PC Gaming

4 readers

1 users here now

Discuss Games, Hardware and News on PC Gaming **Discord** https://discord.gg/4bxJgkY **Mastodon** https://cupoftea.social **Donate** https://ko-fi.com/cupofteasocial **Wiki** https://www.pcgamingwiki.com

founded 1 year ago

Nvidia's DLSS upscaling really does need those AI-accelerating Tensor cores after all (www.pcgamer.com)

submitted 1 year ago by geosoco@kbin.social to c/PCGaming@kbin.social

4 comments fedilink hide all child comments

Investigation into Nvidia GPU workloads reveals that Tensor cores are being hammered, just incredibly briefly.

...

An intrepid Reddit poster, going under the handle Bluedot55, leveraged Nvidia's Nsight Systems GPU metric tools to drill down into the workloads running on various parts of an Nvidia RTX 4090 GPU.

Bluedot55 ran both DLSS and third party scalers on an Nvidia RTX 4090 and measured Tensor core utilisation. Looking at average Tensor core usage, the figures under DLSS were extremely low, less than 1%.

Initial investigations suggested even the peak utilisation registered in the 4-9% range, implying that while the Tensor cores were being used, they probably weren't actually essential. However, increasing the polling rate revealed that peak utilisation is in fact in excess of 90%, but only for brief periods measured in microseconds.

you are viewing a single comment's thread
view the rest of the comments

[–] lustrum@sh.itjust.works 1 points 1 year ago (1 children)

I'm sure this wasn't that unknown. It's almost free to run on the tensor cores asynchronously. While running on shaders is less efficient and uses up their time that could be, well, shading.

[–] conciselyverbose@kbin.social 1 points 1 year ago* (last edited 1 year ago) (1 children)

It was the central premise of DLSS from the start lol.

Tensor cores can be used other ways, but the entire reason they were part of RTX was to accelerate DLSS, and they told us that.

[–] lustrum@sh.itjust.works 0 points 1 year ago (1 children)

Definitely but I'm just saying it wasn't unknown it could have probably ran on shaders.

[–] conciselyverbose@kbin.social 1 points 1 year ago

In the sense that shaders are capable of replicating the operations, sure. But the reason for tensor cores is the same reason as for any other hardware feature. It's obscenely faster and more efficient to do math you'll do frequently with dedicated hardware.