pca006132

9 ms·

61.
by pca006132 2y ago
Probably doesn't make sense. Busy wait is fast when you can dedicate a core to the task, but this means that you cannot have many tasks in parallel with a small set of physical cores. When you oversubscribe, performance will quickly de
62.
by pca006132 2y ago
why not infinite brains so you can have more computational power than these GPUs?
63.
by pca006132 2y ago
Yeah, triangle mesh library and NURBS are very different, they do different things. Interesting to know that the solvespace folks wanted to try integrating manifold. Feel free to open an issue or discussion if needed, there were many change
64.
by pca006132 3y ago
Yeah randomization in algorithms and crypto serve different purpose. But overall introducing randomness can allow you to do amazing things that deterministic algorithm/schemes cannot do.
65.
by pca006132 3y ago
Don't see how this can work. How would you encourage collaboration between projects when they are kind of competing for resources? How to track the 'contribution' of certain projects? Some projects may be more important to so
66.
by pca006132 3y ago
No, there are complexity classes for random algorithms. And randomness is also crucial for some cryptographic protocols, typically you want to show that the probability the adversary can do what they want is very slim, e.g. 2^-n for some la
67.
by pca006132 3y ago
Never knew this before! Thanks for sharing.
68.
by pca006132 3y ago
If I understand correctly, for most publishers you are allowed to host a copy on your personal web as well as the university library catalog?
69.
by pca006132 3y ago
but it isn't hard to add thermocouples to a heatsink? perhaps the reason for not doing this is because the improvement is not visible in standard benchmarks? it is not improving the overall performance, not reducing the peak noise, jus
70.
by pca006132 3y ago
But Firefox can already do LTO and PGO? I thought that firefox let the underlying cryptography library do its work, and those are already using AVX and AES as well.
71.
by pca006132 3y ago
Well I don't intend to mean they are the same, but optimization to improve the latency can often improve throughput as well. Whatever...
72.
by pca006132 3y ago
Performance optimization covers a lot of topics, it depends on what you are trying to optimize. 1. Latency vs throughput. Oftentimes they are the same, i.e. reduce the time it takes to do something. However, when you passed a certain thresh
73.
by pca006132 3y ago
The problem is musl does not support execinfo. For a simple hello world, I managed to statically link with musl to get something with 7888 bytes.
74.
by pca006132 3y ago
Would be nicer if there is a DE on Linux that supports these fancy screen configurations.
75.
by pca006132 3y ago
I think there is also bubblewrap which flatpak uses, which can limit access to folders, and also XDG desktop portal which allows GUI applications to access files that are not available to them. Not sure if these are widely used though, iirc
76.
by pca006132 3y ago
Nice, so now we have OpenAI and OpenSubtitles that has 'Open' in their name and not open :)
77.
by pca006132 3y ago
Feel that it is also important to know when iterators/references to elements will be invalidated, in addition to knowing the complexities of the operations.
78.
by pca006132 3y ago
I remember it works for libc++ (partially, see https://libcxx.llvm.org/Status/PSTL.html ), but forgot when I linked against libc++ last time...
79.
by pca006132 3y ago
Are they fresh graduates? It is very important to understand the workload distribution for any optimization. Even if small things can sometimes get large, optimizing for the small case can often yield large gain as they may occur frequently
80.
by pca006132 3y ago
Clang does support parallel stl already (requires either TBB or OpenMP). Our project https://github.com/elalish/manifold made use of this to speed up mesh processing algorithms a lot.
81.
by pca006132 3y ago
This is targeting hobbyist and not the typical kind of production usage for embedded systems. I rarely heard of embedded people encouraging customers to hack their production system...
82.
by pca006132 3y ago
No, static analysis can also analyze multiple functions. What you are looking for is called interprocedural analysis and is more expensive than single function analysis (aka intraprocedural analysis). More generally what you want is context
83.
by pca006132 3y ago
I am really curious about what kind of optimizations are enabled when you know the microarchitectural design of various CPUs, but without writing assembly by hand. I only know the basics such as optimizing data structure for better cache lo
84.
by pca006132 3y ago
Yeah I don't think college should be abolished, I just think that college should adapt to this trend of online education, instead of just acting like 100 years ago where there is no internet. And they should probably try to make better
85.
by pca006132 3y ago
While I also attended college, I sometimes wonder if a college degree is really needed to learn these general stuffs. I learn a lot while doing projects completely unrelated to schoolwork, before registering for the course. I still have to
86.
by pca006132 3y ago
While this would be nice, I don't think enough people working on these algorithms are trained to use formal theorem proving tools.
87.
by pca006132 3y ago
Wondering why no BRep support means it is not a CAD. It seems that OpenVSP doesn't use BRep but uses parametric surface [1]. I wonder if mesh-based modeling + some constraints solving will get you a CAD, or are there other requirements
88.
by pca006132 3y ago
I doubt if it is possible to have a practical small kernel. Take mesh processing in https://github.com/elalish/manifold for example, we encountered a lot of problems when trying to deal with inexact floating point arit
89.
by pca006132 3y ago
I think the interesting thing about CAD kernel is that there are different representations and limitations to each representation. You have triangular (or polygonal) mesh, BREP which uses NURBS, SDF which is based on functional representati
90.
by pca006132 3y ago
I guess that a lot of people are uncomfortable thinking about vector instructions, and dealing with masks manually? And for vector instructions you need to align things properly, pad the arrays such that they are of the right size, that peo

More ›