I did some research when I designed the AGP Power Monitor and did not get any conclusion for AGP1.0 back then either. I just went with the 2.0 spec, giving my max to be 10A (for safety margin) and be good with it. Since the connectors are physically the same, I'd assume similar specs.

The real current delivery capability would more depend on the mainboard itself, meaning if all rails come from the ATX power supply or in older cases some are generated on the board.

Thanks tehsiggi.

tehsiggi wrote on 2026-08-19, 11:38:

... Since the connectors are physically the same, I'd assume similar specs ...

^ Yep, that stands to reason.

Interestingly the max current limits for VCC3.3, VCC5 & VCC12 are the same in AGP 3.0:

6 amps @ 3.3V are 19.8 watts.

Some of those "Shevalier's favourites" 😀 linear regulator cards blatantly exceeded this limit:

shevalier wrote on 2026-08-18, 09:53:

... Now, an FX5500 on linear regulators — that’s impressive.It probably draws around 50 watts. 😀 ...

Provided extra good cooling, you can get away with some overloading ... but only up to some point before the RBMK reactor starts melting down 😀 .

TELVM wrote on 2026-08-20, 08:55:

Interestingly the max current limits for VCC3.3, VCC5 & VCC12 are the same in AGP 3.0:

Yeah, the connector didn't change. I suppose the limit is also the amount of pins you get per voltage. 12V for example are a single pin.

TELVM wrote on 2026-08-20, 08:55:

6 amps @ 3.3V are 19.8 watts.

Some of those "Shevalier's favourites" 😀 linear regulator cards blatantly exceeded this limit:

Well, the APM that havli is using is not a grade A calibrated tool, which is the same as for mine (though I roughly calibrated mine with a fixed load before) - But it's measurement, so there's always some room for error. I'd say those cards are probably within spec 😉