WIP: Karaktu's (and Zap's!) Unofficial Guide to mobile Celeron overclocking success!

[BigBadBiologist]

Originally posted by: Zap
I made my own spacer today in under 1 minute. I took a floppy disk with a metal shutter and trimmed the metal a bit. Worked well.

BTW, my mobile Celeron in my main machine locks up when it goes somewhere over 52ºC (that's what it says in BIOS after immediate reboot). Guess my low RPM NMB fan isn't enough... only happens when I play Zero Hour.

Aren't the floppy shutters made from steel? What are your CPU temps when using this?

 

[Zap]

Flashing would be good also.

Steel should still transmit heat. Got the one I did here in a Shuttle SS61G, CPU is a mobile Celeron 1.6GHz running at around 2.52GHz at 1.39v. Machine has been running, but sitting idle for a few hours. BIOS reports 36ºC system temperature and 38ºC CPU temperature.

 

[greenwar]

<quote>Sounds like a good excuse to try it w/ the Spark7. I've used aluminum roof flashing in the past, but not since I've found the few hsfs that work straight up. Green: is there alot of pressure when you use the Spark7 or is it barely making contact?</quote>

I am not sure about that actually. Seemed to be it was barely touching. But when I take the hsf off, i see that arctic silver was thoroughly spread and as i mentioned, the cpu runs at 43 degress on full load. So I assume the hsf has fairly enough pressure applied to the core.

 

[TStep]

Tried an old school Aopen MX46 with a Celeron 1.8/256k

-Would not initially recognize the cpu. Cleared CMOS and then it booted right up. No pin mod needed.
-Has CAS adjustment for memory. FSB adjust from 100-150 in 1mhz increments. DDR dividers. No voltage controls. Pretty basic, but has the essentials
-Onboard video has pretty sharp 2D. Not the best, but better than many of the nVidia cards I've tried on this LCD.

 

[Zap]

I just ordered a Shuttle AB60R (or whatever model, with the 865PE chipset) from Mwave. They had it on a $25 special, ended up costing about $34.XX after tax and shipping. I expect to have to do a pin mod. Don't know if I'll have time to play with it thoroughly before... the next few hectic weeks...

 

[Minot]

I got a steal of a deal for a SL6M5 Mobile Celeron 1.5 GHz C1 stepping on Ebay. With shipping it cost me $15.50! Anyway, I paired it with an Asus P4S800-MX Ver 2.0 and 2 x 256MB PC3200 Kingston Ram. The CPU defaulted to 1.65v for Vcore and booted right up at 15 x 100 MHz. I ran Memtest86 for one pass and all was well. I installed Windows 2000 Professional and installed some rudimentary benchmark programs. This is what I installed:

- PCMark2002
- PCMark2004
- Prime95

After I completed installing the drivers and configuring the OS, I went into my benchmarks at stock settings:

1500 MHz
PCMark2002
- CPU: 3676
- MEM: 3765
- HDD: 889
PCMark2004
- System: 1806
Prime95
- 2048K FFT: 127.696 ms
- 4096K FFT: 372.868 ms

I then bumped the system up to 15 x 200 MHz for 3.0 GHz. The system appeared stable and booted on the first attempt. I ran the same benchmarks.

3000 MHz
PCMark2002
- CPU: 7187
- MEM: 4831
- HDD: 883
PCMark2004
- System: 3094
Prime95
- 2048K FFT: 82.706 ms
- 4096K FFT: 247.198 ms

I was very impressed with these scores. I am running this system with a stock Intel HSF which came with a Celeron D CPU. It gives a good whine when the system is benchmarking or underload. It appears my CPU is getting to about 45°C during continuous Prime95 testing. That's a bit hot for my preferences, but not too bad.

The only problem with this is that the system is giving me a "Fatal Error: Rounding was 0.5, expected less than 0.4" during the blended Prime95 Torture Test. I went back and ran three passes of Memtest86 and my memory looks good (no errors). Any suggestions on a way I can make this system more stable? I'm noticing some good fluctuations on the Vcore during benchmarking. It appears to be dropping to 1.55v. Could it be the system is drawing more current than my PSU is ready to dish it? I have a PC Power and Cooling PSU in another system I can move to test. Do you think it is worth it? I also don't see any way to tweak the Vdimm to 2.6v using this motherboard. Not sure if that would matter as the memory appears stable and is running at stock. Thoughts? Thanks for your help. Even if I can't get this problem resolved, I've got to say these Mobile Celeron's sure do rock!

-MINOT-

 

[Karaktu]

ASUS boards tend to show that fluctuation...really doesn't matter what PSU you use. A 200-watt PSU is sufficient for your setup (depending on your video). CPU-Z will show you the actual voltage.

I'm of the opinion that if it's stable in everything but Prime95, I wouldn't worry about it a whole lot. But other than doing a wire mod for more voltage, you don't have many options with this motherboard. It's likely that another .25 - .50v will make that Prime95 failure go away.

45c is perfectly acceptable for these CPUs. There are much better heat sinks out there (and much quieter) that will keep them cooler, though.

Originally posted by: Minot
I got a steal of a deal for a SL6M5 Mobile Celeron 1.5 GHz C1 stepping on Ebay. With shipping it cost me $15.50! Anyway, I paired it with an Asus P4S800-MX Ver 2.0 and 2 x 256MB PC3200 Kingston Ram. The CPU defaulted to 1.65v for Vcore and booted right up at 15 x 100 MHz. I ran Memtest86 for one pass and all was well. I installed Windows 2000 Professional and installed some rudimentary benchmark programs. This is what I installed:

- PCMark2002
- PCMark2004
- Prime95

After I completed installing the drivers and configuring the OS, I went into my benchmarks at stock settings:

1500 MHz
PCMark2002
- CPU: 3676
- MEM: 3765
- HDD: 889
PCMark2004
- System: 1806
Prime95
- 2048K FFT: 127.696 ms
- 4096K FFT: 372.868 ms

I then bumped the system up to 15 x 200 MHz for 3.0 GHz. The system appeared stable and booted on the first attempt. I ran the same benchmarks.

3000 MHz
PCMark2002
- CPU: 7187
- MEM: 4831
- HDD: 883
PCMark2004
- System: 3094
Prime95
- 2048K FFT: 82.706 ms
- 4096K FFT: 247.198 ms

I was very impressed with these scores. I am running this system with a stock Intel HSF which came with a Celeron D CPU. It gives a good whine when the system is benchmarking or underload. It appears my CPU is getting to about 45°C during continuous Prime95 testing. That's a bit hot for my preferences, but not too bad.

The only problem with this is that the system is giving me a "Fatal Error: Rounding was 0.5, expected less than 0.4" during the blended Prime95 Torture Test. I went back and ran three passes of Memtest86 and my memory looks good (no errors). Any suggestions on a way I can make this system more stable? I'm noticing some good fluctuations on the Vcore during benchmarking. It appears to be dropping to 1.55v. Could it be the system is drawing more current than my PSU is ready to dish it? I have a PC Power and Cooling PSU in another system I can move to test. Do you think it is worth it? I also don't see any way to tweak the Vdimm to 2.6v using this motherboard. Not sure if that would matter as the memory appears stable and is running at stock. Thoughts? Thanks for your help. Even if I can't get this problem resolved, I've got to say these Mobile Celeron's sure do rock!

-MINOT-

 

[Minot]

Originally posted by: Karaktu
ASUS boards tend to show that fluctuation...really doesn't matter what PSU you use. A 200-watt PSU is sufficient for your setup (depending on your video). CPU-Z will show you the actual voltage.

I'm of the opinion that if it's stable in everything but Prime95, I wouldn't worry about it a whole lot. But other than doing a wire mod for more voltage, you don't have many options with this motherboard. It's likely that another .25 - .50v will make that Prime95 failure go away.

45c is perfectly acceptable for these CPUs. There are much better heat sinks out there (and much quieter) that will keep them cooler, though.

Thanks for the reply, Karaktu. I ran Prime95 Tortue Test again with CPU-Z in the foreground. The default Vcore appears to be 1.68v. During Prime95 it drops to 1.58v-1.61v. That strikes me rather peculiar. I can hear a difference in the whine of the CPU fan when the test starts making me wonder about my PSU. If I get a chance, I'll pull out the other PSU and give it a whirl.

Do you know if you can up the Vcore in the Bios of the Asus P4S800-MX? I didn't see an option for this.

I do agree with you on the system stability, however. So far this system appears to be rock solid. I can't think of a better value I've gotten out of a CPU. Even at 3.0 GHz the CPU idles around 28°C. Now that's awesome. I'm going to change the HSF to an all copper cooler when I have some time next week and see what happens.

Seperate question: Has anyone tried a Foxconn board based on a SiS or Intel chipset with these CPUs? Do they recognize them during the POST sequence? I'd like to see what performace boost I get out of a dual channel memory setup (SiS 655FX). The Foxconn 655A01-FX-6LRS has my interest perked.

-MINOT-

 

[ender11122]

Originally posted by: Minot

Originally posted by: Karaktu
ASUS boards tend to show that fluctuation...really doesn't matter what PSU you use. A 200-watt PSU is sufficient for your setup (depending on your video). CPU-Z will show you the actual voltage.

I'm of the opinion that if it's stable in everything but Prime95, I wouldn't worry about it a whole lot. But other than doing a wire mod for more voltage, you don't have many options with this motherboard. It's likely that another .25 - .50v will make that Prime95 failure go away.

45c is perfectly acceptable for these CPUs. There are much better heat sinks out there (and much quieter) that will keep them cooler, though.

Thanks for the reply, Karaktu. I ran Prime95 Tortue Test again with CPU-Z in the foreground. The default Vcore appears to be 1.68v. During Prime95 it drops to 1.58v-1.61v. That strikes me rather peculiar. I can hear a difference in the whine of the CPU fan when the test starts making me wonder about my PSU. If I get a chance, I'll pull out the other PSU and give it a whirl.

Do you know if you can up the Vcore in the Bios of the Asus P4S800-MX? I didn't see an option for this.

I do agree with you on the system stability, however. So far this system appears to be rock solid. I can't think of a better value I've gotten out of a CPU. Even at 3.0 GHz the CPU idles around 28°C. Now that's awesome. I'm going to change the HSF to an all copper cooler when I have some time next week and see what happens.

Seperate question: Has anyone tried a Foxconn board based on a SiS or Intel chipset with these CPUs? Do they recognize them during the POST sequence? I'd like to see what performace boost I get out of a dual channel memory setup (SiS 655FX). The Foxconn 655A01-FX-6LRS has my interest perked.

-MINOT-

My voltages dropped the same way. I think it might be normal with these setups.

 

[Zap]

Well, last week before I took off for Oregon I had a few minutes to fiddle with the Shuttle AB60R board. This board is on special at Mwave for $25 plus shipping. It does NOT support Prescott. It uses the 865PE chipset and has additional Silicon Image RAID and Firewire, making it an awesome value for cheap mobile Celeron/P4 setups. Haven't tried a mobile Celeron on it yet but it works fine with the mobile P4, detecting it as 1.5V. IIRC it will allow undervolting. Overvolting is available up to around 1.5875v thereabouts. Strange that it doesn't detect at 1.6V like most other boards that overvolt it on default. It couldn't reach as high a FSB as my Asus P4S800-MX board. The Asus board will POST at a higher terminal speed than the Shuttle board. Still, not a bad setup for around $70 out the door - motherboard capable of dual channel RAM, P4 running at around 2.6GHz (800+ FSB) and an Arctic Cooling 4 TC that works without a spacer.

 

[TStep]

greenwar: rechecked my spark7. You are in fact correct, it works perfectly. I had a heat sensor taped on it, and just that little bit of taped would not allow it to seat properly. So, as to not misinform, I stands corrected.

 

[greenwar]

Tstep, I bought the spark xaser used. Probably the guy before took it off. I didnt have it attached

By the way, little story on D1 stepping. Got a cheap D1 sl6vj 2.0ghz running at 3.5 ghz prime95 stable on my Abit AI7. Temps are idle 36C and full load on prime95 at 46C using a zalman copper 92mm HSF.

Runs pretty fast. I am extremely happy with it. May be can push is further, but the Abit guru software started showing red color on the vcore window so chickened out. btw running 1.6V on the core. with 170MHZ FSB using wintec el cheapo ram. No benchy yet, but the way doom3 and nfsu2 was running i am pretty sure 3dmark score increased. Had to do the usual pin mod to make the AI7 boot and recognize the cpu.

BTW, TSTEP, remember bidding on that A64 3000+ DTR chip on ebay where the guy had a wrong description? I saw your name there and was surprised to see a known nick on ebay. I got the proc for 67 buck. Heck of a deal I thought. Will be ocing it in near future.

 

[TStep]

Yeah, that was a good deal. You should have fun with that one.

 

[hamsterman]

Well thanks to all of you I decided to have a play with Mobile Celerons for my new HTPC.
I tracked down a 1.8GHz mobile cpu and combined it with a Asus P4S800-MX and 1Gb of Corsair value ram... and the little ah heck posted first go without any mods!

After tweaking the FSB I soon found the UN-stable values that would not post! but After gradual tweaking and stability testing with Prime95 I have crept up to 2974mHz which seems to be rock stable and only running at 42 Degrees (C) at full load with a Copper Zalman cooler!

I didn't think such a cheap CPU would overclock 65% OVER its rated speed and still run so cool! So a big thanks to all of you who contributed your findings for the rest of us to follow!!

 

[BATCH71]

I have to say that the ASUS VINTAGE is a fantastic little barebones overall and I personally had better results than others here. It does in fact have an AMI bios instead of the award but....... 1.6 mobicel running at 166fsb no problem on 3 outta 3 units. They do not like to run at anything above that except 200 and that does not have enough juice to stay stable.

Dave

 

[Karaktu]

I got a couple of those FIC/eCube machines that were on sale for $99 + ship at www.cheapguys.com...I'm pretty sure it's an EG61. Anyway, NICE machine. Based on the Intel i845 chipset, but sees mobile Celerons at 1.55v without a bent pin. Just break the 133FSB pin to get overclocking up to 160FSB. And if you DO bend the pin, it defaults to 1.15v! I have two of them with 1.8s running at 133FSB / 2.4GHz and 1.12 - 1.15v for practially silent operation. I'll post more info later, but if you can find these machines at $99 or better, they're highly recommended!

 

[delux1980]

I got SL75J (Cel 2.4) and I cannot get it running on ATI IGP 7000. The board just wouldn't boot. I also tried in asus board, chipset Sis 648 with no success board gives continous modulated sound just as if I wouldn't put cpu at all.

Tried with VID4 bent too - same thing.

Do you think this cpu is dead?

 

[Zap]

Potentially dead. The Asus board sounds like the P4S800, which though glitchy with mobiles should at least POST every now and then.

 

[delux1980]

I found the board is P4S8X-X. I dont use a POST card to get bios codes I just watch the screen to see if bios is up - is it sufficient?

 

[Zap]

Yeah, sufficient. By POST I mean Power On Self Test. I don't think anyone else has tried that motherboard with a mobile, so at this point we couldn't say if it were an incompatible board or the CPU. You'll need to find a board that is known to work and test the CPU in it. This means either buying it, or finding someone with one. If you live within driving distance of El Centro, CA, then I'd invite you over. After messing with mobiles for over a year I have a bunch of boards that will run them.

 

[aidanjm]

Is it a good idea to do the pin mod giving 1.25 volts, and then adjust voltage to suit using the bios settings? (i.e., on the mobos with a good range of voltage adjustments, is it possible to do the pin mod to set the default to 1.25, and then increase the voltage further, to suit, using the bios settings? Or are you locked at 1.25 volts?)

 

[Zap]

If a board has adjustable voltages in BIOS, you can change voltage regardless. The pin mod just changes the default detected voltage. If you can change voltages, then you may not have to pin mod. Some boards (such as my dying Albatron 845PEV Pro) allow you to lower voltage, not just raise it. My board detects at around 1.65v but I have it lowered to around 1.35v in BIOS.

 

[blckgrffn]

Let me know if the mobile celeron works in your shuttle, zap, my 1.6 wouldn't post in mine with the 133 FSB pin mod. Speaking of that, I still have that processor just sitting around somewhere...

 

[Zap]

I guess I can fiddle around with that, seeing as I have a mobile Celeron with the 133MHz mod and a few more unmodded. Might have time this weekend... the modded CPU is in the system I'm using right now and it being my "main" rig, don't want to go around pulling stuff out at random times.

 

[blckgrffn]

lol, come on! It's all in the interest of science or some such...

Nat