p0nce

14 ms·

151.
by p0nce 4y ago
My uncle's job was to clean elevators using TCE and died at about 25yo from this.
152.
by p0nce 4y ago
Fast software rasterizers are not slow at drawing text in the first place.
153.
by p0nce 4y ago
D has a memory @safe subset that would effectively prevent the exploit.
154.
by p0nce 4y ago
Unfortunately the cost of learning C++ has increased.
155.
by p0nce 4y ago
Honestly I think the "interface" vs "implementation" part of a Pascal file is a genius idea.
156.
by p0nce 4y ago
It is actually the most popular DAW at this moment.
157.
by p0nce 4y ago
So.... Bing is more like AI Dungeon, instead of boring ChatGPT. That certainly makes me want to use Bing more to be honest.
158.
by p0nce 4y ago
I've extended QOI for 10-bit here: https://github.com/AuburnSounds/gamut (and also grey and grey+alpha). So it's a separate incompatible codec yes. The problem with a QOI that would have 16-bit is that lossle
159.
by p0nce 4y ago
Just want to point out malloc(0) might gives you any pointer, and you still have to call free() on it.
160.
by p0nce 4y ago
C++ code has a lot of copies actually.
161.
by p0nce 4y ago
Basically if your C++ function takes a pointer you need to write one for unique_ptr, or raw pointer (documenting the ownership). With a GC, there is a global owner so you don't have to think about ownership for everything, this breaks
162.
by p0nce 4y ago
But this is exactly why we have C++ in the first place, you could port your C code to cfront without doing any effort.
163.
by p0nce 4y ago
A lot of C++ software out there could afford to use a GC.
164.
by p0nce 4y ago
No they rely on a product that has market-fit.
165.
by p0nce 4y ago
I did start my own business, 8 years later I earn 2x what I was paid (not FAANG level by any means mind you), work perhaps 3x less, the work is a lot more enjoyable. I was a SE. You need to provision enough time (years) to hit market-fit, a
166.
by p0nce 4y ago
Something I appreciate a lot on Mastodon is simply its federated nature. It seems to me like a nice middle-ground between: 1. niche communities like Discord, and 2. "everyone at the same time" like Twitter.
167.
by p0nce 4y ago
Try books.
168.
by p0nce 4y ago
All of these QOI derivatives improves upon the coding efficiency AFAIK. In QOIP, opcodes themselves are stored in the format, LZ4 or zstd compression https://github.com/chocolate42/qoipond QOIR adds: color profile, EXI
169.
by p0nce 4y ago
It's always about reorganizing workers output, and never about why there are so many bosses to produce for.
170.
by p0nce 4y ago
arm64 is one reason.
171.
by p0nce 4y ago
I did not kept track of every experiment, just saying that if you do make a new opcode in QOI that encode a predictor with a few bits + a few bits of diff, then it's hard to make an improvement at all. It's easier if you remove th
172.
by p0nce 4y ago
I'm not sure why your whole scanline would need a predictor that is top pixel, left pixel, or average. Shouldn't this be for 16 or 32 pixel max? The most popular QOI modification has been to take the average of top and left decode
173.
by p0nce 4y ago
Also extended QOI to make QOIX, a triple codec that can also encode 10-bit and 1 and 2-channels images. So you could say it's 3x the complexity. https://github.com/AuburnSounds/gamut
174.
by p0nce 4y ago
ChatGPT can be useful for casual directions in research, it is very good to repeat hear-say and dominant methods. But it will nicely present falsehoods as facts, and also the fringe methods do not exist in its view. Language models believe
175.
by p0nce 4y ago
So they are actually creating maybe more added value that way. Working as intended.
176.
by p0nce 4y ago
Exactly what is going on really. Employers unable to envision that maybe it might not be worth it to work for your company are getting punished as information flows faster. The youth is not lazy, just has better information.
177.
by p0nce 4y ago
> To clarify, are you saying the entire app was slower with AVX than it was with SSE4? We were optimizing loop by loop, and some loops converted in AVX could well be slower yes (at this point of time). AVX is probably more often a win no
178.
by p0nce 4y ago
Yes. It also makes the masked instruction more portable to arm or non-x86, which is great. It also avoids the side channel?
179.
by p0nce 4y ago
But it's true at some companies, it's always the busy season.
180.
by p0nce 4y ago
And if there is failure in the 10 years following shipping, the company fixes it with no additional charge (this is a law around here).

More ›