Is there a difference between a 100HF and a 100GF? The biggest difference I can see is that the HF has a heatsink attached and the GF doesn't.

Pic can be found here, Wikipedia doesn't allow hotlinking to my knowledge

http://en.wikipedia.org/wiki/File:IBM-5x86.jpg

sliderider wrote:

Is there a difference between a 100HF and a 100GF? The biggest difference I can see is that the HF has a heatsink attached and the GF doesn't.

That is the main difference I noted as well, however in my actual tests, I found the -HF models to overclock to 133 MHz quite a bit better. It may have just been luck of the draw though.

feipoa wrote:

sliderider wrote:

Is there a difference between a 100HF and a 100GF? The biggest difference I can see is that the HF has a heatsink attached and the GF doesn't.

That is the main difference I noted as well, however in my actual tests, I found the -HF models to overclock to 133 MHz quite a bit better. It may have just been luck of the draw though.

Is it because of the heatsink or some other factor? If it's just a matter of cooling, then attaching a really good fansink combo to the GF should produce the same results or better, right?

Perhaps. But, I have two -100HF units. One of them overclocks to 133 MHz at a low lower voltage the the other. One needed 4.0V, which makes it run too hot, while the other did pretty well at 3.75 V.

To be honest, I only used a generic 486 heatsink/fan on the -100GF unit, so I cannot really comment on what a larger heatsink/fan will accomplish.

sebaz_ri wrote:

Now it even not POST's, so the IBM 5x86C@133 is only for your Biostar mobo

Now that i think about it, changing the psu could give different results? I have a spare AT psu laying around and thought about trying it, which psu did you used when you overclocked the IBM 5x86-100HF?

I have 3 generic AT PSU's I use on my 486 systems. The one sitting in front of me is a Green Line ES-250T.

I suppose one day it would be prudent to open the case and check the caps. Just recently, the AT PSU in my 386 popped and fizzed. This was the original AT PSU which came with the 386 computer. If this is any indicator, the 486 PSU's would be next in line to go.

feipoa wrote:

I have 3 generic AT PSU's I use on my 486 systems. The one sitting in front of me is a Green Line ES-250T.

I suppose one day it would be prudent to open the case and check the caps. Just recently, the AT PSU in my 386 popped and fizzed. This was the original AT PSU which came with the 386 computer. If this is any indicator, the 486 PSU's would be next in line to go.

I also use generic AT PSU's in most of my systems. They all have good caps, but some came with bad caps and i replaced them. As i have already said before, most of my systems are dumpster finds and i need to check everything before i press the power button unless i want some fireworks

Just for the heck of it, the last system i found was a PCChips M507 with a Pentium 100 processor and... fake L2 cache like usual, and of course, it was working.
And anyway, most of the systems i found are PCChips because they were cheap back in the day

Small update to this thread. I hope this information will save someone the headache I just went through. If you plan to run your IBM 5x86c chip at 133 MHz on the Biostar MB-8433UUD at 2x66, you need to set your memory read wait state to 2 (or greater?) for long-term stability, not 1. I found that the memory write wait state needed to be left at 0.

While I have adapted my motherboard to accept 1024 KB of double-banked L2 cache, this is not a viable option for most, and as such, I think the most stable system is with 256 KB double-banked L2 cache, set in write-through mode, with a single stick of 64 MB FPM (5V). Alternately, 512 KB, single-banked L2 cache with 2x64 MB sticks is also performing well with these settings in write-through mode. In each case, set the L2 timing to 3-2-2 and the PCI-to-FSB divisor at 1:1/2.

My system is stable with 3.65 V or 3.7 V to the CPU, with 3.5 V being the minimum to POST. The extensive tinkering I did with the CPU voltages was probably not necessary as the long-term instability was related to the memory read wait state in the BIOS.

Not all IBM 5x86c-100 chips will run well at 133 MHz. Only 2 in 6 I tested would run reliably. Some Cyrix 5x86-120 chips may also work.

As a comparison to an actual Cyrix 5x86-133/4x running on the same board, the
L1 = 184 MB/s
L2 = 58.9 MB/s
RAM = 44.0 MB/s

Will try your settings on my 100GP running at 120 (3x40)

At 120 MHz, you can leave the memory read wait state at 0 and usually the cache at 2-1-1. PCI-to-FSB divisor at 1:1, typically.

I'm running 3x50mhz on a 2.0 8433UUD-A
I can get to 80.09Mb/s Memory bandwidth but cannot get it to boot windows.
I have EDO RAM that does not like low WS..
I can get it to run speedsys, but I can't get it to load Windows.

This motherboard works best with FPM memory.

I am confused with what you are attempting. It is unlikely you will get your Cyrix 5x86-100GP to run at 3x50 MHz. Do not even bother with this.

Why are your CMOS settings set strange? If you now have a 50 MHz FSB, set the HOST/PCI at 1:2/3. Set L1 at Wr-Back. Load Setup Defaults. Then set your L2 to 3-2-2-2, DRAM read wait to 1ws and DRAM write wait to 0ws. This board seems to work better with DRAM write wait states set to 0 ws, even with a 66 MHz FSB.

feipoa wrote:

Why are your CMOS settings set strange?

Because I don't know what I am doing ! 😀

Here are the settings @ 120Mhz with EDO ram that are stable and fast :

This thread is really for discussing getting a Cyrix 5x86 running at 133 MHz. 120 MHz is an off-the-shelf spec and fairly straight forward to setup.

At any rate, what is your criteria for determining if the system is stable? Did you run memtest 4.0? What about HIMEM and Windows?

Are you sure you need set the EDRO DRAM Read Speed to 3-1-1-1? That setting is usually disabled and set at 4-1-1-1. I do not beleive it has any affect. Are you able to demonstrate otherwise?

Are you sure you can use 7+1 with L2 in Wr-Through mode? Typically you set 7+1 for WB mode and 8+0 for WT mode. So far, I have not been able to demonstrate a stable system with 7+1 WT or 8+0 WB on this motherboard.

IT seems thay my 100GP + mobo will not run at 133. It's either 120 or 150. Unless I am missing something.

how do you achieve 2x66 or 4x33 on a shuttle hot-433?

I've been messing with my IBM 5x86c and shuttle hot-433 ver 4 board and I can get 120 easy but how do you set 2x66mhz? (or cant you) when I set the jumpers for a Cyrix 5x86 X4 and a 33mhz bus I get 66mhz

To achieve 4x33 on a Shuttle HOT-433, you need a 4x-capable Cyrix 5x86 CPU, in which case, you'd set the motherboard with the 2x jumper. 4x-capable Cyrix 5x86 CPUs are difficult to come by. Finding a 4x-capable CPU which runs reliably at 133 MHz is rare indeed.

To achieve 2x66, you need a Cyrix 5x86 or preferably an IBM 5x86c capable of running reliably at 133 MHz. Set the FSB to 66 MHz. I beleive it is:

JP1: 1-2
JP2: 2-3
JP3: 1-2

CLKMUL to 2x. I beleive it is JP18: 2-3

Set the voltage to 3.6 V, JP9: 5-6.

If that fails, try 4.0 V, JP9: 7-8.

I have only been to run an IBM 5x86c at 133 MHz in a stable configuration on an MB-8433UUD at 3.75 V. L2 cache needs to be set at 3-2-2 and memory read wait at 2 WS. Set the PCI divisor at 1/2. From my tests, the HOT-433, M919, PKM-0033S, and 4DPS can only handle up to a 60 MHz FSB reliably.

Good luck.

I have had success in getting an IBM 5x86c-100 running at 133 MHz using a Gigabyte GA486AM/S v2.21 with BIOS date 7 Nov. 1995. Thus far, this is the only other motherboard asside from the MB-8433UUD which I have been able to run at 2x66 MHz. For whatever reason, I was only able to get WinNT 4.0 operational. Win 95c hangs on boot at 133 MHz, although runs fine at 100 MHz. Also, I was only able to get the QFP IBM 5x86c chips to work on this motherboard. 100/133 MHz in NT4 ran fine; 100 MHz in W95c ran OK. Test voltage was 3.740 V, which was what I previously determined was stable for this QFP chips at 133 MHz.

IBM 5x86c-100HF PGA and Cyrix 5x86-133/4x chips would not boot into NT4 at any frequency. I don't know why this is. Am5x86-133 and Intel DX4-100-WB worked fine. I do not have any explaination for the selective behavior of this motherboard for certain IBM/Cyrix 5x86 chips except that these early UM8886AF chipsets have always been more problematic compared to the UM8886BF chipsets.

I've been playing with this board - Gigabyte GA486AM/S Ver 2.21 and can't find correct jumpers for 60/66MHz FSB or my board/CPU won't take it. For the moment I have AMD 5x86 but will try later with the Cyrix one.

Can someone post undocumented FSB settings?

Thanks