tarik wrote on 2025-09-11, 09:14:

Absolutely right, I agree with this too. 😀 Maybe it’ll spark someone’s curiosity and they’ll make a video about it—lowering the core clock through the BIOS first, then reducing the voltage.

To minimize potential risks, it might be better to test this on motherboards that have both PCIe and AGP slots, so the BIOS can be reflashed if anything goes wrong. I’m just putting these ideas forward as suggestions. 😀

I'm not a guy for videos. However I have all the equipment and knowledge needed to perform such a test. The question is? What is it you are looking for? Reduction in power consumption? Temperatures? Influence on stability / longevity is hard to determine without leaving it running for a year. Tbh, my 9800s (if not faulty) have been pretty solid. I never really had a "good day/bad day" experience with them. Not even back in the day.

I've got my free afternoon today and am happy to run some tests and sacrifice a 9800 for that. And if anything fails.. either flashing with a PCI card or eeprom flasher will always work out.

tarik wrote on 2025-09-11, 09:31:

st31276a wrote on 2025-09-11, 09:25:

I agree. […]
Show full quote

The Serpent Rider wrote on 2025-09-11, 09:09:

Majority of old cards actually had pretty hefty headroom, as it should be, that's why overclocking was possible in the first place. Modern GPUs, on the other hand, really push it to the limit.

I agree.

Modern graphics cards are the component that most frequently fail.

From my trident 9000 to the geforce2 mx still work. The geforce 4, 6, 8 and 200 series I had since, has all failed. I haven't even overclocked them.

Haven't purchased a graphics card since, running on stuff I can pick up for free and keep using them until they fail.

That’s exactly why I started this topic 😀 to find new ways to keep these cards alive. Cooling is important of course, but instead of focusing only on that, I’m talking about creating new BIOS versions. The goal should be to ease the load on the cards, ignoring small differences like an 8 FPS drop.

Well for AGP cards there will not be many that would profit from a different bios, since most of them don't have any software voltage control. And just underclocking is worthless, as mentioned.

tehsiggi wrote on 2025-09-11, 09:32:

I'm not a guy for videos. However I have all the equipment and knowledge needed to perform such a test. The question is? What is […]
Show full quote

tarik wrote on 2025-09-11, 09:14:

Absolutely right, I agree with this too. 😀 Maybe it’ll spark someone’s curiosity and they’ll make a video about it—lowering the core clock through the BIOS first, then reducing the voltage.

To minimize potential risks, it might be better to test this on motherboards that have both PCIe and AGP slots, so the BIOS can be reflashed if anything goes wrong. I’m just putting these ideas forward as suggestions. 😀

I'm not a guy for videos. However I have all the equipment and knowledge needed to perform such a test. The question is? What is it you are looking for? Reduction in power consumption? Temperatures? Influence on stability / longevity is hard to determine without leaving it running for a year. Tbh, my 9800s (if not faulty) have been pretty solid. I never really had a "good day/bad day" experience with them. Not even back in the day.

I've got my free afternoon today and am happy to run some tests and sacrifice a 9800 for that. And if anything fails.. either flashing with a PCI card or eeprom flasher will always work out.

tarik wrote on 2025-09-11, 09:31:

st31276a wrote on 2025-09-11, 09:25:

I agree. […]
Show full quote

I agree.

Modern graphics cards are the component that most frequently fail.

From my trident 9000 to the geforce2 mx still work. The geforce 4, 6, 8 and 200 series I had since, has all failed. I haven't even overclocked them.

Haven't purchased a graphics card since, running on stuff I can pick up for free and keep using them until they fail.

That’s exactly why I started this topic 😀 to find new ways to keep these cards alive. Cooling is important of course, but instead of focusing only on that, I’m talking about creating new BIOS versions. The goal should be to ease the load on the cards, ignoring small differences like an 8 FPS drop.

Well for AGP cards there will not be many that would profit from a different bios, since most of them don't have any software voltage control. And just underclocking is worthless, as mentioned.

Reducing system power by 15% already extends hardware lifespan. I’m speaking from 15 years of experience, and this is actually a well-known method—used in systems that run 24/7, like data centers and render farms.

When I say “testing,” I mean being able to lower these values while keeping the system stable. That can be verified with a half-hour stress test. I’ve already done this myself on my own card, an ATI FireGL X3 AGP.

What I’m suggesting here is to open up a new area of exploration for retro enthusiasts—and maybe even develop new BIOS versions for these cards.

I get your point, I am just no sure what you would now want somebody to explore / test there? If it's possible to reduce the power consumption of a 9800 Pro? Certainly, I can do that and bench it incl. a stress test.

This will however not work with a BIOS only, so it's always with a hardware modification.

Nobody ever doubted that reducing the power consumption can help expanding the life-span of something.
But you specifically asked for undervolting of AGP cards (via BIOS), which is possible, but mostly with hardware changes involved.

So we have two distinct cases:
- Cards with BIOS voltage control => easy to undervolt + underclock for power reduction and less requirement in cooling
- Cards with hardware voltage control => harder to undervolt due to the need of a hardware mod, but possible with the same result as for the BIOS voltage variant

tehsiggi wrote on 2025-09-11, 09:44:

I get your point, I am just no sure what you would now want somebody to explore / test there? If it's possible to reduce the pow […]
Show full quote

I get your point, I am just no sure what you would now want somebody to explore / test there? If it's possible to reduce the power consumption of a 9800 Pro? Certainly, I can do that and bench it incl. a stress test.

This will however not work with a BIOS only, so it's always with a hardware modification.

Nobody ever doubted that reducing the power consumption can help expanding the life-span of something.
But you specifically asked for undervolting of AGP cards (via BIOS), which is possible, but mostly with hardware changes involved.

So we have two distinct cases:
- Cards with BIOS voltage control => easy to undervolt + underclock for power reduction and less requirement in cooling
- Cards with hardware voltage control => harder to undervolt due to the need of a hardware mod, but possible with the same result as for the BIOS voltage variant

Yes, you might be right. I’ll share my own test data at a suitable time.

tehsiggi wrote on 2025-09-11, 09:44:

I get your point, I am just no sure what you would now want somebody to explore / test there? If it's possible to reduce the pow […]
Show full quote

I get your point, I am just no sure what you would now want somebody to explore / test there? If it's possible to reduce the power consumption of a 9800 Pro? Certainly, I can do that and bench it incl. a stress test.

This will however not work with a BIOS only, so it's always with a hardware modification.

Nobody ever doubted that reducing the power consumption can help expanding the life-span of something.
But you specifically asked for undervolting of AGP cards (via BIOS), which is possible, but mostly with hardware changes involved.

So we have two distinct cases:
- Cards with BIOS voltage control => easy to undervolt + underclock for power reduction and less requirement in cooling
- Cards with hardware voltage control => harder to undervolt due to the need of a hardware mod, but possible with the same result as for the BIOS voltage variant

Now I understand what you mean. Even if we lower the voltage through the BIOS, the hardware will still draw the same actual voltage. You’re saying this is because the power input isn’t adjustable at the hardware level. Yes, I hadn’t thought about that. That’s an important point—thank you.

Just to clarify:
What I'm trying to tell is that there isn't even a way to influence a voltage in the BIOS in the first place. So there's nothing we can do in the BIOS with regards to voltage to begin with.

The voltage regulators are completely isolated from any GPU control. Software control is a principle that isn't present on most AGP cards. So it's just a switching regulator with a fixed feedback loop that sets a voltage, done. Only the feedback loops pair of resistors influences the output voltage.

Having read the discussion... my opinion - people often get way, way too obsessed with things like temperatures without any real evidence it does anything meaningful in specific case .

Yes, theoretically less heat is always good, practically arbitrary 15% less heat may very well mean a difference between 50000 hours and 50010 hours or something.

It is better to keep things... sensible. When there is a way to reduce voltage a bit without losing performance - why not? When you start downclocking or messing with hardware - nope. Losing performance for uncertan gains makes no sense. And I've seen way too many hardware broken by people trying to "extend its life" - everything from liquid metal to bent PCBs, cracked dies, etc, etc.

If the goal is to reduce heat and extend longevity, would over-speccing and underclocking not be an option?

Unless one is building a period correct retro system, often times people use graphics cards that are beyond specs for the period in question. For example, using GeForce 4 or FX cards in Windows 98 builds.

In such scenarios, there is more than enough performance headroom to reduce performance of the cards while still maintaining relatively high performance for those use cases.

Shponglefan wrote on 2025-09-11, 10:37:

If the goal is to reduce heat and extend longevity, would over-speccing and underclocking not be an option?

Unless one is building a period correct retro system, often times people use graphics cards that are beyond specs for the period in question. For example, using GeForce 4 or FX cards in Windows 98 builds.

In such scenarios, there is more than enough performance headroom to reduce performance of the cards while still maintaining relatively high performance for those use cases.

Underclocking alone does not reduce power consumption very much. I can add some numbers for that later on.

Shponglefan wrote on 2025-09-11, 10:37:

If the goal is to reduce heat and extend longevity, would over-speccing and underclocking not be an option?

Unless one is building a period correct retro system, often times people use graphics cards that are beyond specs for the period in question. For example, using GeForce 4 or FX cards in Windows 98 builds.

In such scenarios, there is more than enough performance headroom to reduce performance of the cards while still maintaining relatively high performance for those use cases.

Yes, deliberately creating a bottleneck could also be a method for this. Nice idea.