Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
muricula
searching Neon…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
6 ms
·
1.
▲
by
muricula
1mo ago
Price is confounded by VC subsidies, economies of scale, and inference optimizations. I think a more interesting chart would be ARC AGI vs forwards pass flops or ARC AGI vs training tokens. Of course we don't have those numbers for the
2.
▲
by
muricula
2mo ago
I'm no expert but it seems like a descendent of LSTMs. There's a series of papers which show how to reformulate attention as RNNs which arrives at linear attention. Then they add a decay term to get mamba2. Then they add modified
3.
▲
by
muricula
6mo ago
Same Armv8 ISA. And it's the same ISA Android Linux has run on for over a decade.
4.
▲
by
muricula
7mo ago
So you don't care if injustice happens to other people, only if it could happen to you?
5.
▲
by
muricula
8mo ago
The host kernel and probably the host VMM can see guest memory, so I wouldn't use it for that.
6.
▲
by
muricula
9mo ago
Python has that property when you don't bring C extensions into the conversation. Data races exist, but can never cause memory corruption due to the GIL.
7.
▲
by
muricula
9mo ago
Usually people are talking about race conditions. When you say contrived you're thinking races conditions are difficult to win and unrealistic but attackers who have a lot of money on the line spend the time to win all sorts of wild ra
8.
▲
by
muricula
9mo ago
I've played with something similar with my M1 using Apple's MLX framework. The problem is I'm compute bound. I've never managed to get my M1 Max's GPU to process more than ~7.8k tokens per second at bf16 precision,
9.
▲
by
muricula
11mo ago
Is there a whitepaper or ISA manual change describing the feature?
10.
▲
by
muricula
1y ago
Does anyone know what the block dot at 12 o'clock on the image is? https://chipwise.tech/wp-content/uploads/2025/09/A19_SoC_die...
11.
▲
by
muricula
1y ago
I could be mistaken, but I think the mac studio comes with either an M3 ultra or an M4 max, and the ipad comes with an M4 chip. I think they decided not to make an ultra for the M4 generation, but don't take my word for it.
12.
▲
by
muricula
1y ago
I think the audio quality gives this recording character. What could be more cyberpunk than hearing the quirky artifacts resulting from ripping an obsolete recording medium?
13.
▲
by
muricula
1y ago
Look up iMessage's "blastdoor" sandbox: https://support.apple.com/guide/security/blastdoor-for-messa...
14.
▲
by
muricula
1y ago
That might have been true in the early 2000s when they were growing the Hillsborough Oregon campus but most new fabs are opening in Arizona for taxation and political reasons. I don't have the numbers to back it up, but based on articl
15.
▲
by
muricula
1y ago
The article is written for a different audience than you might be used to. oregonlive is the website for the newspaper The Oregonian, which is the largest newspaper in the state of Oregon. Intel has many of its largest fabs in Oregon and is
16.
▲
by
muricula
2y ago
Is that a CPU based inference build? Shouldn't you be able to get more performance out of the M3's GPU?
17.
▲
by
muricula
2y ago
You may be interested in the new Clang -fbounds-safety extension https://clang.llvm.org/docs/BoundsSafety.html
18.
▲
by
muricula
2y ago
Tax breaks and politics. https://www.nytimes.com/2023/02/22/technology/arizona-chips-...
19.
▲
by
muricula
2y ago
You might double check that the way you number the grades is the same as the way the parent comment numbers them. For example, as I understand it, in the US most kindergarteners are 5 years old and most first graders are 6, while in NZ most
20.
▲
by
muricula
2y ago
No dereferences, just casts. It shouldn't generate any loads/reads from memory.
21.
▲
by
muricula
2y ago
`h` is for "hidden" layers.
22.
▲
by
muricula
2y ago
Is it just me or does it seem like modules are a good idea which is totally dead on arrival? It's been four years, and we've had a whole new edition of c++ standard in the meantime, and Clang and GCC still don't have full sup
23.
▲
Exploits for Sale: Exploiting the NT Kernel in 24H2: New Bugs in Old Code
(exploits.forsale)
5 points
by
muricula
2y ago
|
0 comments
24.
▲
Intel Meteor Lake's NPU
(chipsandcheese.com)
3 points
by
muricula
2y ago
|
0 comments
25.
▲
by
muricula
3y ago
Teddy Roosevelt is both on Mt. Rushmore and in one of the photos generated, so it's definitely not 1 of 4, but 2 of 4 were correct.
26.
▲
by
muricula
3y ago
It's not setting the fire which causes taxpayers to lose money, it's writing off the loss against taxes you would have otherwise paid on profits made from other movies.
27.
▲
by
muricula
3y ago
What is this language's approach to memory safety?
28.
▲
by
muricula
3y ago
Got an error when visiting the page, here's an archived version: https://web.archive.org/web/20240205213635/https://johnnyssw...
29.
▲
by
muricula
3y ago
FYI I use Safari with a Mozilla produced ad blocker extension. It comes along with their FireFox Focus browser.
30.
▲
by
muricula
3y ago
Is there a specific paper or something you can point me to? Or are you talking about like llama.cpp? Because I thought that referred to the fact that it was originally one c++ file named llama.cpp?
More ›