I know P3's aren't really taxing on the boards, but if you're going to try to overclock them so far then why not recap the motherboards with lower-impedance capacitors, perhaps even polymers, even go as far as to refresh the power-supply with higher quality caps so as to increase the stabilty of the boards? Even recapping some video cards will improve their stability, especially while clocking.

Usually I underclock instead of overclock CPUs. But with these oldest of Pentium II's, Klamath core, it is easy to forget that speeds like 350MHz are already too much to ask. 😉
Would like to know the speed rating of the L2 cache chips, but I don't want to break another plastic CPU package.
(Correction on my previous post: SL2HD is a 233MHz part)

133MHz FSB is overclocking the mainboard. Official specification is either 66 or 100MHz. Although it is generally found that the i440BX chipset itself is fine with a 133MHz FSB, other mainboard components and expansion cards may not be, especially AGP video cards at 2/3 divider.

Recapping is certainly a thing to consider, if only because these mainboards and PSU's are from 1998-2000. But lets hope they will hold out fine for a few years more.

I am thinking: In my last test with a Klamath core at 2.0x multiplier, the L2 cache throughput is near that of the SDRAM. So at that multiplier the later Deschutes core Pentium II's are practically performing the same, despite their non-functional L2 at 2.0x: More reason for using the cooler and faster Deschutes core over the Klamath.

As to the multiplier jumper settings on all tested unlocked slot-1 and socket 370 intel CPUs: XXXX and OOOO always gives a rather odd result. As if these two settings were somehow not intended for normal use, some kind of 'test' mode.

Running a bx board at 133 mhz can couse dmg for the mainboard?since bx doesn't officialy support 133.

Worst case is that 133MHz can cause something to get out of sync. But while temperatures and currents remain within spec I am not worried about any damage.

Also I only switch to 133MHz for SVGA games, and usually switch gradually: Boot at 50 MHz, then to 100, next to 133MHz. Many enthousiasts have ran a BX at 133MHz+ FSB for years, combined with much higher CPU speeds.

Your Jetway mainboard also does not support K6-2+ 'officially' right.

Asus P2B rev 1.04 has a small issue with 50 MHz FSB: it can boot up with 50, or switch to 50 once, but when you change FSB after that it usually hangs. I cannot explain that. Gigabyte GA-6BXC and Shuttle HOT-661 rev.1 boards work+switch fine at 50MHz FSB.

Actually it officialy supports the k6 3,and the k6 2+ beeing a k6 3 with 128mb cache,i guess it supports that 1 too 😜
The reason why i was asking is becouse i got a Gigabyte BX 2000 mainboard and altough the mb has the option to set 133mhz fsb,manual says it's not advisable,and i got a slot 1 pentium 3 at 800mhz and was wondering if it's safe to run it,without breaking the mainboard.

Worst case is more like BIOS fuckup and HDD (or in my case, CF) corruption. Been there, done that, BX-133 is nothing I'm going to attempt again 😀

To be fair, the AGP video card plays a big role in it. But the ones that had the most value for my build didn't like it. It's not like any 2000 game would choke on a 100 MHz FSB P3 anyway.

Nahkri wrote:

altough the mb has the option to set 133mhz fsb,manual says it's not advisable,

The manufacturer has to put that in the manual so if you break something while overclocking they won't have to honor the warranty.

Nahkri wrote:

The reason why i was asking is becouse i got a Gigabyte BX 2000 mainboard and altough the mb has the option to set 133mhz fsb,manual says it's not advisable,and i got a slot 1 pentium 3 at 800mhz and was wondering if it's safe to run it,without breaking the mainboard.

Because that pentium 3 will remain at the locked 8.0x multiplier, it would run at 1066MHz (=8.0x133). This is too much of an overclock IMHO. There are pentium 3 CPU's with a much lower multiplier, usually the socket 370 ones aimed for 133MHz FSB use. The most extreme example is the pentium III 533EB, which was sold to be run at 4.0x133MHz, there is both a socket 370 and slot-1 version of that one.

d1stortion wrote:

Worst case is more like BIOS fuckup and HDD (or in my case, CF) corruption. Been there, done that, BX-133 is nothing I'm going to attempt again 😀
To be fair, the AGP video card plays a big role in it. But the ones that had the most value for my build didn't like it.

My Geforce MX440 does not like it either. Regarding the other problems, was the PCI bus at 33MHz?

For 133MHz FSB The board should have a good PLL and solid VRM set. Then I use one stick of PC-133 RAM CL2 at the most relaxed latencies. The North bridge AGP divider has to be set 2/3, Which is what intel calls '100MHz operation' in their datasheet, contrary to '66MHz operation' . The North bridge should have a small heatsink.
The early ICS PLLs I know automatically switch the PCI bus to the proper 33MHz. The ISA bus is then also at the proper speed. The IDE controller is on the PCI bus.
In my view only the AGP bus remains as a cause of trouble. Its at 89 instead of 66MHz, and this cannot be fixed. I found the Voodoo3 AGP and some Geforce-2 MX AGPs won't complain. For troubleshooting I keep a Voodoo3 PCI in reach.

The GA-6BXC motherboard with Voodoo3 AGP + Tualatin ES CPU has been totally reliable so far. Asus P2B 1.04 and Shuttle HOT-661 rev.1 lacked options to begin with, and I am still 'having fun' with these.

d1stortion wrote:

It's not like any 2000 game would choke on a 100 MHz FSB P3 anyway.

True, yet SVGA mode DOS games benefit from a Fast FSB + Write Combining more then anything: Double the FSB gives about double the FPS.

Mine is a pentium 3 800EB,it needs 133 mhz fsb to run at full speed,that's why i was wondering if running the mb at 133 all the time will couse any problems.
I noticed on the the mb 2 modes for setting 133 mhz fsb,first 1 says pci will run at 44 mhz,second says pci will run at 33 mhz,so i prolly use second 1 in order to avoid problems with ide controler.
Too bad there isn't any settings to keep the agp at normal speed.

Ok. So you are at 600MHz currently. If you are not interested in the speed switching then the benefit is less. Depends on what you want to achieve. Getting a plain 8.0x100MHz CPU may be a lot easier, Whilst getting a tualatin+adapter gives even more speed.

I made a few benchmarks with Quake v1.08, it shows the average FPS.
From pure DOS: Video mode 640x480, timedemo 2, nosound, nojoy.
System: GA-6BXC with Tualeron QID2QS CPU. 128MB RAM. Voodoo3 video, either AGP or PCI variant. AGP divider=2/3.
With and without 'MTRRLFBE LFB WC' (write combining).

1Speed MHz   PCI+WC  AGP+WC  PCI    AGP
24x_50=200   25,9    26,0    13,7   13,7
34x_66=266   34,3    34,4    18,2   18,2
44x100=400   50,5    50,9    25,0   27,0
54x133=533   66,4    67,5    27,0   35,8
68x_66=533   55,7    56,3    21,0   21,1

Edit: added missing benchmark entries

gerwin wrote:

My Geforce MX440 does not like it either. Regarding the other problems, was the PCI bus at 33MHz?

Yes it was. The mobo in question is a CUBX-E, so certainly not some cheapskate board. When I bought it it had an 800EB in it, so I assume the previous owner ran it like that without problems. Or he just settled for 600 MHz, who knows? 🤣

Nahkri wrote:

Actually it officialy supports the k6 3,and the k6 2+ beeing a k6 3 with 128mb cache,i guess it supports that 1 too 😜

The K6-2+ is a K6-3+ with 128kB cache, it needs a lower voltage than a K6-3 and BIOS support to enable all features.

Yesterday I wanted to see if I can undervolt the Pentium II and III processors. At least for the speeds up to 533 MHz.

The Pentium III Tualatin core (default 1,5V) was on 1,3V already and I have currently no means to lower it further. Works fine even up to 1000MHz. I remember Swaaye wrote about his Tualatin running fine with just 1,1V.

The Pentium III Coppermine core impressed me: SL3VA (default 1,65V) accepted all voltages, it ended up at the lowest setting: 1,3V (measured 1,325V). Let it run Quake for an hour without any problems. I tape-modded the other two coppermine core CPUs (in the picture on the right), as native slot-1 leaves me no jumpers. They behave similar, but need more testing.
The benefit is that they don't really need a fan at 1,3V at 533MHz or less. So up to 533 MHz the Coppermine core is practically the same as a Tualatin core. Making SL3VA, SL3XG and especially the ES more interesting for my use.

Next in line was a Pentium II Deschutes core (default 2,0V). 1,8V with 400MHz was not a reliable combination for the SL2QF. Raised it back to 1,9V which did not cause any problems yet. This processor was originally in the package on the left of it. I removed the package to read the L2 cache specs, and add L2 cache cooling and a new fan.

Also shown is the Pentium II Klamath core SL2HD with its heavy heatsink. Defaults to 2,8V. Maybe I will mess with it later.

Good day
Colleagues
It's an old article, but I would like to contribute my observations and ask if one or more people can check it and, if necessary, confirm or refute it.
Gigabyte ga-6bxc ver.1.7 450mhz cpu 256mb ram,
Geforce 6200 524MB ram
win98se and ME
only possible with 16 colors and a resolution of 640x480.
Error message:
Graphics card and motherboard use the same system resources.

My question is, are there any users who successfully use ATI cards on the system?
Because I only own Geforce cards and don't want to write off a working board.

Have a nice day.

Gustl wrote on 2023-10-05, 08:25:

Geforce 6200 524MB ram

It would have been better to start a new topic. The only resemblance to this topic is the motherboard series, but nothing else.
That graphics RAM size is oddly high. Look here instead:
512MB Geforce 6200 AGP in Windows 98SE?
As an addendum to that: Most BIOS'es for GA-6BXC cannot disable ACPI, there was a modded BIOS posted on Vogons, adjusted to allow for disabling ACPI. I have not tried it myself yet.
Link: Re: Powerleap PL-IP3/T and GA-6BXC MB
More Here: http://www.vogonsdrivers.com/getfile.php?fileid=727

Thanks