Sphere478 wrote on 2022-03-19, 09:37:

See the p5a simple mod thread. […]
Show full quote

LocalBus wrote on 2022-03-19, 08:49:

Maybe @Chkcpu can help out with some BIOS modifications 😀 […]
Show full quote

Sphere478 wrote on 2022-03-19, 02:01:

Some bios editing utilities can enable hidden Settings. […]
Show full quote

Some bios editing utilities can enable hidden Settings.

Awdbedit I think is one, but I’m told it corrupts.

I think it would be really cool to make a socket 3 to socket 7 adapter.

We could be running dx4s in mvp3 and ali aladdin V boards :p

Maybe @Chkcpu can help out with some BIOS modifications 😀

Judging by the brochure for OPTi 571/572 it should indeed support "up to 2 MB of Write-
back cache".

https://archive.org/stream/bitsavers_optibroc … e_1993_djvu.txt

I think the main thing holding this board back though, is the 32-bit DRAM bus.

The VLB-to-PCI might not help the cause either, but the main chipset design is just not keeping up.

When comparing benchmark results for # doom -timedemo demo3 (screen size maxed out), I get 2550 realtics. I bet a 430 chipset based P100 would have better results. Fun fact is that MBF204 runs significantly faster, where the same demo3 results in 1869 realtics = 40.0 frames per second.

Regarding 486 compatibility with this board (provided some kind of socket adapter) - I know @Snufkin looked into it. Might be some pins that would need to be hot-wired to the chipset since those are not present on Socket4, 5 or 7. Who knows, some day ...

See the p5a simple mod thread.

In that thread there is a way to toggle write back/write through with a hardware mod

K6-2/3+ Success on Asus P5A 1.06 - simple mod

Thanks for sharing! Hmm, I wonder if pulling pin AJ19 high toggles both L1 and L2 cache to write-back mode? L1 is already set to WB and this is possible to change from BIOS (tested WT as well with less performance as a result).

Small update, updated the BIOS ROM with modbin45 and unlocked "write-back" mode for L2 cache. Well it was probably locked for a reason because it doesn't work 😀 It doesn't boot past the main BIOS screen with it enabled.

This may be the slowest P100 system, an achievement in itself!

2550 realtics in Doom 😂 It feels very similar in performance to my old 486 DX2 66 VLB system back in the days.

In MBF204 in get much better results with page_flip=0, which gives 1858 realtics.

With page_flip=1 gives 2472 realtics.

Quake timedemo demo1 gives 18.4 fps.

LocalBus wrote on 2022-03-22, 18:30:

Small update, updated the BIOS ROM with modbin45 and unlocked "write-back" mode for L2 cache. Well it was probably locked for a […]
Show full quote

Small update, updated the BIOS ROM with modbin45 and unlocked "write-back" mode for L2 cache. Well it was probably locked for a reason because it doesn't work 😀 It doesn't boot past the main BIOS screen with it enabled.

20220322_185805.jpg

This may be the slowest P100 system, an achievement in itself!

2550 realtics in Doom 😂 It feels very similar in performance to my old 486 DX2 66 VLB system back in the days.

In MBF204 in get much better results with page_flip=0, which gives 1858 realtics.

With page_flip=1 gives 2472 realtics.

Quake timedemo demo1 gives 18.4 fps.

Maybe a later cpu will respond differently? Or maybe your chips only support one mode?

Sphere478 wrote on 2022-03-22, 21:32:

LocalBus wrote on 2022-03-22, 18:30:

Small update, updated the BIOS ROM with modbin45 and unlocked "write-back" mode for L2 cache. Well it was probably locked for a […]
Show full quote

Small update, updated the BIOS ROM with modbin45 and unlocked "write-back" mode for L2 cache. Well it was probably locked for a reason because it doesn't work 😀 It doesn't boot past the main BIOS screen with it enabled.

20220322_185805.jpg

This may be the slowest P100 system, an achievement in itself!

2550 realtics in Doom 😂 It feels very similar in performance to my old 486 DX2 66 VLB system back in the days.

In MBF204 in get much better results with page_flip=0, which gives 1858 realtics.

With page_flip=1 gives 2472 realtics.

Quake timedemo demo1 gives 18.4 fps.

Maybe a later cpu will respond differently? Or maybe your chips only support one mode?

L1 cache works in WB mode. Regarding the L2 cache I guess it is simply down to the chipset design.

I do have a P54CS P200 CPU on the way, which is probably as high as this board can take, given the 3.3V regulator.

You could always get an interposer and put a K6-2 or III or 2+ or 3+ in it.

cyclone3d wrote on 2022-03-23, 01:13:

You could always get an interposer and put a K6-2 or III or 2+ or 3+ in it.

No pun intended 😉

LocalBus wrote on 2022-03-22, 23:15:

Sphere478 wrote on 2022-03-22, 21:32:

LocalBus wrote on 2022-03-22, 18:30:

Small update, updated the BIOS ROM with modbin45 and unlocked "write-back" mode for L2 cache. Well it was probably locked for a […]
Show full quote

Small update, updated the BIOS ROM with modbin45 and unlocked "write-back" mode for L2 cache. Well it was probably locked for a reason because it doesn't work 😀 It doesn't boot past the main BIOS screen with it enabled.

20220322_185805.jpg

This may be the slowest P100 system, an achievement in itself!

2550 realtics in Doom 😂 It feels very similar in performance to my old 486 DX2 66 VLB system back in the days.

In MBF204 in get much better results with page_flip=0, which gives 1858 realtics.

With page_flip=1 gives 2472 realtics.

Quake timedemo demo1 gives 18.4 fps.

Maybe a later cpu will respond differently? Or maybe your chips only support one mode?

L1 cache works in WB mode. Regarding the L2 cache I guess it is simply down to the chipset design.

I do have a P54CS P200 CPU on the way, which is probably as high as this board can take, given the 3.3V regulator.

you could always try forcing it in hardware.. but iI suspect that the bios is setting it and it just doesn't work. see Socket 5/7/SS7 (Motherboard) Tweaker (Released) thread

you can do pentium 200 mmx overdrive, or k6-233 Thoughts on k6-233 on socket 5

past that, need a powerleap

The TIP127 PNP transistor used as part of the VRM circuit is probably the weakest link here. The TDP rating between a P100 vs P200 is 50% higher: ~10W vs ~15W.

Better keep an eye on the temperature and consider at least a fan over the VRM ❄

LocalBus wrote on 2022-03-23, 11:51:

The TIP127 PNP transistor used as part of the VRM circuit is probably the weakest link here. The TDP rating between a P100 vs P200 is 50% higher: ~10W vs ~15W.

Better keep an eye on the temperature and consider at least a fan over the VRM ❄

good eye, yeah that heatsink looks a little small. I wonder if you could wire another transistor in parallel?

Get two from the same batch if you do that, you want them to be as identical as possible.

Interesting idea, so simple as hooking them up in parallel to let them share the load with the same switching speed 🤔

I will definitely order a few TIP127 to have as spares as a start. Who knows when the original one fails...

It should be able to cope with 5A continuously with adequate cooling, which really puts it at its max with a P54CS @200 MHz.

LocalBus wrote on 2022-03-23, 13:54:

Interesting idea, so simple as hooking them up in parallel to let them share the load with the same switching speed 🤔

I will definitely order a few TIP127 to have as spares as a start. Who knows when the original one fails...

It should be able to cope with 5A continuously with adequate cooling, which really puts it at its max with a P54CS @200 MHz.

If there is some kind of driver chip then you must be sure not to overload it. But being a linear regulator.. idk if it has that…?

Sphere478 wrote on 2022-03-23, 14:14:

LocalBus wrote on 2022-03-23, 13:54:

Interesting idea, so simple as hooking them up in parallel to let them share the load with the same switching speed 🤔

I will definitely order a few TIP127 to have as spares as a start. Who knows when the original one fails...

It should be able to cope with 5A continuously with adequate cooling, which really puts it at its max with a P54CS @200 MHz.

If there is some kind of driver chip then you must be sure not to overload it. But being a linear regulator.. idk if it has that…?

Looks like a linear design to me? I would otherwise expect some chokes and capacitors to filter the output.

The large electrolytic capacitor sitting next to it is on the Vcc (5V) side I belive.

The 41DR IC is anonymous, haven't found a datasheet for it yet, but it plays a role in the voltage regulation. Maybe similar to this -> https://datasheets.maximintegrated.com/en/ds/ … 626-MAX1627.pdf

download/file.php?id=125264&mode=view

LocalBus wrote on 2022-03-23, 14:54:

Looks like a linear design to me? I would otherwise expect some chokes and capacitors to filter the output. […]
Show full quote

Sphere478 wrote on 2022-03-23, 14:14:

LocalBus wrote on 2022-03-23, 13:54:

Interesting idea, so simple as hooking them up in parallel to let them share the load with the same switching speed 🤔

I will definitely order a few TIP127 to have as spares as a start. Who knows when the original one fails...

It should be able to cope with 5A continuously with adequate cooling, which really puts it at its max with a P54CS @200 MHz.

If there is some kind of driver chip then you must be sure not to overload it. But being a linear regulator.. idk if it has that…?

Looks like a linear design to me? I would otherwise expect some chokes and capacitors to filter the output.

The large electrolytic capacitor sitting next to it is on the Vcc (5V) side I belive.

The 41DR IC is anonymous, haven't found a datasheet for it yet, but it plays a role in the voltage regulation. Maybe similar to this -> https://datasheets.maximintegrated.com/en/ds/ … 626-MAX1627.pdf

download/file.php?id=125264&mode=view

looks like there are better heatsinks you could get also I think this one is designed to stand up but you can prob modify it

Most of these early Pentium boards use a circuit with an LDO voltage regulator alongside a transistor like TIP127 to boost the output current. I have a Soyo Socket 5 board that supports only P90/P100 and yet it uses quite a beefy heatsink for the transistor (it still gets quite hot):

TheMobRules wrote on 2022-03-24, 00:47:

Most of these early Pentium boards use a circuit with an LDO voltage regulator alongside a transistor like TIP127 to boost the output current. I have a Soyo Socket 5 board that supports only P90/P100 and yet it uses quite a beefy heatsink for the transistor (it still gets quite hot):

TIP127.jpg

You know what, idk why I didn’t think of this earlier,

Re: vrm module project thread roundup, share ideas, make new designs

Interesting mods for sure 🤔

Gigabyte GA-586HX (Rev. 1.53)

This one is very similar to my setup so I guess you could play around with resistors for the voltage reference part of the circuit to say reach 2.8V. But that would also affect the I/O voltage.

I will be happy with getting this board accepting the P54CS CPU.

Placed a fan next to the TIP127 and been running quake demo loop for an hour now with my P100.

With the fan I am not that worried that it can even cope with the 15W TDP of the P54CS @ 200 MHz.

You don’t happen to have a powerleap do you? A powerleap bypasses all but I/O voltage

If you do have one, you could put a k6-3+ in it 🤣 It should be possible, you could get ahold of Jan and he can probably update the bios to support all the newer cpus. Re: Diy modding support for k6+And 120gb hard drives into bios roms

Cool heatsink, what model is it? can you still buy them?

Sphere478 wrote on 2022-03-24, 11:43:

You don’t happen to have a powerleap do you? A powerleap bypasses all but I/O voltage

If you do have one, you could put a k6-3+ in it 🤣 It should be possible, you could get ahold of Jan and he can probably update the bios to support all the newer cpus. Re: Diy modding support for k6+And 120gb hard drives into bios roms

Cool heatsink, what model is it? can you still buy them?

Haha go nuts! I love it 😀

No PowerLeap at disposal, maybe for the better in this case... could be a fun project in another life to design a custom transposer myself, I do have a couple of spare ZIF Socket-7 parts.

Naah I will not kill this board, in fact it was intended for something 486-ish in performance. Maybe I hit the sweet spot already 😀

But I will for sure give the P200 a spin just to see how severely limited it would be by the chipset 😉

Maybe Jan can help me here with the CPUID, since I guess it will report as something else due to being 2 years later CPU model?

In fact I even had to tune back the L1 cache to write-through mode for Sam and Max to work with General Midi support (someone ought to patch this...).

Regarding the heatsink, I got it from a Super Socket 7 system I bought a while back, and where due to its heavy weight; the whole ZIF Socket 7 was ripped out of the board during shipping... hence I have spare ZIF Socket 7 parts 😀 Many bent pins on that K6-2+ CPU but I managed to straighten them out again.

It is a Zalman HS targeting Socket 370 I believe:

http://www.eprom.com/home/Zalman/zm_cnps6000cu.html

No idea if there are some new old stock available somewhere?

After reading up on this thread:

Doom in DOS: Original vs Source Ports

I can draw the conclusion that at least VGA Mode X is significantly slower as compared to the corresponding VESA mode (320x200) on my system.

Comparing DOOM 1.9 with both MBF 2.04 and BOOM 2.02 consistently shows ~700 realtics better performance for the source ports. This puts my system more on par with P90 instead of say 486DX2 performance bracket 😀

Key seems to be page flip enabled | disabled. With page_flip = 0 in MBF 2.04 I get a significant performance boost. With page flip_flip = 1 I would say the performance is more or less on par with the original. To my understanding this also toggles between VGA Mode (page_flip=1) and VESA mode (page_flip=0).

With BOOM 2.02 I did not explore any configuration, but got similar performance as with MBF 2.04 with page_flip=0.

Isn't page_flip supposed to help performance?

AFAIK the Cirrus Logic GD-5446 is no slouch when it comes to VGA modes, or?

I did see that for Rendition Verite there is some kind of tool to remap mode13 to VESA modes, but not sure if there is any generic method for this for other adapters as well.