Intel has oneAPI.
https://www.intel.com/content/...
The A770 was unique because it had 16GB memory at a street price of $279 and sometimes even less. I really hope we'll see a B7xx with at least 16GB RAM.
Here is llama accelerated on intel HW:
https://github.com/intel-analy...
TSMC is now a monopoly... that can't be good.
Intel is not closing down the fabs. It merely splits them up into a separate company.
This act doesn't diminish TSMC competition. In fact, it introduces more competition because until now Intel fabs were used almost exclusively by Intel, but now they'll compete in the open market.
1. TOPS are like MIPS. You need to measure actual performance, not just multiple number of MACs by the clock rate. You network has other parts beyond the MAC units, and if you are not efficient, it is going to slow you down, no matter how many TOPS/s your MACs do, similar to Amdahl's law for multiprocessing.
2. The IMX500 has been announced on May 2020, and has a mature toolset. It's just this specific RPi product that is new.
3. Check the TOPS/W, not only TOPS, but even TOPS/W numbers are skewed (https://semiengineering.com/lies-damn-lies-and-tops-watt/).
4. Even better, check actual benchmarks: https://arxiv.org/pdf/2307.078...
> It's indiscriminate because they are not targeting anyone, they are just deploying hundreds of these things and hoping a few end up in the hands of their targets.
What a joke.
This was a highly targeted attack against a specific pager order by Hezbolla for its terrorists who ordered those pagers because they didn't want to use its phones as they felt those phones were insecure.
The PYPL table (2nd one) was copied incorrectly and dropped C++ at 5 (That's why the table ends at #9).
Meaningless at the moment.
Apple is already offering the M2 model on its website.
Two models, both with 8GB RAM and 256GB/512GB SSD.
https://www.apple.com/shop/buy...
Maybe higher-end models will come with more RAM.
From what I see, his github account is alive and kicking,
https://github.com/Marak
but:
Once the repo was cleaned and push-forced with a single commit, all his 100+ github-forks were reset as well.
How did this happen? I assumed that if you fork on github, nothing can happen to your fork.
I would take that as a lesson not to rely on github forks as a way to guarantee access to code.
If you want to really fork, start your own repository with a real copy!
Anyhow, a new public fork has been formed:
https://github.com/faker-js/fa...
The hash represents a specific commit, and actually a specific state of the repository at a specific commit.
If you checkout a specific hash, you get the exact version you expect. period.
So they extract the plastic and then what? Make more useless plastic products from it, who were probably not bought if they were not made.
If you really want to recycle, find an existing manufacturer who will use it to *replace* its existing plastics source, don't create yet another manufacturer of plastic products, because this will not reduce plastic consumption by existing manufacturers.
It's all a question of complexity. If the current algorithms runs for hours or days and gets you worse results than AI algorithms that runs much faster, then it is an improvement. You can't "debug" either the traditional or the new output.
What you usually do is take the new netlist and formally compare it to the original logical design written in Verilog (formal equivalence tools).
If you want to be sure, you extract the netlist from the final GDS-II layout.
In both cases , AI or not, you use the same techniques and tools to formally verify your output.
No problem, it's just the physical layout that's AI-driven.
It's like saying you can't debug your code because the compiler is AI-driven,
and for chip design, if you need to debug actual cells on the synthesized netlist, it is already quite messy even without AI.
Logic gets optimized out if possible, duplicated when you need to drive large number of inputs, and signals are retimed if you allow your tool to do it,
which brings it whole new logical content not found in your design.
I beg to differ.
The chip designers I know are still hanging onto 20 years old Verilog coding styles, and hardly knows about the RTL capabilities of SystemVerilog.
I'm not to blame them, because both Synopsys and Cadence took years to implement those features after they were standardized.
In this case we are talking about AI driven placement, which is transparent to the peasant physical design engineer. He runs the same script, perhaps adding a new command line flag there, and that's all he needs to know. If he gets better results more quickly than before, all the better. It doesn't change his workflow a bit.
Found an even earlier reference to NetBSD on May 19th, 1993.
https://groups.google.com/g/comp.os.386bsd.questions/c/PniapyOpRKQ/m/iOrsSxp0p8oJ
There are earlier references to freeBSD, but these mostly refer generically to a free version of BSD (mostly to distinguish the free 386BSD from the BSDI commercial 386BSD). They do not specifically refer to a project named "FreeBSD", or simply propose this name doe 386BSD.
https://groups.google.com/search/conversations?q=freebsd%20before%3A1993%2F6%2F1
Without going deeper into the debate, the 1st mention of both FreeBSD and NetBSD I could find on Usenet was on July 1993.
https://groups.google.com/g/comp.os.386bsd.misc/c/0BOG_ZNBv2Y/m/cyrQD1jQ9XkJ
"What man has done, man can aspire to do." -- Jerry Pournelle, about space flight
