-
```
While integrating LZ4 into my Pcompress utility
(http://moinakg.github.com/pcompress/), I did some small experimentation using
SSE intrinsics for the high compression version and got a slight sp…
-
```
While integrating LZ4 into my Pcompress utility
(http://moinakg.github.com/pcompress/), I did some small experimentation using
SSE intrinsics for the high compression version and got a slight sp…
-
Our current HashMap uses cryptographically secure hashing but it is slow. Sometimes this is a big problem. Implement another hash map that is very fast.
brson updated
9 years ago
-
Running fontforge in valgrind looking for a different bug produces warnings about this:
```
==32563== Source and destination overlap in strcpy(0x7feffa080, 0x7feffa083)
==32563== at 0x4A08038: str…
ghost updated
11 years ago
-
Somewhat like #416, except that
1. Only reproduces with syscall buffering enabled
2. Only reproduces 1/10 or 1/20 runs or so
3. Reproduces at what looks like fairly arbitrary points in execution, as o…
-
randn might be slower than it should be in some environments. I see:
```
julia> [@time (rand(1,10000000); gc()) for x = 1:10];
elapsed time: 0.030134916305541992 seconds
elapsed time: 0.0309810638427…
-
Some reading material on [Threads in Java](http://en.wikibooks.org/wiki/Java_Programming/Threads_and_Runnables)
I was remembering incorrectly; `run()` is on a user subclass of `Runnable`, not of `Thr…
jjfiv updated
11 years ago
-
I am trying to reimplement some Matlab code in Julia, but there is a big performance gap when using complex numbers.
As an example:
m = rand(1024,1024);
@time n = sum(m);
elapsed time: 0.00701022148…