Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Update hpc.md #2234

Open
wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Update hpc.md #2234

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 12 additions & 0 deletions jsoc/gsoc/hpc.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,3 +13,15 @@ There are projects now that host the building blocks: [DaggerFlux.jl](https://gi
**Skills:** Familiarity with UCX, representing execution models as DAGs, Flux.jl, CUDA.jl and data/model parallelism in machine learning

**Mentors:** [Julian Samaroo](https://github.com/jpsamaroo), and [Dhairya Gandhi](https://github.com/DhairyaLGandhi)

## Optimizing GPU scheduler in Dagger.jl with Multistreams

**Difficulty:** Hard (350h)

This project aims to explore and enhance GPU performance by integrating [Dagger.jl](https://github.com/JuliaParallel/Dagger.jl), Julia’s high-performance parallel computing framework, with GPU multistream capabilities. Dagger.jl enables task-based parallelism, allowing complex computations to be broken down into smaller, manageable tasks that can be efficiently scheduled across computing resources. By incorporating GPU multistreams, students will investigate how multiple streams can be used to overlap data transfers with kernel executions, enabling concurrent operations on the GPU. This overlapping reduces idle times, as data movement and computations occur simultaneously, thus maximizing GPU resource utilization. The project will focus on designing and implementing parallel workflows where independent tasks are executed concurrently, leveraging Dagger’s dynamic task scheduling and GPU’s ability to manage multiple streams effectively. Students will experiment with different workload patterns, measure performance improvements, and analyze the impact of multistream execution on throughput and latency. Through performance benchmarking and optimization, this project will provide hands-on experience in GPU programming, parallel algorithm design, and high-performance computing, equipping students with valuable skills for tackling real-world scientific and data-intensive applications.

There are projects now that host the building blocks: [DaggerGPU.jl](https://github.com/JuliaGPU/DaggerGPU.jl) and [Dagger.jl](https://github.com/JuliaParallel/Dagger.jl) which can serve as jumping off points.

**Skills:** Familiarity with GPU, representing execution models as DAGs, CUDA.jl

**Mentors:** [Julian Samaroo](https://github.com/jpsamaroo), and [Rabab Alomairy](https://github.com/Rabab53)