AMD Infinity Cache patent hints at RDNA 2 secret sauce

[ik9000]

Hey I'm not into livestreaming food being fed to people lol

yeah we'd figured that much out

[3dcandy]

yeah we'd figured that much out

Though it IS a tempting take on the whole thing...

[Corky34]

It's just a big L2 cache with a fancy name. Technologically there's nothing new here. A bigger cache means fewer trips to the GPU memory, reducing the dependence on bandwidth. There's nothing here that GPUs don't already do, so no reason to think it'd need per-game optimisation.

Having read the TomsHardware story i think it maybe (big maybe there) more than just a bigger lvl2 cache, going on that article they were talking about sharing the lvl1 cache between cores, they mentioned something about some research that showed something like 15-50% performance improvement.

EDIT: this is the section..

There's a possibility that the Infinity Cache may be related to a patent that AMD filed last year on (PDF>)Adaptive Cache Reconfiguration Via Clustering. Subsequently, the authors (PDF>)published a paper on the topic. It talks about the possibility of sharing the L1 caches between GPU cores.

Traditionally, GPU cores have their own individual L1 cache, while the L2 cache is shared among all the cores. The suggested model proposes that each GPU is allowed to access the other's L1 cache. The objective is to optimize the caches' use by eliminating the replicated data in each slice of the cache. The results are pretty amazing. Across a suite of 28 GPGPU applications, the new model improved performance by 22% (up to 52%) and energy efficiency by 49%.

I've read neither of those PDF's so don't have a clue if they could be related to this Infinity Cache thingy.

[Iota]

Having read the TomsHardware story i think it maybe (big maybe there) more than just a bigger lvl2 cache, going on that article they were talking about sharing the lvl1 cache between cores, they mentioned something about some research that showed something like 15-50% performance improvement.

EDIT: this is the section..

I've read neither of those PDF's so don't have a clue if they could be related to this Infinity Cache thingy.

If so it sounds like a free win, increased performance while not massively increasing power consumption (like Ampere).

[3dcandy]

Having read the TomsHardware story i think it maybe (big maybe there) more than just a bigger lvl2 cache, going on that article they were talking about sharing the lvl1 cache between cores, they mentioned something about some research that showed something like 15-50% performance improvement.

EDIT: this is the section..

I've read neither of those PDF's so don't have a clue if they could be related to this Infinity Cache thingy.

That's the basic idea - Infinity Fabric bus to share the cache. NV partnered with Micron for GDDR6X and it is a major reason for Ampere being hot and power hungry, but it is fast

[lumireleon]

as new tech is becoming bloated and saturated (boring) AMD/INTEL/NVIDIA should go the Tesla way and "reduce manufacturing costs by finding new ways of manufacturing"

[ik9000]

as new tech is becoming bloated and saturated (boring) AMD/INTEL/NVIDIA should go the Tesla way and "reduce manufacturing costs by finding new ways of manufacturing"

Nvidia already are. By not making enough of anything.

[3dcandy]

Nvidia already are. By not making enough of anything.

That IS the Tesla way lol

PS Does their boss smoke weed on investors Skype calls too?

[Zhaoman]

Yeah what else has Jensen got baking in his oven... Guaranteed high (performance numbers)

[ik9000]

Yeah what else has Jensen got baking in his oven... Guaranteed high (performance numbers)

unobtanium?