Ampere vs Turing wattage equalized benchmarks

[Guest]

Are there any reviews on the Ampere architecture cards (3080 or 3090) that reduced the power of those cards to equalize them with Turing i.e. a 2080ti at 250w vs 3080 at 250w, or a 3080 at 200w vs a 2080ti at 200w.

I would be really interested in those numbers if someone knows a site or a video where that was tested.

[Jumballi]

Atm no, and we probably won’t be seeing any as it’s both a waste of time and difficult to achieve.

 

Anyone buying a card will be attempting to use the card at the boundaries set by that card. If your system only has 240w to give, and the more powerful card is 2x performance per watt but uses 300w, then you can’t go with it regardless.

[Guest]

3 minutes ago, Jumballi said:

Anyone buying a card will be attempting to use the card at the boundaries set by that card. If your system only has 240w to give, and the more powerful card is 2x performance per watt but uses 300w, then you can’t go with it regardless.

That's why I specified the architecture in my question. I'm more interested in that than specific gaming GPUs. I guess I have to wait until lower TDP cards release like the 3070 to get those numbers.

[boggy77]

15 minutes ago, Medicate said:

That's why I specified the architecture in my question. I'm more interested in that than specific gaming GPUs. I guess I have to wait until lower TDP cards release like the 3070 to get those numbers.

there are fps/watt benchmarks out there

[Guest]

4 minutes ago, boggy77 said:

there are fps/watt benchmarks out there

what website is this?

[Blademaster91]

You'll have to undervolt the 3080 to get close to 2080Ti power usage.

[boggy77]

4 minutes ago, Medicate said:

what website is this?

https://www.techspot.com/review/2099-geforce-rtx-3080/

[porina]

47 minutes ago, Jumballi said:

Atm no, and we probably won’t be seeing any as it’s both a waste of time and difficult to achieve.

It depends on the level of precision desired but it can be as simple as changing a slider in MSI Afterburner. If you can control for other factors it can be a good indication of the changes between architectures. The biggest hurdle here is that generally speaking the more execution units there are, the more power efficient it gets (to some point). And that can be a bigger impact than the architectural differences.

[SkilledRebuilds]

9 hours ago, Medicate said:

Are there any reviews on the Ampere architecture cards (3080 or 3090) that reduced the power of those cards to equalize them with Turing i.e. a 2080ti at 250w vs 3080 at 250w, or a 3080 at 200w vs a 2080ti at 200w.

I would be really interested in those numbers if someone knows a site or a video where that was tested.

You could probably find 2080Ti power modded data (300-320w and higher)

While not exactly what you are after... 320w vs 320w videos exist..

 

 

 

 

[Ankerson]

9 hours ago, Blademaster91 said:

You'll have to undervolt the 3080 to get close to 2080Ti power usage.

 

Actually they are about the same and or the same depending on the actual cards.

 

Some 2080Ti's pull every bit of 375W and more depending on the card...

 

 

[SkilledRebuilds]

1 hour ago, Ankerson said:

 

Actually they are about the same and or the same depending on the actual cards.

 

Some 2080Ti's pull every bit of 375W and more depending on the card...

 

 

I put a 400w Vbios on mine before I settled in at a 380w vBios (Altho I only require 320-340w for my targets to be held)

It used whatever I gave it if I let it... 400w easy (stock clocks (boosting still to 1850-1900) with 400w unlocked power)
More watts is one thing... pumping the extra wattage increases temps, drops clocks,.. its better to stick around 300-350w (on AIR) and not drop as much on the clocks while figuring out maxboost along the way with testing.