skaarg wrote on 2023-04-13, 20:24:

Meatball wrote on 2023-04-12, 21:25:

7950GT 256MB (for an upcoming Windows ME/Windows 2000 dual boot machine)

I didn't believe Geforce 7 series ran on Windows 9X/ME. Is there a modified driver for it?

You can get 9000 series running, even:

https://theretroweb.com/drivers/96

Meatball wrote on 2023-04-13, 21:16:

skaarg wrote on 2023-04-13, 20:24:

Meatball wrote on 2023-04-12, 21:25:

7950GT 256MB (for an upcoming Windows ME/Windows 2000 dual boot machine)

I didn't believe Geforce 7 series ran on Windows 9X/ME. Is there a modified driver for it?

You can get 9000 series running, even:

https://theretroweb.com/drivers/96

Have you actually tried this? Newest I have been able to actually get working on 9x is the 7 series. 8 series (at least the 8800 Ultra gives a BSOD as soon as drivers are installed, even with the R.Lowe patch.

cyclone3d wrote on 2023-04-13, 22:21:

Meatball wrote on 2023-04-13, 21:16:

skaarg wrote on 2023-04-13, 20:24:

I didn't believe Geforce 7 series ran on Windows 9X/ME. Is there a modified driver for it?

You can get 9000 series running, even:

https://theretroweb.com/drivers/96

Have you actually tried this? Newest I have been able to actually get working on 9x is the 7 series. 8 series (at least the 8800 Ultra gives a BSOD as soon as drivers are installed, even with the R.Lowe patch.

I haven't, nor would I bother. I don't even stall DX10 cards (for regular use) on XP, let alone 9X.

I found this video card which seems unusual to me is 4Mb of memory and PCI connector,
and it's working after a little fix.
U-T975g1 Trident 3DImage9750 chip

PC@LIVE wrote on 2023-04-13, 20:01:

An AMD K6-2+ 400MHz arrived today, a cpu with integrated L2 cache (128KB), this can be used on motherboards with VCORE lower tha […]
Show full quote

An AMD K6-2+ 400MHz arrived today, a cpu with integrated L2 cache (128KB), this can be used on motherboards with VCORE lower than 2.0V, but I know that they can also be used with voltages up to 2.1V, obviously if you is below 2.0V will be better.
A lot depends on the motherboard, usually the minimum selectable is 2.0V (without jumper), other older cards don't go below 2.5-2.8V, but there are some cards that can go down to 1.8V or less. Obviously having one of these cards, the 1.6V CPU can be installed without any problems, the slight 0.2V overvolt will be tolerated without creating problems.
Otherwise, what to do?
Here the simplest solution is to use it in boards with a minimum voltage of 2.0V (without jumpers), but what if instead we wanted to go below 2.0V? What could be done? The solution could be found in some old site, assuming it is still online, or having knowledge of electronics, you could make some changes. Well, personally I don't have such knowledge, and therefore I wouldn't know if a component can be replaced with another, but over time I have made numerous measurements on the mosfets, so I assume that when you measure on the 2.0V mosfet, that is the voltage that reaches the CPU, so I thought, usually after the mosfet there is an inductor, if I detach one side and insert a resistor in series, I could lower the voltage to 1.6V or 1.7V (depends on the value used), if this were actually the case, it would be perfect!
The same thing could be done with motherboards that do not go below 2.8V, bringing the voltage to 2.2V or 2.0V, or even 1.5V or 1.6V, you could install newer K6 CPUs, or a 2.2V Cyrix (assuming you have one), perhaps for the sole purpose of benching.

On some boards multiple jumpers on the Vcore setting block will bring in lower Vcores than documented due to connecting several resistors in parallel, for lower effective resistance, which "sets" the voltage regulator.

For something of an overview of common arrangements, many socket 7 boards work like this, though exact values vary by design and voltage regulator part numbers, take a look at all the Vcore articles for the M571 and variants here...
http://www.m571.com/m571/franczabkar.htm

cyclone3d wrote on 2023-04-13, 22:21:

Meatball wrote on 2023-04-13, 21:16:

skaarg wrote on 2023-04-13, 20:24:

I didn't believe Geforce 7 series ran on Windows 9X/ME. Is there a modified driver for it?

You can get 9000 series running, even:

https://theretroweb.com/drivers/96

Have you actually tried this? Newest I have been able to actually get working on 9x is the 7 series. 8 series (at least the 8800 Ultra gives a BSOD as soon as drivers are installed, even with the R.Lowe patch.

Unified shader cards wont work at all under Win9x except as very basic SVGA cards using the MS SVGA driver only. (the 7950GT AGP was the last nVidia card that "could" be made to work under 9x and the X1950XTX was the last ATI card)
Not even hacked drivers will get around this issue, its a pure hardware handling limitation of 9x, so even if you find "hacked" drivers saying they support unified cards under 9x I would take that with a truck load of salt and be very skeptical of it. (Most of these hacked drivers simply allow 9x to id the GPU via inf fudging but don't actually do anything else)

EDIT - The 7950 GX2 was an odd card and IIRC also doesn't work under 9x but this is due to nVidia SLI not being supported under 9x and not the above 9x hardware handling problem, there are also no drivers available for it to function as a singular GPU. I only mention it here because someone will likely point it out if I don't.

PC@LIVE wrote on 2023-04-13, 20:01:

An AMD K6-2+ 400MHz arrived today, a cpu with integrated L2 cache (128KB), this can be used on motherboards with VCORE lower tha […]
Show full quote

An AMD K6-2+ 400MHz arrived today, a cpu with integrated L2 cache (128KB), this can be used on motherboards with VCORE lower than 2.0V, but I know that they can also be used with voltages up to 2.1V, obviously if you is below 2.0V will be better.
A lot depends on the motherboard, usually the minimum selectable is 2.0V (without jumper), other older cards don't go below 2.5-2.8V, but there are some cards that can go down to 1.8V or less. Obviously having one of these cards, the 1.6V CPU can be installed without any problems, the slight 0.2V overvolt will be tolerated without creating problems.
Otherwise, what to do?
Here the simplest solution is to use it in boards with a minimum voltage of 2.0V (without jumpers), but what if instead we wanted to go below 2.0V? What could be done? The solution could be found in some old site, assuming it is still online, or having knowledge of electronics, you could make some changes. Well, personally I don't have such knowledge, and therefore I wouldn't know if a component can be replaced with another, but over time I have made numerous measurements on the mosfets, so I assume that when you measure on the 2.0V mosfet, that is the voltage that reaches the CPU, so I thought, usually after the mosfet there is an inductor, if I detach one side and insert a resistor in series, I could lower the voltage to 1.6V or 1.7V (depends on the value used), if this were actually the case, it would be perfect!
The same thing could be done with motherboards that do not go below 2.8V, bringing the voltage to 2.2V or 2.0V, or even 1.5V or 1.6V, you could install newer K6 CPUs, or a 2.2V Cyrix (assuming you have one), perhaps for the sole purpose of benching.

Have a look into the datasheet of the voltage regulator of your specific board. It will tell you what the regulator can do. Most regulators starting at 2.0V have four inputs ("bits") for setting the CPU core voltage. If it has five it can go down to 1.3V in most cases. There are boards which use a five-bit voltage regulator but only implement four jumpers. The fifth one is tied to a specific level, in that case to allow only the higher range of voltages to be set (i.e. >2.0V).
The way to go here is to look if it is possible to add a fifth jumper. Done on a Jetway J-542B.
Sometimes there are only four-bit regulators implemented but there are pin-compatible five-bit regulators available. Then it would be possible to exchange these, implement the fifth jumper or even a dip-switch and voilà. Done on a Chaintech CT-5SIM.

Other than that, better don't touch the voltage regulator circuit.

juancho wrote on 2023-04-13, 22:48:

I found this video card which seems unusual to me is 4Mb of memory and PCI connector, and it's working after a little fix. U-T97 […]
Show full quote

I found this video card which seems unusual to me is 4Mb of memory and PCI connector,
and it's working after a little fix.
U-T975g1 Trident 3DImage9750 chip

What seems so unusual? The Trident 9750 was usually shipped with 4MB, so that's unsurprising (to me at least).

Pretty broken early 3D accelerator from 1997, with performance similar to the S3 Virge DX (i.e. frequently slower than software rendering). One thing Trident did get right was TV-out: their implementation was one of the most trouble-free out there, simply duplicating whatever was sent to VGA.

Another big tower system... I try to avoid them (with mixed results) because I don't really have space for them, but I couldn't pass this originally black with dual-color MHz display. Internally it came with DFI MVP3 board, K6-2 500MHz and Creative Graphics Blaster TNT. Case also has the metal stand (not on pics).

HanJammer wrote on 2023-04-14, 08:23:

Another big tower system... I try to avoid them (with mixed results) because I don't really have space for them, but I couldn't pass this originally black with dual-color MHz display. Internally it came with DFI MVP3 board, K6-2 500MHz and Creative Graphics Blaster TNT. Case also has the metal stand (not on pics).

they sure do take up a lot of space, i only have one - lots of room inside, enough for two computers!. It's an impressive beast though

BitWrangler wrote on 2023-04-14, 01:34:

On some boards multiple jumpers on the Vcore setting block will bring in lower Vcores than documented due to connecting several […]
Show full quote

PC@LIVE wrote on 2023-04-13, 20:01:

An AMD K6-2+ 400MHz arrived today, a cpu with integrated L2 cache (128KB), this can be used on motherboards with VCORE lower tha […]
Show full quote

An AMD K6-2+ 400MHz arrived today, a cpu with integrated L2 cache (128KB), this can be used on motherboards with VCORE lower than 2.0V, but I know that they can also be used with voltages up to 2.1V, obviously if you is below 2.0V will be better.
A lot depends on the motherboard, usually the minimum selectable is 2.0V (without jumper), other older cards don't go below 2.5-2.8V, but there are some cards that can go down to 1.8V or less. Obviously having one of these cards, the 1.6V CPU can be installed without any problems, the slight 0.2V overvolt will be tolerated without creating problems.
Otherwise, what to do?
Here the simplest solution is to use it in boards with a minimum voltage of 2.0V (without jumpers), but what if instead we wanted to go below 2.0V? What could be done? The solution could be found in some old site, assuming it is still online, or having knowledge of electronics, you could make some changes. Well, personally I don't have such knowledge, and therefore I wouldn't know if a component can be replaced with another, but over time I have made numerous measurements on the mosfets, so I assume that when you measure on the 2.0V mosfet, that is the voltage that reaches the CPU, so I thought, usually after the mosfet there is an inductor, if I detach one side and insert a resistor in series, I could lower the voltage to 1.6V or 1.7V (depends on the value used), if this were actually the case, it would be perfect!
The same thing could be done with motherboards that do not go below 2.8V, bringing the voltage to 2.2V or 2.0V, or even 1.5V or 1.6V, you could install newer K6 CPUs, or a 2.2V Cyrix (assuming you have one), perhaps for the sole purpose of benching.

On some boards multiple jumpers on the Vcore setting block will bring in lower Vcores than documented due to connecting several resistors in parallel, for lower effective resistance, which "sets" the voltage regulator.

For something of an overview of common arrangements, many socket 7 boards work like this, though exact values vary by design and voltage regulator part numbers, take a look at all the Vcore articles for the M571 and variants here...
http://www.m571.com/m571/franczabkar.htm

I know that site, thanks, besides the M550 I also have an M549 and an M506-507?, plus other S.7 cards, in most of the cards I have fixed jumpers, i.e. one selects the desired voltage (for example 2.2 or 2.8 or others ), in cases with total 8-pin jumpers (4x2), more jumpers can be used to select particular voltages (2.3V K6-2-550), in some boards I tried undocumented combinations, in a VX with a minimum voltage of 2.8V removing all the jumpers the VCORE drops to 1.25V, perhaps too few to start a K6 or a K6-2?
Unfortunately I don't understand how these voltage regulator circuits work, but if you want I could put some pictures, with details of the jumper and mosfet area, if this can help you understand more.

ChrisK wrote on 2023-04-14, 06:33:

Have a look into the datasheet of the voltage regulator of your specific board. It will tell you what the regulator can do. Most […]
Show full quote

PC@LIVE wrote on 2023-04-13, 20:01:

An AMD K6-2+ 400MHz arrived today, a cpu with integrated L2 cache (128KB), this can be used on motherboards with VCORE lower tha […]
Show full quote

An AMD K6-2+ 400MHz arrived today, a cpu with integrated L2 cache (128KB), this can be used on motherboards with VCORE lower than 2.0V, but I know that they can also be used with voltages up to 2.1V, obviously if you is below 2.0V will be better.
A lot depends on the motherboard, usually the minimum selectable is 2.0V (without jumper), other older cards don't go below 2.5-2.8V, but there are some cards that can go down to 1.8V or less. Obviously having one of these cards, the 1.6V CPU can be installed without any problems, the slight 0.2V overvolt will be tolerated without creating problems.
Otherwise, what to do?
Here the simplest solution is to use it in boards with a minimum voltage of 2.0V (without jumpers), but what if instead we wanted to go below 2.0V? What could be done? The solution could be found in some old site, assuming it is still online, or having knowledge of electronics, you could make some changes. Well, personally I don't have such knowledge, and therefore I wouldn't know if a component can be replaced with another, but over time I have made numerous measurements on the mosfets, so I assume that when you measure on the 2.0V mosfet, that is the voltage that reaches the CPU, so I thought, usually after the mosfet there is an inductor, if I detach one side and insert a resistor in series, I could lower the voltage to 1.6V or 1.7V (depends on the value used), if this were actually the case, it would be perfect!
The same thing could be done with motherboards that do not go below 2.8V, bringing the voltage to 2.2V or 2.0V, or even 1.5V or 1.6V, you could install newer K6 CPUs, or a 2.2V Cyrix (assuming you have one), perhaps for the sole purpose of benching.

Have a look into the datasheet of the voltage regulator of your specific board. It will tell you what the regulator can do. Most regulators starting at 2.0V have four inputs ("bits") for setting the CPU core voltage. If it has five it can go down to 1.3V in most cases. There are boards which use a five-bit voltage regulator but only implement four jumpers. The fifth one is tied to a specific level, in that case to allow only the higher range of voltages to be set (i.e. >2.0V).
The way to go here is to look if it is possible to add a fifth jumper. Done on a Jetway J-542B.
Sometimes there are only four-bit regulators implemented but there are pin-compatible five-bit regulators available. Then it would be possible to exchange these, implement the fifth jumper or even a dip-switch and voilà. Done on a Chaintech CT-5SIM.

Other than that, better don't touch the voltage regulator circuit.

I'll try to take a look, I hope to find useful info, and I think I have the J-542, in the worst case I could try to replicate the change by adding the fifth jumper.
But I try to make as few changes as possible, for example I've seen in the past, that to get a lower VCore you could use jumpers modified with the addition of a resistor, I think it could be like an additional resistor, following this thread, I thought of make a similar modification on the inductor of the VCORE, detaching a pin (from the MB), and connect a resistor of a certain value in series (it depends on the volts you want to obtain), in this way you would not touch anything else, and you could put a bridge instead of the resistor, if you want to bring the VCORE back to the original voltage.
This is obviously an idea, but it could work, right, I guess so, but I could be proven wrong, a lot depends on how the VCORE regulation loop is made.

HanJammer wrote on 2023-04-14, 08:23:

Another big tower system... I try to avoid them (with mixed results) because I don't really have space for them, but I couldn't pass this originally black with dual-color MHz display. Internally it came with DFI MVP3 board, K6-2 500MHz and Creative Graphics Blaster TNT. Case also has the metal stand (not on pics).

That is one beautiful chonk!

dionb wrote on 2023-04-14, 07:28:

juancho wrote on 2023-04-13, 22:48:

I found this video card which seems unusual to me is 4Mb of memory and PCI connector,
and it's working after a little fix.
U-T975g1 Trident 3DImage9750 chip

What seems so unusual? The Trident 9750 was usually shipped with 4MB, so that's unsurprising (to me at least).

Pretty broken early 3D accelerator from 1997, with performance similar to the S3 Virge DX (i.e. frequently slower than software rendering). One thing Trident did get right was TV-out: their implementation was one of the most trouble-free out there, simply duplicating whatever was sent to VGA.

I agree. They are fine with basic 2D acceleration within Windows and can be used with software mode games. Maybe even some early Direct2D titles as well. They are not a good option for DOS, and have problems with certain games. S3 is always a better option between the two, for DOS. Both are fine for Windows.

Got an ATi Rage 128 AIW (PCI) shipped from the US to Germany:

Despite the not-so-great packaging, the card seems to have survived.
First, it didn't output any signal. Then I put it into my system together with a Rage 128 AGP. It was then detected by Windows 98 and after a reboot, it just works fine by itself:

Old hardware can be so temperamental sometimes. 😁

That is some kind of luck that it survived packed like that

Yah, when you get the pack thin enough 25mm I think, it usually travels lettermail though, which is like it's put in a tote partially full of paperback novels and shaken around a bit, whereas if it has to go parcel, you gotta pack it to survive being put in a dumpster half full of anvils that's picked up and shaken around a bit. Sometimes there's a small packet tier which is in between. So it's kind of a choice there, pack it thinner and it gets lighter treatment or pack it thicker to survive "whatever", where the first usually travels rapid for lowish cost and the second is slower and typically more expensive (as you end up paying volume unless you just stick a stamp on a brick.)

Got some ancillary parts this morning, found a DKVM-4K nice and cheap, seemed to be new open box, box was a bit worn, but everything inside very crisp and new. Good to get a 4 port, got a couple of 2 ports and a monster 10 port plus USB (and likely too modern for pre-2000) and the 2 porters never seemed enough for one "cluster". This one seems pretty good, doesn't have to claim any particular OS compatibility, does the double tap on scroll lock plus function shifting, and M key for reset of keyboard and mouse if they get out of synch.

Also got an external drive PSU this one labelled by iOmega, maybe originally for a zip, got the 4 pin 5v/12V miniDin. I don't know if there's an actual standard for these, but a large amount of vendors used the same pinout if there wasn't. Anyway, had 3 for 4 drives, so now one per.

Then couldn't resist a 10 pack of floppies, "microfloppies" no less, so prolly pre-millenial, maxell brand. List compatibility with PS2 up to model 50... whooo! So yah, dunno how long they kept that box for but feels like earlier 90s when they designed it.

schmatzler wrote on 2023-04-14, 16:09:

Got an ATi Rage 128 AIW (PCI) shipped from the US to Germany: rage128-packaging.jpg rage128.jpg […]
Show full quote

Got an ATi Rage 128 AIW (PCI) shipped from the US to Germany:
rage128-packaging.jpg
rage128.jpg

Despite the not-so-great packaging, the card seems to have survived.
First, it didn't output any signal. Then I put it into my system together with a Rage 128 AGP. It was then detected by Windows 98 and after a reboot, it just works fine by itself:
rage128-3dmark2001.jpg

Old hardware can be so temperamental sometimes. 😁

Ah that's a nice one! I didn't remember that they produced Rage 128-based AIW cards. Do you actually plan to use this part of the card or is it just for the collection?
I remember having an AIW Radeon 8500, the most unstable card (drivers...) I ever had 🙁

Minutemanqvs wrote on 2023-04-14, 18:45:

Ah that's a nice one! I didn't remember that they produced Rage 128-based AIW cards. Do you actually plan to use this part of the card or is it just for the collection?

I have the AGP version of that card, too. They are amazing with the latest drivers, really solid Win98 cards.
This one was planned to be used with an old Asus Socket 7 board (P55SP4), but it doesn't boot with it, sadly.
(Probably because the PCI is 5V only)

I also have a 9250SE on my way which might work with that board, though.

I probably won't use the TV Tuner and video output options ever. 😀