slashdave

4 ms·

31.
by slashdave 2mo ago
Why do think they would want to be vindictive? The point is to prevent leaks, leaving lapses is exactly counter to that goal.
32.
by slashdave 2mo ago
It's not hard if you are honest
33.
by slashdave 2mo ago
Those that claim their opponents are in denial are most likely in denial themselves
34.
by slashdave 2mo ago
Since when does realism become hope? (It's the other way around)
35.
by slashdave 2mo ago
Shh! Don't upset the hype train
36.
by slashdave 2mo ago
No, we do this kind of thing regularly. Optimizations often depend on the model that is using them. Namely, fusion techniques.
37.
by slashdave 2mo ago
Indeed. Compilers do have to make compromises, and it is not possible from code alone to understand the tradeoffs.
38.
by slashdave 2mo ago
This SV mentality that technology can solve all problems is rather tiring, really.
39.
by slashdave 2mo ago
Have your agent take care of it
40.
by slashdave 2mo ago
> We only know how to do it by means of considerable sacrifice Little sacrifice actually > Solving the issue would be doing it without sacrifice So... you are expecting magic? LLMs cannot create resources out of thin air.
41.
by slashdave 2mo ago
Yeah, a corner case (streaming) and good developers working very hard. The original SSE (SIMD) extensions were actually written with this use case in mind. When they first came out, the compilers were awful. They've since caught up.
42.
by slashdave 2mo ago
Except for some odd corner cases, it is very hard for a good engineer to beat compilers these days (they are very good now). Not to mention, tuning depends on the platform, and often the exact model of chip.
43.
by slashdave 2mo ago
Show it the disassembler...
44.
by slashdave 2mo ago
I'm... confused. Don't we use macro assemblers anymore?
45.
by slashdave 2mo ago
Is AWS listening?
46.
by slashdave 2mo ago
Seriously? We already know how to solve all of these issues. What we lack is collective political will.
47.
by slashdave 2mo ago
What? No. Frontier physics is experiment driven.
48.
by slashdave 2mo ago
It will not happen with existing LLM techniques.
49.
by slashdave 2mo ago
> we are also seeing incredible advances in software performance Incredible? > open ai announced like 15% improvement by fixing gpu kernel issue That is... ordinary software optimization.
50.
by slashdave 2mo ago
> The sooner people can be broken out of their denial There is irony here
51.
by slashdave 2mo ago
Cursor achieved fast adoption by making it simple to move from Visual Studio code. Double-edged sword. It is also simple to move back to VS code and agent extensions.
52.
by slashdave 2mo ago
Don't forget: resources
53.
by slashdave 2mo ago
... we don't know, because this one can't?
54.
by slashdave 2mo ago
So.. you don't live in Taiwan?
55.
by slashdave 2mo ago
> why not just let it fix everyone's code? That's the stated idea. Fix code before releasing to the public.
56.
by slashdave 2mo ago
Does "value" have to be "efficient"? I am not in the habit of judging other people's values. If it makes someone happy, why not? Of course: capitalism. So slippery slope and all that.
57.
by slashdave 2mo ago
A million dollars really isn't so much these days
58.
by slashdave 2mo ago
In a simple way, money = value. If someone is "paying", there is demand.
59.
by slashdave 2mo ago
Just make the contents of CLAUDE.md "@AGENTS.md" Now, if only I could get Codex to read rules.
60.
by slashdave 2mo ago
Will it make generous use of parentheses?

More ›