News Enthusiast hacks FSR 4 onto RX 7000 series GPU without official AMD support, returns better quality but slightly lower fps than FSR 3.1

I wouldn’t call it a slight drop in performance, the 7000 series do not come with Tensor cores or whatever AMD calls them that’s why the performance hit is so high.
 
Still looks like you get a 30% bump or so from native, so it still helps in that regard. The fact that it looks a little better is just a nice bonus. I wonder how AMD's own porting efforts are coming along at this point.
 
Well, this explains why we don't have anything official out for the 7000 series. Performance matters...and it's not worth a 30% hit to support FSR4 on these cards. If it's THAT important to people, upgrade!
 
Well, this explains why we don't have anything official out for the 7000 series. Performance matters...and it's not worth a 30% hit to support FSR4 on these cards. If it's THAT important to people, upgrade!
There are no laptop dGPU or iGPU that support FSR4, even for the Strix Halo beast, so... only a possibility for desktop people.
 
FSR4 variation might be possible on the NPU.
Microsoft has ASR, but it only works on Snapdragon for now.
https://videocardz.com/newz/microso...er-resolution-system-level-upscaling-for-npus
If MS can do it, I don't see how AMD couldn't.
Well, to be perfectly honest, a lot of this is WAY over my head...but I would have to say that software technology like FSR 4.0 is DESIGNED from the outset to take advantage of the new hardware that is only available on the new GPUs...and if it takes resources to develop this tech for older GPUs, I would be 100% against them even trying. I would prefer that ALL of their efforts go into making their drivers better for the new cards and to improving the experience on RDNA4 and UDNA cards, not catering to (lets face it) a small number of owners of previous gen cards.
 
Well, to be perfectly honest, a lot of this is WAY over my head...but I would have to say that software technology like FSR 4.0 is DESIGNED from the outset to take advantage of the new hardware that is only available on the new GPUs...and if it takes resources to develop this tech for older GPUs, I would be 100% against them even trying. I would prefer that ALL of their efforts go into making their drivers better for the new cards and to improving the experience on RDNA4 and UDNA cards, not catering to (lets face it) a small number of owners of previous gen cards.
I looked into it, and supposedly FSR4 works off of Tensor cores, or FP8 and sparsity.
FP8 is something an NPU has, but the ones in RDNA4 are built into the GPU pipeline per CU, where as Strix Halo only has the NPU as a separate part of the same I/O tile.
 
  • Like
Reactions: John Nemesh
I looked into it, and supposedly FSR4 works off of Tensor cores, or FP8 and sparsity.
FP8 is something an NPU has, but the ones in RDNA4 are built into the GPU pipeline per CU, where as Strix Halo only has the NPU as a separate part of the same I/O tile.
I am a HUGE fan of what AMD is doing with their APUs and the Strix Halo is a great product...but their APUs GPU tech is always going to trail what they are doing with their dedicated graphics cards, unfortunately.

I CAN see a day soon, though, when APUs will be able to completely eliminate the need for what is currently now a "60 series" dedicated graphics solution (nVidia or AMD). It's not quite here with Strix Halo, though.
 
  • Like
Reactions: Notton
I am a HUGE fan of what AMD is doing with their APUs and the Strix Halo is a great product...but their APUs GPU tech is always going to trail what they are doing with their dedicated graphics cards, unfortunately.

I CAN see a day soon, though, when APUs will be able to completely eliminate the need for what is currently now a "60 series" dedicated graphics solution (nVidia or AMD). It's not quite here with Strix Halo, though.
I doubt APUs can really do that, not unless we somehow manage to make them just as cheap as other solutions. A Strix Halo laptop can cost the same as 2 or 3 laptops with a RTX 4060.
 
I doubt APUs can really do that, not unless we somehow manage to make them just as cheap as other solutions. A Strix Halo laptop can cost the same as 2 or 3 laptops with a RTX 4060.
Yes, but that's not a direct reflection of what it costs to manufacture, rather AMD saying "These are worth more in the marketplace".

APUs are going to be the future of computing...look at Apple for a perfect example. There are HUGE benefits to having the GPU integrated with the CPU as opposed to a dedicated card that has to use PCIe. This won't happen overnight, but eventually, separate cards will go the way of math co-processors (remember those?).