overclocking FSB overclocks everything on it the only way around that is if there are dividers in the bios or by jumpers here is the manual for that board http://oldschooldaw.com/asus/1997/p2l97s-105_1997.pdf

just fyi most pci cards can handle 66mhz but i would say your biggest issue in ocing will be ram and hdd corruption if u go to high as the ide will be overclocked also but i wouldnt be scared to try 75 or 83mhz and run a bench test

also most agp cards can handle 83mhz

with ram depending on what u are using edo or sdram u get higher clocks from more loose timings
been a while but isa from what i remember can overclock decent last isa i oc's was on a amiga 2000 and that was 8bit converted to 16bit but the real issue for u with that card is can it be overclocked hard for me to give advice as u didnt state what kind of isa card it was it does make a difference knowing

If you need speed on an LX motherboard, you could just buy a fast celeron mendocino, put it on a slotcket and use it in your mobo.
You may even be able to run coppermine celerons if your motherboard accepts them

I have a P2L97 (non SCSI) buried away somewhere. According to some old notes it is a rev2.04, and has an ICS9148BF-04 clock generator IC on it. The following may not necessarily be true for your board if it's different than mine.

From looking at the manual, and comparing it with the clock generator datasheet, it's apparent how the jumpers match up with the signals that control the clock generator. It seems that the clockgen gets it's configuration from values stored at some memory address, but apparently the critical 3 bits involved here are defined with the 3 FSB jumpers on the board.
Based on the documentation, it appears that the clock generator can drive a 75MHz FSB without overclocking PCI - but *not* with the jumpers set the way Asus has documented in the version of the manual that I have.

This is Asus' jumpering

and this is from the datasheet for the clock generator

Matching these up:

1Asus             ICS9148B-04
2FS0              bit 4
3FS1              bit 5
4FS2              bit 6

setting a jumper at 1-2 is apparently a logical "1", and setting the jumper at 2-3 is apparently a logical "0".
Unless I've overlooked something, the above is the only way to interpret these jumpers that will always match between both the Asus manual and the clockgen datasheet.

Note that the Asus manual defines 75MHz and 83MHz with exact opposite jumpering. This is one of the clues of how the jumpers map out, and shows which settings Asus told people to use.
The 75MHz setting in the Asus manual matches up with the clockgen setting "101", which runs PCI at a 1/2 ratio (overclocked at 37.5MHz).
For 83MHz they are apparently using the "010" setting, which is still a 1/2 ratio (overclocks PCI to 41.6MHz).

If you instead set the jumpers to the "001" setting then FSB should be 75MHz but with PCI at only 32MHz:
FS0 = 1-2
FS1 = 2-3
FS2 = 2-3

I don't know where the 32MHz is coming from, it doesn't seem like any obvious ratio of 75MHz. Maybe Asus had a good reason for not using this setting.
This doesn't say anything about the ISA clock. It seems sensible that the ISA clock might be generated by the southbridge at 1/4 of the PCI clock, but I don't know that.

This is all just theory that appears to fit the available documentation, and as mentioned, it's possible that your board revision is different.

WOW, thanks for a ton of useful info! Mine is the Rev.1.05 board and does appear to overclock the PCI in standard configuration - just judging by "finger temperature" of the RAM chips on my poor Voodoo Banshee. They are definitely hotter at 75MHz (although the card still seems stable.)

I have exactly the same clock generator chip, so I'm going to assume it works the same way. I'll try out your revised jumper config a little later today.

EDIT: hmm, nope. Just tried it now and it didn't seem to want to POST in that config (1-2, 2-3, 2-3.) Maybe I'll just invest in some little heat sinks for the RAM on the Banshee and call it a day.

I just tried ALL the settings and the only ones that don't work are "100" (83/33) and "001" (75/32). Both refuse to POST. It looks like the non 1/2-divider modes are not implemented on this board.

OTOH this does give me access to the undocumented 68.5/34.25 mode as a compromise between 66 & 75MHz. I actually haven't had any stability problems at 75/37.5 yet, although I haven't really stress-tested the system.

I suspect the Banshee is actually fine at 37.5MHz even though its RAM feels hot to the touch. It's the VIP card for the VFX-1 in the ISA slot I'm worried about, but I haven't put it in yet.

dont forget to give the ram slower timing on those 2 are what ever is safe for whatever type you are using

^^ I'm using PC133, I imagine it would be fast enough to handle anything this board can do at standard timings? It's just set to auto in the bios.

Just in case you aren't already aware,
On the plain P2L97, rev1.05 was one of the boards that had an onboard 3.3v regulator which couldn't provide enough current and might get damaged by high powered AGP cards. Asus published a rework procedure for it, at the time the card they said was prompting issues was the Riva TNT.
https://web.archive.org/web/20010311151314/ht … tnt-rework.html

That page didn't mention if it affects the P2L97-S version, but this 3rd party page says it does:
http://plasma-online.de/english/upgrade/tweak … x_asus_tnt.html

Sounds like this isn't an issue with the current setup, but just something to keep in mind if you didn't already know that limitation.

Great info. I have one of these boards on the way. Not sure what version it is. Basically just wanted to mark this thread so I can come back and look more easily.

shamino wrote:

Just in case you aren't already aware, On the plain P2L97, rev1.05 was one of the boards that had an onboard 3.3v regulator whic […]
Show full quote

Just in case you aren't already aware,
On the plain P2L97, rev1.05 was one of the boards that had an onboard 3.3v regulator which couldn't provide enough current and might get damaged by high powered AGP cards. Asus published a rework procedure for it, at the time the card they said was prompting issues was the Riva TNT.
https://web.archive.org/web/20010311151314/ht … tnt-rework.html

That page didn't mention if it affects the P2L97-S version, but this 3rd party page says it does:
http://plasma-online.de/english/upgrade/tweak … x_asus_tnt.html

Sounds like this isn't an issue with the current setup, but just something to keep in mind if you didn't already know that limitation.

Yeah I've run into that already trying to run a Radeon 9250 256MB. I need the PCI Banshee right now because it's about the best thing I have with a feature connector but I'll keep the mod in mind if I go back to an AGP card for some reason. Fortunately it's not that hard to do.

The ISA bus has separate clock generator, overclocking FSB won't affect it.

The Serpent Rider wrote:

The ISA bus has separate clock generator, overclocking FSB won't affect it.

^^ ...aaaaand that's what I was looking to hear. 75MHz it is. 😜