So RTX 5080 is slower than 4090...

[Vegasskill]

...and about 20% faster in Vulkan, 6% faster in OpenCL, and 9% faster in Blender than RTX 4080.
In my opinion it's effectively the same generation as RTX 40xx, which is why they are leaning so heavily on DLSS 4. Without DLSS 4 RTX 50xx series is nothing.

I can bet that DLSS 4 could work on 40xx series but NVIDIA will block this new multi frame generation option in drivers.

Link: https://www.techpowerup.com/331599/new-leak-reveals-nvidia-rtx-5080-is-slower-than-rtx-4090


So guys. Those of you who have RTX 4070, 4080 and 4090 cards, will you buy 5080?

[D. Flame]

Ursprünglich geschrieben von A&A:

Ursprünglich geschrieben von D. Flame:

WRONG.

If both GPUs are limited to 60FPS, then the amount of power each uses to reach that 60 FPS is their power efficiency.

If the 5090 is more power efficient, then it should be able to lower its power draw to match that lower FPS ask.

If you are running both cards at full tilt, then you are not testing their power efficiency.

This is a subjective question.
If we take something like the 4060 and compare it to the 5090 at 1080p 60FPS, no ray tracing and no DLSS, this is what happens.

You have tensor cores and ray tracing acceleration cores staying in standby and wasting power. The RTX4060 has 17k less streaming cores, and if 3k is enough, the rest will stay in standby. Again wasting power.

Yes, because they lack power efficiency. If they were power efficient, they would be able to scale back and turn off what they are not using to conserve power.

That is how power efficiency works. It uses as much power as it needs to get the job done, and it tones it down or turns it off when it isn't needed.

Like if it costs 500W to get 60 FPS, but you uncap it and suddenly get 66 FPS for 500W, your power draw didn't suddenly get 10% more efficient. Your power draw never changed.

[A&A]

Ursprünglich geschrieben von D. Flame:

Yes, because they lack power efficiency. If they were power efficient, they would be able to scale back and turn off what they are not using to conserve power.

That is how power efficiency works. It uses as much power as it needs to get the job done, and it tones it down or turns it off when it isn't needed.

Like if it costs 500W to get 60 FPS, but you uncap it and suddenly get 66 FPS for 500W, your power draw didn't suddenly get 10% more efficient. Your power draw never changed.

So you want the size of the current chip to double, which will make very little difference and won't appeal to your wallet, whereas the RTX4060 has the same lack when it comes to lighter workloads, and so you want to see power efficiency on a generational rather than individual scale.
The same goes for CPUs. Ryzens may be praised for power efficiency, but gaming on a Threadripper is stupid and it is less power efficient as the core count increases and you have more of them sitting in standby.