AndrettiGTO wrote on 2026-05-23, 00:42:

It's a shame so many of these specialty cards no longer have the matching breakout cables. No cable for my ATI X800 XT AIW and almost unobtanium. :\

I think that one uses the diagram on the pinouts site. Can get male to male, pin to pin "dupont" cables, jab one end in the card and one end into a female/female VGA gender changer, then when it works, glob both ends with hotglue to keep the pins in place.

BitWrangler wrote on 2026-05-22, 23:07:

MattRocks wrote on 2026-05-22, 21:26:

Those are an entirely different generation and a different product tier. They have the same 28-pin connector, but appear to have […]
Show full quote

BitWrangler wrote on 2026-05-22, 19:54:

Dug up some links about some "go arounds" with the freaky ATI connector that may be fully or partially related to that pinnacle […]
Show full quote

Dug up some links about some "go arounds" with the freaky ATI connector that may be fully or partially related to that pinnacle one...

ChrisK posted about his dongle-less card here...
Re: What retro activity did you get up to today?

We had a few more posts about things, and then...
Re: What retro activity did you get up to today?

But also there's a connector pinout for what might be an X800 or X1100 or something AIW or VIVO connector...
https://pinoutguide.com/Audio-Video-Hardware/ … e5_pinout.shtml

I know the 9x00 AIWs were coming with Pinnacle Studio at the time, not sure if that is enough of a link to infer connector compatibility. Not sure what you had in mind to plug your box into.

Those are an entirely different generation and a different product tier. They have the same 28-pin connector, but appear to have very different pinout so might not won't work with a Pinnacle breakout box.

Like most companies, Pinnacle had multiple product tiers:

• Pinnacle Studio - Consumer or OEM product for home video
• Pinnacle Liquid Edition v5 - Enthusiast retail for film editing listed at $699 retail (came with a PCI firewire card for DV camcorder capture)
• Pinnacle Liquid Edition Pro v5 - Prosumer / Studio product listed at $999 retail (this is the box that contained an AGP card featuring a 28 pin breakout connector)
• Pinnacle Silver (and above) - Colour critical workloads for film and broadcasting costing around $25k

Pinnacle Liquid Edition Pro v5 matters because it was the world's first editor to feature realtime full-resolution previews (powered by the GPU on the hot path, not by CPUs with an indirection). And, Pinnacle Liquid Edition Pro v5 is central to the Pinnacle-ATI partnership. It was an investment in technology innovation, not an OEM software freebie.

They were competing primarily against Apple Final Cut Pro, which was a closed ecosystem powered by PPC G4/G5 dual processor workstations. Final Cut Pro was boasting realtime previews, but it was actually only near-realtime (processed in software) and output at reduced "preview resolution". The generation that matters historically is specifically Pinnacle Liquid Edition Pro v5, which was powered by a Radeon 8-series GPU and ATI Rage Theatre. Importantly, the original Pinnacle pin out supports DV, Audio, and a lot of Video on the breakout box for studio work.

Next generation: When you compare the later ATI 9-series, those are paired with a different Theatre chip and their pinout appears to not support DV or Audio on the breakout box/cable. They also bundled a budget consumer Pinnacle Studio editor, which is more than a step down from the original Pinnacle bundle. On the Pinnacle side, their later Liquid Edition Pro v6 came with a lower cost USB breakout box so they were also stepping down. The moment of focus on GPU accelerated studio workflows had passed - it was still present, just no longer new.

And as previously said, its very interesting because AMD and Avid maintain no records of Pinnacle Liquid Edition Pro v5 or the AGP card it shipped with, save for one line in Avid Liquid 7 Pro manual that acknowledges the card exists only by stating it is no longer supported! Two important things there: First, the confession it really exists. Second, the confession that they don't want to pay to for a DX9 driver update. I believe to power the effects the hardware actually only only ever needed to provide fixed function DX7 T&L, but that's not what the marketing said. There's also questions to ask about why ATI did not release an 8-series VIVO reference board. It's all very curious.

Wow, that's kinda wild. Was that 8 series the one they sold also as the AIW 8500DV with the purple dongle with square plug https://theretroweb.com/expansioncards/s/ati- … r-radeon-8500dv ? I have one of those but it came devoid of software. It would have seemed more likely that the 9 series stuff would be more advanced though as the GPU had very limited GPU compute capabilities, not sure exactly what, probably stuff with matrix bit shifts on the shaders or something. Early days for that though compared with OpenCL/Cuda I met a guy once though who was doing some computational chemistry on one.

Yes, the ATI AIW 8500 DV and Pinnacle Liquid Edition Pro accelerator occupied the retail shelves at the same time - those two cards are siblings. But, the AIW is an AIW, while the Pinnacle card is a more specialised VIVO for mid-range studio workflows.

As its the exact same GPU, you can load Catalyst/Omega drivers onto the Pinnacle and play games but that is not what makes the card unique - or, maybe that is what makes it unique? Thinking about it, not many specialised professional cards play games properly 😉

Speculation only but the analogue stage around the Rage Theatre might be different, there is another chip between breakout port and Theatre chip that probably deserves a closer look, and the hardware or driver configuration of the Rage Theatre might be different. There's very little commentary but old discussions on missing drivers suggest the ATI and Pinnacle implementations might be subtly different.

And the missing drivers today is a whole other interesting story around Pinnacle-Avid, Pinnacle-ATI, and general film production security. I'll finish the essay and publish shortly..

MattRocks wrote on 2026-05-23, 06:22:

Yes, the ATI AIW 8500 DV and Pinnacle Liquid Edition Pro accelerator occupied the retail shelves at the same time - those two ca […]
Show full quote

BitWrangler wrote on 2026-05-22, 23:07:

MattRocks wrote on 2026-05-22, 21:26:

Those are an entirely different generation and a different product tier. They have the same 28-pin connector, but appear to have […]
Show full quote

Those are an entirely different generation and a different product tier. They have the same 28-pin connector, but appear to have very different pinout so might not won't work with a Pinnacle breakout box.

Like most companies, Pinnacle had multiple product tiers:

• Pinnacle Studio - Consumer or OEM product for home video
• Pinnacle Liquid Edition v5 - Enthusiast retail for film editing listed at $699 retail (came with a PCI firewire card for DV camcorder capture)
• Pinnacle Liquid Edition Pro v5 - Prosumer / Studio product listed at $999 retail (this is the box that contained an AGP card featuring a 28 pin breakout connector)
• Pinnacle Silver (and above) - Colour critical workloads for film and broadcasting costing around $25k

Pinnacle Liquid Edition Pro v5 matters because it was the world's first editor to feature realtime full-resolution previews (powered by the GPU on the hot path, not by CPUs with an indirection). And, Pinnacle Liquid Edition Pro v5 is central to the Pinnacle-ATI partnership. It was an investment in technology innovation, not an OEM software freebie.

They were competing primarily against Apple Final Cut Pro, which was a closed ecosystem powered by PPC G4/G5 dual processor workstations. Final Cut Pro was boasting realtime previews, but it was actually only near-realtime (processed in software) and output at reduced "preview resolution". The generation that matters historically is specifically Pinnacle Liquid Edition Pro v5, which was powered by a Radeon 8-series GPU and ATI Rage Theatre. Importantly, the original Pinnacle pin out supports DV, Audio, and a lot of Video on the breakout box for studio work.

Next generation: When you compare the later ATI 9-series, those are paired with a different Theatre chip and their pinout appears to not support DV or Audio on the breakout box/cable. They also bundled a budget consumer Pinnacle Studio editor, which is more than a step down from the original Pinnacle bundle. On the Pinnacle side, their later Liquid Edition Pro v6 came with a lower cost USB breakout box so they were also stepping down. The moment of focus on GPU accelerated studio workflows had passed - it was still present, just no longer new.

And as previously said, its very interesting because AMD and Avid maintain no records of Pinnacle Liquid Edition Pro v5 or the AGP card it shipped with, save for one line in Avid Liquid 7 Pro manual that acknowledges the card exists only by stating it is no longer supported! Two important things there: First, the confession it really exists. Second, the confession that they don't want to pay to for a DX9 driver update. I believe to power the effects the hardware actually only only ever needed to provide fixed function DX7 T&L, but that's not what the marketing said. There's also questions to ask about why ATI did not release an 8-series VIVO reference board. It's all very curious.

Wow, that's kinda wild. Was that 8 series the one they sold also as the AIW 8500DV with the purple dongle with square plug https://theretroweb.com/expansioncards/s/ati- … r-radeon-8500dv ? I have one of those but it came devoid of software. It would have seemed more likely that the 9 series stuff would be more advanced though as the GPU had very limited GPU compute capabilities, not sure exactly what, probably stuff with matrix bit shifts on the shaders or something. Early days for that though compared with OpenCL/Cuda I met a guy once though who was doing some computational chemistry on one.

Yes, the ATI AIW 8500 DV and Pinnacle Liquid Edition Pro accelerator occupied the retail shelves at the same time - those two cards are siblings. But, the AIW is an AIW, while the Pinnacle card is a more specialised VIVO for mid-range studio workflows.

As its the exact same GPU, you can load Catalyst/Omega drivers onto the Pinnacle and play games but that is not what makes the card unique - or, maybe that is what makes it unique? Thinking about it, not many specialised professional cards play games properly 😉

Speculation only but the analogue stage around the Rage Theatre might be different, there is another chip between breakout port and Theatre chip that probably deserves a closer look, and the hardware or driver configuration of the Rage Theatre might be different. There's very little commentary but old discussions on missing drivers suggest the ATI and Pinnacle implementations might be subtly different.

And the missing drivers today is a whole other interesting story around Pinnacle-Avid, Pinnacle-ATI, and general film production security. I'll finish the essay and publish shortly..

There seems to be something whack about the drivers for the 8500DV as well, I was seeing some suggestions on digital video forums that the mainstream catalysts did not fully support it, and for full functionality the 7500 AIW drivers had to be installed.... which I am guessing is for the Rage Theatre implementation side.

BitWrangler wrote on 2026-05-23, 12:00:

MattRocks wrote on 2026-05-23, 06:22:

Yes, the ATI AIW 8500 DV and Pinnacle Liquid Edition Pro accelerator occupied the retail shelves at the same time - those two ca […]
Show full quote

BitWrangler wrote on 2026-05-22, 23:07:

Wow, that's kinda wild. Was that 8 series the one they sold also as the AIW 8500DV with the purple dongle with square plug https://theretroweb.com/expansioncards/s/ati- … r-radeon-8500dv ? I have one of those but it came devoid of software. It would have seemed more likely that the 9 series stuff would be more advanced though as the GPU had very limited GPU compute capabilities, not sure exactly what, probably stuff with matrix bit shifts on the shaders or something. Early days for that though compared with OpenCL/Cuda I met a guy once though who was doing some computational chemistry on one.

Yes, the ATI AIW 8500 DV and Pinnacle Liquid Edition Pro accelerator occupied the retail shelves at the same time - those two cards are siblings. But, the AIW is an AIW, while the Pinnacle card is a more specialised VIVO for mid-range studio workflows.

As its the exact same GPU, you can load Catalyst/Omega drivers onto the Pinnacle and play games but that is not what makes the card unique - or, maybe that is what makes it unique? Thinking about it, not many specialised professional cards play games properly 😉

Speculation only but the analogue stage around the Rage Theatre might be different, there is another chip between breakout port and Theatre chip that probably deserves a closer look, and the hardware or driver configuration of the Rage Theatre might be different. There's very little commentary but old discussions on missing drivers suggest the ATI and Pinnacle implementations might be subtly different.

And the missing drivers today is a whole other interesting story around Pinnacle-Avid, Pinnacle-ATI, and general film production security. I'll finish the essay and publish shortly..

There seems to be something whack about the drivers for the 8500DV as well, I was seeing some suggestions on digital video forums that the mainstream catalysts did not fully support it, and for full functionality the 7500 AIW drivers had to be installed.... which I am guessing is for the Rage Theatre implementation side.

As they use the same Rage Theatre chip, the 7-series drivers will bind to 8-series cards, but that will not guarantee the VIVO ports are behaving properly. The 7-series predates the 28-pin connector so it's fairly unlikely they will behave.

MattRocks wrote on 2026-05-23, 12:10:

BitWrangler wrote on 2026-05-23, 12:00:

MattRocks wrote on 2026-05-23, 06:22:

Yes, the ATI AIW 8500 DV and Pinnacle Liquid Edition Pro accelerator occupied the retail shelves at the same time - those two ca […]
Show full quote

Yes, the ATI AIW 8500 DV and Pinnacle Liquid Edition Pro accelerator occupied the retail shelves at the same time - those two cards are siblings. But, the AIW is an AIW, while the Pinnacle card is a more specialised VIVO for mid-range studio workflows.

As its the exact same GPU, you can load Catalyst/Omega drivers onto the Pinnacle and play games but that is not what makes the card unique - or, maybe that is what makes it unique? Thinking about it, not many specialised professional cards play games properly 😉

Speculation only but the analogue stage around the Rage Theatre might be different, there is another chip between breakout port and Theatre chip that probably deserves a closer look, and the hardware or driver configuration of the Rage Theatre might be different. There's very little commentary but old discussions on missing drivers suggest the ATI and Pinnacle implementations might be subtly different.

And the missing drivers today is a whole other interesting story around Pinnacle-Avid, Pinnacle-ATI, and general film production security. I'll finish the essay and publish shortly..

There seems to be something whack about the drivers for the 8500DV as well, I was seeing some suggestions on digital video forums that the mainstream catalysts did not fully support it, and for full functionality the 7500 AIW drivers had to be installed.... which I am guessing is for the Rage Theatre implementation side.

As they use the same Rage Theatre chip, AIW 7500 drivers will bind - but that will not guarantee the VIVO ports are behaving properly.

Is this relevant to the driver story? Apparently XP service pack 2 nerfed the supplied drivers, seemingly ATI washed their hands and said it was Pinnacle's problem... https://community.avid.com/forums/p/14122/79460.aspx

And I know SP2 was the big security patch for XP as it was hole-ier than a pair of fishnets.

Error: double post

Error: double post

BitWrangler wrote on 2026-05-23, 12:43:

MattRocks wrote on 2026-05-23, 12:10:

BitWrangler wrote on 2026-05-23, 12:00:

There seems to be something whack about the drivers for the 8500DV as well, I was seeing some suggestions on digital video forums that the mainstream catalysts did not fully support it, and for full functionality the 7500 AIW drivers had to be installed.... which I am guessing is for the Rage Theatre implementation side.

As they use the same Rage Theatre chip, AIW 7500 drivers will bind - but that will not guarantee the VIVO ports are behaving properly.

Is this relevant to the driver story? Apparently XP service pack 2 nerfed the supplied drivers, seemingly ATI washed their hands and said it was Pinnacle's problem... https://community.avid.com/forums/p/14122/79460.aspx

And I know SP2 was the big security patch for XP as it was hole-ier than a pair of fishnets.

Yes, but does that error implicate ATI's Catalyst installer?

Speculation only but I think that particular nugget is a sign of bad faith from Avid and/or ATI, coupled with user ignorance. I think the real technical problem is a minor one. Liquid 7 just needs DX9 compliant drivers, not DX9 hardware acceleration - so, any 8500 GPU drivers from DX9 era should be enough.

However, instead of paying for the DX9 update, Avid and ATI removed all references to the AGP card from their support sites - that is not cool and it left film studios stumped.

It's unclear who owned responsibility, but the signs are that Pinnacle owned support responsibility because Pinnacle had been distributing driver updates during the DX8 era. Hence, I speculate Avid inherited that responsibility. ATI also did not maintain the drivers, but being generous they might have been contractually constrained by whatever was agreed in the Pinnacle-ATI partnership.

Drivers for the Radeon 8500 GPU were maintained, as were drivers for the Rage Theatre, so I suspect using the INF bare drivers instead of full Catalyst install would sidestep the install problem. But, enterprise users can't hack INF files while complying with corporate policies! Other forums show Avid support were being very unhelpful and directed customers to buy "any" AGP card with 64Mb RAM.

That was really bad advice and shows Avid misunderstood the Pinnacle product because you can’t put a 64MB Matrox in the AGP slot without disabling the real-time T&L previews.

And it’s really bad advice to give professional film studios who need locked down information security and support guarantees - those customers can't have random coax input on the workstation, or bundled OEM demos in the box, or second hand parts. It's as though Avid had absolutely no clue who Pinnacle's clients were!

And, all of the complaining users were trying to stay "current," but that is a moot point from a retro perspective - we just need drivers that work in their intended context (Liquid 5/6, DX8, Win2K/XP).

Not super retro, but I bought a Quadro FX 3500 (the Quadro equivalent of the GeForce 7900 GS) for $19 US shipped on eBay (seller's photo). I'd like to see if I can get better performance in Windows NT 4.0 than my GeForce 6800 GS, but I'm worried the Quadro card might be a touch too new. Nvidia G70 chips seem to work with the last Quadro driver for NT 4, but the Quadro FX 3500 uses the G71 chip which Nvidia released shortly after their last NT 4 driver.

Regardless, I might have a use for this Quadro card in a Windows 98 build as long as I can get the system stable with later Nvidia drivers.

AndrettiGTO wrote on 2026-05-23, 00:42:

It's a shame so many of these specialty cards no longer have the matching breakout cables. No cable for my ATI X800 XT AIW and almost unobtanium. :\

Yup, and even more annoyingly - many of these things could be acquired or worked around with general retail stuff - if only there were decent documentation for the pin assignments. A thing that is rather obligatory with any professional tech...

AndrettiGTO wrote on 2026-05-23, 00:42:

It's a shame so many of these specialty cards no longer have the matching breakout cables. No cable for my ATI X800 XT AIW and almost unobtanium. :\

Yes... its very frustrating. I also have an X800 AIW. Beautifull card but the cable just doesn't come up anywhere. One solution is to buy in box X800 with cable but that costs hundreds of dollars...

fosterwj03 wrote on 2026-05-23, 14:03:

Not super retro, but I bought a Quadro FX 3500 (the Quadro equivalent of the GeForce 7900 GS) for $19 US shipped on eBay (seller's photo). I'd like to see if I can get better performance in Windows NT 4.0 than my GeForce 6800 GS, but I'm worried the Quadro card might be a touch too new. Nvidia G70 chips seem to work with the last Quadro driver for NT 4, but the Quadro FX 3500 uses the G71 chip which Nvidia released shortly after their last NT 4 driver.

Regardless, I might have a use for this Quadro card in a Windows 98 build as long as I can get the system stable with later Nvidia drivers.

As I understand, g7x was simply a die shrink.
The down side of that is it enters bumpgate territory.

myne wrote on 2026-05-23, 15:20:

fosterwj03 wrote on 2026-05-23, 14:03:

Not super retro, but I bought a Quadro FX 3500 (the Quadro equivalent of the GeForce 7900 GS) for $19 US shipped on eBay (seller's photo). I'd like to see if I can get better performance in Windows NT 4.0 than my GeForce 6800 GS, but I'm worried the Quadro card might be a touch too new. Nvidia G70 chips seem to work with the last Quadro driver for NT 4, but the Quadro FX 3500 uses the G71 chip which Nvidia released shortly after their last NT 4 driver.

Regardless, I might have a use for this Quadro card in a Windows 98 build as long as I can get the system stable with later Nvidia drivers.

As I understand, g7x was simply a die shrink.
The down side of that is it enters bumpgate territory.

Well, I've found that die-shrunk chips on GPUs (regardless of the manufacture) also include subtle differences that the driver often needs to accommodate. The difference between the Radeon 9000 and 9200/9250 would be an example where earlier drivers can't identify the 9200/9250 even after modifying the .INF file. Best case scenario is that the driver ignores any differences if it can recognize and interface with the chip at all (that's the trick).

As for bumpgate, I haven't had any experience with it. The GPU's in my collection either aren't in the age-range or cool enough that the solder balls don't migrate. I'll have to see with this Quadro card.

fosterwj03 wrote on 2026-05-23, 15:31:

As for bumpgate, I haven't had any experience with it. The GPU's in my collection either aren't in the age-range or cool enough that the solder balls don't migrate. I'll have to see with this Quadro card.

With a suitable cooler, the G71 can be kept below 60°C even under heavy load, and this temperature should definitely prevent any potential damage.

fosterwj03 wrote on 2026-05-23, 15:31:

myne wrote on 2026-05-23, 15:20:

fosterwj03 wrote on 2026-05-23, 14:03:

Not super retro, but I bought a Quadro FX 3500 (the Quadro equivalent of the GeForce 7900 GS) for $19 US shipped on eBay (seller's photo). I'd like to see if I can get better performance in Windows NT 4.0 than my GeForce 6800 GS, but I'm worried the Quadro card might be a touch too new. Nvidia G70 chips seem to work with the last Quadro driver for NT 4, but the Quadro FX 3500 uses the G71 chip which Nvidia released shortly after their last NT 4 driver.

Regardless, I might have a use for this Quadro card in a Windows 98 build as long as I can get the system stable with later Nvidia drivers.

As I understand, g7x was simply a die shrink.
The down side of that is it enters bumpgate territory.

Well, I've found that die-shrunk chips on GPUs (regardless of the manufacture) also include subtle differences that the driver often needs to accommodate. The difference between the Radeon 9000 and 9200/9250 would be an example where earlier drivers can't identify the 9200/9250 even after modifying the .INF file. Best case scenario is that the driver ignores any differences if it can recognize and interface with the chip at all (that's the trick).

As for bumpgate, I haven't had any experience with it. The GPU's in my collection either aren't in the age-range or cool enough that the solder balls don't migrate. I'll have to see with this Quadro card.

Don't worry. There are dozens of videos and other reports of g7x working fine with infs.

MattRocks wrote on 2026-05-21, 20:23:

Thanks. I'm off the painkillers now. […]
Show full quote

Cuttoon wrote on 2026-05-21, 18:31:

Yep, show us! […]
Show full quote

Law212 wrote on 2026-05-21, 17:10:

Show us!

Yep, show us!

Also, maybe you should ban yourself from taking painkillers?

(Sorry, I hope you don't actually need them and wish you the best of health!)

Thanks. I'm off the painkillers now.

Attached is the only relevant photo I have on my phone right now. It's one of the better items - an AGPx4 card with blower-like cooler. The rear of the blower is closed, and the fan forces air along the vanes. But, I may actually have bid on this before going into hospital, and simply forgotten what happened when.

If it worked this would be my favourite GeForce2 Ti. It looks great in the photo, but the moment it initialised in my motherboard there was a great bang as the two caps by the backplate popped. So, it's currently unusable and currently not my favourite.

I'll share more tomorrow.

Haha that's crazy, my Leadtek GF2 Titanium also popped those two caps when I first turned it on. Thankfully it was undamaged and worked fine after replacing those caps - that blower cooler design is terrible because it lacks a way to direct the airflow over the vanes of the heatsink, since it's not a blower fan and is open. I made a shroud to improve it, perhaps it will help your card! (assuming you have a 3d printer)

Post with picture of it: Re: What retro activity did you get up to today?

It's also my favourite GF2 card with the interesting heatsink, thermal sensor and status LEDs. Leadtek tried but if you read old reviews for the card, its cooling system didn't work too well out of the box.

myne wrote on 2026-05-23, 23:21:

fosterwj03 wrote on 2026-05-23, 15:31:

myne wrote on 2026-05-23, 15:20:

As I understand, g7x was simply a die shrink.
The down side of that is it enters bumpgate territory.

Well, I've found that die-shrunk chips on GPUs (regardless of the manufacture) also include subtle differences that the driver often needs to accommodate. The difference between the Radeon 9000 and 9200/9250 would be an example where earlier drivers can't identify the 9200/9250 even after modifying the .INF file. Best case scenario is that the driver ignores any differences if it can recognize and interface with the chip at all (that's the trick).

As for bumpgate, I haven't had any experience with it. The GPU's in my collection either aren't in the age-range or cool enough that the solder balls don't migrate. I'll have to see with this Quadro card.

Don't worry. There are dozens of videos and other reports of g7x working fine with infs.

Thanks. I'm excited to find out.

Thermalwrong wrote on 2026-05-24, 00:55:

Haha that's crazy, my Leadtek GF2 Titanium also popped those two caps when I first turned it on. Thankfully it was undamaged and […]
Show full quote

MattRocks wrote on 2026-05-21, 20:23:

Thanks. I'm off the painkillers now. […]
Show full quote

Cuttoon wrote on 2026-05-21, 18:31:

Yep, show us!

Also, maybe you should ban yourself from taking painkillers?

(Sorry, I hope you don't actually need them and wish you the best of health!)

Thanks. I'm off the painkillers now.

Attached is the only relevant photo I have on my phone right now. It's one of the better items - an AGPx4 card with blower-like cooler. The rear of the blower is closed, and the fan forces air along the vanes. But, I may actually have bid on this before going into hospital, and simply forgotten what happened when.

If it worked this would be my favourite GeForce2 Ti. It looks great in the photo, but the moment it initialised in my motherboard there was a great bang as the two caps by the backplate popped. So, it's currently unusable and currently not my favourite.

I'll share more tomorrow.

Haha that's crazy, my Leadtek GF2 Titanium also popped those two caps when I first turned it on. Thankfully it was undamaged and worked fine after replacing those caps - that blower cooler design is terrible because it lacks a way to direct the airflow over the vanes of the heatsink, since it's not a blower fan and is open. I made a shroud to improve it, perhaps it will help your card! (assuming you have a 3d printer)

The attachment Leadtek-GF2-WinFast-Titanium-FAN-SHROUD-v1.zip is no longer available

Post with picture of it: Re: What retro activity did you get up to today?

It's also my favourite GF2 card with the interesting heatsink, thermal sensor and status LEDs. Leadtek tried but if you read old reviews for the card, its cooling system didn't work too well out of the box.

Thank you! 😀

I agree the cooling is shroud would improve GPU core temps, but before berating the WinFast design..

I also recall an acute challenge of the era was accommodating any card in PCI slot one while suffering case airflows driven by a single 80mm PSU fan. NVidia AGP coolers were growing deeper. So, maybe it wasn't just about cooling.

Several coolers of the time just didn't look comfortable with a big PCI card suffocating them. I remember magazines suggesting to leave the first PCI slot empty to give AGP fans breathing space, but that often meant dropping a capability. Maybe the WinFast design was also about giving the user their PCI slot back?

Arranging PCI cards was hardest in the GF2Ti era when a full fat PC had modem card, sound card, SCSI card, network card, USB card, Firewire card, TV capture card, and legacy 3Dfx cards. Everything was compromised. I remember preferring the cheapest £5 Realtek office NIC to accommodate an AGP fan, when the performance advice was £20+ Intel workstation/server NIC.

Cooling wasn't really a big concern back then.
The heatsink on my winfast ti200 was hilariously bad by all measures but it was plenty good enough.

Why bad? It was chromed, felt like plastic, was probably the cheapest junk aluminium in existence, the fins were tiny bumps, the main area above the die was smooth to fit the fan, the bottom contacting the ihs wasn't even close to flat and the whole thing was held in by 2 pushpins

myne wrote on 2026-05-24, 09:21:

Cooling wasn't really a big concern back then.

Hmm..

The space between AGP and PCI was tiny. The active cooling on cases was not targeting the AGP/PCI slot. Those were real inherited issues and the GF2Ti was the hottest card of the time.

Enthusiasts were leaving side panels off and vacating PCI slots. That shows real concern.