Thank you for this! Great work.

Awesome work. I've just been messing around with a K6-3/400 on a Gigabyte 5AX. This comes really handy!

Am386DX-40 wrote on 2022-02-03, 21:54:

Awesome work. I've just been messing around with a K6-3/400 on a Gigabyte 5AX. This comes really handy!

Could you share some results? I'm just in the process of setting up a GA-5AA.

From testing it seems the MVP3 chipset is a bit faster.

Oh! This is invaluable!! Thank you so much for doing this!!!

Has anyone done a comparison for mvp3 or sis530?

mwdmeyer wrote on 2022-02-05, 05:20:

Am386DX-40 wrote on 2022-02-03, 21:54:

Awesome work. I've just been messing around with a K6-3/400 on a Gigabyte 5AX. This comes really handy!

Could you share some results? I'm just in the process of setting up a GA-5AA.

From testing it seems the MVP3 chipset is a bit faster.

I’ve been playing with a mvp3 freeway board lately. Performance seems very close possibly better than my p5a 1.04. But there is no question that the ali V has the better memory controller for adding more ram. Thr mvp3 is suuuper picky about what ram is used. But the memory speed, seems to depend on the board and bios used? I’ve had really good results on the freeway, but also impressive results on the p5a.

Between the two I prefer the freeway. But I have a 503+ and I prefer the p5a over it. I also have the 5aa but it’s a early version. Agp is picky on it.

My personal ranking:
Freeway
P5a 1.04
503
Early version 5aa
P5a 1.06

Why early 5aa? I’ve got a version 3.2 and also a 1.1

mwdmeyer wrote on 2022-02-05, 06:38:

Why early 5aa? I’ve got a version 3.2 and also a 1.1

I haven’t been too happy with stability and overclockability on my early 5aa.

I think mine is a 2.something. So maybe calling it a mid version is more accurate.

The 3.something got better power conversion and later chipset versions I believe.

I’m just ranking what I have. Also it seems like p5a and p5b had better performance. But oddly the later versions give lots of trouble on those. I had a 1.06 and had nothing but issues. The mod worked but board was still unstable.

Oh right, I thought you meant you preferred 1.1 over the 3.2 rev.

I've got a bunch of benchmarks with my MVP3 here.
https://www.vogonswiki.com/index.php/Super_So … et_7_Benchmarks

I like the Soltek SL-54U5, is stable and fast. Just having issues finding a PS2 header for it.

I'm struggling to get close to the Soltek performance wise on the 5AA. For me 128mb is enough for a K6 2/3. Once you install the memory patch for the MVP3 it is very quick.

Sphere478 wrote on 2022-02-05, 05:41:

Oh! This is invaluable!! Thank you so much for doing this!!! […]
Show full quote

Oh! This is invaluable!! Thank you so much for doing this!!!

Has anyone done a comparison for mvp3 or sis530?

mwdmeyer wrote on 2022-02-05, 05:20:

Am386DX-40 wrote on 2022-02-03, 21:54:

Awesome work. I've just been messing around with a K6-3/400 on a Gigabyte 5AX. This comes really handy!

Could you share some results? I'm just in the process of setting up a GA-5AA.

From testing it seems the MVP3 chipset is a bit faster.

I’ve been playing with a mvp3 freeway board lately. Performance seems very close possibly better than my p5a 1.04. But there is no question that the ali V has the better memory controller for adding more ram. Thr mvp3 is suuuper picky about what ram is used. But the memory speed, seems to depend on the board and bios used? I’ve had really good results on the freeway, but also impressive results on the p5a.

Between the two I prefer the freeway. But I have a 503+ and I prefer the p5a over it. I also have the 5aa but it’s a early version. Agp is picky on it.

My personal ranking:
Freeway
P5a 1.04
503
Early version 5aa
P5a 1.06

If you were asking me, yes, I did the comparison also for the via4in1 for the mvp3 I posted the result in another topic.
I never tested the sis since it's not a real ss7 chipset.

mwdmeyer wrote on 2022-02-05, 05:20:

Am386DX-40 wrote on 2022-02-03, 21:54:

Awesome work. I've just been messing around with a K6-3/400 on a Gigabyte 5AX. This comes really handy!

Could you share some results? I'm just in the process of setting up a GA-5AA.

From testing it seems the MVP3 chipset is a bit faster.

Mine was the 5AX though, and I didn't do any serious benchmark, just compared the k6-2 to the k6-3 clock for clock and 3-400 vs 2-500. Don't even remember what 3d card I used.
What I do remember though, is that I discovered something weird going on with the performance depending on how you populate the memory slots. I was benchmarking all the way from 16mb up to 512mb to see the effects of the "cacheable range" (and see if the k6-3 fixed that) but I found that there were HUGE dips in performance when dimms were installed in a certain way. Don't know why, can't explain, but for example (I'm making this up, but you get the idea) you install a 64mb dimm on slot 1 and a 128mb dimm on slot 2 and you get good/normal performance, now you add a 64mb dimm on slot 3 and you get horrible performance. You switch the 128mb to slot 3 and the 64mb in slot 3 to slot 2 and performance is restored. Sometimes you had to leave 1 slot free, it varied a lot.

I can try to reproduce those again as I have the board at hand.