If gasoline engines burned their fuel as efficiently as possible, they would produce three by-products: water vapor (H2O), carbon dioxide (CO2) and nitrogen (N2).
Unfortunately, engines do not run perfectly, and as a result, they also produce three by-products commonly referred to as the "terrible trio" of automotive pollutants. This trio includes the following:
* Carbon monoxide (CO) – An odorless, tasteless, poisonous gas, carbon monoxide can cause a variety of health problems and even death. Many urban areas experience critically high levels of carbon monoxide, especially during the cold winter months when engines take longer to warm up and run cleanly
* Unburned hydrocarbons (HC) – Responsible for causing a variety of respiratory problems, unburned hydrocarbons can also cause crop damage and promote the formation of smog
* Oxides of nitrogen (NOX) – Like unburned hydrocarbons, oxides of nitrogen cause respiratory problems and promote the formation of smog
I can't even play the decent free games I got because I can't find them in the UI. It doesn't have sort by rating (or any other popularity metric) so you have to wade through the junk. Imagine paying for that experience...
I think my i9 was released right after the Spectre and Meltdown mitigations in 2019, but I seem to remember even more recent vulns in that family… so that could also be a factor.
If you get the "You're absolutely right!" response from an LLM that screwed up on a field you're familiar with and still let them play with your health, you're...courageous to say the least.
The original unit was the [slightly smaller] Roman mile which was standardized with the military in mind, i.e. able-bodied men in their prime. Seems like the average for men today is 2.5 feet or so which is more or less still on the money for 2k steps/1k paces to a mile.
Because most ARM SBCs are still limited to whatever linux distro they added support to. Intel SBCs might underperform but you can be sure it will run anything built for x86-64.
So what? Are you suggesting that Apple would have switched to RISC-V?
I like RISC-V (it's my job and I'm very involved in the community) but even now it isn't ready for laptops/desktop class applications. RVA23 is really the first profile that comes close and that was only ratified very recently. But beyond that there are a load of other things that are very much work in progress around the periphery that you need on a laptop. ACPI, UEFI, etc. If you know RISC-V, what does mconfigptr point to? Nothing yet!
Anyway the question was why would anyone switch from one proprietary ISA to another, as if nobody would - despite the very obvious proof that yes they absolutely would.
But it's still something grafted onto the existing architecture, of many grids with many blocks with many warps, and lots and lots of coordination and passing intermediary results around. It's only a 4x4x4 unit, afaik. There's still a lot of main memory being used to combine data, a lot of orchestration among the different warps and blocks and grids, to get big matrices crunched.
The systolic array is designed to allow much more fire and forget operations. It's inputs are 128 x 128 and each cell is its own compute node basically, shuffling data through and across (but not transitting a far off memory).
TPU architecture has plenty of limitations. It's not great at everything. But if you can design work to flow from cell to neighboring cell, you can crunch very sizable chunks of data with amazing data locality. The efficiency there is unparalleled.
Nvidia would need a radical change of their architecture to get anything like the massive data locality wins a systolic array can do. It would come with massively more constraints too.
reply