Deschide articolul original într-un tab nou Schimbarea acestei opțiuni se aplică imediat și reîncarcă pagina.
LECTURA
Înapoi la articole →

Zhenyu Sha

zhenyu.github.io/feed.xml
Vizitează site-ul

Cel mai recent articol

Ultima verificare

Urmăritori 0 urmăritori
Articole: 7

  1. Why a GPU Profiling Capture Turned into a Distributed Systems Problem

    (Zhenyu Sha)

    Starting from the ideas behind Meta’s MAIProf, I redesigned capture ownership, process discovery, and trace delivery for Kubernetes.

  2. Starting from the GPU Roofline: Tuning the Data Path for Multimodal Video Training

    (Zhenyu Sha)

    In a previous post, I wrote about design patterns for large-scale multimodal training data pipelines: metadata + blobs, distributed streaming DAGs, and pre-sharded training artifacts.

  3. How to Get Tensor Core Metrics from a SageMaker Training Job

    (Zhenyu Sha)

    W&B will happily tell you your GPU is at “100% utilization.”

  4. Is Kubernetes Autoscaling Missing a Capacity Intent Layer?

    (Zhenyu Sha)

    A friend recently made a simple observation about Kubernetes autoscaling: Kubernetes can scale Pods, but turning those Pods into the right Nodes still feels like something every DevOps or platform team has to solve by hand.

  5. When We Talk About Workflow, What Are We Actually Talking About?

    (Zhenyu Sha)

    I’ve been having the same conversation a lot lately.

  6. MultiKueue Solves Dispatch, Not Multi-Cloud GPU Placement

    (Zhenyu Sha)

    MultiKueue is often presented as a natural answer for multi-cloud GPU scheduling.

  7. Design Patterns for Large-Scale Multimodal Training Data Pipelines

    (Zhenyu Sha)

    I wrote this mostly to clarify my own thinking.