Came across another couple of pieces of my (retro) computing history today, connected to the Amstrad PC2386/65 I once owned.

Firstly, I found the paperwork for it - a base unit with the 12" colour monitor (order form dated 31st May 1989 - boy that was a lot of money!!!, guarantee card and the on-site upgrade invoice from when they retrofitted the WD1006 controller). Secondly, I found the install disks for it - a copy of MSDOS 4.01 and Windows/386 2.1x, so I may throw together some old kit and see if I can get them installed and working.

Toblerone wrote:

Today I installed Hercules Stingray 128 / 3D from the 3Dfx Voodoo Rush family and flashed a newer bios into it. Tried an absurd number of reference drivers and played the first level of Tomb Raider countless time trying to iron out the kinks. Getting that son-of-a-gun working properly was a CHORE. Finally the manufacturer's drivers which I happened to stumble upon were the best fit.

Could you either upload or post a link to the drivers you used? I have one of these cards too and when I tested it it worked in some games but didn't work in others. I know the Rush cards weren't great, but I had hoped for better compatibility than I was getting.

For my own retro activity today, I just tested a Radeon 9600 (OEM passively cooled) AGP card I had laying around and it started artifacting rather quickly in basic DX9 tests. I touched the heatsink and it was barely even warm, so I figured it was a contact problem. I shut the system down, removed the heatsink, and the dried paste looked like it probably wasn't contacting the core at all any more. I cleaned it off a bit with Articlean and applied some Arctic Alumina (its decent and cheap so I have tons of it) and reattached the heatsink. I ran the same tests as before (3 times), 3DMark2003 and other tests and it works flawlessly. Sweet! 😀

Ended up discovering something extremely useful -

Played about some more with the setup for my newly acquired CGA graphics card for my XT build. It wouldn't output in colour correctly on either of my LCD TV sets (flickering & horizontal distortion), but would just about work on the Plasma TV albeit with occasional image distortion and image jumping & scrolling. (I'm in PAL land here just for reference.)

I wasn't prepared to settle for that and I don't own any CRT TVs nor any Composite capable monitors. So I thought about trying something I did have an abundance of - old VHS VCRs that claim to be NTSC 4.43 capable.

I wasn't sure it would work, but sure enough the first VCR I pulled from storage - plugged in the composite RCA cable to the VCR and connected the VCR to the TV via SCART and voila it worked. Perfectly stable image and in colour! Sure, the colours are wrong by default: the Plasma was able to compensate 100% with the Tint control and looked amazing, matching screenshots online. The LCD TV's colours are a bit off, even with Tint set to max, but probably par for the course with LCD in general if I'm not mistaken? Still, I can settle for those colours as the smaller LCD is more practical than a 42" Plasma. 🤣

Screenshot of the 22" Widescreen LCD Monitor/TV I'm currently testing with:

For fun and reference purposes at some point this week, I have another 4 VCRs that I'll test the same procedure with. All models from the late 1990s/early 2000s.

The VCR I'm testing with in the screenshot above was a Sanyo VHR-899 E, bought in the late 1990s.

I've noted that the Power Mac 9600 kept having the Finder crash and restart randomly when just idling, so I decided to start taking a closer look at the DIMMs. Two of the six it came with were just plain dead and not showing in System Profiler, which wasn't reassuring.

Now, since this thing supports interleaved memory, I had two in slots A1 and A2, and two more in slots B1 and B2. It would pass POST just fine, good chime. But then I thought "Hey, what if I moved all of these to the A slots, non-interleaved?"

It didn't chime after that - neither a good power-on chime nor a Chime of Death. I started pulling DIMMs 'til I found the culprit and it started chiming good again, then added some of the others I pulled in the process. Seems like I have only three good DIMMs here.

Unfortunately, that puts a wrench into running in interleaved mode due to the odd amount. I only have 192 MB now in the form of three 64-pin DIMMs, and I can't go swapping with the Power Mac 6500 I used to have because I sold that at VCFSE right after getting this 9600. Just goes to show why you shouldn't rush to replace a known good system with an untested one!

At least they still make 168-pin 5V EDO DIMMs for me to cram into this thing. I wish I didn't have some sizeable expenses coming up, because all this retro computer restoration is putting a massive dent into my wallet.

Also, I crammed an AGP Radeon 9000 into this MDD2003 Power Mac G4 that I'm restoring; no issues there, thankfully. All that remains with that system is the need for a new power supply, or a recap of the existing Samsung one. I already see a blown fuse and a bulging big CapXon in there, who knows what else has gone bad on it...

if you need/want some edo dimms, let me know. I have a box full of the smaller ones.

luckybob wrote:

if you need/want some edo dimms, let me know. I have a box full of the smaller ones.

Smaller than 64 MB, which is the lowest density I'd want to bother with in a 9600? Heck, if I could get a full set of twelve 128 MB DIMMs, that'd max it out nicely, but the cost is rather impractical compared to just using my MDD, which is also much, much more stable in general.

Even after plucking out the bad DIMM and even a good DIMM to bring it down to 128 MB interleaved, I still can't do things like add an IDE hard drive as a slave to the DVD drive that's already in there, which still locks up the machine while it's searching for System Folders to boot from. Or, for that matter, copy over files to my FireWire drive enclosure without suddenly halting midway, which is quite obvious when said enclosure's drive activity light isn't lit. Mouse still moves around, but nothing else's happening and pressing the power button on the front won't shut it off. Gotta pull the plug.

This must be what owning some vintage exotic European sports car is like: wicked fast, but also nowhere near reliable enough to use as a daily driver.

I would run 192 MB non-interleaved, but the performance penalty for doing so is fairly significant. I suppose further testing will reveal if that's what I ultimately have to do in order to make this 9600 stable.

That said, smaller DIMMs of the FPM variety would've come in handy if I had bothered to take home some of those DOS compatibility cards back at VCFSE, but I had already hauled enough hardware home and spent too much money as it was. (Note to prospective VCF attendees: save up a few hundred bucks beforehand, you just might need it!)

UPDATE: No improvement in stability with the 128 MB interleaved setup, I'm afraid. This thing still has a tendency to freeze and crash at random, including during the startup process. I'm going 192 MB non-interleaved for now, eating the performance penalty and seeing what happens.

Ozzuneoj wrote:

Toblerone wrote:

Today I installed Hercules Stingray 128 / 3D from the 3Dfx Voodoo Rush family and flashed a newer bios into it. Tried an absurd number of reference drivers and played the first level of Tomb Raider countless time trying to iron out the kinks. Getting that son-of-a-gun working properly was a CHORE. Finally the manufacturer's drivers which I happened to stumble upon were the best fit.

Could you either upload or post a link to the drivers you used? I have one of these cards too and when I tested it it worked in some games but didn't work in others. I know the Rush cards weren't great, but I had hoped for better compatibility than I was getting.

For my own retro activity today, I just tested a Radeon 9600 (OEM passively cooled) AGP card I had laying around and it started artifacting rather quickly in basic DX9 tests. I touched the heatsink and it was barely even warm, so I figured it was a contact problem. I shut the system down, removed the heatsink, and the dried paste looked like it probably wasn't contacting the core at all any more. I cleaned it off a bit with Articlean and applied some Arctic Alumina (its decent and cheap so I have tons of it) and reattached the heatsink. I ran the same tests as before (3 times), 3DMark2003 and other tests and it works flawlessly. Sweet! 😀

The manufacturer's driver I am now using is s128w133.exe, which I downloaded from http://driverzone.com/drivers/hercules/stingray/ website. I haven't really mapped out the compatibility with them yet, as I have only tried out Tomb Raider and Screamer Rally. Both worked though. What games you had compatibility issues with? I could try them as well out of curiosity. The newest manufacturer's drivers as far as I know are s3d-9x-301s.exe for the single board model. But for now I am sticking with s128w133.

Setting up a Socket 478 testrig, in order to test two identical TNT2-Ultra 32mb cards. So far both cards have posted, so the next step is to do a Win98 install and run 3D mark 99.

Improving DOSBox host compatibility with old operating systems
DOSBox Compilation Guides

Research has been fun and found out some things that can potentially improve other projects.

Mission Impossible. Looking for additional information and especially software and drivers for this nice multimedia machine, based on Core2, media center desktop factor, as you can see from the photos. The recovery partition is not present, for now I'm running AntiX Linux, but I have some problems with the audio section. Cool and quiet torrent box.

The back of the machine is literally FULL of connections, like tuners, radios, audio in all shapes, etc.

The model is an ACER Idea D510. There seems to be no more driver lost in the vastness of the Net, and I could read that by default it came with Windows Vista on board. Soon I'll try to use external audio via a Live! 24bit USB.

PTherapist wrote:

I wasn't sure it would work, but sure enough the first VCR I pulled from storage - plugged in the composite RCA cable to the VCR and connected the VCR to the TV via SCART and voila it worked. Perfectly stable image and in colour!

The VCR I'm testing with in the screenshot above was a Sanyo VHR-899 E, bought in the late 1990s.

I remember that our Sony VCR had a switch somewhere to allow playback of NTSC cassettes on PAL TV.
I think I had to use it only once, but it worked.

Wohoo..... :-p Both TNT2's went through 3Dmark99 without any glitch. Just a little over 6000 marks in 640x480, drivers set to max image quality. These Compaq TNT2's never let me down.

Speaking of Compaqs, upgraded my Presario 5000 from a 700MHz Duron to a 1.1GHz Thunderbird. Runs well!

brostenen wrote:

Wohoo..... :-p Both TNT2's went through 3Dmark99 without any glitch. Just a little over 6000 marks in 640x480, drivers set to max image quality. These Compaq TNT2's never let me down.

There manufactured by Creative for Compaq/Gateway.

I'm actually debating which to put into my Gateway GP7-800 right now.

Option 1 is a Creative/Gateway TNT2 Ultra 32MB, which would be fastest in Direct3D and OpenGL. It's one of those passively cooled ones with the huge black heatsink.
Option 2 is an STB 3DFX Voodoo3 3000 AGP, which would be the fastest at GLIDE.
Option 3 is a Ensoniq 3DFX Voodoo Banshee AGP, which would have the widest range of support for GLIDE titles.

I'm leaning towards the Bansee ATM because the game I'm building this system to play is Half Life and it's the best fit for the time frame plus I haven't used the Banshee very much at all since I purchased it. Plus I know Half Life's MiniGL driver support it.

TheAbandonwareGuy wrote:

There manufactured by Creative for Compaq/Gateway. […]
Show full quote

brostenen wrote:

Wohoo..... :-p Both TNT2's went through 3Dmark99 without any glitch. Just a little over 6000 marks in 640x480, drivers set to max image quality. These Compaq TNT2's never let me down.

There manufactured by Creative for Compaq/Gateway.

I'm actually debating which to put into my Gateway GP7-800 right now.

Option 1 is a Creative/Gateway TNT2 Ultra 32MB, which would be fastest in Direct3D and OpenGL. It's one of those passively cooled ones with the huge black heatsink.
Option 2 is an STB 3DFX Voodoo3 3000 AGP, which would be the fastest at GLIDE.
Option 3 is a Ensoniq 3DFX Voodoo Banshee AGP, which would have the widest range of support for GLIDE titles.

I'm leaning towards the Bansee ATM because the game I'm building this system to play is Half Life and it's the best fit for the time frame plus I haven't used the Banshee very much at all since I purchased it. Plus I know Half Life's MiniGL driver support it.

Yup. They are basically creative cards. Yet they are special, as they have a non standard video output alongside s-video and standard vga.

They feel more premium than a standard noname production. Thicker pcb and heavy too. Going to put one of them up for sale locally, so I will check out sold items on ebay, and set the price a bit lower.

Regarding the cards you are choosing between. I would say the Voodoo3 card.

Not today, but actually yesterday.. It was the first really sunny day of May, so I decided it was time to try my hand at retrobrighting for the first time, on my most 'grandma's rotten teeth' yellow plastic stuff:

The results were not bad for a first go:

There is some streaking on the front bezel, I noticed I had bubbles under the streatch wrap too late - the salon cream had evaporated.. I quickly renewed the coat and re-filmed the thing, which helped alleviate the situation a bit but did not fix it 100%. Still, it's a million times better than before, and a second attempt would likely fix the issue (no, I won't bother with it).

Azarien wrote:

I remember that our Sony VCR had a switch somewhere to allow playback of NTSC cassettes on PAL TV.
I think I had to use it only once, but it worked.

Yeah I remember having one with that too, though most were automatic. I only ever owned a couple of NTSC videos, as by the time I had enough disposable income to start importing them, DVD had really taken off with the prices becoming reasonable, so I switched to DVD instead.

Practically every 1 of my 90s/2000s VHS machines can play the tapes via NTSC 4.43 so I figured if they can do that then they might just accept a regular NTSC 3.58MHz input and it works a treat as the VCRs are acting to stabilise the image from a CGA Composite output for modern TVs.

Testing time again today:

- Went through a pile of IDE disks, with only two 40GB Maxtors (the least interesting in the lot) dead. The 6.4GB Seagate was working fine, which was the main thing.

- Also some soldering: fixing caps on a motherboard and GPU. Both working now - an MSI MS-6168 Ver2 (uATX i440BX with onboard Voodoo3) and a Leadtek Winfast A170 DDR Gf4MX (with nice but noisy heatsink-fan). The MS-6168 somehow doesn't support CuMine CPUs despite an identical board accepting them fine with the exact same BIOS (opened a separate thread for that), but it's happy enough with a P3-500 Katmai.

- Tested my latest CPU, a Covington cacheless Celeron 266. Even BIOS POST is noticeably slow on that thing. But it works.

- Unfortunately my Intel MU440EX (uATX i440EX with onboard Rage Pro and Yamaha 740) is dead, with bad corrosion around and on the - soldered, non-standard - Intel BIOS flash ROM. Odd place far away from the battery, but wherever it came from it's dead now :'(

- Also messed around with some sound cards. My CT3670 (SB32 RAM-less AWE with SIMM slots) works as does my CT4500 (AWE64), although the games I tried didn't like my K6-3+ system at all. Laser Squad insisted that no sound blaster could be detected, even with correct SET BLASTER and explicitly feeding it the address and IRQ. Ascendancy detected the cards fine but hung after a few minutes the first time I ran it, subsequently crashed during startup with what sounded like a C error about not being able to assert planets.somethingorother. And just for good measure Civ1 had no problem with sound, but crashed with divide by zero error due to CPU being too fast.

Oh well, enough for now. High time for bed.

dionb wrote:

Testing time again today: […]
Show full quote

Testing time again today:

- Went through a pile of IDE disks, with only two 40GB Maxtors (the least interesting in the lot) dead. The 6.4GB Seagate was working fine, which was the main thing.

- Also some soldering: fixing caps on a motherboard and GPU. Both working now - an MSI MS-6168 Ver2 (uATX i440BX with onboard Voodoo3) and a Leadtek Winfast A170 DDR Gf4MX (with nice but noisy heatsink-fan). The MS-6168 somehow doesn't support CuMine CPUs despite an identical board accepting them fine with the exact same BIOS (opened a separate thread for that), but it's happy enough with a P3-500 Katmai.

- Tested my latest CPU, a Covington cacheless Celeron 266. Even BIOS POST is noticeably slow on that thing. But it works.

- Unfortunately my Intel MU440EX (uATX i440EX with onboard Rage Pro and Yamaha 740) is dead, with bad corrosion around and on the - soldered, non-standard - Intel BIOS flash ROM. Odd place far away from the battery, but wherever it came from it's dead now :'(

- Also messed around with some sound cards. My CT3670 (SB32 RAM-less AWE with SIMM slots) works as does my CT4500 (AWE64), although the games I tried didn't like my K6-3+ system at all. Laser Squad insisted that no sound blaster could be detected, even with correct SET BLASTER and explicitly feeding it the address and IRQ. Ascendancy detected the cards fine but hung after a few minutes the first time I ran it, subsequently crashed during startup with what sounded like a C error about not being able to assert planets.somethingorother. And just for good measure Civ1 had no problem with sound, but crashed with divide by zero error due to CPU being too fast.

Oh well, enough for now. High time for bed.

Don't copper mines use a different voltage setting? I would think some sort of board revision that doesn't support it or a bad capacitor or something causing it to be too far out of specification to be the most likely cause.

TheAbandonwareGuy wrote:

Don't copper mines use a different voltage setting? I would think some sort of board revision that doesn't support it or a bad capacitor or something causing it to be too far out of specification to be the most likely cause.

Yes, they run on 1.6-1.75V as opposed to 1.9V-2.1V. There is an earlier revision of this board (Rev:1) that does not support them, but the Rev:2 definitely should, and all VRM-related components are identical down to IC or transistor part number. It also doesn't really look like a marginal/failing voltage issue; the error code is very reproducible and stable and behaviour is identical between a P3-600 and a P3-1000 that draws almost 50% more current.

I'm open to the idea it's something failing though, but question is where to start looking...