Matrix Multiplication On GPU: Part 2, Tiling
Breaking down large matrix multiplications into tiles
8 min read
4 Oct 24
Matrix Multiplication on GPU: Faster than Nvidia, Sometimes
It works on my laptop
10 min read
1 Oct 24
Gradients of Softmax and Logsumexp
Essential functions for categorical distributions and attention mechanisms in machine learning
4 min read
4 May 24
C++: Pattern Matching Template Types
How to check if a template type matches a pattern? Something like is_like_v<T, vector<int,_>>.
6 min read
2 Mar 24
C++: Overloading the Spaceship Operator, A Recipe
How to overload the three-way comparison (spaceship) operator<=>, and a reminder to overload operator== as well.
3 min read
9 Feb 24
C++: Check if a type is an instantiation of a given class template
How to implement an is_instance_of type trait.
3 min read
5 Feb 24
C++: Forwarding references, overload resolution, and taking back control
Consider merging overloads into one function with forwarding reference parameters
5 min read
27 Jan 24
C++: Disable implicit conversion in specific contexts only
You get one implicit conversion, so burn it with a wrapper
2 min read
14 Nov 23
C++: Revisiting combinatorial instantiation of templates with std::variant
Compiler optimizations can break it, function attributes can fix it.
3 min read
16 Aug 23
C++: Combinatorial instantiation of templates with std::variant
An alternative to explicit instantiations and macros.
3 min read
11 Jun 23
C++: Avoiding Argument Dependent Lookup
A little trick using an extra namespace and cross-import.
2 min read
30 Apr 23
Sums of Discrete Random Variables as Banded Matrix Products
Zero-stride catch and a custom CUDA kernel.
10 min read
16 Mar 23