I've been having a play with Speedsys and managed to get my overall memory throughput up from ~80MB/s to around 111MB/s. It seems stable, it'll run three iterations of 3DMark 99 time after time. Oh, and I've also been playing Quake 2. It's runs perfectly but I had to record a 1 second silent .wav file and use the Quake PAK tool to replace the 'Computer updated!' sound that plays EVERY 3 seconds and was extremely annoying😃.

Some Pentium III Coppermine 1000 overclocking with MSI 815E Pro and ATI Radeon 8500 retail. Actually I didn't find the CPU max and hit the motherboard's roof. 7.5x166 is the max I can get from BIOS and it works without issues. I would guess this wont divide PCI/AGP clocks beyond 133. Now I understand why this MB had clock-mod when I bought it about a year ago (some in the picture below). It's been used in some serious overcloking by previous owner, nevertheless works like perfect. I removed all mods and tweaks, I like it more as original.

3DMark2001SE
7.5x133: 6117
7.5x166: 7427
+Radeon 8500 oc 300/300: 7618

This PIII really likes higher FSB and performance increase is significant. Too bad I could not find the max this time.

Is there any part of that IC being used?

Yesterday I was able to repair an old 256k OAK VGA/EGA card - turns out one of the RAM chips were bad. It was, of course, the last one I checked. The same computer's floppy failed, just last week I cleaned the heads and re-greased the sliders. Suddenly, general failure warnings. I tried to move the heads around, but it doesn't even attempt to read past the first track. Heads are probably gone. Into the pile it goes, maybe if I get lucky I'll find another one with some other failure and get this one going again.

As expected from my previous post today was the theme of reviving the ASUS A8V motherboard from its USB Over Current bios bug that was preventing it from functioning.
I still do not fully understand what causes the error, but i know i have a pretty high chance of triggering it after a cmos reset as i needed multiple cmos resets in a row to tame the board while some worked fine in between and likely stayed working if i had not tested that cmos software again. The situation was as follows, you reset the cmos (with a tool or physically on the board with a jumper) and either it will work completely fine and stay that way until something triggers a cmos reset. Or, it boots up but with severe drive access defects where either the internal drive and a usb stick with MS-Dos 7.1 would be unable to access any files and then one reboot later it would be all screwed up again with the USB Over Current error preventing a boot.

So a few cmos resets later i finally had a working MS-Dos 7.1 session ready to flash firmware 2.19, and this has tackled the issue. I booted up Windows XP aaaand.... frozen. Its a multi-boot time machine project, so plenty of other things to try. So i tried Windows 7 which of cou... frozen again. Wonder what 98 will do, alright 98 boots and is updating the drivers to fit the new bios aaand frozen! This was not a good sight and got me a little nervous. Had the error been correct all along and did something die on my board? Perhaps this bios was not as stable and compatible as i thought and i'd have to choose the least of two evils. Many reboots later i did get a working XP session, so time to test the sound card as i had to remove it to get to the cmos battery. Instant bluescreen.

So this could not be a coincidence, the sound card on use crashed, and my system only freezes when the drivers load with no working audio. I did have to remove it from the system briefly after all. One reseat of the sound card later it was testing time again. And this time....? Fully stable. My overclock works, my bios is stable, no weird things happen when i reset the cmos. The USB ports are working fine again. The stability has returned to this build.

Lessons learned? If you dare to play with your cmos or bios settings to much your going to have a day full of trouble shooting if it goes wrong even if it seemingly has no reason to go wrong. And if you can avoid it do not update your ASUS A8V past 2.19 because this version is definitely way more stable now that the bad check is no longer part of the system.

With the case open i also wanted to try a dual GPU experiment since the theme of this build is trying to cover as many operating systems as i can get away with on the hardware. Windows 3.11 works reasonably well but the NVidia 6800GT has no working driver for it that isn't a buggy generic svga driver. I still have an unused S3 Trio64 so i added that to the build to see what would happen. Could i reliably switch in between? Turns out, no. Windows 98 absolutely hates this and while i kind off got it working i really disliked the instability this caused with the two display drivers conflicting. I was not going to sacrifice the stability of all the other operating systems to maybe improve Windows 3.11.

So the S3 is back in the box waiting for a project i can use it in or a time i need it for debugging. My retro PC is stable once again, and will be getting an extra SSD in it soon for when i want to run modern operating systems to really see what it can do in terms of performance (While the main harddrive remains the period correct western digital drive with the multiboot solution, i am aiming for period correct with some extra hidden touches).

Trying to fit Zida 4DPS into ITX case. 3D printing bunch of adapters in progress. Bottom of the case was originally used for SFX PSU. Todo: fitting laptop CD and FDD drives and MHz display in space meant for 5,25" DVD drive (there is a spring loaded "DVD" flap in the front), making custom short cables and 3D printed backplate for IO... There is also a USB flap on the top, i want to put CF card reader in there.
Voodoo 2 for scale (too big to fit inside 😁).

henk717 wrote on 2021-04-11, 18:25:

Lessons learned? If you dare to play with your cmos or bios settings to much your going to have a day full of trouble shooting if it goes wrong even if it seemingly has no reason to go wrong.

Sometimes, the admirable drive to play with old frail hardware can result in a circumstance where you thought you were carefully arranging a soft blanket, but actually you were poking and prodding at a sleeping bear. You know??? 😀

Coaxing the bear back to sleep, can result in quite a longer, more frustrating day than was originally anticipated!

Personally, I have started writing a text file with all changes, and testing one or two changes at a time. That way, if things go south, I know what works and what doesn't. Of course, there are some changes that will work almost every time, so I can change them first. When overclocking, I never change too much in one session.

fool wrote on 2021-04-11, 17:47:

Some Pentium III Coppermine 1000 overclocking with MSI 815E Pro and ATI Radeon 8500 retail. Actually I didn't find the CPU max a […]
Show full quote

Some Pentium III Coppermine 1000 overclocking with MSI 815E Pro and ATI Radeon 8500 retail. Actually I didn't find the CPU max and hit the motherboard's roof. 7.5x166 is the max I can get from BIOS and it works without issues. I would guess this wont divide PCI/AGP clocks beyond 133. Now I understand why this MB had clock-mod when I bought it about a year ago (some in the picture below). It's been used in some serious overcloking by previous owner, nevertheless works like perfect. I removed all mods and tweaks, I like it more as original.

3DMark2001SE
7.5x133: 6117
7.5x166: 7427
+Radeon 8500 oc 300/300: 7618

This PIII really likes higher FSB and performance increase is significant. Too bad I could not find the max this time.

7.5x166 seems like a pretty good overclock on a Coppermine!

Finally got around to replacing the hinges in my IBM P70 today... I'd previously tried epoxying them, which worked sort of ok for a while, but I could tell it wasn't holding anymore the last time I tried using the system as the screen had gone crooked again. This is a practically universal problem with these systems, but I found someone selling 3D printed hinges on Ebay (probably someone here, I'm guessing) and they seem to work fine. My screen's at least straight again now, and it moves the way it should, not like someone with a hyperextended knee. Now I just need to replace all the other plastic bits behind the screen somehow, since it no longer locks into place. All the plastic clips have just gotten loose over the years. One of the bottom ones has metal reinforcement so it's still ok, but the other side doesn't, and the top clip is just a single tiny piece of plastic on only one side meant to hold against two fairly strong springs (which are now working properly). After 30 years, it just can't do it anymore. The whole thing is really just one of IBM's worst designs ever, and it's unfortunately in one of their most interesting and unique computers.

I did fire it up and bathe in the warm glowing warming glow of its orange gas plasma screen, though. Really one of my favorite screens on any computer ever.

adalbert wrote on 2021-04-11, 22:25:

Trying to fit Zida 4DPS into ITX case. 3D printing bunch of adapters in progress. Bottom of the case was originally used for SFX PSU. Todo: fitting laptop CD and FDD drives and MHz display in space meant for 5,25" DVD drive (there is a spring loaded "DVD" flap in the front), making custom short cables and 3D printed backplate for IO... There is also a USB flap on the top, i want to put CF card reader in there.
Voodoo 2 for scale (too big to fit inside 😁).

Thats a Chiftec BT-02B if my eyes are not deceiving me!
Lovely little chassi, and way ahead of its time when launched.

A relative still uses my old chassi.
Now with an old i7-2700K, a tiny Quadro GPU, a slim DVD, 2 laptop HDD's at top and a large 2.5" SSD at the bottom.
And ofc the PSU has been replaced.
Tight fit, but it works =), I belive some of the scars on my fingers are from that very build!

bofh.fromhell wrote on 2021-04-12, 10:11:

Thats a Chiftec BT-02B if my eyes are not deceiving me!
Lovely little chassi, and way ahead of its time when launched.

Yes, that's it 😀 I removed PSU and will go with a Pico PSU for this mainboard, that way it is possible to fit some baby AT and micro ATX boards with 4 expansion slots in there 😀

I set up a RetroPie on a spare Pi 3B+ I had and my wife and daughter ecstatically played Rad Racer (NES) and Ms Pac-Man (NES) all afternoon long. 😀

Finished setting up TLS 1.2 and removing outdated/insecure protocols from RetroZilla + ading a user script to help with site rendinger. I'm kind of shocked how fast it is now in WIndows 95 on my Versa M/75 486 DX4 after all the tuning + facebook and other social media works. It's pretty insane. Everything has UniVBE now and Links 2.22 installed for DOS mode.

I also spent a lot of time tuning the M/75's 95 install, made the swapfile 120MB fixed size for now (though I could probably turn it off completley, that computer really does not churn the drive very much, kind of surprising for a 486 running 95, so the 40MB of RAM might help). changed the drivers for the HDD controller to IDE w/ Bus Mastering, seems that turned on DMA in the background because boot times went down from 1.5minutes to around 55 seconds. Used TweakUI to rip out more garbage. Now I'm sitting nice at 97% resources upon boot and aside from loading RetroZilla - everything's pretty snappy and even Pentium-era stuff is quite playable.

Bought 3 more ATA-133 drives for the Versa, got another Aironet card in (this time it works, and came with an ANTENNA!). I want to build the same multi-boot setup for the P/75, but with Windows 98 SE using 98Lite. I'm doing this in part for my wife who wants to play with AIM Phoenix, Retrozilla, and some others in Windows 9x like she used to. I might also see if I can tune The Sims to run on that machine. I had The Sims on it once and it was only a few frames away from comfortably playable - I think my tuning from the M/75 above mixed with the UniVBE driver might make it more comfortable. The PCMCIA Door fell off, so I might be putting the Aironet card with antenna in that computer with a velcro mount for the antenna - and use that WiFi manager that adds WPA compatibility to the laptop using that card.

I'm looking at buying a battery for the 40EC and V/50 next and getting those all WiFi'd up. Right now the V/50 is running FreeDOS and hte 40EC is running Windows. I'm thinking the next 40EC incarnation will have DOS 7.1 with WFW311 on it - and the V/50 will have another ginormous setup of FreeDOS on it using my new menu system implimentation.

Thermalwrong wrote on 2021-04-09, 00:03:

~snip~

Nice, always love seeing another classic Toshiba see the light of day.
My T4400 has a nice display, though it takes a little to "warm up." Once it does the colors are nice, sharp and the viewing angle is nice.

The floppy drive also works, but if I stick the wrong floppy in there the belt starts slipping again. I should take care of that next time I go in there.
I only wish I could find a 32mb ram card for it. Because, theoretically, you could stuff a POD in there since it's socketed.
Unlike the newer 486 (and pentium 4900&4950) Toshibas that have soldered processors.

And a T4900, I wish I could find one. I'm only stuck with this 4800 that I don't have the time and money right now to fix up.

My SCSI disk came in:

https://www.ebay.ca/itm/IBM-74G7007-74G6984-D … ve/223815981443

This is going to be for my 486 EISA system. I spent a few hours figuring out why it wouldn't work out of the gate - first I had to set a jumper to get it to spin up, then it would not detect it per-se (as in, it would not load it 80h and assign it a letter). I tried changing negotiations, speeds, duplex, cables, active terminators... I tried it in the EISA system as well as my 233MMX, I thought the drive didn't survive the trip from the middle east. Then I decided to do a media analysis and I got this:

After decoding (and learning about all this stuff, at least enough to get through the error) the ASC ASCQ table I found online showed 31h/01h DZ RO B FORMAT COMMAND FAILED
from this site: https://www.t10.org/lists/asc-num.htm. But I wasn't doing a format. So why not do one and see what happens?

Well, after 30 mins of the HDD activity LED staying pretty much on, I got a *FORMAT COMPLETE* message from the BIOS. I then did a Media Analysis. It found one error somewhere along the line. I then rebooted, set its ID to 2, and used my main SCSI drive to boot 98SE, where I partitioned and formatted the drive. I then brought it downstairs to the EISA system and it booted - sort of. I forgot to set the active partition. THEN it booted.

Hurray! My error was 1) I assumed a partition table was in tact 2) I did not know (or simply forgot) that you need to have a valid format done to the drive before the BIOS attempts to mount it (which now that I type it out, it makes sense).

This is going to be a fun build.

Sweet, got my two Cyrix 66MHz 486 CPUs (5V this time) - can't wait until next week after finals to build some retro systems!

I like to live dangerously so I thought I would test a few things at once:

1. Biostar MB-8433UUD-A Rev 2.0 motherboard that would not POST due to a dead Dallas RTC, from which I desoldered the said RTC, soldered in a socket and socketed a DS12C887+ with a 2013 date code.
2. Untested IBM 5x86-100 (at 3x40MHz no less)
3. 4 sticks of untested 16MB FPM-72 memory
4. Untested PCI S3 Virge

Amazingly enough, they all worked.

Shockingly enough, I appear to have gotten started on my Super 486 build..

appiah4 wrote on 2021-04-13, 20:30:

I like to live dangerously so I thought I would test a few things at once: […]
Show full quote

I like to live dangerously so I thought I would test a few things at once:

1. Biostar MB-8433UUD-A Rev 2.0 motherboard that would not POST due to a dead Dallas RTC, from which I desoldered the said RTC, soldered in a socket and socketed a DS12C887+ with a 2013 date code.
2. Untested IBM 5x86-100 (at 3x40MHz no less)
3. 4 sticks of untested 16MB FPM-72 memory
4. Untested PCI S3 Virge

Amazingly enough, they all worked.

MB8433UUD-A Cx5x86-120.jpg

Shockingly enough, I appear to have gotten started on my Super 486 build..

One of those days when the stars align. Buy yourself a lottery ticket 😊

Yeah you likely wouldn't have such luck if it was (for example) gear from the early 2000's, imo!

If my past year's experiences with several such systems are any indication.

I'm now tossing up whether or not to bother going to the trouble of recapping Northwood and Prescott systems. Purely from a return point-of-view (i.e. what benefit or use I'll get out of the resulting computers) it's not really worth it, but from a hobbyist and disliker-of-creating-waste perspective, it's a Maybe.

appiah4 wrote on 2021-04-13, 20:30:

I like to live dangerously so I thought I would test a few things at once: […]
Show full quote

I like to live dangerously so I thought I would test a few things at once:

1. Biostar MB-8433UUD-A Rev 2.0 motherboard that would not POST due to a dead Dallas RTC, from which I desoldered the said RTC, soldered in a socket and socketed a DS12C887+ with a 2013 date code.
2. Untested IBM 5x86-100 (at 3x40MHz no less)
3. 4 sticks of untested 16MB FPM-72 memory
4. Untested PCI S3 Virge

Amazingly enough, they all worked.

MB8433UUD-A Cx5x86-120.jpg

Shockingly enough, I appear to have gotten started on my Super 486 build..

Shotgun approach. Pretty good considering the 25 year old hardware.