Mu0n wrote on 2021-07-02, 05:37:

Onto the wonderful (ugh) world of MS-DOS 6.22 network capability installation. […]
Show full quote

Onto the wonderful (ugh) world of MS-DOS 6.22 network capability installation.

I'm following this page: http://www.dmp.com.tw/tech/vortex86dx/
DOS NDIS driver: pci-ndis.zip

and this page:
https://www.legroom.net/howto/msdos

I'm stuck fairly early in the process: with MS Network Client v3.0 (which I've extracted to a temporary install folder), it simply will refuse to take my path to the NDIS driver taken from the first link. It complains it can't find any driver. I've tried a few others (including one from RealTek's website (NDIS2 for dos: https://www.realtek.com/en/component/zoo/cate … et-pci-software)

Try this http://www.brutman.com/mTCP/. IRC client, Ftp server/client and more.

The MS Dos Client with the TCP/IP protocol sucks up far to much conventional memory

mTCP ftp server on retro-pc and a modern client in modern system is the way to go
all you need is the DOS packet driver for the mTCP suite to work : load the driver, do your network thing, unload driver , enjoy life 😀

All hail mTCP!

Thanks guys, that did the trick for some quick and massive ftpserv transfers.

on my modern win10 machine, I used WINSCP in ftp mode
on the Vortex86DX, I enabled mtu 1500 but not most the other optional settings.

I hit some snags when there were a lot of files and/or subdirectories to transfer, nothing that clicking on Retry wouldn't solve.

Next up in the distant future, figuring out ipx multiplayer

Most of the fm synth I've played sounds just fine, even inside Dune 1, but there's this heavily distorted part when you go into the planet view, or the encyclopedia, where the main instrument is butchered. Take a listen.

As I wrote before, it's not that big of a deal when it happens so rarely, but it's an improvement possibility, unless the chip is notorious for sounding like that in that specific situation.

Yeah it's fun to get MS Network Client working just for fun but it's pretty rubbish for actually transferring files, pretty much every time there's a new Windows update it breaks something.

The ODI ones are what you want for network gaming

I'm keeping my momentum still! I'm on a roll.

Win98se:
Tried TCP Optimizer which was talked about here : "Win98 poor network performance... advice needed \ VOGONS" Win98 poor network performance... advice needed

But it doesn't improve network transfers under my Windows installation. I keep getting a few seconds of rise up to 1 mb/s, then a downfall to 0 and stagnation for 3-5 seconds before another cycle restarts. I estimate this makes single cd iso transfers take up to an hour. The speed setting in the program always resets to 3 mbps for some reason.

Also, I'm gonna look for a braindead cheap solution ready-to-buy or make my own breadboard circuit at least, of a simple audio 2x1 audio mixer, so I can make this box mix the fm/sb signal to the output of a Roland module, which would be great for all those games which decouple the sound device choices between these two. I don't think I want individual volume control but I might change my mind. Most solutions on Amazon are $30+ things and it ought to be cheaper than this. Basically, the absence of a physical line in socket forces me to look for my own external hardware solution, which could come handy in other projects.

You get these sorts of things on ebay -

Mu0n wrote on 2021-05-27, 14:36:

I went on and calculated my total costs in the past months for the project. I set out to build 2 units (highlighted below in the chart) to save on costs. Keep in mind these purchases:

It is an interesting project, but is it correct to say that it cost upwards of $600 (CAD) to build the first unit? That seems quite steep! I didn't notice it in that post, but did you mention the cost per unit just for the SOM?

640K!enough wrote on 2021-07-07, 17:20:

Mu0n wrote on 2021-05-27, 14:36:

I went on and calculated my total costs in the past months for the project. I set out to build 2 units (highlighted below in the chart) to save on costs. Keep in mind these purchases:

It is an interesting project, but is it correct to say that it cost upwards of $600 (CAD) to build the first unit? That seems quite steep! I didn't notice it in that post, but did you mention the cost per unit just for the SOM?

As I wrote, the barrage of shipping and duties costs is the downside aspect of it all. Your mileage may vary elsewhere in the world. It's possible further costs may be saved by having PCBWay (or others) provide the majority of the components already assembled.
ICOP sells 1 or 2 SOMs for 177 USD apiece (shipping and duties not included)

Yeah it's not a cheap project (especially in Canada, apparently) but if someone wanted to manufacture these in quantity the price would come down significantly. I'm just a hobbyist really so I'm not the person for that job.

I do have a cheaper Vortex86 SoM (the SOM-128-EX) but I'm burned out on building mini DOS PCs for now. Perhaps I'll investigate it in a few months.

The "best" solution would be to design a motherboard that had the vortex86 SoC directly on it, but that would probably involve commisioning DM&P to design it, and then MOQs would really become a problem.

Compatibility notes:

King's Quest VI:

CD movies stutter weirdly, but it's fixed from a replacement vga320.drv from GoG

under maximum cpu speed settings, the midi initialisation for mt-32 goes to a checksum error (and then none of the instruments have been customized right so it sounds pretty bad) , so I made that work with cpu clock division by 16 at least.

Next bug happens semi-randomly when sound effects have to play, namely, the narrator and speech audio clips, it will hang and infinitely play the current waveform. I've tried removing L2 cache and maybe it's helping but I need more testing.

Edit:
Still crashes with L1 disabled. The floppy version has no stability problems so it may be my fallback.

So I think I'm going to recommend people use a 16KB eeprom (24LC16 is what I used) then they can just use the Orpheus flashing tools.

Note that you need to edit CSMPU330.ini to enable Adlib - add the text "IFM" to the GlobalConfigBits field (so it ends up being "GlobalConfigBits = IFM SPS WTEN" ).

If you're using unisound, it's probably a good idea to also change "DeviceName" from Orpheus to something else (I used weecee). If unisound detects an orpheus it will just disable adlib again (although you can override it).

King's Quest VI: solution found - just transplant audblast.drv from the floppy version (which is rock solid on its own, for hours) to the CD version. So far, 15+ minutes of gameplay and no crash so far.

Also, I discovered I can just use general midi with my sound canvas and mix in my sound blaster audio in my module itself and it sounds pretty good compared to the MT-32. Not all games will do this, of course, so I have a passive mixer meant to feed a powered set of speakers arriving on Monday.

More compatibility notes (I'll bring 'em all in one text file at some point):

Descent 1 Anniversary edition: wouldn't run with the included dos4/gw, so I used dos32a in its place and it works perfectly with absolutely decent FPS:
https://www.youtube.com/watch?v=qgh2hGS8Tig (sorry for the quality of the video, as you can see, phone filming without a tripod is shoddy)

Hi-Octane: hard crash whatever I did. It uses an external dos4/gw, so I tried with dos32a and it runs well....TOO well that it's unplayable (even in 640x480), so I tried with cpu divide by 8 and it's close to what it should be (I'll try the others later). I can't make the sound setup recognize my GM roland sc-88st without a hard crash, softmpu being present or not doesn't change things. All I have left to try is to dos32a the sound setup utility as well.

TLDR: both DOOM, Quake and 3dBench 1.0c all point towards an equivalent somewhere around Pentium 1 MMX 200 MHz to 233 MHz

DOOM comparison: https://thandor.net/benchmark/32
Quake timedemo comparison: https://thandor.net/benchmark/33
3DBench 1.0c comparison score: http://img708.imageshack.us/img708/8820/3dbenchdatabase.png

Using Phil's Computer Lab 2017 dos benchmark suite:

1) 3DBench 1.0: 66.6 fps
2) 3DBench 1.0c: 151.0 fps
3) Chris' 3D Benchmark: 108.9 fps
4) Chris' 3D Benchmark 640x480: 46.1 fps
5) PC Player Benchmark: 72.9 fps
6) PC Player Benchmark 640x480: 28.8 fps

a) DOOM minimal: 324.8 fps
b) DOOM maximal: 79.8 fps
c) Quake timedemo: 44.6 fps
d) Quake timedemo 360x480: 20.4 fps
e) Quake timedemo 640x480: 17.2 fps

l) Landmark System Speed Test 6.00:
type: Intel 80486DX, 1.221 MHz, Intel 80487DX, Video: Tseng 4, cpu 2924.7 MHz, FPU: 4085.0 MHz, Video 16 661.69 chr/ms
m) TOPBENCH 3.0: 520
n) Speedsys 4.78:
Vesa mem: 20088 KB/s, Cyrix Cx486SLC 240 MHz, CPU 183.55, L1 (16kb) 493.16 MB/s, L2 (256 kb) 256.28 MB/s, Mem thru: 163.80 MB/s, HD: 9774.53
Rnd Access Time: 0.91 ms, Buffered Rd Speed: 20476 KB/s, Linear verify speed: 256371 KB/s, Linear read speed: 14757 KB/s
s) CHKCPU:
Vendor: unknown, Int CPU Speed: 799.5 MHz, Clock Mult: unknown, CPU-ID Str: Vortex85 SoC, CPU-ID Signature: 000522, CPU Mode: Real, Internal L1: enabled write back
t) CACHECHK:
conv_mem: 640 kb, ext_mem: 65,535K, Total RAM: 639K, Clocked 605.5 MHz
L1 16KB, 1118.3 MHz 0.9 ns/byte (567%) (267%) 2.2 clks
L2 256KB, 418.7 MHz 2.5 ns/byte (212%) (100%) 5.8 clks
Main mem: 197.0 MB/s 5.3 ns/byte (100%) 12.3 clks
Effective RAM access read 21 ns
Effective RAM access write 16 ns

Nice tests!

I did some comparitive benchmarks of the various other vortex86 chips in this thread if you're interested - Vortex86 series benchmarks

Welp, my dos-dedicated machine #1 still works pretty well, despite needing a slight reflow of the micro USB power socket to solve a misconnection problem where power would cut unless the cable was angled from above..

But, machine 2 gets full power for a second, then lowers to a 2.34 V state, which still seems to power the network socket as usual, but makes the blue LED on the SOM shine way less brighter. I'm looking for ideas for what to test.

Edit : oh no, without the SOM I get 5.22V..

Edit2: after reseating the SOM for a couple of times, I get a bright blue led again, but the voltage goes between 4.40 and 4.80 (jumps around periodically), even under minimal load (no ps2 peripherals, no vga, no network, etc)

King's Quest VIII on the Vortex86DX in Win98se: Installation of Eternity:

-mount the .iso as a virtual drive
-wait up the 1-hour installation to hard disk
-Game launches, animation plays, start new game
-copy more files on the hard disk, takes 10 minutes
-loads the first level (Daventry)
-black screen
-install 1.3 patch
-black screen
-install dx9.0c, complains about cpu and aborts
-install dx6
-black screen
-runs nicode 'shim' problem fixer from https://svn.nicode.net/kqmoefix/bin/
-black screen
-uininstall game
-use modern kq8 installer (made for xp/7/8/10), doesn't seem to fuss
-game won't launch, requires a mscoree.dll (from .net framework...which is not present in a stock win98se build)
-attempt .net framework 2.0 install, complains about lack of windows installer 2.0 and internet explorer 5.0.1 (stock win98se has 5.0)
-install installer 2.0
-install ie5.0.1
^^ that's where I'm at

-install .net framework 2.0

edit-

.net framework 4.0 is needed but will definitely not run in win98se. So I can't use the modern installer which gives modern launchers, that's strictly for Vista/7/8/10.
I ended up going with modern solution and running a patched installer for a win10 machine with glide disabled, works well enough when you remember to up the fps to 60 from the config file. It's WILD to me that win10 runs this game better than a period correct configuration.

Mu0n wrote on 2021-08-03, 13:55:

Edit2: after reseating the SOM for a couple of times, I get a bright blue led again, but the voltage goes between 4.40 and 4.80 (jumps around periodically), even under minimal load (no ps2 peripherals, no vga, no network, etc)

Does the SOM work in your other board? Just trying to narrow it down between the SOM and the rest of the board.

Also voltage dropping to 4.4 often means the USB cable or charger isn't strong enough, I often have to experiment to find a combination of cable and charger that works 100%