Celeron 1.2 oc observations and question

I recently overclocked two different Celeron 1.2 cpus in the same setup (Abit st6 with Thermalright AX-7 w/ Panaflow 43 cfm fan, 512K PC133, Maxtor 30G 7200 rpm HD, CD Rom, CD R/RW, floppy, SB512 PCI, nvidia tnt pci [since changed to geforce 4 ti4200)

I note the database which has many 1.2 running at 1600 (133 or better fsb). I note many threads, the gist of which is 90% of the 1.2 will only go up to 125 fsb (and no higher). My limited experience mirrors what I read in the threads.

One cpu was fully stable at 127 fsb at 1.575 v. Temps are 40 idle / 47 load (sandra stress test).

Increasing the fsb got increasingly unstable performance. Would post and go into windows at 128 (1.60 v), 129 (1.65 v) and 130 (1.725 v). However, increasing errors with "explorer.exe generating errors and closing" when hitting the START button. At 131 (1.75 v.) would post and then crash (message was "MODE-EXCEPTION_NOT_HANDLED")

Increasing voltage did not increase stability at any of the fsb. It only brought higher temps with no seeming affect on stability. Lowering voltages would result in windows crashing on boot up.

The other cpu was fully stable at 124 fsb (1.60 v). Progressive instablity until refusing to post at 128. Again increasing voltage did not increase stabilty. It only increased heat.

I do not know what caused the "explorer.exe" and "MODE-EXCEPTION_NOT_HANDLED" errors. Any input would be appreciated.

I have the same errors when trying to run my 1.2 at higher fsb than 130. Doesn't matter what vcore I set, it's unstable anyway. Had really low temp in my room one day and the errors in windows dissappeared (it was stable at 135fsb even), so it's the temps for me that must be lowered. Better cooling is the answer for me anyway.

youll find out that many information regarding the cel 1.2 ppl use to post in threads about the 1.0a and 1.1a.
you just dont trust the database!

i asked for help on choosing the tual cel and all i heard was about the 1.2 would not overclock as well as those both.
the setup you got is pretty much better than half the setups we see in the database, and there r ppl there saying they got 133FSB with stock cooling.
i dont know whats wrong with the database but im sure thats not all right - at least from everything ppl told me and things i readed here.
i guess that you got is ok.

but there is something you might want to try if you can. as Brosken said about the temperatures: it seems like the temperature stable limit for most the tualatin celerons is about 40-45ºC. its really hard to get lower than that with an overclocked cpu so its possible the limit for the 1.2 is something lower than that range.
i cant think of other reason why the cel 1.2 would not overclock as well as the 1.0 and 1.1a (to at least 133MHz).
perhaps you could post us some reply about this. you never talked about temps. if you find out a way to lower temp even more it would help us to know if that about the temp range is really correct.

My notes on temperatures were (using Sandra Stress test):

100 fsb - 38 idle / 42 load (1.475 v)

127 fsb- 40 idle / 47 load (1.575 v)

128 fsb- 43 idle / 48 load (1.60 v)

I didn;t right down the higher fsb temps - but I'm sure they are higher. These are the lowest voltages that would load windows (and at 127 run stable). Higher voltages just increased the temps.

I did run case off (in a cool garage) and it was stable at 128 fsb - 39 idle / 45 load. (Can't keep it in the garage w/o case though.)

Those temps are too high. You need way better cooling to determine if the temperatures are holding you back. You know your cooling is adequate if your system loads below [email protected] at any fsb. You're only at 1.6v and your temps are exceeding 40c+.

For the most part, if your Tualatin doesn't overclock well, if it's curable, it's usually temperature issues.

umm youre not making good contacvt with the heatsink or something, there is no way an AX7 could run the hot with a Celeron...

Almost sounds like he doesn't have any thermal compound on these chips or he has way too much.

What are your ambient temps? If you say you can run it stable in the garage, your ambient temps are too high. You also need to consider case cooling along with cpu cooling.

rayik said:

My notes on temperatures were (using Sandra Stress test):

100 fsb - 38 idle / 42 load (1.475 v)

127 fsb- 40 idle / 47 load (1.575 v)

128 fsb- 43 idle / 48 load (1.60 v)

I didn;t right down the higher fsb temps - but I'm sure they are higher. These are the lowest voltages that would load windows (and at 127 run stable). Higher voltages just increased the temps.

I did run case off (in a cool garage) and it was stable at 128 fsb - 39 idle / 45 load. (Can't keep it in the garage w/o case though.)

Click to expand...

rayik
I have almost the same situation as you do with my new ST6 mobo and a 1.3 chip. I am begining to take all of tempeature quotes with a grain of salt. I have 3 running systems 1 ST6 one Vh6t and one Msi6388 and my temps are flat out 10c higher in the st6 motherboard. I do not know wich one is the more accurrate of the three but the msi and vh6t read almost the same with the same chip but the st6 motherboard wont even give me a reading of less than 41c at first boot and default speed and vcore settings so I dont trust it. I have been discussing this in my original thread vh6t vs st6 and have got a lot of different opinions about it. It is possible that most other motherboards use cheaper sensors than the st6 or the newer revision boards just read high but it makes no sense to me since I get the same performance out of all of my mobos as far as heat is concerned. The bottom line is in my case in the st6 mobo my temps are 43-45c idle and 47-50c load and anything over 55c causes problems. In the other boards it reads 33-35c idle and 37-40c load and anything over 45c in them causes problems. Many will say I have a heat problem. I say no I have a temp probe problem because I have excellent cooling in all of my cases and am running a copper hsf/delta38 fan that sounds like a vacume cleaner running and yes I use as3 compound and have done this many times before I am no beginner at this. This is not the first time someone said that the st6 mobo is known for giving higher readings and I am sure it wont be the last.

davefred99 said:

rayik
I have almost the same situation as you do with my new ST6 mobo and a 1.3 chip. I am begining to take all of tempeature quotes with a grain of salt. I have 3 running systems 1 ST6 one Vh6t and one Msi6388 and my temps are flat out 10c higher in the st6 motherboard. I do not know wich one is the more accurrate of the three but the msi and vh6t read almost the same with the same chip but the st6 motherboard wont even give me a reading of less than 41c at first boot and default speed and vcore settings so I dont trust it. I have been discussing this in my original thread vh6t vs st6 and have got a lot of different opinions about it. It is possible that most other motherboards use cheaper sensors than the st6 or the newer revision boards just read high but it makes no sense to me since I get the same performance out of all of my mobos as far as heat is concerned. The bottom line is in my case in the st6 mobo my temps are 43-45c idle and 47-50c load and anything over 55c causes problems. In the other boards it reads 33-35c idle and 37-40c load and anything over 45c in them causes problems. Many will say I have a heat problem. I say no I have a temp probe problem because I have excellent cooling in all of my cases and am running a copper hsf/delta38 fan that sounds like a vacume cleaner running and yes I use as3 compound and have done this many times before I am no beginner at this. This is not the first time someone said that the st6 mobo is known for giving higher readings and I am sure it wont be the last.

Click to expand...

Dude check this jpg.



So you think that this cel-t at default vcore and voltage was actually running 10 deg. C cooler? I don't think so! That would be room temp dude. If you want to read the whole article go ahead!

The ST6 is not reading temps wrong!
http://www.overclockers.com/tips716/

ol'man
I am not saying that your mobo or anyone elses is reading wrong temps. I am saying mine is fllat out wrong. I have tried both of my cely chips and the readings are consistantly higher in MY St6 mobo. I believe it is either a probe problem or a bios problem but there is no way my 1.1a chip @ default vcore/speed and a good hsf is running @ 41c at first boot idle and my case temps at boot are reading about 36c. My 1.3 chip boots @43c all in the ST6 mobo. The same two chips on average read 10c cooler in my vh6t mobo. All readings are at a room temp of 75deg(air conditioned) Also I read the article and noted that the listed mobo was running some special bios. I am going to re-flash my bois to an older version and see if it makes a difference. How else can you explain such a big difference in temps between different mobos ???? ol'man you are the man when it comes to these ST6 mobos and cely chips but I think this time I disagree with you at least in my case.

davefred99 said:

ol'man
I am not saying that your mobo or anyone elses is reading wrong temps. I am saying mine is fllat out wrong. I have tried both of my cely chips and the readings are consistantly higher in MY St6 mobo. I believe it is either a probe problem or a bios problem but there is no way my 1.1a chip @ default vcore/speed and a good hsf is running @ 41c at first boot idle and my case temps at boot are reading about 36c. My 1.3 chip boots @43c all in the ST6 mobo. The same two chips on average read 10c cooler in my vh6t mobo. All readings are at a room temp of 75deg(air conditioned) Also I read the article and noted that the listed mobo was running some special bios. I am going to re-flash my bois to an older version and see if it makes a difference. How else can you explain such a big difference in temps between different mobos ???? ol'man you are the man when it comes to these ST6 mobos and cely chips but I think this time I disagree with you at least in my case.

Click to expand...

Well you could have a faulty probe. I just don;t know what to add about this anymore.

How are you measuring your case temps? If you are using the on board thermal sensor then you can bet that is always going to be high. Check this!

If you want to do well with the cel-t you want to keep temps around 40 deg. C at first till you get a feel for how high your chip can go. I know the ST6 is not telling lies. This chip runs almost at load temps all the time cause idle software does not work with it for some reason. win2k has this imbedded in it but it does not work on the cel-t. I have been playing around with these tualatin chips for about 1 year now since they first came out. Youcan take my experience and advice if you want or you can try to figure it out on your own. I could right a book about how much I know about the cel-t and its charechteristics

Anyway. Get your temps down! I do all of my testing without the case in the equation. If you have bad case flow I can guarentee your chip is going to be running hot! These chips are running 40w~50w which is really not that low of a number to think they can be cooled really easy. Especially with such a small core and hopefully Intel got the IHS situated right too!

I just tested my aux thermal sensor in a glass of water with a digital thermometer and it said the temps were 92.3 when my digital thermometer said 92.9. These temps are very close so I don;t doubt what they are saying. I think my digital thermometer can goto to 115 so I may try it on the HS too and see what I get!

Like I said above if you think your board temp is your case temp you may be very well mistaken!

I use a plugin temp probe with a thermister on the end. I was very surprised by how well it gave temps. It is left over from my BE6-II. Funny thing is that ABIT doesn't give you these anymore.

Against the outside edge of the HS I got a temp of 39 deg. C with a digital thermometer so I doubt the temp probe on die is telling lies with my mobo. The core according to sandra was around 44 deg. C. You can't expect the core of the cpu and the outside edge of the HS to be the same temp.

I just did some figuring and my chip is making 50w at 1.6GHz/1.65v.

A 1500MHz cel-t @ 1.8v would be producing 56w which is alot! To think a retial HS could keep up is absurd!

Here are some specs for this chip @ 1500MHz and different vcores.

1.5v=39w
1.55v=41w
1.6v=44w
1.65v=47w
1.7v=50w
1.75v=53w
1.8v=56w
1.85v=59w

Now we are getting into athlon territory!

You have to understand also this is pushing alot of watts through a much smaller area than the cumine too so to compare the two chips is absurd also by its own rites!

I used to run my cel-t at 1.8GHz/1.85v that would equal 70w!

I must also reiterate what some people think about the cel-t 1.2GHz. Many have said that 90% of the time it can only hit 1.6GHz(One person at first and others are jumping on the boat now .) I trust the data base here for the most part and personally I think with proper techiniques that number is closer to 50~70% should hit 133fsb as many used a retail cooler.

Many here like to point out that the 1.1a and 1.0a usually crap out around 1500~1525MHz so they automatically think that is what the 1.2GHz should too. I have owned three 1.2GHz chips and they all did above 1575MHz. When I had those chips also my cooling was not as good as it is now also so that number may have been higher. 1 did ~1.6 @ 1.85v which is pushing it for watts produced and conventional cooling methods. The other did 1.57GHz on 1.75v and would not go higher no matter what. And the last beauty does 1650 on ~1.75v but I keep it at 1.6GHz just cause it runs way cooler and quiet this way and I also keep my PCI in 100% spec.

Like I have said over and over the cel-t 1.2 retail philly chip is the one to get.

I have a 300w psu. I also have a 250w psu and it could not get my chip stable above 1.5 or so GHz. PSU's are a big importence with this chip!

Another big importence is the IHS and how good your thermal interface is!

Another big importence is how well the IHS is touching the core inside of your IHS. I have taken 5 of these IHS's off the cel-t/PIII tualatin and I fond one of them to have a nick in the IHS with a raised edge that apparently was inhibiting 100% core contact. If this happend with one of them it may have happend to more than mine.

One of the largest importences is how good of cooling you have and if you have copper next to the core or aluminum. Aluminum hsa only about 2/3~1/2 the thermal conductivity of copper so with such a small core as the tualatin you want to have copper next to it.

And the last big importence is that your diodes are working correctly and telling the true temps of your chip

ol' man said:

I just did some figuring and my chip is making 50w at 1.6GHz/1.65v.

A 1500MHz cel-t @ 1.8v would be producing 56w which is alot! To think a retial HS could keep up is absurd!

Here are some specs for this chip @ 1500MHz and different vcores.

1.5v=39w
1.55v=41w
1.6v=44w
1.65v=47w
1.7v=50w
1.75v=53w
1.8v=56w
1.85v=59w

Now we are getting into athlon territory!

You have to understand also this is pushing alot of watts through a much smaller area than the cumine too so to compare the two chips is absurd also by its own rites!

I used to run my cel-t at 1.8GHz/1.85v that would equal 70w!

Click to expand...

Well ol'man no one can accuse you of not doing your home work on these chips and I certainly think you know what you are saying and agree with you on the fact that most people dont really understand just how much oc'ing these tullys to 1500+ at vcores of up to 1.85v taxes there system. Most read a couple of threads or look at the data base and figure its a no brainer to get these things to there max oc. The truth is it takes many hours of work and a lot of proper planning to do it right. I think that your figures on heat are pretty much right on if everything is taken into consideration. I dont know what else to say about my temp readings though because I have tried just about every trick I can think of and it comes down to either my probes on my st6 are wrong or my probes on my other two mobos are lieing to me. I still cant figure how my temps based on your figures of roughly 1500 @ 1.5v (39W) can produce temps as high as I am getting (47-49c load) with a good copper hsf/38delta and IHS removed.
I am going to try and find a decent temperature gauge and do some measuring to compare. Any other thoughts.

arent you both forgetting the fact the mainboard bioses use to apply compensation to the temp the sensor reads?

sandra also reports temps with the compensation applied. compensation can vary from 10º to 15ºC more over the real temp; and some monitoring programs will just show the same temp + compensation (as the mainboard manufacters probe does).
i dont really trust snadra to read my temps because it does read them wrongly. right now sandra reports my cpu at 17ºC, MBM 33. and snadra still applies the compensation.

so i didnt see you talking about Motherboard Monitor, if you havent it, get it! it will give you the control over the compensation and youll be able to choose right each sensor you want it to read from.

davefred99 said:

Well ol'man no one can accuse you of not doing your home work on these chips and I certainly think you know what you are saying and agree with you on the fact that most people dont really understand just how much oc'ing these tullys to 1500+ at vcores of up to 1.85v taxes there system. Most read a couple of threads or look at the data base and figure its a no brainer to get these things to there max oc. The truth is it takes many hours of work and a lot of proper planning to do it right. I think that your figures on heat are pretty much right on if everything is taken into consideration. I dont know what else to say about my temp readings though because I have tried just about every trick I can think of and it comes down to either my probes on my st6 are wrong or my probes on my other two mobos are lieing to me. I still cant figure how my temps based on your figures of roughly 1500 @ 1.5v (39W) can produce temps as high as I am getting (47-49c load) with a good copper hsf/38delta and IHS removed.
I am going to try and find a decent temperature gauge and do some measuring to compare. Any other thoughts.

Click to expand...

Your temps sound about right with only a slight portion of the HS touching the core. When ever my HS was at a angle that is about where my temps were at. Funny they were not higher. Anyway I guess when the HS is a little off kilter I guess these chips awill still run but run hot with upper limits compromised.

oTTO said:

arent you both forgetting the fact the mainboard bioses use to apply compensation to the temp the sensor reads?

sandra also reports temps with the compensation applied. compensation can vary from 10º to 15ºC more over the real temp; and some monitoring programs will just show the same temp + compensation (as the mainboard manufacters probe does).
i dont really trust snadra to read my temps because it does read them wrongly. right now sandra reports my cpu at 17ºC, MBM 33. and snadra still applies the compensation.

so i didnt see you talking about Motherboard Monitor, if you havent it, get it! it will give you the control over the compensation and youll be able to choose right each sensor you want it to read from.

Click to expand...

Mine does not apply compensation with the thermistor I plug into it. In fact temps were right on in sandra with a digital thermometer I used!