jdashg

15 ms·

61.
by jdashg 5y ago
The nutrition problem is a social problem, not a technical problem.
62.
by jdashg 5y ago
Filling 8MB (32bpp 1080p) in 10us would be 800GB/s, or 6400gbps. An RTX 3090 has 19.5gbps of on-device bandwidth. GPUs are FAST, but frame data is BIG.
63.
by jdashg 5y ago
Reference is a type qualifier, pointer is a type. Best to just internalize it, even if you'd rather it be different.
64.
by jdashg 5y ago
This is true, and while you can still change @name in the options, it is sort of buried in there.
65.
by jdashg 5y ago
Not in full-range. Narrow-range "tv" is [16,235] for Y and [16,240] (128 +/- 112) for Cb and Cr. Full-range is [0,255] and [1,255] (128 +/- 127) respectively. The spec isn't that bad to read, though it's not lo
66.
by jdashg 5y ago
C is not a "portable assembly language" even if you really, really want it to be.
67.
by jdashg 5y ago
Honestly a big reason is that there simply aren't tests.
68.
by jdashg 5y ago
One of the major user stories is "I have CSS and a video on my page, and my colors should match". Another is the desire for product photos to look relatively true-to-life, so you don't order something and it's too dark,
69.
by jdashg 5y ago
I would say that much of it is actually solved in theory, just not in practice. (I.e. we know what it should display, but browsers get it wrong still) There are both outstanding issues with browsers being accidentally wrong/incomplet
70.
by jdashg 5y ago
All-zeros isn't even really a valid ycbcr value, since even "full-range" "pc" is 1-255 for cb and cr! (Only full-range y has a range of [0,255])
71.
by jdashg 5y ago
I guarantee there were other articles on this topic posted elsewhere with a less provocative title, but they don't end up getting hits, here or elsewhere. C'est la vie.
72.
by jdashg 5y ago
I don't think people are confused by this in their initial solutions, but rather this after-the-fact problem modification happens to match the incorrect intuition people have. The instinct is strong that "I am right" so &quot
73.
by jdashg 5y ago
Worth mentioning that Ryzen's branch predictor does in fact have an NN component.
74.
by jdashg 5y ago
For a more macro/systemic discussion of slack, I liked this SSC article: https://slatestarcodex.com/2020/05/12/studies-on-slack/
75.
by jdashg 5y ago
Vulkan Compute is the alternative, and you should look again if you're serious.
76.
by jdashg 5y ago
Browser vendor here: We're definitely still in the W3C and collaborating in a number of areas there. The WHATWG situation is complicated, but we have not "left the W3C".
77.
by jdashg 6y ago
The current roadmap for that is "never". :) There just isn't enough ROI for rewriting all old code.
78.
by jdashg 6y ago
No, 1200 is too low to generally approve like this. 1200 is commonly stated as the absolute minimum for women not under medical observation. The consensus for "is 1200 enough" seems to be "ask your doctor" not "it&#
79.
by jdashg 6y ago
At FIRE levels, exposure to healthcare-related bankruptcy is basically none, as long as your rising-with-age insurance premiums don't price you out. It's still not great, but insurance does do its job. (especially when you have th
80.
by jdashg 6y ago
> Unfortunately, programming in C++ is really brutal for people who would rather learn by example than by carefully reading documentation. This is a great observation! There's probably a ton of this that feeds the "$lang/a
81.
by jdashg 6y ago
The debate was settled long ago. The resolution was WGSL: https://gpuweb.github.io/gpuweb/wgsl.html
82.
by jdashg 6y ago
If you take a profile (perfht.ml) it'll show you the names of the processes as part of the profiling report.
83.
by jdashg 6y ago
There's no argument you can successfully make that, because the Basilisk is unconvincing, therefore Rationalists are intellectually bankrupt. The connection is stillborn, because there's no path towards bridging the gap between th
84.
by jdashg 6y ago
Calls for location-blind "pay based on my value" share a remarkable parallel with the desire for house prices that track a house's uninflated intrinsic value: The market doesn't care what you think it should pay.
85.
by jdashg 6y ago
On the Doylist side, it's here because it was linked from Scott Siskind's response to the NYT article about himself: https://astralcodexten.substack.com/p/statement-on-new-york-... I read them in that order,
86.
by jdashg 6y ago
Rationalists aren't Basilisk believers in general, so the connection you draw here is stillborn.
87.
by jdashg 6y ago
I know that the value here is that it saves people effort to pre-digest this knowledge, but oh man this is the stuff that high school math definitely taught us to solve. Heck, maybe this makes it a useful interview question scenario?
88.
by jdashg 6y ago
Methyline Blue is tricky to test because it really does turn things super blue. Interestingly, this is sometimes used as a powerful placebo in other studies, since it sure does convince people they aren't taking a sugar pill! https:
89.
by jdashg 6y ago
1440p at 27" with no scaling is perfect for my desktop. I wish it were more popular, because it's a great compromise between quality and performance for gaming. The relatively common use of "2k" to mean 2560x1440 has alw
90.
by jdashg 6y ago
It does look surprisingly good, but it's still blurring unless they changed it this year. Alternating-color pixel grids always moire there.

More ›