Generally I agree, but the bindless approach can give you both ergonomics and performance in some cases (not all). Certainly it can reduce CPU overhead - passing a single pointer to a shader is much faster than heavy bind group updates. The trade off is that you might have more pointer indirection on the GPU, which can introduce stalls.
Being able to use pointers and u64s for resources also makes having a shared header between shader and c++ source easier, though you still probably need some macro trickery to smooth over language differences. Metal is really good for this since the shading language is closest to C++.
Yes, you can get very close to that API with this extension + existing Vulkan extensions. The main difference is that you still kind of need opaque buffer and texture objects instead of raw pointers, but you can get GPU pointers for them and still work with those. In theory I think you could do the malloc API design there but it's fairly unintuitive in Vulkan and you'd still need VkBuffers internally even if you didn't expose them in a wrapper layer.
I've got a (not yet ready for public) wrapper on Vulkan that mostly matches this blog post, and so far it's been a really lovely way to do graphics programming.
The main thing that's not possible at all on top of Vulkan is his signals API, which I would enjoy seeing - it could be done if timeline semaphores could be waited on/signalled inside a command buffer, rather than just on submission boundaries. Not sure how feasible that is with existing hardware though.
From a cursory look at the code, I don't see any use of fused-multiply-add (FMA) which would likely help with precision issues in a number of places with the float version. The "problem in more detail"[1] readme specifically calls out computation of `x^2 - y^2` as a source for error, and that has methods that dramatically reduce error with FMAs[2].
Not sure how you arrived at 60% - aren't federal capital gains topped off at 20%, and this bill only 7% for gains over $250,000?
Even if it is 60%, I think that's reasonable. In your case, why is the employee selling all of their stock in one year? If they sell over multiple years, they could get gains of up to $250,000 without paying any state tax, which is plenty.
I'd also be in favor of a state income tax, since the sales tax regime in Washington is very regressive.
I wonder if you've given any thought to generalizing to fractional differential equations? My intuition tells me that the dynamics that you're learning are "local" in the sense that the ODE solvers depend only on the current state (and maybe some recent history), whereas learning the dynamics of a fractional system could give the system a larger "history" in the case of your time-series models.
I'm curious what you think the problems are that unbounded indices causes? Unsigned overflow is well defined in C++, so I don't see the problem here. AFAIK, the only issue would be with requiring power of 2 sizes.
I've seen articles in the past few days about the email leaks on the washington post, and looking on CNN right now I see an article on the Podesta emails on the politics page. I've seen a lot of talk about hacking and emails this election. I would say the media is talking more about Trump's sexual assault accusers right now because either:
a) The emails haven't been verified (as far as I know)
b) There's nothing in them as scandalous as a sex scandal (as far as I can tell)
c) Trump keeps saying scandalous things, which draws attention to himself.
A lot of the problems with the media, as far as I can tell, stem from the fact that no one is paying for it anymore. This leads the media to going for attention grabbing headlines, which means a sex scandal will always win out over any allegations of corruption.
This is the research project of another PhD student in my lab. It takes an audio file and a script and generates a facial animation to lip synch to the audio. The key contribution (in my view) is that the animation curves it generates are much simpler than those generated by motion capture, so even when the results aren't perfect, they are easily tweaked by animators.
"Defendants appear to insist that the higher the utility value of speech the less like speech it is. An extension of that argument assumes that once language allows one to actually do something, like play music or make lasagne, the language is no longer speech. The logic of this proposition is dubious at best."
I love this - definitely not a side to the argument that I'd considered before, but I find it very compelling. Well written article all around.
Not a chance. Human interaction is crucial to rehabilitation - what do you think the recidivism rates would be like for a person who hasn't had a real interaction with someone in months or years? The goal should be to make prison more like the world outside, not less. This sounds more inhumane than the existing model, and that's saying something.