Slot 1 upgrade problems

[Sooty]

Replacing Celeron 333mhz (SL2WN) with a Pentium lll 600mhz (SL3JM) –purchased used, off eBay.

Board is an ASUS P2B Rev 1.10, with BIOS 1008.

SL3JM is officially supported on this board, with no BIOS upgrade required above 1008, according to ASUS.

The SL3JM came with a large Compaq Proliant heatsink:



The first problem came when swapping the CPU’s. It doesn’t seem compatible with the existing slot retention clips. The CPU goes into the slot okay (as far as I can tell), and the retention clips hold it in place to some degree, but not as intended, and not 100% securely.

Looking on Wiki I found this:
>>> “Some Pentium-II 350MHz and 450MHz processors, and all slot-1 Pentium-III's, came in the improved SECC2 variety. The SECC2 package varied from the original SECC package by removal of the heat-plate connected to the CPU via a thermal interface, used less raw material, was lighter, and could be produced at lower cost.

It should be noted that while SECC2 retention clips will hold a SECC package, SECC clips will not hold an SECC2 package.â€

[Vicotnik]

It's not difficult to replace the retention clips on a slot1 board. So if you could find a SECC2 retention clip you would be fine. I had lots laying around a year or so ago, but I moved and threw it all away since slot1 isn't so hot today. Otherwise I could have mailed you a few.

Since the system is so old and worth next to nothing, perhaps you could use the old SECC clip and if needed secure the CPU further with hot glue.

edit: The CPU is reported as 400MHz (6x66MHz and not 6x100MHz) since you run it at 66MHz FSB, like the Celeron. You need to change the FSB to 100MHz. This is controlled by a jumper block on the board, see the manual for details.
I have actually built so many P2B systems I can still remember the jumper settings for the FSB as well as the locations for the front panel switches and LEDs. I had a few rev 1.10, one 1.12 and a ton of 1.02. Great boards.

[bonestonne]

have you looked at the CPUs next to each other to try and find a difference in their design?

if the existing retention method doesn't hold the new CPU then there's got to be an obvious difference in their design. i'd suggest maybe checking the width of the actual cartridge design first...it could be that simple.

i haven't really messed with Slot 1 in a long time, and i'm more familiar with Slot 2. but i think that its just a size difference.

[IsaacKuo]

On my slot-1 systems, I simply remove the retention clip hardware altogether. The only thing holding my processors in place is friction.

I have issues with 66mhz FSB on one of my motherboards. I have no manuals for any of my motherboards; they are all pieces of "junk" I've acquired from random places.

[Sooty]

have you looked at the CPUs next to each other to try and find a difference in their design?

if the existing retention method doesn't hold the new CPU then there's got to be an obvious difference in their design. i'd suggest maybe checking the width of the actual cartridge design first...it could be that simple.

i haven't really messed with Slot 1 in a long time, and i'm more familiar with Slot 2. but i think that its just a size difference.

Yes, ideally I should of compared physical dimensions, but it’s a working PC so I didn’t have much time to make the swap. However it’s now clear there are at least 3 different sized packages: SEPP (my Celeron), SECC and SECC2. Why that's not mentioned in this Ultimate P2B Upgrade Guide, that I was using as guidance, I don’t know.

It's not difficult to replace the retention clips on a slot1 board. So if you could find a SECC2 retention clip you would be fine. I had lots laying around a year or so ago, but I moved and threw it all away since slot1 isn't so hot today. Otherwise I could have mailed you a few.

Since the system is so old and worth next to nothing, perhaps you could use the old SECC clip and if needed secure the CPU further with hot glue.

If you had managed to mail me some of those retention clips it would of been appreciated, but thanks for the heads up. If it’s an easy mod then hopefully I can find some clips to do that. Failing that: hot glue is one option, super sticky tape maybe another, but I did find this method, using plastic cable ties:

edit: The CPU is reported as 400MHz (6x66MHz and not 6x100MHz) since you run it at 66MHz FSB, like the Celeron. You need to change the FSB to 100MHz. This is controlled by a jumper block on the board, see the manual for details.
I have actually built so many P2B systems I can still remember the jumper settings for the FSB as well as the locations for the front panel switches and LEDs. I had a few rev 1.10, one 1.12 and a ton of 1.02. Great boards.

Well this P2B has seen me through good and bad times, and never let me down, so that’s why I’m treating it to some TLC in it’s old age.

The manual lists many jumper settings for many CPU’s, but the SL3JM isn't listed. Any idea which settings I should use? There appear to be two sets required: ‘CPU Bus Frequency’ (FS0, FS1, FS2, FS3) and ‘CPU Core: BUS Frequency Multiple’ (BF0, BF1, BF2, BF3).

[IsaacKuo]

Instead of using hot glue, you could simply remove the old retention hardware and rely on friction alone. I've used friction alone to support a heavy P3 with a huge heatpipe based VGA cooler (this was before I discovered that such a large heat sink was not necessary).

That computer has gone on an interstate road trip in the trunk of a car, and the CPU card was still firmly seated within the slot. It didn't budge at all, despite the bulk and mass of its cooler.

[Sooty]

Well that’s an option worth considering then, Isaac, and I don't plan any road trips for it either

The way the clips are pressing up against it, they could well be exerting a twisting force, resulting in the CPU not seating properly. The cable-tie method (pics above), may or may not help in that regard.

[Sooty]

The manual lists many jumper settings for many CPU’s, but the SL3JM isn't listed. Any idea which settings I should use? There appear to be two sets required: ‘CPU Bus Frequency’ (FS0, FS1, FS2, FS3) and ‘CPU Core: BUS Frequency Multiple’ (BF0, BF1, BF2, BF3).

I've had a look at the manual again, and it looks pretty simple really. I have to set the CPU Core:BUS Frequency Multiple to 6.0x and the CPU External Clock (BUS) Frequency Selection to 100.3mhz? Can someone confirm that's correct? Upgrade CPU is a SL3JM.

[Sooty]

I've had a look at the manual again, and it looks pretty simple really. I have to set the CPU Core:BUS Frequency Multiple to 6.0x and the CPU External Clock (BUS) Frequency Selection to 100.3mhz? Can someone confirm that's correct? Upgrade CPU is a SL3JM.

Can anyone confirm?

[Vicotnik]

That's correct. You could set the multiplier to anything though since the CPU is multiplier locked.

[Sooty]

Thanks Vicotnik!

[Sooty]

I put the SL3JM back in. I’m 99% sure it’s seating properly in it’s slot. Set the mobo jumpers to 6x and 100mhz. The post info confirms it’s running at 598mhz, and it feels much quicker too. But, the Dr. Watson problem persists. Appears only to happen when on the web. Happens on about 1 in 3 websites, including on SPCR. This is the message I get from Dr. Watson:

An application error has occurred and an application error log is being generated.

Firefox.exe

Exception: illegal instruction (0xc000001d), Address: 0x301106cc


The illegal instruction codes seem to be consistent (as above). I’ve Googled those codes and can find nothing.

What do you reckon: duff cpu or not seating properly...or what?

[Gojira-X]

try looking here for how to get hold of the error logs that dr watson produces.

looks like the problem is a firefox one - if you can get firefox to start look up info about running Mozilla Talkback - you might then find out what the problem is and be able to report the possible bug to Mozilla.

[Vicotnik]

What do you reckon: duff cpu or not seating properly...or what?

I would suspect memory problems. Does the system pass a few hours of Memtest86? Memtest86 is included on the Ultimate Boot CD btw.

If Memtest86 reports errors after a while, you could try tweaking the timings for the memory. This is controlled by BIOS settings (under Chipset Features Setup). SDRAM Configuration should normally be set to 'By SPD' but if the memory has problems it can help stability to manually loosen the timings a bit. It costs you a little bit of performance, but I do't think that will matter much in your case.