Can't set the SL2Z4's multiplier to 4x on a GA-5AX

[CHips]

I just got a cool Pentium MMX 267 MHz 1.9 V SL2Z4 and would like to run it at 4 x 100 MHz, or maybe even more. I have a Gigabyte GA-5AX (Rev 5.2, F4 Bios) for this.

So far I don't have any trouble to make it work at 3.5 x 100 MHz @1.7 V, but when I set the jumpers (or actually, the Dip Switches) in order to get 4x, it boots as 2x...

SuperPi is indeed slower (so this is not just a wrong speed recognition).

However, with a K6-2+ instead, the multiplier is correct.

I tried all the 8 combinations but can only set to 2x, 2.5x, 3x, or 3.5x. The CPU doesn't work at all in my Asus P5A (1.06).

Any idea why I can't set to 4x? This multiplier should work as the CPU is intended to work at 4 x 67 MHz...

Thanks!

[frag_]

If you set pins BF2,BF1,BF0 as 1,0,0 (for the K6-2 it's x2.5) and resulting multiplier is not x4.0, then it's faulty or fake.

[CPUShack]

They seem to run ok at 4.0x100 on a Tekram P5MVP-A4 (but at tooo high a voltage)


Be strange to be a fake (what would it be faked from that was happy at 350MHz?

[debs3759]

Have you tried running it at 4 x 67 MHz to see if it can run at 4 x?

[CHips]

Yeah I really don't think that it is a fake... Cpu-Z reports the right MyA0 Stepping.

Yes, doing "4 x 67 MHz" also gives the wrong 2x multiplier (the CPU works at 133 MHz in this case).

I guess that I will stick to 3.5x and try to increase bus over 100 MHz, though it seems to be really hard with 768 MiB of RAM. I should probably either reduce or get PC133 CL2 sticks...

Still a neat CPU and 350 MHz @1.7V is already nice overclock. Intel, AMD or any other CPU company should have made 180-130 nm 1 GHz Super Socket 7 CPUs with locked 10x multiplier :p ...
Pentium III OverDrive or so
Or who knows if someone will be able to mod Socket 370 CPUs to make them work in Socket 7 !

[H3nrik V!]

[CPUShack]

[CPUShack]

I wonder if an older BIOS may work better?

[CHips]

3.5 x 115 MHz = 402.5 MHz @2.2 V seems to work well too with 512 MiB PC100 CL2. Both settings work in Fanless, though @1.7 V there is much less heat. 120 MHz don't seem to work well (I set the timings to the worst possible) ...

I might get more than 115 MHz but I don't want to have less RAM, using 133 CL3 doesn't really seem better and I don't have any 133 CL2 nor 150/167 CL3. Or maybe it is just the motherboard that can't handle more than 115 MHz.

At least at 350-400 MHz Windows 7 is already almost usable with a Pentium MMX ! Sadly Aero doesn't seem to work without some K6 or Pentium 2 instructions, even with a Wddm GPU (there is a Blue Screen that is not present if I use a K6-2+). It is unfortunate as one of my (crazy) projects was to trigger the Fdiv Bug while using Aero, with a Sx960. I can't run the WinSat as well as it is crashing (I think that I got 0.9 for the CPU in a 1.0-7.9 scale with a Pentium III 450 MHz Lol).

I think that I will stick to either 3.5 x 115 with fan or 3.5 x 100 without and then play with Debian 8.
4x multiplier might not even be that useful as I would likely need dangerous voltages to do 4 x 115 MHz = 460 MHz or more.

[cuttingedgecs]

From memory, the GA-5AX, like many boards of the time used the same jumper settings for different multipliers on different CPUs. For example, when a P55C in inserted, a set jumper combination will give a different multiplier than that same combination when a K6-2+ is inserted. This was done as I understand to allow the greatest number of supported chips/multipliers for the smallest amount of jumpers. One of the common settings Asus had at the time would see a K6-2 400 run at 6x66MHz, where the same settings would run a P54C 133 at 2x66MHz. If this board was designed with the expectation of a P55C CPU only reaching 233MHz (and I suspect it was), it would make sense that a jumper setting that would yield a 4x multiplier for a K6-2+ CPU, would be defined as something else for your chip.

[cuttingedgecs]

Just taking a quick look at the manual, and you can see there are 10 available multipliers, and only 8 possible switch positions to select them. There is definitely some CPU detection going on there in order to select the correct multiplier. You will see that the position for 1.5x and 3.5x are the same, and the position for 2x and 6x are the same. It's possibly sloppy coding, but the board isn't expecting a Pentium at 266MHz, so it hasn't been coded to deliver a 4x multiplier for a Pentium, even if that's what you ask.

[CHips]

Makes sense. Maybe the Bios could be modded.

[H3nrik V!]

That's not quite how it works. The motherboard doesn't "send" a multiplier to the cpu - at least not in those generations .. It just sends the jumper setting to the cpu, which then interprets the selected multiplier. An example: A p55c interprets the 1.5 multiplier as 3.5 ..

[max1024]

I had run SL2Z4 on TMC MI5VP4 (VIA MVP4) with x4, detailed there.

[CHips]

Very cool page (I used Google Translate to read it)!

So this is not a Bios problem but something at the hardware level ? But then I could get x4 if I bind the "W33" and "W35" pins?

And Wow, 124 x 4 @2.6 V? I don't think that I could reach this with less than 3 V as I already need 2.2 V (or maybe 2.1 V with fan) to do only 402 MHz...
Either your CPU won the lottery or I got the worst draw...