HN user

blakepelton

128 karma

Check out my blog for summaries of recent CS research:

https://danglingpointers.substack.com/

thedanglingpointers@gmail.com

Posts62
Comments43
View on HN
danglingpointers.substack.com 13d ago

Pipeline Parallel Decompression

blakepelton
3pts0
dl.acm.org 1mo ago

Tiered Network Buffer Architecture for Fast Networking in Chiplet-Based CPU

blakepelton
2pts1
danglingpointers.substack.com 1mo ago

Binary Compatible Critical Section Delegation

blakepelton
1pts0
dl.acm.org 2mo ago

Radshield: Software Radiation Protection for Commodity Hardware in Space

blakepelton
2pts1
dl.acm.org 2mo ago

Cacheman: A Comprehensive Last-Level Cache Management System

blakepelton
1pts1
danglingpointers.substack.com 2mo ago

Hapax Locks: Scalable Value-Based Mutual Exclusion

blakepelton
1pts0
danglingpointers.substack.com 3mo ago

Scalar Interpolation: A Better Balance Between Vector and Scalar Execution

blakepelton
1pts0
danglingpointers.substack.com 3mo ago

Flexible I/O for Database Management Systems with XNVMe

blakepelton
1pts0
dl.acm.org 3mo ago

A Digital Compute-in-Memory Architecture for NFA Evaluation

blakepelton
8pts1
danglingpointers.substack.com 4mo ago

Better Memory Tiering, Right from the First Placement

blakepelton
1pts0
dl.acm.org 4mo ago

Efficient Lossless Compression of Scientific Floating-Point Data on CPUs and GPU

blakepelton
3pts1
danglingpointers.substack.com 4mo ago

Contiguitas: The Pursuit of Physical Memory Contiguity in Datacenters

blakepelton
1pts0
dl.acm.org 4mo ago

EDM: An Ultra-Low Latency Ethernet Fabric for Memory Disaggregation

blakepelton
2pts1
danglingpointers.substack.com 4mo ago

Re-architecting End-host Networking with CXL

blakepelton
6pts0
dl.acm.org 4mo ago

An Analysis of User-Space Idle State Instructions on x86 Processors

blakepelton
22pts6
danglingpointers.substack.com 4mo ago

Forest: Access-Aware GPU UVM Management

blakepelton
2pts0
dl.acm.org 4mo ago

UPP: Universal Predicate Pushdown to Smart Storage

blakepelton
13pts1
danglingpointers.substack.com 5mo ago

RayN: Ray Tracing Acceleration with Near-Memory Computing

blakepelton
1pts0
danglingpointers.substack.com 5mo ago

Shadow: Simultaneous Multi-Threading Architecture with Asymmetric Threads

blakepelton
2pts0
dl.acm.org 5mo ago

Light-Weight Cache Replacement for Instruction Heavy Workloads

blakepelton
2pts1
danglingpointers.substack.com 5mo ago

LoopFrog: In-Core Hint-Based Loop Parallelization

blakepelton
1pts0
dl.acm.org 6mo ago

CHERIoT RTOS: An OS for Fine-Grained Memory-Safe Compartments

blakepelton
4pts1
danglingpointers.substack.com 6mo ago

Dynamic Load Balancer in Intel Xeon Scalable Processor

blakepelton
1pts0
dl.acm.org 6mo ago

The XOR Cache: A Catalyst for Compression

blakepelton
1pts1
danglingpointers.substack.com 6mo ago

Rearchitecting the Thread Model of In-Memory Key-Value Stores with μTPS

blakepelton
1pts0
dl.acm.org 6mo ago

CHERIoT: Complete Memory Safety for Embedded Devices

blakepelton
3pts1
danglingpointers.substack.com 6mo ago

Morsel-Driven Parallelism

blakepelton
1pts0
danglingpointers.substack.com 7mo ago

Rearchitecting the Thread Model of In-Memory Key-Value Stores with μTPS

blakepelton
2pts0
danglingpointers.substack.com 7mo ago

Tai Chi: A General High-Efficiency Scheduling Framework for SmartNICs

blakepelton
2pts0
dl.acm.org 7mo ago

Oasis: Pooling PCIe Devices over CXL to Boost Utilization

blakepelton
17pts2
  Location: Seattle, WA
  Remote: Yes
  Willing to relocate: No
  Technologies: C++, Graphics, GPU, FPGA, HW/SW interface, Performance Optimization
  Résumé/CV: https://drive.google.com/file/d/1Ru26-85bPD9NMI2oKHSkhvrxmvhLjZ2v/view
  Email: blakepelton@hotmail.com
20+ years of software experience in low-level domains where performance is critical.
No Graphics API 7 months ago

Great post, it brings back a lot of memories. Two additional factors that designers of these APIs consider are:

* GPU virtualization (e.g., the D3D residency APIs), to allow many applications to share GPU resources (e.g., HBM).

* Undefined behavior: how easy is it for applications to accidentally or intentionally take a dependency on undefined behavior? This can make it harder to translate this new API to an even newer API in the future.