Skip to content

matmul, ones and zeros#9

Merged
purushoath02 merged 3 commits intomasterfrom
ops-n-logic
Mar 15, 2025
Merged

matmul, ones and zeros#9
purushoath02 merged 3 commits intomasterfrom
ops-n-logic

Conversation

@kingjuno
Copy link
Owner

@kingjuno kingjuno commented Mar 14, 2025

This adds codes for ones, zeros, and also matmul
matmul is ~ to naive but this one takes advantage of cache so expect it to be a little bit more faster.

we need to have benchmark of some sort to test these. ofc we will improve matmul in future but lets make it less priority for now since we are will implement this in cuda as well (which is what we will be using)

but please check if everything is working or not.
thanks

updates:
after benchmarking with numpy
a = lm.ones((1000, 1000))
b = lm.ones((1000, 1000))
a@b

took 8sec while numpy took just 0.05 seconds
now i have done some optimizations and made it 4seconds and now 0.1 seconds. I will look into somemore optimizations but ig for now its fine. You can go ahead and merge if you think everything is alright.

@kingjuno kingjuno requested a review from purushoath02 March 14, 2025 13:52
@purushoath02 purushoath02 merged commit c423d48 into master Mar 15, 2025
3 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants