Lucid HyperFormance

"Lucid has made it clear here that the principle feature of HyperFormance is that it’s reducing the rendering workload on the dGPU by intercepting and removing some rendering tasks within a frame."

"The significance here is that while HyperFormance would be looking at a frame based on draw commands, "

"in which case it’s more likely that it’s an attempt to calculate and only render parts of an image that have changed from a previous frame – a GPU form of delta/inter-frame compression."

Little thought.
If an object is already on screen (say a car) would it be possible for lucix to render it to texture then instead of issuing the draw calls/pixel shading ect for it every frame could lucix use the texture as a sprite for a few frames ?
 
My guess is that they use vertical split frame rendering to divide work between GPUs and they drop strips to push up their framerate and then bullshit a bit about strips not being necessary, which is obviously bullshit for benchmarking purposes. Now there is some method to the madness, if you are relatively sure a frame won't be done to show on the vsync you might as well drop it and start on the next one instead of only showing a bit of it on the teared frame (assuming vsync is off) ... but meh, it's a dangerous guess to make. You might be working on a complex frame which would always take much longer than previous frames, in which case you throw away useful work (but it will ALWAYS increase the benchmarks even when throwing away useful work).

Personally I think vsync should be on always, even when benchmarking ... so all their magic is useless.
 
Last edited by a moderator:
just found this

One thing Lucid has made very clear is that while HyperFormance technically improves framerates, this is not the intention of the technology. Indeed it's largely a side effect of the fact that with HyperFormance the GPU is rendering some (or none) of a frame while still reporting to the CPU that the frame has been rendered and the buffers flipped.

So its a tech based on one of the oldest methods known of improving performance - Lying ;)
they say the main goal of their tech is reducing input lag, could that be like the trick with nv cards when setting frames to render ahead to 0 or 1 (cant remember exactly, but the default was 3)
 
BTW one thing I quite don't understand. This project seems to be not ONLY focus on hybird gpu system. But why all media are talking about hybrid with Lucid MVP?
 
Last edited by a moderator:
We published a story today about Virtu and HyperPerformance. During a quick analysis, my colleague Guillaume had a look at what they do in 3DMark by checking draw calls. Basically for some of the frames they don't clear the framebuffer, they kill all draw calls but 4 of them. Guess what these 4 draw calls do ? They draw fresh fps, frame and time counters on the previous frame :eek:
 
They should put slider in the driver setting how much frames in a second not to bother rendering! :LOL:
People could then set amazing 3DMark scores using low end GPU's and argue on forums that they are better than newest designs from both AMD and nVidia ...


On a serious note - what's the point of this? Reducing input lag? What for if I can't see frame rendered.
 
Back
Top