Seems like the SOM304RD-PI (Vortex86DX as well) would be a good candidate for an ITX or mATX board. No onboard VGA, but it does have native PCI and ISA available at the same time. No branch prediction on that CPU though. The only other one DM&P has with both PCI and ISA is the Vortex86SX which is slower, has no FPU, and I think also no branch prediction.
(I made a comparison chart here)

@ExplodingLemur

How do you test for branch prediction on a CPU?

Eivind wrote on 2023-06-22, 13:21:

Also, the SD card speed was quite hampered by its PIO mode: IMG_0607.jpeg […]
Show full quote

Also, the SD card speed was quite hampered by its PIO mode:
IMG_0607.jpeg

..but after turning DMA mode on, it's much better:
IMG_0610.jpeg
IMG_0608.jpeg

This SOM uses the R1012, as opposed to the R1011 seen previously on the DX SOM. Does this one appear to perform normally with those drivers installed? Is there no compatibility mode file system warning on the Performance tab?

I'm beginning to wonder if the purported DMA bug on the other chips has been fixed, is still present, if there only appeared to be a bug because they were testing higher DMA modes without an 80-conductor cable, or if there really is a flaw in its CRC-generation component. No sign of file system corruption so far?

Duffman wrote on 2023-06-23, 01:10:

When I get my wee86DX board built I think it'll perform pretty similarly to your board. my theory in the weecee's case is that it's not the Vortex86DX CPU thats too slow for the unreal engine, it's the 2D only volari Z9 GPU thats too slow.
I haven't built it yet, so I don't know for sure, it's just my theory.

Well, I mean the DX is a good deal faster than the EX....but of course a Voodoo 3 (or any other hardware accelerated Glide/D3D card) will crush the Z9 in software rendering mode completely, it's not a fair comparison. :p

ExplodingLemur wrote on 2023-06-23, 03:01:

Seems like the SOM304RD-PI (Vortex86DX as well) would be a good candidate for an ITX or mATX board. No onboard VGA, but it does have native PCI and ISA available at the same time. No branch prediction on that CPU though. The only other one DM&P has with both PCI and ISA is the Vortex86SX which is slower, has no FPU, and I think also no branch prediction.
(I made a comparison chart here)

Nice chart! 👍
I have a couple of suggestions for additional columns, if you're interested; 1. approx. price, 2. BIOS support (AMI or Coreboot+SeaBIOS).
Also, the EX is often spec'ed at 300 or 400 MHz - but in my experience it runs cool and rock solid at 500 MHz with a little bit of active cooling. And that additional speed does make a difference.
I haven't tried any of the other variants, but I imagine they could be "overclocked" a bit with active cooling as well (but we're probably hindered by their AMI BIOSes I guess?).

Duffman wrote on 2023-06-23, 03:18:

How do you test for branch prediction on a CPU?

You read the datasheet 😀

Eivind wrote on 2023-06-23, 05:21:

Nice chart! 👍 I have a couple of suggestions for additional columns, if you're interested; 1. approx. price, 2. BIOS support (AM […]
Show full quote

Nice chart! 👍
I have a couple of suggestions for additional columns, if you're interested; 1. approx. price, 2. BIOS support (AMI or Coreboot+SeaBIOS).
Also, the EX is often spec'ed at 300 or 400 MHz - but in my experience it runs cool and rock solid at 500 MHz with a little bit of active cooling. And that additional speed does make a difference.
I haven't tried any of the other variants, but I imagine they could be "overclocked" a bit with active cooling as well (but we're probably hindered by their AMI BIOSes I guess?).

I'll see if I can dig up info on the supported BIOSes. Is that linked directly to the SoC, or more the SoM?
For prices I'll list a range for the SoMs carrying each CPU.

640K!enough wrote on 2023-06-23, 04:09:

This SOM uses the R1012, as opposed to the R1011 seen previously on the DX SOM. Does this one appear to perform normally with those drivers installed? Is there no compatibility mode file system warning on the Performance tab?

I'm beginning to wonder if the purported DMA bug on the other chips has been fixed, is still present, if there only appeared to be a bug because they were testing higher DMA modes without an 80-conductor cable, or if there really is a flaw in its CRC-generation component. No sign of file system corruption so far?

Huh, interesting about those integrated RDC components!
There is a file system compatibility warning under "Performance", but only for the A: drive. Picture below.
As for signs of fs corruption - haven't seen any! 😀 Happy to run tests, if you have anything specific in mind?

ExplodingLemur wrote on 2023-06-23, 05:37:

I'll see if I can dig up info on the supported BIOSes. Is that linked directly to the SoC, or more the SoM?
For prices I'll list a range for the SoMs carrying each CPU.

Well, strictly speaking a BIOS must be compatible for the whole system, so SoM... but since it's DMP making both the SoC and SoM - and I've never seen different BIOS types for the same SoC - I guess we could generalize and say SoC, in practice.
To the best of my knowledge, DMP have only used Coreboot/SeaBIOS for the EX and EX2 - all the rest are AMI builds.

Eivind wrote on 2023-06-23, 05:09:

Well, I mean the DX is a good deal faster than the EX...

I wonder why they chose to make the EX slower than the DX?
You would think with the EX being the newer CPU they would want it to be faster than the old one.

Duffman wrote on 2023-06-23, 06:58:

I wonder why they chose to make the EX slower than the DX?
You would think with the EX being the newer CPU they would want it to be faster than the old one.

I think they're just aimed at different purposes / markets. The EX can function more like a beefed up microcontroller, with all its customizable I/O. Probably draws less power as well, I dunno. And it's cheaper.

Now we're getting somewhere...
I got a SATA SSD and Ethernet up and running!

Poor IE5, I'm surprised Vogons loaded at all! 😁

CS4237B chip installed, tested everything in DOS, including using a Pi Zero 2 and a wavetable board for MIDI - works perfectly.
Win98 on the other hand.... hrrrmmf... can't seem to get any driver working, I'm stuck with either Windows crashing or complaining about a missing CWSAUDIO.WCM file at boot.

I know there should be a decent overlap of people on this thread and on the TinyLlama thread - did anyone manage to get a stable CS4237B setup in Win98 on that system?

@Eivind

I got it working in XP if that counts.
I had to assign the resources to it manually to get it to work though.

Duffman wrote on 2023-06-23, 15:59:

I got it working in XP if that counts.
I had to assign the resources to it manually to get it to work though.

Changing resources doesn't bother me, as long as the driver doesn't crash the whole system! 😀 Not sure if win xp drivers are applicable, though... 😒 Which driver was it?

@Eivind

It was just the inbuilt driver that came on the XP SP1 ISO.

Finally got it working using this driver!
First I went into the Resources section of each of the three CS4237B entries in Device Manager and made sure they were set correctly, ie.,
- DMA channel 1 and IRQ 7 for the codec, all port were fine.
- Port 330 without IRQ 9 for the MPU401 device.
...and then chose "update driver" for each and pointed windows to the "CWD-v286-1998 win-latest" folder from the driver download. Et voilà!

After thinking about it, I could probably change the defaults in the cs4237b.asm file I use to program the crystal chip's eeprom - to avoid having to muck around with the resource settings.

Update on real PS/2 devices:
Got my hands on a PS/2 keyboard and mouse (thanks, @perhenden!) - plugged them in, and they didn't work! 😮
Then I discovered I'd made a huge mistake and mirrored the traces to the pins. My own fault, of course, but dammit Kycon - who uses a "bottom view" on the PCB layout section of a datasheet?? One of those life lessons, I guess... Thankfully the HIDman's USB-to-PS/2 works a treat.

@Eivind

You could compare your eeprom asm file to the one rasteri used for the weecee. the sound on the weecee works flawlessly.

Duffman wrote on 2023-06-24, 01:39:

@Eivind

You could compare your eeprom asm file to the one rasteri used for the weecee. the sound on the weecee works flawlessly.

In this instance, he'll probably want to limit the possible configurations that are advertised, given that there are fewer resources available on the platform's restricted ISA implementation. If you give the Plug and Play sub-system too many choices, it's liable to choose a combination that you didn't expect, and on this platform, it may not be a usable one.