Firadeoclus

While you can use uniform registers to implement the uniform keyword from shading languages, the two are not the same. Uniform registers are not constants, and they are only uniform/shared across one warp. Nvidia architectures before Turing did not have uniform registers.

What makes a judge a "real" judge is that a sufficient number of people capable of enforcing judgments share the view that the judge is a real judge.

Visual programming languages are not prevented from using textual symbols, though. And many people use diagrams to communicate ideas more efficiently than they could with text alone.

> Any language can be expressed in S-expressions.

I'm not sure this is true for the C preprocessor, though, where macros can represent partial structure.

This is a really interesting system, and I'm excited to give this a try!

Without knowing whether the following cases would actually be useful/relevant, I'm curious if these things apply to Unison:

- Is there a way to "pin" a symbol to a specific hash/version so it won't automatically update when the referred function gets a change? I.e. I could write: A = B@ ++ "world" and when I store and retrieve it, it becomes (example syntax): A = B@a3yx ++ "world"

- Is there a way to refer to a function/symbol by name rather than by hash? I.e. a call site which uses B by looking up which hash is currently associated with the name "B", such that if I do two simultaneous renames, B->D and C->B, the code would now refer to the function previously known as C?

- Are there ways in which the way function updates "bubble up" through the codebase (I assume updating a function means updating every function that calls it, recursively) could become a problem? It would seem that changing one little function could have a knock-on effect that requires re-hashing most of the codebase.

> It's not a matter of identity, but explanatory power. To reiterate, the self-acknowledgement/'diagnosis' as autistic (not ASD) is 'useful from a self-compassion and tolerance standpoint'. So instead of, why do I find X difficult when everyone else finds it easy (or even laughs/condescends/points it out), I can say, I'm autistic, that's not easy for me.

It bothers me a bit that, if you turn it around, the lack of an explanation would be an impediment to self-compassion and tolerance. "I find X difficult" ought to be enough, whether that coincides with any other traits (autistic or otherwise) or not.

Which could be a coordination problem, though. Where initial high costs are due to a lack of scale whereas if everyone participated, costs would come down fast.

Putting fans on radiators should also increase the amount of heat that can be transferred, right?

>The whole point of interfaces is that you're not supposed to care how a method is implemented, only what it does which is explained by a combination of context, naming and documentation.

There are, however, cases where code is a better explanation of "what it does" than naming and documentation. Both naming and documentation are hard and can become out-of-sync. Code is less ambiguous than natural language.

One way to get deterministic output is to use integer/fixed point math. Quantised models already do that for matrix multiplication, but things like softmax may still be implemented using some floating point math. It's possible to replace that, just takes a bit of extra work and is probably slower than using the GPU's native float ops.

This. Being able to expand macros lets you learn what it does in context, and once you're more familiar with the abstraction it gives you the benefit of hiding the details and highlighting the moving parts within an otherwise static structure.

This makes sense but to be honest, how I'd really like to represent data to go along with code is as a table or a tree (i.e. presentation/interaction more like a spreadsheet).

> "More people with cards tend to be white and therefore the cards are racist" is such a weak take - there are more white people than minorities, for one.

But this is quite different from what the article actually says. What they're saying is P(Card|White) > P(Card|Minority), not P(White|Card) > P(Minority|Card).

>> White drivers in the borough, the complaint said, “are significantly more likely to have courtesy cards than minority drivers”. As a result of a ticketing quota system, this means “police officers are forced to disproportionately ticket minority drivers”.

Certainly, getting those elements in larger quantities will be important, but by the time we have the cargo capacity to build colonies on the moon it seems likely that someone would be very happy to trade, say, 100t of Earth-sourced C/N/H for 100t of Moon-mined Au, shipping included.

You could call the way the curriculum is organised "spaced repetition", but this isn't personalised and thus missing the feedback part of spaced repetition. Generally, children aren't taught the method of spaced repetition itself. "Learning how to learn" should be its own subject early on.

There is no reason public transport should aim to be financed entirely by fares, or to make a profit from fares. It benefits people who rarely or never use it.

A "generic raw text editor" is a tool. The vast majority of text editors people use are actually quite complex, way beyond the complexity that would be needed to implement a "generic structured data editor". And the latter could make many very-nice-to-have features simpler to implement so it would very likely end up with less complexity than a full-featured "text editor".

The main problem to be solved is that structured data editors (and structured data query tools) are not ubiquitous, even though structured data formats are. Which really is a shame, considering the amount of time and effort sunk into issues (TFA, tabs vs. spaces, unmatched delimiters and other structural typos, syntax extensibility, etc.) that wouldn't even have to exist if we weren't dealing with raw text.

> They almost convinced me somewhere in my career. But the hard truth I learnt is that most people are saying this because they aren’t capable of verbalizing what they are programming.

I'd say that's true, and it's worth noting at this point that expressing certain things in natural language is hard. The strict rules of programming languages mean that you can reason about programs to a complexity level that would otherwise be unreachable. Notation as a tool of thought. The corollary is that there may not be a simple natural language equivalent of the code you're writing, and that adding documentation might be more effort than it's worth.

A text editor is a tool. Text files are only "human readable"/"human editable" in the sense that such tools are ubiquitous whereas structured data editors are not.

Which is a real shame. Entering, manipulating and validating small amounts of structured data is such a fundamental task when using computers that it deserves dedicated tools, and the sheer amount of issues that only exist because such tools are not ubiquitous is absolutely mindboggling.

What you might do is train using dense matrices, then sparsify those (pick the 2 out of each set of 4 weights that are closest to zero, mask them out), then do a few more training iterations with the mask in place.

It turns out that even without the extra training iterations you often lose surprisingly little in terms of quality of output. In reality you can sparsify a lot more, but 2 out of 4 is so simple and easy to implement in hardware, more complex schemes are much harder to justify.

However, small matmuls (say, <2048 bytes in the K dimension) won't get anywhere near 2x performance.

Note that AMX can achieve roughly double the FLOPS with FP16, and 8 TFLOPS for the GPU is only about 77% of peak. You can do better than that, especially using FP16 90+% is possible (which is >9.4 TFLOPS).

That's a pretty huge amount of processing power hidden away! Are these experimentally confirmed performance numbers?

So this is 16x16 single precision fused multiply-adds at 3.2GHz with a throughput of one per clock, right? (16 x 16 x 2 x 3.2e9 = 1.6384e12) And does using fp16 quadruple throughput again? That would put AMX well above the GPU for fp16 matrix multiplication! (2.6 TFLOPs fp16/fp32 for 8-core GPU, 128 multiply-add per core at 1.278GHz)

How does it compare in terms of power, can it sustain 3.2GHz indefinitely or does it hit power/thermal limits fairly quickly?

> the terminal provides a very fast feedback loop where I can iteratively examine the output of a command and refine the filters I apply. As a result, I am able to prototype commands quickly and accurately.

That's what any REPL does. And it works great with typed data. Sometimes even with graphical output.

Though its impact may be limited, this is some nice work!

Now could someone look at how video codecs can produce excellent high-detail scenes and motion in 4k resolution while at the same time making a blocky mess out of soft gradients, especially in dark scenes with slow movement?

Moreover, an ANN parameter is not equivalent to a human neuron.

> Electric motors haven't become remarkably more efficient --- the starting point was already quite high.

Batteries have, though. And if we're talking about personal transport, the passenger mass to total mass ratio in today's cars leaves room for huge efficiency gains in principle.

>No, it is not. It is the physically correct way to do it. 50% more red light is emitted than blue or green. That is a red tint.

That is a red tint. And that's why it's not the correct way to do it. A subpixel antialiased white line that is seven subpixels wide, on a black background, should not produce (8-bit RGB triplets): 0-0-0, 255-255-255, 255-255-255, 255-0-0.

> Why would I ever go to a woman for "emotional support"? Her lived experience and frame of reference will be so radically different from my own that her advice is unlikely to be applicable. When I need support, I call my guy friends with ("Hey, I haven't encountered this problem in my life before. I believe you had a similar scenario in the past. Suggestions?"), and they do the same.

Emotional support goes beyond advice and suggestions. In fact, for many people (including many men), advice is the least important part and often unwelcome at first. What people are looking for is empathy, being heard and validated. Therapists don't share the same lived experience as most of their clients, yet they can still be exceptionally skilled at providing emotional support.

It depends on what the converted energy is used for. If you use it to power a device in the house then it's quite likely most of that energy will end up as heat in the house again.

What exactly does it mean to "perceive a bicycle"? Noticing shapes and colours and recognising them as a distinct object? Recognising an obstacle? Noticing qualities like smoothness and straightness and associating the concept "man-made"? Being able to explain its purpose? Predicting how it might move, if ridden by a person?