set the fsb all the way up to 1.85v (as high as my motherboard would go) because any lower and it wasnt very stable at all..and at that it got through all of the 3dmarks2k1 benchmark but when i try and run a game like tribes2 it freezes up in less than a minute at that..my full load temp was only around 40C load at 2ghz so i doubt it is a heat issue, i am running 256mb of samsung rdram and an abit th7-raid mobo with a 450w enermax power supply..can anyone help me with this problem? it would be greatly appreciated!. thanks in advance. What you are seeing is the standard reaction of the !.7 to overclock. The most successful 2.0 overclocker to my knowledge is TC. I believe he did a pin wrap to achieve that. I have gone as High as 2.125 but not stablely. When I play games I turn it back to 1.7. Like Burning Phoenix I am working out some cooling solutions and haven't run 3DMARK2001. I am not familiar with your motherboard or it's bios options. I am using the ASUS P4t. It has a 3x and 4x mutiplier option in the advanced settings area and I was able to reach 2125 using 3x. The ram is also being overclocked with the FSB so you do whats necessary to keep it as close to 800mhz as possible. With 4x I could only get the FSB to 115 unstable. I believe that puts 920mhz on the ram. At 2125/3x my FSB is 125 which is 375 FSB but the ram is running at 750mhz. I think I got that right anyway you get the idea. I can do everything but games at that speed and I'm keeping it there for burnin purposes. My idle temp is 40c/104F. I don't want to heat it up too much with this Gorb. I don't know if it's qualified for that speed. I'm considering water cooling for the first time after finding out that quick disconnects are being made available. We use them for fairly high air pressure on robots so I'm confident they won't leak if installed properly. Hopefully TC will see your thread and beam you up. Hope my input has been helpful and I hope you have that multiplier option in your bios.:)

[glennfoster]

I just wanted to say how lucky you all are that you have a decent motherboard. Just wait for the roars of laughter as I post my message in a couple of weeks....tried to overclock my 1.5GHz P4.....on my Microstar Motherboard.... :(

Still, if I save up some money....


Good luck with the overclock.

[TC]

Hey Touser,

The only special thing I did was a pin wrap. The Asus P4T has a bug where it does not raise the core voltage when you change it in the bios. You can select a higher voltage, but in reality it stays the same at whatever the default is. Using the spec sheets from Intel I located the voltage identifier pins on the bottom of the chip. I used a very fine wire and wrapped this around the pins according to the voltage I wanted the cpu to run at - in my case 1.85. Other than that I have done nothing. In fact I'm using the Intel oem heat sink and fan. Temps under load are around 44C. I did apply arctic silver to the northbridge, but I do that on all mobo's. I suspect I lucked out with a good chip too because I have booted into windows as high as 2.2GHz with a homemade water block. I wouldn't call 2GHz 100% stable though. I can run 3D Mark in a loop for a few hours and it will exit with an error at some point. I don't have trouble with anything else and Seti runs 24/7. Rest of system specs:

640MB Samsung PC800
Gainward GF3
40GB Maxtor DMax 60
Linksys nic

[touser]

thank you guys for the help it looks like the highest i can go right now is 1955mhz..i set in my bios the core voltage to 1.85v but under the health section the cpu voltage was only 1.74v :( is there anything else i can do? i remember reading somewhere to turn off rdram turbo mode? how do i do this as i dont see anywhere on my motherboard an option to do so :(

[Wicked Klown]

I know bout a week ago I was having probs with 3Dmark2001. It keep freezing on either game 2 or game 3. I took it to the shop to have them ghost a HDD for me and look at the prob with the video card and all he did was install the latest driver and now I can run it with no probs. I don't know if this will help you but it might.

[Big Lar]

I don't have a 1.7, nor do i have the abit board, But, what I do have is= The Asus P4T, and a 1.5gig, and I run mine daily at 1800, at 4x on the rdram, and in Turbo Mode. Thats 120fsb, The cpu will do 1949, Stable, however with a 1.8 on the way I don't want to take the risk of blowing this chip, and is my nature to run at Under the max the chip will do, besides, I get better bandwidth at 1800. As to Pin Wrapping, I have heard that "some" asus boards won't change voltage, however mine does just fine, reported in bios and mobo monitor when running 1949 at 1.85v ( actually reports 1.83 to 1.84v). Cooling is the name of the game tho, No peltier needed, just water cooling with cool water, not ambient, cooled. It also helps to have air blowing over the Rdram.

[Burning Phoenix]

I have talked about my old system 700@1022 with a homemade waterblock and peltier but i had problems with that waterblock leaking alot. So i'm now waiting on dangerden's MAZEP4-2 waterblock. I would like to reach 2.0 or better with my 1.7 chip. Now that you mentioned it my MBM seems to show the incorrect core V. No matter how high i put the voltage it doesn't seem to make a difference.

TC do you leave the wire wrap on our cpu all the time now or was it just for experiment? Send me a link or something on how to do that.

[TC]

Originally posted by touser
thank you guys for the help it looks like the highest i can go right now is 1955mhz..i set in my bios the core voltage to 1.85v but under the health section the cpu voltage was only 1.74v :( is there anything else i can do? i remember reading somewhere to turn off rdram turbo mode? how do i do this as i dont see anywhere on my motherboard an option to do so :(

Here's the link to how to do the wire wrap for you folks that want to try it. My P4T has the bug changing the vcore also, so the wire has to be left on all the time. It forces the board into thinking the cpu is asking for 1.85 by default.

P4 Vcore Mod (http://www.vr-zone.com/guides/p4mod/)

[touser]

TC thank you for posting that link..one question tho sence all the pins have to go into a little socket wouldnt running a wire from one pin to the other stop those pins from going into the hole?

[JetMech]

Originally posted by Big Lar
I don't have a 1.7, nor do i have the abit board, But, what I do have is= The Asus P4T, and a 1.5gig, and I run mine daily at 1800, at 4x on the rdram, and in Turbo Mode. Thats 120fsb, The cpu will do 1949, Stable, however with a 1.8 on the way I don't want to take the risk of blowing this chip, and is my nature to run at Under the max the chip will do, besides, I get better bandwidth at 1800. As to Pin Wrapping, I have heard that "some" asus boards won't change voltage, however mine does just fine, reported in bios and mobo monitor when running 1949 at 1.85v ( actually reports 1.83 to 1.84v). Cooling is the name of the game tho, No peltier needed, just water cooling with cool water, not ambient, cooled. It also helps to have air blowing over the Rdram. It is my understanding that rather than blowing the chip if it overheats, the P4 throttles down like the speedstep mobile processors. I have not seen this in writing. I have though experience a lockup unlock, lockup unlock sequence as if the chip locks due to heat and after a couple of minutes resumes operation until it locks again. Eventually it locks solid. The whole time this was going on the cpu temp was 134f.

[Big Lar]

Well, I have no fear of a heat issue, as I keep my cpu at about 55f, and the highest it hits is about 75f. I cool my water. I have found from previous experience, at least on my behalf, that excessive running at the "Max" does hurt the cpu over time. I know this has been highly debated, but that is My experience anyhow, hence I find the Max of the cpu, and drop er back a bit, I would run at 1875 daily, but at 1800 the bandwidth is better at 4x.

[Burning Phoenix]

you may have something there for my 1.7 ghz locks up or shuts down after which i remember each time the threshold of heat for the chip happens to be for around 134 F also! Could this be it max?

[JetMech]

I do't think this is the max is the max; however it is not the best place to be. I'm looking at a cooling solution (Alpha Pal or Swiftech 462 ) so I can work below that number. I can see your logic tho. There has to be a limit temp the at which the throttle is triggered. You might be right. WE've got hte flag up. Maybe some of the other guy's will monitor this in their attempts.

[Burning Phoenix]

just to much of a common # in which my computer shuts down during overclocking , hard gaming or seti

[TC]

Originally posted by touser
TC thank you for posting that link..one question tho sence all the pins have to go into a little socket wouldnt running a wire from one pin to the other stop those pins from going into the hole?

Yes it can prevent the bottom of the chip from going all the way down flat and level in the socket - so you need to use very fine wire. I've not had any problems with mine, but I used 26 AWG gold wire. If you're worried about this there are a couple things you can do. 1) you can use a dremel tool to shave the socket down slightly where those wires will be, and that gives them some room to fit into the top surface of the socket and allow the cpu to fully seat. 2) you can wire the same pins on the back of the mobo instead of the cpu.

If you wire the cpu it's very important that the connections are good. Easiest way to do this is form the wire into a triangle (if you're wrapping three pins) and twist the ends together. Loop the circle over the proper pins, then use needle nose pliers or a small vise grip to grasp the twisted ends and turn while gently tugging away from the cpu. Twist until the wire is tightly bound around the pins, and carefully cut away the excess wire. Make sure the twisted ends are not long enough to touch an adjacent pin.

[JetMech]

Originally posted by TC


Yes it can prevent the bottom of the chip from going all the way down flat and level in the socket - so you need to use very fine wire. I've not had any problems with mine, but I used 26 AWG gold wire. If you're worried about this there are a couple things you can do. 1) you can use a dremel tool to shave the socket down slightly where those wires will be, and that gives them some room to fit into the top surface of the socket and allow the cpu to fully seat. 2) you can wire the same pins on the back of the mobo instead of the cpu.

If you wire the cpu it's very important that the connections are good. Easiest way to do this is form the wire into a triangle (if you're wrapping three pins) and twist the ends together. Loop the circle over the proper pins, then use needle nose pliers or a small vise grip to grasp the twisted ends and turn while gently tugging away from the cpu. Twist until the wire is tightly bound around the pins, and carefully cut away the excess wire. Make sure the twisted ends are not long enough to touch an adjacent pin. If you use the dremel method be sure to use air and blow the pin sockets out to ensure no foreign matter interfers with current flow. Also be sure to stay grounded if working the backside of the board. This may seem like a no brainer but some may not be aware.

[TC]

Originally posted by r_wesleyjr
If you use the dremel method be sure to use air and blow the pin sockets out to ensure no foreign matter interfers with current flow. Also be sure to stay grounded if working the backside of the board. This may seem like a no brainer but some may not be aware.

Yeah good idea. Get a static strap if you're going to work on the board - well of course you should have one when you work with any electronic part of your computer. Definitely blow out the socket if you work on that.