hajile

13 ms·

181.
by hajile 1y ago
https://www.pewresearch.org/internet/fact-sheet/mobile/ 15% and increasing quickly. That’s nearly 1 in 5.
182.
by hajile 1y ago
A19 has WAY less bandwidth on its 64-bit bus than desktop chips with 128-bit busses . AMD’s strix halo is also slower despite a 256-bit bus. Pushing this point further, x86 chips are also slower when the entire task fits in cache. The real
183.
by hajile 1y ago
NOTHING produced good code for the original Itanium which is why they switched gears REALLY early on. Intel first publicly mentioned Poulson all the way back in 2005 just FOUR years after the original chip was launched. Poulson was basicall
184.
by hajile 1y ago
Even if you had that environment today, how would you replace the function of JS? Web development moves faster than any software in history. A lot of this is on the back of JS being ergonomic/fast to code and having very good performan
185.
by hajile 1y ago
The frontend would be so much better if Eich had shipped scheme. ES6 wouldn’t have been needed. CSS-in-JS would have become standard decades ago. HTML would have almost disappeared too because parsing s-expressions would have been native. J
186.
by hajile 1y ago
You are complaining about the libraries around it rather than react itself. function MyForm() { const [isHidden, setIsHidden] = useState(false) return ( <> <button onClick={()=>setIsHidden(!is
187.
by hajile 1y ago
In JS land, transpile was used to distinguish something like elm->JS from ES6->ES3. One was the same language with different versions and the other was different source languages. Yes, all transpilers are compilers, but not all compil
188.
by hajile 1y ago
Doesn't MS still have screens rendered like Windows 3.1 or Win95 in some corners of the OS?
189.
by hajile 1y ago
I don't remember the source, but worldwide, most really old people have a couple things in common. First is that they live in countries with some kind of pension plan. Second, they generally come from poor neighborhoods where all the p
190.
by hajile 1y ago
If the LLM isn’t getting its data about coding projects from those projects and their surrounding documentation and tutorials, what is it going to train with? Maybe it also needs some amount of other training data for basic speech patterns,
191.
by hajile 1y ago
I was somewhat surprised to find that Zed still doesn't have a way to add your own local autocomplete AI using something like Ollama. Something like Qwen 2.5 coder at a tiny 1.5b parameters will work just fine for the stuff that I want
192.
by hajile 1y ago
There's a lot to break down here. FP/SIMD vs int/ALU, package vs core power, percentage vs total, branching vs unbranching code, average IPC, etc. Let's start with package vs core power. Package power is a terrible metri
193.
by hajile 1y ago
Cheese and Chips makes some bad arguments in that article. Their claim that ARM decoders are just as complex wasn't true then and is even less true now. ARM reduced decoder size 75% from A710 to A715 by dropping legacy 32-bit stuff. Co
194.
by hajile 1y ago
In Haswell, 4.8w out of 22.1w for the core were used for the decoder for integer/ALU instructions[0]. According to this[1] analysis of the entire ubuntu repository, 89% of all instructions were composed of just 12 instructions (all int
195.
by hajile 1y ago
Apple never used a uop cache in their designs. ARM dropped uop caches when they removed 32-bit support. Qualcomm also skipped uop cache. uop made sense with 32-bit support because the 32-bit ISA was so complex (though still simple compared
196.
by hajile 1y ago
Apple made a big deal about this, but other iGPUs have done this for years.
197.
by hajile 1y ago
2. uops are a cope that costs. That uop cache and cache controller uses tons of power. ARM designs with 32-bit support had a uop cache, but they cut it when going to 64-bit only designs (look at ARM a715 vs a710) which dramatically reduced
198.
by hajile 1y ago
Intel was working with Nokia to heavily invest into Meego OS until it was killed by Elop+Microsoft. If it hadn't been killed, it may have become something interesting today.
199.
by hajile 1y ago
Intel has proposed APX to address this. It does away with some of the 32-bit garbage that complicates design for no good payoff. Most importantly, it increases from 16 to 32 registers and allows 3-register instructions (almost all x86 instr
200.
by hajile 1y ago
M4 Pro was a massive step back in perf/watt over M3 Pro. To my knowledge, there aren't any M4 die shots around which has led to speculation that yields on M4 Max were predicted to be really bad, so they made the M4 Pro into a binn
201.
by hajile 1y ago
I'm not talking about the plane. I'm talking about the weapon itself. Imagine a next-generation fire-and-forget weapons with radar and broad-spectrum camera arrays and an AI trained on a fused version of all this data. Typical def
202.
by hajile 1y ago
The upcoming generation of weapons is going to use realtime sensor fusion done by AI. Cutting edge chips will matter for those weapons.
203.
by hajile 1y ago
The major patents on all the most important parts of x86 expired years ago now. Nobody wants to take on a legacy ISA with tons of footguns everywhere when newer ISAs have learned a lot of lessons from x86 about how to do things better.
204.
by hajile 1y ago
I believe the most modern TSMC fabs outside of Taiwan are in Arizona. They are just moving to 4nm which is nearly 5 years old and just a revision of 5nm which is getting close to 7 years old. TSMC aims to have N3 in Arizona by 2028 at the e
205.
by hajile 1y ago
> Social media really seems to have settled into a means of following public figures. Individuals generally seem to interact with each other via group texts. I wish Google circles were still a thing.
206.
by hajile 1y ago
File System Access API has some serious issues. To quote Mozilla's position on the topic > There's a subset of this API we're quite enthusiastic about (in particular providing a read/write API for files and directorie
207.
by hajile 1y ago
I'd say that the best is probably the book "Lisp in Small Pieces". Be aware that it's some 500 pages and makes several interpreters then a couple of compilers as I recall. If you read it though, you'll come out the
208.
by hajile 1y ago
It used to be said that "Lisp programmers know the value of everything and the cost of nothing." This applies to MOST devs today in my experience and doubly to JS and Python devs as a whole largely due to a lack of education. I&#x
209.
by hajile 1y ago
It seems like none of the really good language proposals have much traction. Proper Tail Calls have been in the language for TEN YEARS now, but v8 and Spidermonkey still violate the spec and refuse to implement for no good reason. Record&#x
210.
by hajile 1y ago
I've heard both assertions. I tend to believe that Carbon Fiber is a fine material to use. This submersible used untested techniques. They didn't adhere the layers together properly and apparently never bothered to X-ray the tube

More ›