If a C770 or C970 card can prove to be a contender like B580 is at the respective price point then people will buy them.
Of course these are parts that have been in the works for several years now. They’ve had time to see what Nvidia and AMD are doing with their competitive products
I can’t tell if there’s just no stock, or huge demand, but either way it seems like a great product. If only you could buy it.
Intel still has to publish several additional manuals, to cover Xe in Meteor Lake, Xe in Arrow Lake H, Xe 2 in Lunar Lake and Xe 2 in discrete GPUs, and then Xe 3 in Panther Lake, to bring their documentation up to date.
You can extract most of the GPU information from the open-source Linux drivers and LLVM compiling back-end, but this is a poor substitute for actual GPU ISA documentation.
The compiler is LLVM
https://github.com/llvm/llvm-project/tree/main/llvm/lib/Targ...
It's all there. For the driver/runtime everything is also open and upstreamed into the Linux kernel (register map and pcie bar and etc). The packet protocol isn't well-documented but documented enough that tinygrad managed to build their own driver from scratch.
edit: they are scary huge.
Why can't AI come up with some kind of fast universal computation machine that doesnt have the need for the siliconized version of ifdefs?
One such method was an improvement in matrix matrix multiplication in 2022. So I’m sure there is a potential discovery of many efficient numerical algorithms that will be uncovered with machine learning.
https://www.nature.com/articles/s41586-022-05172-4
My parent comment was a knee jerk response to the over confidence and wishful thinking of computer work people to a magic AI box solving all the hard physical problems like its some simple conversation prompt away.
Is it surprising that silicon does too?
Given that human brains and silicon has nothing in common, then yes, it is surprising. What makes you think otherwise?
I get that your take qualifies as stoner philosophy, so I won't push too far, but really. Do give some concrete examples to support your ideas rather than GPT fluff.
When you write programs do you not use heuristics? And tests?
Because fixed silicon inevitably has fixed "baked in" choices.