oivey

8 ms·

91.
by oivey 1y ago
That wouldn’t make a program written in it “more” correct.
92.
by oivey 1y ago
The program won’t be “more” correct. What would that even mean? Writing correct programs might be easier (or not) with more “constrained” (ill defined) typing.
93.
by oivey 1y ago
You need cards like the A100 / H100 / H200 / B200. GPUs aren’t fundamentally worse at f64. Nvidia just makes it worse in many cards for market segmentation.
94.
by oivey 1y ago
That’s not how economics works. Local energy price surges will also drive global prices up. The US is part of the global energy market. I guess you could ban export of energy and institute price controls, though.
95.
by oivey 1y ago
Enshitification would imply they were ever good. Police brutality has been a constant throughout US history, particularly against minorities, among many other things.
96.
by oivey 2y ago
For some cases maybe that makes sense, but in a very large percentage it does not. As example, what if I want to build and use/deploy a Python app that needs the latest NumPy, and the system package manager doesn’t have it. It would be
97.
by oivey 2y ago
It’s still shocking to me that in this whole ordeal many reviewers escaped scrutiny. Getting a cut of a sale of a product that you portray yourself as impartially reviewing is insanely immoral. Who cares if these people scammed themselves i
98.
by oivey 2y ago
That flexibility is a fantastic advantage of software over hardware, not a liability. You still can deliver software as an unchanging thing, but generally no one does because incrementally giving users something useful sooner is more valuab
99.
by oivey 2y ago
Not really true. The legal liability is generally associated with the Professional Engineer designation. In electrical engineering, for example, my experience is that almost no one uses/needs to be a PE unless they work in infrastructu
100.
by oivey 2y ago
I agree with your post, but thought I’d point out that the problems those labs worked on weren’t transistors, Unix, etc. Those were the solutions.
101.
by oivey 2y ago
And yet, it was never called that, right? It would have been a Mafia task force, a task force against a particular family, or a task force against organized crime. You wouldn’t call it an Italian task force because you’re probably hoping fo
102.
by oivey 2y ago
No, that would be ridiculous. An “Italian squad” might be, though.
103.
by oivey 2y ago
The README says the savings is ~75 MB. In most notebook workflows you’re at most running a couple at once. Saving <1% of my system memory doesn’t let me do anything I couldn’t do before. This also isn’t going to add concurrency/para
104.
by oivey 2y ago
What’s the advantage of this? It isn’t obvious to me that reducing memory usage and CPU of an empty/idle kernel is all that meaningful if the actual Python code in your notebook uses far more resources. It’s also not obvious to me how
105.
by oivey 2y ago
I’m not sure what you’re trying to say. How is what you’re describing different than immersive video or the non-immersive sports apps that do stuff like project a diagram of the field and advanced stats around you?
106.
by oivey 2y ago
Are either of those numbers relevant? US auto companies heavily outsource component manufacturing to many, many other smaller companies. Seemingly you’d want to compare to compare the supply chain included job numbers along with the number
107.
by oivey 2y ago
You cut off the part of the quote that says you can’t trust people by default, misrepresenting what who you quoted was saying.
108.
by oivey 2y ago
What you’re talking about is good design but has nothing to do with encapsulation. From Wikipedia: > In software systems, encapsulation refers to the bundling of data with the mechanisms or methods that operate on the data. It may also r
109.
by oivey 2y ago
Encapsulation arguably isn’t a good part, either. It encourages complex state and as a result makes testing difficult. I feel like stateless or low-state has won out.
110.
by oivey 2y ago
The magic of AlexNet has little do with Nvidia. It’s a more fundamental breakthrough than just the compute hardware used. Further, it’s not related to Nvidia either. Instead, it showed the promise of GPGPU. It just so happens that no one at
111.
by oivey 2y ago
I think they were highlighting that that phrase is bullshit. It’s trivial to escape many compile time checks.
112.
by oivey 2y ago
This particular example explicitly dodges compile time checking for some ad-hoc (but likely safe) runtime behavior. It’s not a strawman at all. It’s a classic example of how sometimes the compiler can’t help you, and even worse, how program
113.
by oivey 2y ago
Besides the horrendous formatting, some stuff in this article seem incorrect or irrelevant. Like, is this even possible? > A single Anaconda distribution may have multiple NumPy versions installed at the same time, although only one will
114.
by oivey 2y ago
The time spans for doing it properly or shipping are maybe a bit muddled here, but if you can make a million dollars by shipping something now, you should probably ship it now. If the code is that bad (ie can end up costing way more to main
115.
by oivey 2y ago
The whole reason CUDA/GPUs are fast is that they explicitly don’t match the architecture of CPUs. The truly sophomoric attitude is that all compute devices should work like CPUs. The point of CUDA/GPUs is to provide a different se
116.
by oivey 2y ago
The blue one is pretty significantly less noisy, and the animation showing the two 3D reconstructions is much clearer. The noise reduction probably has a lot to do with that.
117.
by oivey 2y ago
This isn’t the traditional thought process on this topic. Traditionally, it was thought that the ills of resource allocation under democracy were far outweighed by the automatic optimization of the free market. The Soviets did a lot to sugg
118.
by oivey 2y ago
That’s a weird way to account for their business size. There isn’t a significant marginal cost per transaction. They didn’t sell $12T in products. They facilitated that much in payments. Their profits are fantastic.
119.
by oivey 2y ago
True. That’s really a bridge too far for people who believe a Boeing executive who says tactical soap dispensers really do cost $700.
120.
by oivey 2y ago
> Because they bought it. Fair enough, but I was referencing people taking a Boeing exec’s claim that somehow regulations cause military soap dispensers to cost that much at face value. > I assume you talk about the government here? N

More ›