winwang

22 ms·

331.
by winwang 3y ago
Here's an article on syncing in CUDA via cooperative groups: https://developer.nvidia.com/blog/cooperative-groups/ There's also explicit warp synchronization, i.e. __syncwarp(). More on warp primitives h
332.
Show HN: Faster sorting with register shuffling in CUDA(github.com)
3 points by winwang 3y ago|0 comments
333.
by winwang 3y ago
Agreed. Was working on a parallel algo whose width could grow exponentially. It was clear that it made more sense to analyze from a space perspective.
334.
by winwang 3y ago
Shoulda used Dirac notation :P
335.
by winwang 3y ago
Oh sure, it was more of a joke app idea, or in another sense, for those who want philosophically "perfect" randomness. After all, assuming a single bit sample, anything but a hadamard state has nonoptimal entropy. Of course, it ma
336.
by winwang 3y ago
...and then you break out the actuarial models :) And at some point, you just gotta do it.
337.
by winwang 3y ago
I'm using the term "analog computing" to mean using non-digital (or even digital but in a nonstandard way). A quantum processor is not digital as each qubit has an uncountable number of states. A quantum computer would likely
338.
by winwang 3y ago
Cloud RNG number streaming is interesting but costly, no? I did have the idea to serve truly random numbers via a quantum computer (trivial by just preparing the simplest state and measuring). Anything else can't be said to be truly ra
339.
by winwang 3y ago
I mean a lawful(ish) monadic type/api, i.e. `LLM a` in FP-land. Otherwise, every sequential machine/program can be said to be monadic.
340.
by winwang 3y ago
And now I'm waiting for the "LLM" monad.
341.
by winwang 3y ago
Skimmed the litepaper. Has the flavor of: you can do "simulated" annealing by literally annealing. I like the idea of using raw physics as a "hardware" accelerator, i.e. analog computing. fwiw, quantum computing can be s
342.
by winwang 3y ago
Compilation comes to mind. Efficient distributed compute also usually prefers vertical scaling (i.e. more cores per node).
343.
by winwang 3y ago
Note that Amdahl's law doesn't apply if you change the entire algorithm (program) to fit better on parallel hardware (i.e. the task is not fixed). From my dealings with parallel algorithms, there is much less sequential dependency
344.
Sorting with CUDA (SIMD) Shuffling(winwang.blog)
4 points by winwang 3y ago|1 comments
345.
by winwang 3y ago
Seems that people like SIMD/SWAR stuff. Public writing is scary -- feedback is much appreciated!
346.
by winwang 3y ago
I'm having a lot of fun imagining this being said to a kid who's trying to buy some folders for school.
347.
by winwang 3y ago
"a sufficient smart compiler...." (/s) I had a question recently: would an extra 65th bit in some registers have a good use case with stuff with SWAR?
348.
by winwang 3y ago
If people like stuff like this, the simdjson paper uses similar techniques, is extremely well-written and has great examples. Paper: https://arxiv.org/abs/1902.08318 Github: https://github.com/simdjson&
349.
by winwang 3y ago
Not sure, but functions are serializable.
350.
by winwang 3y ago
Kind of reminds me of a unikernel.
351.
by winwang 3y ago
(not googler) Google's codebase is idiomatic to Google due to their strict language tooling. e.g. their C++ code stays away from advanced features. The tooling teams at Google have very strong say.
352.
by winwang 3y ago
Great comment! I didn't mean to suggest Monte Carlo isn't incredibly clever and nice. I just get intimidated by cool-sounding names :)
353.
by winwang 3y ago
Also did a bit of QC research. Since I'm tipsy, I skimmed the Twitter post. Here's a pdf on arxiv about HMC (2017) which I found in 20 seconds: https://arxiv.org/abs/1701.02434 Seems like Claude3 is decently
354.
by winwang 3y ago
That would make more sense in the alternate history where Intel and AMD never try to make GPUs anymore.
355.
by winwang 3y ago
Was there any performance difference between R and Python in your case?
356.
by winwang 3y ago
Shouldn't we think in terms of the post-pruning ethics? :D
357.
by winwang 3y ago
Very abstract-algebra-y (loci being probably some generalization of a metric space). ...Why don't we just import he rest of algebra here :)
358.
by winwang 3y ago
Triple agree. I am also curious about how HN handles (or wants to handle) "profanity"/"taboos".
359.
by winwang 3y ago
I've thought a bit about this. Some revenue sharing for top mods and corporate oversight (with statistical tooling) to incentivize continued good moderation. Public moderation actions and votes on them, etc.
360.
by winwang 3y ago
"poison scales with dex, drowning is a weapon passive"

More ›