TheAbandonwareGuy wrote on 2023-03-18, 03:28:

The 7000 series cards are the last to use traditional pixel/vertex pipelines as opposed to a unified shader architecture. The 79 […]
Show full quote

Ozzuneoj wrote on 2023-03-17, 15:15:

TheAbandonwareGuy wrote on 2023-03-17, 15:03:

Well I just bought 3 Quadro 3500s (GeForce 7900GS) for $5 each. SHIPPED. Going to try to figure out how to flash them to GeForce cards.

Seller has at least 10 more: https://www.ebay.com/itm/266132993924

(I think this is one time an exception to the eBay link rule would fit? Pretty unambiguously a great deal and there's no need to fight over them with the quantities the seller has)

Wow... that's super cheap. Can you give me a potential use case for this? I know the 6000 series is the last to have any kind of "normal" support within Windows 9x. Is it possible to make this work? If not, what would be the benefit of running this over something newer? Any specific games\programs that prefer a DX9 level card that won't even be happy with an 8600GT, 9500GT or 9800GT?

The 7000 series cards are the last to use traditional pixel/vertex pipelines as opposed to a unified shader architecture. The 7900GS is the 5th fastest among them (only 7900 GTO, 7900GTX, 7900GX2 and 7950GT are fast. None of them by a large margin). You also have to keep in mind that as you start using newer drivers (which your forced to use newer drivers the newer of hardware you use) the more driver based compability breaks you see with older software. Around GeForce 8 you see NVIDIA driver start to balloon up in size and feature set. You can really see GeForce 7 as more of the end of an era at NVIDIA than anything else.

For $5 its worth stocking up on a few cards just to throw them cheap Pentium D/Core2 Windows XP machines. I own an actual 7900GS (along with almost every other 7000 series card) but 7000 series cards are all on borrowed time so I don't like to use mine in builds, I just keep them around for when I need benchmark data or some specific function of them for a test. I won't be pissed if a $5 Quadro dies on me.

I bought 2 for an SLI Pair, the third is to experiment with flashing them to 7900GSs so I can run them as true GeForce cards. Also I don't think there are as many protections to prevent going from Quadro to GeForce as the other way around. Quadros were massively more expensive, no one was buying them back then to use for gaming. That would be stupid. I think the main way NV prevented GF to Quadro flashing was locking out the high precision float and memory self checking (not ECC, different tech) that Quadros use.

No, I spent a few hours digging around and the G71 Quadro cards are hardware locked to prevent flashing, the 7900GS can "reportedly" be flashed to a Quadro though, but whats the point of that ..the 7900GS doesn't have the extra Quadro features unlocked on the silicon as nVidia was lasering them off during production. (I guess its possible it was Vbios lock)

That said, I don't see the point in flashing a Quadro FX3500 even if it was possible, there is no performance difference between the two cards (identical mem/core clocks, shaders, ROPs and TMUs) and the Quadro has the better feature set, even with production drivers it'll play any game you throw at it without skipping a beat.

Is there any advantage to the 7800GS over the 6000 series? I skipped the whole 7000 series, as I felt that the 6800 GT was good enough for everything I wanted to throw at it until the 8800 GT rolled around.

Kahenraz wrote on 2023-03-18, 04:37:

Is there any advantage to the 7800GS over the 6000 series? I skipped the whole 7000 series, as I felt that the 6800 GT was good enough for everything I wanted to throw at it until the 8800 GT rolled around.

Not really, there are a lot of other issues that would make you not want one but that also applies to the 8000 series cards.

TrashPanda wrote on 2023-03-18, 04:28:

TheAbandonwareGuy wrote on 2023-03-18, 03:28:

The 7000 series cards are the last to use traditional pixel/vertex pipelines as opposed to a unified shader architecture. The 79 […]
Show full quote

Ozzuneoj wrote on 2023-03-17, 15:15:

Wow... that's super cheap. Can you give me a potential use case for this? I know the 6000 series is the last to have any kind of "normal" support within Windows 9x. Is it possible to make this work? If not, what would be the benefit of running this over something newer? Any specific games\programs that prefer a DX9 level card that won't even be happy with an 8600GT, 9500GT or 9800GT?

The 7000 series cards are the last to use traditional pixel/vertex pipelines as opposed to a unified shader architecture. The 7900GS is the 5th fastest among them (only 7900 GTO, 7900GTX, 7900GX2 and 7950GT are fast. None of them by a large margin). You also have to keep in mind that as you start using newer drivers (which your forced to use newer drivers the newer of hardware you use) the more driver based compability breaks you see with older software. Around GeForce 8 you see NVIDIA driver start to balloon up in size and feature set. You can really see GeForce 7 as more of the end of an era at NVIDIA than anything else.

For $5 its worth stocking up on a few cards just to throw them cheap Pentium D/Core2 Windows XP machines. I own an actual 7900GS (along with almost every other 7000 series card) but 7000 series cards are all on borrowed time so I don't like to use mine in builds, I just keep them around for when I need benchmark data or some specific function of them for a test. I won't be pissed if a $5 Quadro dies on me.

I bought 2 for an SLI Pair, the third is to experiment with flashing them to 7900GSs so I can run them as true GeForce cards. Also I don't think there are as many protections to prevent going from Quadro to GeForce as the other way around. Quadros were massively more expensive, no one was buying them back then to use for gaming. That would be stupid. I think the main way NV prevented GF to Quadro flashing was locking out the high precision float and memory self checking (not ECC, different tech) that Quadros use.

No, I spent a few hours digging around and the G71 Quadro cards are hardware locked to prevent flashing, the 7900GS can "reportedly" be flashed to a Quadro though, but whats the point of that ..the 7900GS doesn't have the extra Quadro features unlocked on the silicon as nVidia was lasering them off during production. (I guess its possible it was Vbios lock)

That said, I don't see the point in flashing a Quadro FX3500 even if it was possible, there is no performance difference between the two cards (identical mem/core clocks, shaders, ROPs and TMUs) and the Quadro has the better feature set, even with production drivers it'll play any game you throw at it without skipping a beat.

Well that's stupid if so. First off what if they needed to issue a vBIOs update to fix an actual BIOs level issue? Second off what product were they protecting with the flashing block? Like I said the Quadros were the most expensive product they offered at the time. People buying them and flashing them into GeForce would have only meant they made MORE money.

IDK, I'll look into this more when the cards show up.

TheAbandonwareGuy wrote on 2023-03-18, 05:43:

TrashPanda wrote on 2023-03-18, 04:28:

TheAbandonwareGuy wrote on 2023-03-18, 03:28:

The 7000 series cards are the last to use traditional pixel/vertex pipelines as opposed to a unified shader architecture. The 7900GS is the 5th fastest among them (only 7900 GTO, 7900GTX, 7900GX2 and 7950GT are fast. None of them by a large margin). You also have to keep in mind that as you start using newer drivers (which your forced to use newer drivers the newer of hardware you use) the more driver based compability breaks you see with older software. Around GeForce 8 you see NVIDIA driver start to balloon up in size and feature set. You can really see GeForce 7 as more of the end of an era at NVIDIA than anything else.

For $5 its worth stocking up on a few cards just to throw them cheap Pentium D/Core2 Windows XP machines. I own an actual 7900GS (along with almost every other 7000 series card) but 7000 series cards are all on borrowed time so I don't like to use mine in builds, I just keep them around for when I need benchmark data or some specific function of them for a test. I won't be pissed if a $5 Quadro dies on me.

I bought 2 for an SLI Pair, the third is to experiment with flashing them to 7900GSs so I can run them as true GeForce cards. Also I don't think there are as many protections to prevent going from Quadro to GeForce as the other way around. Quadros were massively more expensive, no one was buying them back then to use for gaming. That would be stupid. I think the main way NV prevented GF to Quadro flashing was locking out the high precision float and memory self checking (not ECC, different tech) that Quadros use.

No, I spent a few hours digging around and the G71 Quadro cards are hardware locked to prevent flashing, the 7900GS can "reportedly" be flashed to a Quadro though, but whats the point of that ..the 7900GS doesn't have the extra Quadro features unlocked on the silicon as nVidia was lasering them off during production. (I guess its possible it was Vbios lock)

That said, I don't see the point in flashing a Quadro FX3500 even if it was possible, there is no performance difference between the two cards (identical mem/core clocks, shaders, ROPs and TMUs) and the Quadro has the better feature set, even with production drivers it'll play any game you throw at it without skipping a beat.

Well that's stupid if so. First off what if they needed to issue a vBIOs update to fix an actual BIOs level issue? Second off what product were they protecting with the flashing block? Like I said the Quadros were the most expensive product they offered at the time. People buying them and flashing them into GeForce would have only meant they made MORE money.

IDK, I'll look into this more when the cards show up.

nVidia still does this, I figure they have their reasons for it but it did at least change it so you can flash cards with their own signed bios updates but back then .. Vbios updates from nVidia were not a thing generally. (Now its simply an encrypted signed Vbios rather than locked and can only be flashed with a compatible signed Vbios)

IIRC the 6800 series was the first generation that had this Vbios locking feature, cards before G40 could be flashed with little issue, if I had to guess they made a corporate decision to prevent tampering of their Quadro cards and Geforce cards in general.

Asininity wrote on 2023-03-18, 04:09:

kixs wrote on 2023-03-17, 20:39:

Something complete :) […]
Show full quote

Something complete 😀

Terratec Wave System SCW001/2

It's always neat to have the full box. How does it sound? I wasn't able to find much on the particular board, but what I did seems like it could have an interesting sound.

Sounds fine by me 😉

Otherwise here is a small review:
http://www.amoretro.de/2014/07/terratec-waves … rboard-2mb.html

BitWrangler wrote on 2023-03-18, 03:52:

Kahenraz wrote on 2023-03-18, 01:27:

I have the same problem. I won't buy any keyboard that isn't a standard layout with a horizontal enter key. It's too confusing o […]
Show full quote

Ozzuneoj wrote on 2023-03-17, 18:08:

My issue is that my muscle memory for entering backslash commands in DOS is totally destroyed by anything without a "normal" horizontal enter key. I have no muscle memory for doing backslash on keyboards with a big fat enter key or an ISO shaped enter key. Really limits my desire to use retro keyboards on retro systems, as many of them have big huge enter keys and I have to slow down and think about every backslash. 🙁

I have the same problem. I won't buy any keyboard that isn't a standard layout with a horizontal enter key. It's too confusing otherwise.

Fun fact. During my childhood, our middle school filled the computer lab with brand new Compaq computers. Compaq did something really bizarre. They split the keyboard into two keys, one of the spacebar, and the other side as the backspace. So if you had already trained your muscle memory to use your thumb on the wrong side, every time you tried to hit the spacebar, you would backspace instead.

vintage-compaq-computer-keyboard_1_7250e36c11a542cc6f0fc48f4fa4f118.jpg

It doesn't phase me much because I've always been on multiple keyboard layouts for decades. Whether it was Amiga or UK to US or Canadian ML to US or whatever. I can use 5 or more slightly different keyboards a day sometimes.

I work just the opposite way around, if I need to type on a keyboard that's just a bit more narrow or wide than my usual, I have to retype most for the first couple of hours if anyone is to understand it 🤣

Bought two CPUs yesterday, one is actually an FPU, to upgrade a 386DX, I already have the same pair on another PC, but the CPU is Cyrix, I don't know if there is any difference between TI and Cyrix, when they arrive I will try doing benches, exchanging CPUs, we'll see how it goes.
I also took three IDE-CF adapters, and two CF cards of only 32MB, for these two memory cards, the use will be exclusively for the first DOS versions, and possibly for Windows 1.0X, to be used on PC 286 or 386SX.

PC@LIVE wrote on 2023-03-18, 09:02:

I also took three IDE-CF adapters, and two CF cards of only 32MB, for these two memory cards, the use will be exclusively for the first DOS versions, and possibly for Windows 1.0X, to be used on PC 286 or 386SX.

most likely pin 28 will be pulling ALE to ground and screwing with 286/386, needs modifying Re: "Fixed" 386sx motherboard works but not with 16-bit VGA card

rasz_pl wrote on 2023-03-18, 09:24:

PC@LIVE wrote on 2023-03-18, 09:02:

I also took three IDE-CF adapters, and two CF cards of only 32MB, for these two memory cards, the use will be exclusively for the first DOS versions, and possibly for Windows 1.0X, to be used on PC 286 or 386SX.

most likely pin 28 will be pulling ALE to ground and screwing with 286/386, needs modifying Re: "Fixed" 386sx motherboard works but not with 16-bit VGA card

I've already tried similar adapters with CF memory cards, on 386SX and 386DX PCs, and didn't have the problems you noted, the memory cards were 8MB, 128MB, but I have others I don't remember trying, they are from 256MB and 512MB, I don't think they can give problems, but I've had them when the disk had to boot, the PC froze as if there was an incorrect size, I don't remember exactly how, but by dint of trying, I managed to get any card to work CF, boot successfully.
Regarding your problem with the 16bit VGA, maybe you have already done other tests, but if the problem is with the IDE-CF, here I remember that with some controllers the PC would not start, after having used others (probably more recent), everything has worked.
So, I can only advise you to make several attempts, changing cards (assuming you have several), you may find the right combination, that everything works normally, maybe on other PCs those cards can go without any problem, but that depends, if they are faulty I think it will be noticeable, and you will see similar problems on other PCs.

PC@LIVE wrote on 2023-03-18, 10:19:

So, I can only advise you to make several attempts, changing cards (assuming you have several), you may find the right combination

It's not about cards. The problem is in CF adapter. There is a mistake in PCB (maybe not on yours, but there is on mine, and on some other too), or designer acted on the assumption that the adapter will be used with the IDE controller that uses cable's pin 28 to signal Cable Select function. The line 28 on pre-CableSelect capable controllers (i.e. all those Multi-I/O ISA cards) carries different signal - the ALE. And the I/O card just passes through the ISA bus ALE signal to the cable. Grounding that line on the CF adapter breaks the ISA bus logic. The CF adapter must not drive the line 28 of the cable, yet it does, and it only works (not messes up with the system) if the IDE controller already grounded the line 28 on its side of the cable to use it as Cable Select line.

Is there a list of problematic CF adapters? I have a few cheap adapters from China, bought in 2013/14 and never had any problems with them. Tested on 286 to at least some Athlon based computer. Different brand CF cards from 16MB to 8GB.

vstrakh wrote on 2023-03-18, 11:00:

PC@LIVE wrote on 2023-03-18, 10:19:

So, I can only advise you to make several attempts, changing cards (assuming you have several), you may find the right combination

It's not about cards. The problem is in CF adapter. There is a mistake in PCB (maybe not on yours, but there is on mine, and on some other too), or designer acted on the assumption that the adapter will be used with the IDE controller that uses cable's pin 28 to signal Cable Select function. The line 28 on pre-CableSelect capable controllers (i.e. all those Multi-I/O ISA cards) carries different signal - the ALE. And the I/O card just passes through the ISA bus ALE signal to the cable. Grounding that line on the CF adapter breaks the ISA bus logic. The CF adapter must not drive the line 28 of the cable, yet it does, and it only works (not messes up with the system) if the IDE controller already grounded the line 28 on its side of the cable to use it as Cable Select line.

Ok I understood the problem, before I thought it was due to a fault, however it is possible that even the ones I have have the same problem, but the symptoms I encountered are different, as I wrote before, some old controllers do not detect the CF , and would it be possible that the PIN28 problem is the cause?
I solved it by changing the controller, but by mounting the IDE-CF adapter on more recent PCs, everything works normally, so no problem with 16-bit ISA video cards, at least with the ones I used, maybe it's just a problem with that motherboard with IDE-CF, because on my AMD 386SX 33MHz it works, ditto on the 386DX-40.

This one is problematic: Re: Troubleshooting 386 motherboard, video memory writes clashes with PCM audio playback
Cutting the copper around pin 28 on the IDE connector solves the issue.

And you don't even know there is a problem, because on most of builds the issue does not manifest. It all depends on the tons of factors - the drive strength of the ALE signal, the speed of address latches on the cards that may survive the shorted ALE because it takes time for the pulse to reach the CF adapter and reflect back. Those +/- nanosecs may be enough to allow everything run smoothly, or it will fail because of edge cases on another build, and you will suspect your "faulty" video card rather than broken logic of the ISA bus.

OK. Great to know if something weird happens. If testing a motherboard in unknown condition it's always better for the first power up to do with only basic setup - mobo + ram + vga.

vstrakh wrote on 2023-03-18, 11:23:

This one is problematic: Re: Troubleshooting 386 motherboard, video memory writes clashes with PCM audio playback
Cutting the copper around pin 28 on the IDE connector solves the issue.

And you don't even know there is a problem, because on most of builds the issue does not manifest. It all depends on the tons of factors - the drive strength of the ALE signal, the speed of address latches on the cards that may survive the shorted ALE because it takes time for the pulse to reach the CF adapter and reflect back. Those +/- nanosecs may be enough to allow everything run smoothly, or it will fail because of edge cases on another build, and you will suspect your "faulty" video card rather than broken logic of the ISA bus.

Ok thanks, I didn't know about the PIN28 problem, as soon as I can I'll take a look at some adapters, I guess they are all connected to ground, I have various versions of adapters, one is dual CF, while the other 3-4 are to connect to the 40 PIN cable (Master or Slave).
Eventually I can try again one of those controllers (old type), which had problems reading the CF, and I could make a change, if I'm not mistaken, isolating Pin 28, maybe if the failure is caused by that Pin, it could work ?, after modifying one of those adapters.

COAST 256kb
PBS-256

I'll admit it was the box art that drew me in. Now I just need to bring myself to actually use it.

kixs wrote on 2023-03-18, 06:39:

Sounds fine by me 😉

Otherwise here is a small review:
http://www.amoretro.de/2014/07/terratec-waves … rboard-2mb.html

In the end, that's really all that matters. I saw the same, I wish there were more reviews on all of the various sound cards. I'd probably try my hand at it if I knew more about the subject.

I find very early 3D accelerators to be a lot of fun, even if they're not all that useful. It was an incredible time to see CPU speeds and software acceleration be able to outpace dedicated 3D acceleration in silicon.

I wonder if some of these products were simply years development, and suddenly all of these companies panicked and threw whatever that had onto the market a quickly as possible to get a return on their research and development costs, before they become completely obsolete. Maybe it was some kind of snowball effect within the industry.

For certain, the display adapter market had an enormous shakedown once DirectX and OpenGL became mainstream. If you couldn't compete with either of these at a minimum, you were completely out of the market.

There was a lot of talk about it in 94 or so and a lot of companies kicked off a project then I think. Hell, even I started designing a geometry and physics engine, broad strokes wise, it quickly became apparent that the proof of concept would fill enough boards to fill a filing cabinet sized rack. Then Voodoo came out and I thought I'd missed the boat... then a year later it dawned on me that it didn't do what my concept did, so I started thinking about it again.. then geforce came out and I thought I'd missed it again.. then a year or two passed and I realized AGAIN that it didn't do what I had in mind... then finally PhysX plus unified shader was doing it. Anyway those with more money to throw into the flames, bashed on and looked for the next threat, and committed "best so far" to silicon every once in a while to try to stay in the game I guess.