High-low-high-low on Award BIOS is a CPU error, but the manual also states:

− Speaker Alarm when detect "CPU FAN Failure" or
“CPU Overheat”.

That gives us two leads:
- maybe the CPU fan is dead, or you didn't hook it up to the CPU fan header
- maybe the CPU is being run at settings it can't run at.

Now, the fan is simplest to check.

Assuming the fan isn't the problem, let's go to the CPU. This board has dipswitches for FSB & multiplier but automatic/BIOS settings for voltage, which is IMHO the worst of both worlds. The manual says "Slot 1 supports Pentium II / III / Celeron processor running at 233-650MHz.". It makes no reference to Katmai vs Coppermine cores, which is suspicious. I see in the pics that your boards are rev 1.9, and I have found references to people running Coppermine CPUs on this revision, but to be safe, lets test with a 100MHz FSB capable Deschutes/Katmai (i.e. 1.8V) CPU. In your list that's the P3-450 and possibly the P3-500 (there's also a P3-500E with Coppermine core). So start testing with the P3-450.

Then the dipswitch settings. You want 100MHz FSB. According to silkscreen (and manual) that's switches 1/5, 1/6, 1/7 and 1/8 all set to the 'off' position. Looking at your pics they are already there. Multiplier settings (switches 1/1, 1/2, 1/3 and 1/4) don't matter as - unless you have an engineering sample - the CPU multiplier will be locked. 4.5x is off-on-off-on, so to be completely sure, set that. As there are no hard settings for voltage, all you can do to rule issues here out is to do a CMOS clear.

Apart from that there's a mysterious "JP11 : System Acceleration" jumper. This isn't explained anywhere in the manual, but I suspect it's the PCI divider, which in that case could be set to 1/3 or 1/4. The setting for "100MHz normal operation" is 2-3, and with 100MHz FSB that's what you want to have it at. It looks like it's there already on both boards in the pic.

Finally RAM - I don't think that's the issue here as you're getting a CPU warning, but to be sure play it safe. SDR-SDRAM has a number of pitfalls in terms of compatibility, in particular this board won't work with (very old) 2-way interleave DIMMs or with (non-JEDEC spec newer low-end) x4 chips, and as i440BX can support max 128Mb chips, 256MB DIMMs with 8 256Mb chips will only be detected at half capacity. There's a very low chance of the former with 64MB DIMMs and a higher chance of the latter two with 256MB DIMMs. Post the chip (not DIMM) brand & model numbers to be sure. If one of the 64MB DIMMs has 8 chips, that's almost certainly a safe bet. Use that for testing unless you get clear memory errors (different beep codes or POST codes).

Then the question: are those CPUs known good? If not, they could be the issue. I'd start with everything as above with the P3-450, then swap it out for P3-500. If that still fails in the same way, stick in the P2-233 and change the FSB switches (5,6,7,8) to on-off-off-on for 66MHz FSB. If all that fails, try the P3-650, but tbh at that point you almost certainly have two or more failures need another known-good slot 1 system to exchange parts with to know for sure.

Thanks for the fast answer, dionb! I will try out all the different CPUs again and answer back 😀

I know that the Pentium III 500 (500/512/100/2.0V SL35E) , the Pentium III 700 (700/256/100/1,65V SL359) and the Pentium II 233 (SL2QA) were most likely tested, as they came from two different commercial sellers (with the German laws, a commercial seller would not write a legal binding "fully tested and functioning", I think); the Pentium III 450 (450/512/100/2.0V SL35D) could be faulty, though (I didn't test that one first, so I'm sure I didn't fry both mainboards)

I speculated that the PSU could also be at fault (maybe not giving enough current on the +5V line / undervoltage on the +5V line for the CPU) - would that also cause the high-low-high-low error?

nicolasm wrote on 2025-05-23, 12:56:

Thanks for the fast answer, dionb! I will try out all the different CPUs again and answer back 😀

I know that the Pentium III 500 (500/512/100/2.0V SL35E) , the Pentium III 700 (700/256/100/1,65V SL359) and the Pentium II 233 (SL2QA) were most likely tested, as they came from two different commercial sellers (with the German laws, a commercial seller would not write a legal binding "fully tested and functioning", I think);

German law offers good protection in case of faulty goods, but doesn't protect the goods themselves. I'd only trust them fully when you've seen them working yourself. The P3-500 is a Katmai, so perfectly good as a generic test CPU that definitely should work with these boards.

the Pentium III 450 (450/512/100/2.0V SL35D) could be faulty, though (I didn't test that one first, so I'm sure I didn't fry both mainboards)

If suspect, check first with the P3-500. Unless there's clear physical damage or burn marks on the P3-450, I'd not be too concerned about it frying motherboards.

I speculated that the PSU could also be at fault (maybe not giving enough current on the +5V line / undervoltage on the +5V line for the CPU) - would that also cause the high-low-high-low error?

Yes, the CPU is probably the first part to fail when power delivery is insufficient. Why are you speculating? Have you measured the voltage on the board? Find any pin that should be showing +5V and compare that (with a multimeter) to GND. Or connect the PSU to a known-good system and see what its BIOS can tell you about voltages.

nicolasm wrote on 2025-05-23, 12:56:

Thanks for the fast answer, dionb! I will try out all the different CPUs again and answer back 😀

I know that the Pentium III 500 (500/512/100/2.0V SL35E) , the Pentium III 700 (i accidentally wrote 650 in the first post] (700/256/100/1,65V SL359) and the Pentium II 233 (SL2QA) were most likely tested, as they came from two different commercial sellers (with the German laws, a commercial seller would not write a legal binding "fully tested and functioning", I think);

German law offers good protection in case of faulty goods, but doesn't protect the goods themselves - the fact you can return stuff doesn't mean it can't be broken. I'd only trust them fully when you've seen them working yourself. The P3-500 is a Katmai, so perfectly good as a generic test CPU that definitely should work with these boards.

the Pentium III 450 (450/512/100/2.0V SL35D) could be faulty, though (I didn't test that one first, so I'm sure I didn't fry both mainboards)

If suspect, check first with the P3-500. Unless there's clear physical damage or burn marks on the P3-450, I'd not be too concerned about it frying motherboards.

I speculated that the PSU could also be at fault (maybe not giving enough current on the +5V line / undervoltage on the +5V line for the CPU) - would that also cause the high-low-high-low error?

Yes, the CPU is probably the first part to fail when power delivery is insufficient. Why are you speculating? Have you measured the voltage on the board? Find any pin that should be showing +5V and compare that (with a multimeter) to GND. Or connect the PSU to a known-good system and see what its BIOS can tell you about voltages.

Did either of the board listings indicate which BIOS version they were running...Gigabyte recommend (even for the P3-500 Katmai) that 1.x revision boards should be updated to at least BIOS version F1 or later (see https://theretroweb.com/motherboards/s/gigabyte-ga-6bxe#bios ) and they don't list slot 1 coppermine cpu support until 2.0 revision boards.

I tested all the CPUs again (starting with the Katmai Pentium III 500 and the Pentium II 233, setting the correct FSB (100 for the p3y 66 for the p2) and multiplier, Jumper 11 is on "normal mode" (fast mode is 133 MHz for the PCI Bus)

Unfortunately, no change from what happened before - I still get the same hugh-low-high-low tone

Funnily enough, you may be onto something with RAM: When I try starting with any of the 64 MB Sticks alone in one Slot, i get a repeating long beep instead (memory error), but with a 256 MB stick alone or more than one of the 64 MB sticks, I get the high-low-high-low siren tone (which honestly makes no sense)

The second motherboard came packaged with the 64MB sticks (x4) and the Pentium III 500 Mhz and was tested (though I don't know the specific bios version)

I attached photos of the RAM sticks - which still would be weird as both motherboards have the same behavior with any RAM 😒

I also ruled out the PSU, as I tested it with a Corsair HX750i (25A on the +5V line with a -5V ATX adapter)

Looks like either all CPUs or all mainboards are faulty (or I'm still doing something wrong)

Ugh, pictures...

Let's see:

1st DIMM (Mosel Vitelic):
V54C3128804VAT7 => 16Mx8 128Mb chip => perfectly compatible with i440BX, chips are 7ns, not 7.5, so actually better than sticker advertises.

2nd DIMM ("PC-133 MQ")
J39S16X8128DT75 => relabeled fantasy-code, but implies 16Mx8 as well. If there are 16 chips it should be compatible assuming it works and the code isn't complete nonsense.

3rd DIMM ("256M-133 PRO")
GSSD1608AT-75 => Procom relabeled fantasy-code, but implies 16Mx8 as well. If there are 16 chips it should be compatible assuming it works and the code isn't complete nonsense.

4th DIMM ("made in Taiwan")
MK5264805T-8 => relabeled fantasy-code, but implies 8Mx8. If there are 8 chips it should be compatible assuming the code isn't complete nonsense.

5th DIMM ("DM901.25")
MK5264805T-8 => relabeled fantasy-code, but implies 8Mx8. If there are 8 chips it should be compatible assuming the code isn't complete nonsense.

6th DIMM ("SpekTek")
524LKT-8A => relabeled fantasy-code => relabeled fantasy-code. Googling shows Polish links (local brand?) indicating 64MB or 128MB, so looks like 8Mx8 chips too, with DIMM capacity depending on number of those chips. If 8: 64MB.

5th DIMM ("DM901.25")
MK5264805T-8 => relabeled fantasy-code, but implies 8Mx8. If there are 8 chips it should be compatible assuming the code isn't complete nonsense.

Tbh this is a pile of very, very fishy memory. Only the first DIMM has actual original markings on the chips. The rest have been re-marked which is usually a sign of second rate DIMM manufacturers buying up batches of rejected chips and trying to run them at lower spec... I would focus testing on the Mosel Vitelic DIMM until such time as you've tested the rest in a known-good system.