wmu

18 ms·

151.
by wmu 10y ago
Really good introduction, well written.
152.
by wmu 10y ago
107x faster than what? This algorithm is well known, I've seen the equation many years ago.
153.
by wmu 11y ago
I would add namespaces to your list. They're extremely useful to organize entities of a program. No more prefixes to functions and types names. Also enum classes are pretty useful. Unlike old enums, the labels do not pollute an outer s
154.
by wmu 11y ago
The worst thing in modern C++ is metaprogramming. It's powerful technique, but the syntax is awful , programs are unreadable & unmaintainable. Moreover, compilers don't help with debugging such code, diagnostics messages rarel
155.
by wmu 11y ago
> If you can accept working with GNU extensions that are available in recent-ish GCC and Clang I do my private project in C++ so it's not a case, but at my current company we use also MSVC. I wish we could abandon that compiler and
156.
by wmu 11y ago
I understand you point, however it not as simple as it seems. Of course, for trivial code transition between different SIMD flavors could be seamless. But the world is cruel. :) Think about shuffling instructions (pshufb), lookup vector for
157.
by wmu 11y ago
Not sure about "single code path". Differences amid SIMD flavors are significant, there are cases when translation one-to-one is either impossible or unpractical. A bright example are AVX2 instructions operating on 128-bit lanes r
158.
by wmu 11y ago
As other said it's not possible with polynomial curves, but you can get good approximation. Only rational curves can form arcs. Wikipedia has a nice explanation https://en.wikipedia.org/wiki/Non-uniform_rational_B-
159.
by wmu 11y ago
I did some tests and it seems that the AVX2 code is faster than popcnt when the input size is 512 bytes or more.
160.
by wmu 11y ago
Hi, author here. You're perfectly right, microbenchmarks have flaws. Procedures should be run for different data sizes (as I did years ago) and code should be compiled with different compilers. And of course the method described in the
161.
by wmu 11y ago
A good point, I didn't consider IDE support.
162.
by wmu 11y ago
I totally agree, type inference speed writing a code, nothing more. For easy cases reading such code is OK (var path = "/usr/bin/whatever" - obvious), but with a more complex code it becomes awful. I'm a C++ pr
163.
by wmu 11y ago
Looks amazing! Really impressive results. Very cool that progressive loading is composed into the format. However I am afraid that without support from biggest companies the format will never gain popularity. Just think how long it took to
164.
by wmu 11y ago
Really nice and useful article - 10/10 :) It is said that trie consumes too much memory. But the author has chosen a naive representation, while it could be done better. There is mentioned paper about balanced ternary trees - does anyb
165.
by wmu 11y ago
Impressive artwork, love way how the animations react on the navigation events. I've never seen such nice webpage.
166.
by wmu 11y ago
Right, I'll fix it.
167.
by wmu 11y ago
"I like how getting 46% gain over already optimized C code is considered "insignificant". Author here (again!) I used SIMD in different algorithms and often speedup was greater than 2, 3, or more. So, from my skewed point of
168.
by wmu 11y ago
Indeed my mistake, thanks for pointing it out. Being own proofreader is not an easy job. :)
169.
by wmu 11y ago
You're right, I haven't aware of that. I'll fix it.
170.
by wmu 11y ago
Hi, author here. Yes, I've considered these instructions, but couldn't find any advantage over a sequence of plain instructions. The sequence properly deals with invalid inputs, pre-validation with pcmpxstrx isn't required.
171.
by wmu 12y ago
Fnv is well-know hashing function, like CRC. Names in the code are not the main problem. 1. Structure of the code is a mess: everything lives in the same namespace, while some stuff could be a private properties of classes/structs. 2.