aitotat wrote on 2022-06-22, 19:08:

Yes, music speed is correct on that video.

There are two different kinds of speed issued with YMF and Cycles. If CPU is little too fast you get distorted audio (it is instantly obvious) and if CPU is way too fast the game won't start. You just get abnormal program termination error message.

How exactly what kind of OS and processor that causes these two types?

Cheers,

I didn't write down when there was abnormal program termination and when only distorted audio. I think de-turbo 1/3 gives bad audio and any faster gives the error message. Very little difference between tested CPUs: Intel WT didn't work with only L1 cache disabled. I used MS-DOS 7.1 (=Win98se without Windows) but Dos version should not matter (except Freedos is not compatible with Cycles).

I did some testing with Cycles on my Pentium MMX 166 rig. Here are my findings:

AWE64 Gold (CQM)

• full speed: game quits with the message: "Bad driver. Abnormal program termination"
• L1 cache disabled: music works correctly (no distortions) and plays at the proper speed

OPTi 82C930 (1:1 copy of YMF289B)

• full speed: game quits with the message: "Bad driver. Abnormal program termination"
• L1 cache disabled: music works correctly (no distortions) and plays at the proper speed

So I got the same result on both cards. This leads me to believe that what you're experiencing might not be just an OPL3 issue, though that could be a part of it. It's also possible that Creative's OPL3 cards are more speed sensitive than OPTi's solution.

Off topic: YMF289 clone - is that integrated in the 82C930 or a discrete clone IC on the card? I always thout C930 had integrated crappy OptiFM..

appiah4 wrote on 2022-06-24, 12:29:

Off topic: YMF289 clone - is that integrated in the 82C930 or a discrete clone IC on the card?

It's in the DXP44Q chip.

I always thout C930 had integrated crappy OptiFM..

They didn't start using their in-house OPTi FM solution until 82C931.

Joseph_Joestar wrote on 2022-06-24, 12:41:

It's in the DXP44Q chip. […]
Show full quote

appiah4 wrote on 2022-06-24, 12:29:

Off topic: YMF289 clone - is that integrated in the 82C930 or a discrete clone IC on the card?

It's in the DXP44Q chip.

I always thout C930 had integrated crappy OptiFM..

They didn't start using their in-house OPTi FM solution until 82C931.

My bad, I confused the two.

I didn't realize that OPL3 stuff is sensitive to speed of processors. I thought I could do that on a pentium 133, 166 and 200 and I also have 486 and 386 computers.

Now what? I was also planning to use OPL3 on a PIII 800, means I cannot do it?

Cheers,

pentiumspeed wrote on 2022-06-24, 14:30:

I didn't realize that OPL3 stuff is sensitive to speed of processors. I thought I could do that on a pentium 133, 166 and 200 and I also have 486 and 386 computers.

Now what? I was also planning to use OPL3 on a PIII 800, means I cannot do it?

Cheers,

You can but the faster you go the likelier it becomes that software and hardware compatibility breaks. You'll have to spend time checking what works with what. I think Audicians and AWE64s are pretty resilient to that class of processors and can go over 1ghz.

Joseph_Joestar wrote on 2022-06-24, 11:49:

So I got the same result on both cards. This leads me to believe that what you're experiencing might not be just an OPL3 issue, though that could be a part of it. It's also possible that Creative's OPL3 cards are more speed sensitive than OPTi's solution.

Or maybe P166 MMX is too fast even for CQM. P166MMX is a lot faster than 486 @ 100. And I think chipset also matters because even during my tests the WT Intel was too "fast" (although not faster than WB Intel) with just L1 cache disabled so chipset matters also. Or at least ISA timings etc.

I did some more tests, this time with Yamaha YMF-718 (Audician), ES1868F (Terratec Gold 16/96) and SB16 CT2290 with integrated real Yamaha OPL.

YMF-718 and CT2290 were just like all the other real OPLs. No difference. ES1868F was just like ES1688F. It too required the "a" parameter with Indy, no matter what speed was set.

I ordered MK8330 and I will test it as well when I get it. I wonder if it only sounds like a real Yamaha but is less speed sensitive or just a clone with speed sensitivity.

pentiumspeed wrote on 2022-06-24, 14:30:

Now what? I was also planning to use OPL3 on a PIII 800, means I cannot do it?

Most games have the necessary waits between data port writes so it will work correctly. But with P3 you are going to need slowdown anyways because of other speed related issues. The OPL speed issue seems to matter most with very old games written for Adlib. Funny, since Adlib has OPL2 and it is much more speed sensitive than OPL3. I think Cycles is meant for slow 286 systems so it doesn't have any OPL delays because code then just worked without.

I recapped the motherboard with Kemet polymer capacitors, well the 10µF caps are polymers but there were two 1µF capacitors (on some earlier post I wrote one 1µF but there were two) and there was also one 4.7µF capacitor between ISA slots. For the 1µF and 4.7µF I just used electrolytics that I happened to have. I wanted to replace the tantalums before they explode and polymers are supposed to be best alternatives for tantalums. I also replaced the single 330µF electrolytic capacitor this motherboard has. Well, everything works but they didn't all fit nicely. I intentionally bend the 1µF capacitor on the left side of heatsink. Even the original tantalum was placed so that it blocked the heat sink clip on the socket. Now it is not a problem anymore. The other 1µF on the right side of the socket is bend a little because it didn't fit otherwise.

I did more retrobright stuff. Here is front of the case and here are the failures closer. And I should have removed the Pinus logo.

I did try to remove the Pinus logo. But it didn't come loose and I didn't want to use force. I thought it was plastic so I just taped it so it wouldn't get affected. But no, it was not plastic. Looks to be copper covered with something shiny. Brass perhaps? Well it did start to corrode. I'm glad I had tape over it. Otherwise it would be too far gone. Now I was able to clean it with IPA and try to fix it with gold marker. But it was too dry so I'll have to get another one.

But the other mistakes. The orange buttons now need to be repainted. And I didn't even like the orange originally but now I think they should be restored. Next the disk drive. It is now too white and the color is uneven. I suppose it is better than badly yellowed ones but... I don't know why it got that bad. I did check it many times and it looked like it is not good enough. Even when I washed it it looked good, actually. But after I left it to dry, this was the result.

And same almost happened to the case front bezel. If you look carefully, there are couple of small white dots there. And one small piece over the 5,25" drive bays is completely white like the disk drive. That small piece was the most yellow to begin with and it became white a lot sooner than the rest of the case was ready. Picture lies a bit. It may look like that it should be little more white but that is not the case. When compared to the painted metal parts of the case, this front bezel is a little bit too white now. But only a very little so it does not matter. I was very lucky with this one. A little more and it might have looked like the disk drive. I'll paint that small white part. I'll need to get paint anyway since I'm going to make new 1.8"HDD-CF-adapter that also includes McCake display and buttons. And since I now need to repaint buttons as well, those McCake buttons are going to match the colors of other buttons.

Finally the combined 1.8" HDD + CF + McCake drive is ready! More details later.

Here are more pictures of the new McCake drive and new HDD-CF-drive. The 1.8" HDD to IDE adapter I now used is better than the old one (I think those are in the Copam thread). The previous adapters have issue with HDD led. It only works when CF card is connected (slave drive). These new adapters are much easier to mount and the improved design is easier to assemble.

I'm quite happy with painting as well. It is not quite correct color but it is actually very close to the painted metal parts of the case. I even considered repainting the case. I bought furniture paint with color code RAL 1013 (oyster white). That was not correct with anything. It was too white. So I made my own mix with whatever leftover paints I had. Color is quite good now but I don't know the color code.

Orange buttons are repainted with gray primer and orange spray paints. The orange color is more red than original but I like this one better. I used the same paint for McCake buttons and I wanted them to look a bit like the reset and turbo buttons. Buttons could have been thicker. I used ice scream sticks as buttons but I should have glued two sticks together. But these work and look nice so no worry. There are some minor glue residues on the display and the display would have been nicer if it would be red or orange. But I'm quite happy with end result.

I did need to remove 3.5" and 5.25" disk drives and replace them with combo drive. 8x Mitsumi CD-ROM-drive is now in use. It is quiet but could be even more quiet so I think 4x or 6x drive would be even better.

That is so beautiful, congrats!

Nice! I've always had LED MHZ displays for my 386/486 machines although the fastest I ever owned was a DX2/66. I owned a Cirrus Logic 5426 VLB card w/1MB ram for my 486. It wasn't probably the best/fastest but it had VESA extensions built-in to its BIOS so I never needed Sci-tech occupying any of my precious DOS RAM. Compushow detected all the proper resolutions as did any other software that required VESA. Were there any other video cards of that era that had VESA BIOS?

Norton Commander wrote on 2022-07-18, 15:56:

Were there any other video cards of that era that had VESA BIOS?

VESA up to 1.2 is common with later ISA and VLB video cards. I don't know if any has 2.0 support. Likely not as there are so many PCI cards without 2.0 support.

MK8330 arrived couple of days ago and as promised, I tested how speed sensitive OPL it has. And it turns out to be identical to real Yamaha OPL3.

Here are some new pictures now that the system is complete (for now).

Cards from top to bottom:
Roland MPU-401AT + McCake
Lo-tech ISA CompactFlash Adapter (for XTIDE Universal BIOS that can be reflashed. 3COM NIC could also hold the ROM but not flash it)
* VLB multi-I/O card with Vision QD6580 IDE controller
* VLB #9 GXE 64 with S3 Vision864
Roland SCC-1A
3Com Etherlink III 3C509B
GUS Extreme

* means original card for this system.

Notice the Corsair 650W ATX PSU. It has been modified to AT PSU with -5V. I used it when I did all the benchmarking previously and it works just fine. I did have to make a bit hole to the case under the 120mm PSU fan.

But what CPU is there now? I decided to go with WB Intel DX4 and it is clocked to 100 MHz so no overclocking. Why this? First I don't want to overclock. Sure, this CPU would be quite fast at 120 MHz but this already runs everything 486 needs to so no need for the extra speed. And I don't like the idea of 40 MHz VLB even thought there were no issues at all with it. Also, since this computer has always used DX4-100 (Intel WT) I didn't really like the idea to use Cyrix (but I was tempted). And the Intel WB is the only CPU that had only pros and no cons compared to the original CPU. This is a bit faster and never slower, slightly lesser power consumption and no need to disable both turbo and L1 cache for cycles. Disabling L1 alone is enough (for Yamaha OPL3. Now that I have ESFM, nothing needs to be done or games that otherwise need slowdown, deturbo 1/3 alone should be enough).

Why is there SCC-1A? Well, no good reason. McCake can do the GM/GS stuff as well and SCC-1A is set to port 332h and IRQ3 to prevent conflicts. Most likely I will later move the SCC-1A to another system. It's been many weeks since I've last coded MIDIto and I don't know when I'll have time to do so. But next version should have option to relocate MIDI IRQ so I will be testing that with the SCC-1A.

Speaking of conflicts, there was a MPU-401 conflict with GUS Extreme.

I really like the CF drive and McCake enclosure. Looking good!

aitotat wrote on 2022-07-16, 14:52:

IMG_20220716_174852_(2250_x_3000_pixel).jpg

Finally the combined 1.8" HDD + CF + McCake drive is ready! More details later.

That’s really cool!