laser
analisis-numerico-computo-cientifico
laser | analisis-numerico-computo-cientifico | |
---|---|---|
6 | 1 | |
261 | 44 | |
1.5% | - | |
3.6 | 0.0 | |
4 months ago | over 1 year ago | |
Nim | Jupyter Notebook | |
Apache License 2.0 | Apache License 2.0 |
Stars - the number of stars that a project has on GitHub. Growth - month over month growth in stars.
Activity is a relative number indicating how actively a project is being developed. Recent commits have higher weight than older ones.
For example, an activity of 9.0 indicates that a project is amongst the top 10% of the most actively developed projects that we are tracking.
laser
-
From slow to SIMD: A Go optimization story
It depends.
You need 2~3 accumulators to saturate instruction-level parallelism with a parallel sum reduction. But the compiler won't do it because it only creates those when the operation is associative, i.e. (a+b)+c = a+(b+c), which is true for integers but not for floats.
There is an escape hatch in -ffast-math.
I have extensive benches on this here: https://github.com/mratsim/laser/blob/master/benchmarks%2Ffp...
-
Benchmarking 20 programming languages on N-queens and matrix multiplication
Ah,
It was from an older implementation that wasn't compatible with Nim v2. I've commented it out.
If you pull again it should work.
> Anyway the reason for your competitive performance is likely that you are benchmarking with very small matrices. OpenBLAS spends some time preprocessing the tiles which doesn't really pay off until they become really huge.
I don't get why you think it's impossible to reach BLAS speed. The matrix sizes are configured here: https://github.com/mratsim/laser/blob/master/benchmarks/gemm...
It defaults to 1920x1920 * 1920x1920. Note, if you activate the benchmarks versus PyTorch Glow, in the past it didn't support non-multiple of 16 or something, not sure today.
Packing is done here: https://github.com/mratsim/laser/blob/master/laser/primitive...
And it also support pre-packing which is useful to reimplement batch_matmul like what CuBLAS provides and is quite useful for convolution via matmul.
-
Why does working with a transposed tensor not make the following operations less performant?
For convolutions: - https://github.com/numforge/laser/blob/e23b5d63/research/convolution_optimisation_resources.md
-
Improve performance with SIMD intrinsics
You can train yourself on matrix transposition first. It's straightforward to get 3x speedup between naive transposition and double loop tiling, see: https://github.com/numforge/laser/blob/d1e6ae6/benchmarks/transpose/transpose_bench.nim#L238
analisis-numerico-computo-cientifico
What are some alternatives?
Arraymancer - A fast, ergonomic and portable tensor library in Nim with a deep learning focus for CPU, GPU and embedded devices via OpenMP, Cuda and OpenCL backends
osmnx-examples - Gallery of OSMnx tutorials, usage examples, and feature demonstations.
nim-sos - Nim wrapper for Sandia-OpenSHMEM
docker-curriculum - :dolphin: A comprehensive tutorial on getting started with Docker!
ParallelReductionsBenchmark - Thrust, CUB, TBB, AVX2, CUDA, OpenCL, OpenMP, SyCL - all it takes to sum a lot of numbers fast!
DeepLearning - Contains all my works, references for deep learning
blis - BLAS-like Library Instantiation Software Framework
vqgan-clip-generator - Implements VQGAN+CLIP for image and video generation, and style transfers, based on text and image prompts. Emphasis on ease-of-use, documentation, and smooth video creation.
JohnTheRipper - John the Ripper jumbo - advanced offline password cracker, which supports hundreds of hash and cipher types, and runs on many operating systems, CPUs, GPUs, and even some FPGAs [Moved to: https://github.com/openwall/john]
Kernels - This is a set of simple programs that can be used to explore the features of a parallel platform.
john - John the Ripper jumbo - advanced offline password cracker, which supports hundreds of hash and cipher types, and runs on many operating systems, CPUs, GPUs, and even some FPGAs
cocp - Source code for the examples accompanying the paper "Learning convex optimization control policies."