For those with a 430TX board and working Tillamook w/L2 enabled, would you be willing to upload the BIOS here?

I am pleased to report that the bridging ADSC# with ADS# allowed my Asus VX97 (i430VX-based) to work with L2 cache enabled. However, if I had BRDY# bridged with BRDYC#, the screen stayed blank at power-on. Unfortunately, the 4x mode did nothing. I couldn't get 4x with or without the 4x mod, and I tried all possible multiplier jumpers on the motherboard.

I specifically remember that to get the Cyrix MII working with 4x, that the BIOS needed to be modified. Jan S. was kind enough to do that for me. Anyway, I suspect a similar case with the VX97.

I'd like to try some known working 430TX-based Bioses on my 430TX board if anyone has a known working one they can share.

Also, the ADS#-to-ADSC# mod offered no assistance to my Biostar 430TX-based motherboard - the screen is still blank at power-up.

I downloaded the v1.3 BIOS (11/27/1997) for the QDP Titanium 1B here, http://www.elhvb.com/mobokive/Archive/Qdi%20- … M_IB/index.html

I tried it on my Biostar 430TX board with a regular P233 MMX and it works fine. I tried it with the Tillamook and the screen remains blank. I tried the Tillamook with 1, 2, and 3 bridge options. So it seems to me that there is something set in hardware on the Biostar board which doesn't like the Tillamook - perhaps a resistor somewhere, or a particular short, who knows.

It would have been nice to have one all working 430TX board. My Biostar works with Cyrix MII 4x CPUs, 128 GB HDD sizes, and K6-III/II+ chips, but no Tillamook.

BIOS of my QDI Titantium IB : http://www.tmeeco.eu/BitShit/PCschit/TITAN1B.BIN
Version 1.3S and most likely patched to support large HDDs as it does't match the dump on Mobokive.

What's the difference between the 1B and 1B+? I am pretty sure I have a 1B+.

Different Super IO setup and probably different voltage and freq setting mechanism, one will brick the mobo by flashing IB+ BIOS to IB or the other way.

Good evening, I finally took some time to play with this cpu.
I actually wired the yellow pins, the cpu is booting in a Epox EP-51MVP3E-M but the results so far are unsatisfactory, how can I be sure i'm having the l2 cpu cache enabled?
It it is so the cache on mb should become l3 cache like how it happens with k6-x+ cpu, right?
Because in all previous screens I see just l1 and l2 cache.

Just use speedsys. The graph shows it

When using the Tillamook, the speedsys graph should only show L1 and L2 cache. No L3 cache.

Yep, but it should show working L2 cache!

Those are my results:
https://i.imgur.com/8f6D9Yz.jpg

Quake 320: 64,8
Quake 640: 16,2

They seems low to me, even worse than the k6-2 350 I was using before.

Your memory timing seems to be quite slow.
Check this.
Try 10MHz FSB and fine-tune the timings.
SDRAM is also a bit faster than EDO-PS2

I'm using 100 mhz with cas 2 and turbo settings...
Do you think the l2 cache is enabled? Is it right that I have worse results than the k6-2 350 mhz?
I thought mmx was faster, especially the fpu.

You can clearly see the l2 is working because if the 3 different plateus of the graph.

l2 cache is on mb, in fact I can see it even with k6-2 that doesn't have any l2 cache in the cpu.

I really wonder why if l2 cache is detected it doesn't show up like how it happens with k6-x+ cpu.

nemo, are you expecting a Pentium 266 MMX chip to beat a K6-2-350? I think that is a fairly unrealistic expectation. Did you look at the Quake results from the Ultimate 686 Benchmark Comparison? While I didn't test a Tillamook, I did run a standard Pentium 233 MMX at 75x3.5, or 262 MHz on a SS7 motherboard.

Quake results at 640x480:
Pentium MMX 262 = 17.3 fps [or 18.7 w/430tx]
K6-2-350 = 18.3 fps

Quake II results at 640x480:
Pentium MMX 262 = 12.3 fps [or 13.7 w/430tx]
K6-2-350 = 15.6 fps

Quake II results with OpenGL at 640x480:
Pentium MMX 262 = 23.8 fps [or 25.1 w/430tx]
K6-2-350 = 27.9 fps

Interesting how the 430TX results are about 10% faster in software mode and 5% faster in accelerated mode.

Actually yes, I thought it could compete against k6-2 at the same frequency, I would expect that having some l2 cache in the cpu (running at cpu speed?) could give it some advantage.
I was probably wrong.
To be sure if the cache is enabled I will test a pentium 233 mmx to 250 and underclock the tillamook to 250 too.

Nemo1985 wrote:

Actually yes, I thought it could compete against k6-2 at the same frequency, I would expect that having some l2 cache in the cpu (running at cpu speed?) could give it some advantage.
I was probably wrong.
To be sure if the cache is enabled I will test a pentium 233 mmx to 250 and underclock the tillamook to 250 too.

mmx has 16kb L1 cache (on cpu) and on your motherboard there is 1024kb L2 cache (1mb).
K6-2 has 32kb L1 cache (on cpu) and your board again has 1024kb L2.
K6-2+ has extra cache on cpu: L1 32kb and L2 128kb So your motherboard cache of 1024kb becomes L3 with this cpu
k6-3+ has 256kb extra cache on the cpu so: L1 32kb, L2 256kb and your motherboard cache again becomes 1024kb L3

Just like the normal k6-2 the mmx (and mmx Tillamook) do not have L2 cache on the cpu and therefor you can only see a L1 and L2. With Tillamook L2 was not working on most boards so only L1 was left and performance was slow. That's why boards need to be modded.

The k6-2 on the same clockspeed might be faster but i don't think speedsys is telling the right story as the differences are way too big there (same as for example 3dmark99 where 3dnow! is heavily favored). Looking at Feipoa's score's i think that it is much more accurate and when clocked both at 350mhz (100fsb) they would be fairly equal?

I compared the Intel mmx 233@300mhz 100fsb and the Tillamook at 350mhz and cache seems to work fine and the Tillamook is indeed faster (due to the extra mhz ofcourse but without L2 it would be slower).

Also posted k6-2, k6-2+ and k6-3 at the same 3,5x 100fsb on the same motherboard (Gigabyte Ga-5aa with Ali V chipset, 512kb cache) for a speedsys comparison.
Notice how they all get the same cpu score while the k6-3 is ofcourse the fastest in real live.

EDIT: sorry about the first picture, it's the k6-2+ but it is blurry. Scores are basicly the same as for the k6-3...

I agree with you meljor, the main difference between vanilla mmx and tillamook is that the second has some oboard cache
How can you say that the l2 cache on tillamook is working if it doesn't show up un speedsys, it should show just as it does with k6 plus cpu, don't you think so?
On your screenshots the results for l2 cache (that is the one on the mb) are the same for k6-2, pentium mmx and tillamook.
While the cpu cache gets correctly detected on k6-x+ cpu.
Could you do a benchmark with vanilla mmx and tillamook at the same frequency with quake?
If it shows different numbers cache is detected, so speedsys is wrong, otherwise cache isn't used despite of cpu modding.
Thank you.