Top all-time · #2318
AMD ROCm Blogs
Company blog · rocm.blogs.amd.comaihardwareprogramming
AMD ROCm Blogs is the company blog at rocm.blogs.amd.com, writing about ai, hardware and programming. Since 2024, 31 different Hacker News readers have linked it 60 times, most often “SPIR-V on ROCm: A Portable IR for AMD GPUs”. It is still writing: the last post was on 2026-10-30, and you can follow it by feed.
Too few sampled comments (12) to say how often HN recommends it.?
Why HN links it
HN readers link AMD's ROCm developer blog as a source on GPU software: DeepSeek inference performance, running DeepSeek locally, SPIR-V support, and C++ parallel algorithms on AMD GPUs.
Summarised by Claude from a sample of 12 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
Most-linked posts?
- SPIR-V on ROCm: A Portable IR for AMD GPUs3 people · 4 mentions · first linked 2026-07-20
- Matrix Core Programming on AMD CDNA Architecture2 people · 7 mentions · first linked 2025-10-02
- Supercharge DeepSeek-R1 Inference on AMD Instinct MI300X2 people · 2 mentions · first linked 2025-03-22
- Unlock DeepSeek-R1 Inference Performance on AMD Instinct MI300X GPU2 people · 2 mentions · first linked 2025-02-23
- AMD MI300 compute and memory partitioning2 people · 2 mentions · first linked 2025-06-01
- A Deep Dive into LDS Optimizations on AMD Instinct MI450 GPUs2 people · 2 mentions · first linked 2026-08-28
What Hacker News says?
We have a replacement for CUDA, it is called C++17 parallel algorithms. It has vendor support for running on the GPU by Intel, AMD and NVIDIA and will also run on all your cores on the CPU. It uses the GPU vendors compiler to convert your C++ to something that can natively run on the GPU. With unified memory support, it becomes very fast to run computations on heap allocated memory using the GPU, but implementations…
Comment linking /software-tools-optimization/hipstdpar/README.html
on “Bend: a high-level language that runs on GPUs (via HVM2)”Submitted Aiter: AI Tensor Engine for ROCm as a story · 179 points
HN discussionIt's great to see vLLM getting faster/better for DeepSeek. I tested vLLM vs SGLang a couple weeks ago and SGLang's DeepSeek support was much better/faster (on 2 x p5 H100 nodes). It's great that no one's standing still, I saw this recent AMD article that reported SGLang perf on MI300X has increased by 4X over the past couple weeks: https://rocm.blogs.amd.com/artificial-intelligence/DeepSeekR... (w/ the extra memory…
Comment linking /artificial-intelligence/DeepSeekR1_Perf/README.html
on “DeepSeek Open Source FlashMLA – MLA Decoding Kernel for Hopper GPUs”Submitted AMD ROCm Software Blogs as a story · 118 points
HN discussionThis reads like a badly done, sponsored hype video on YouTube. So if we look at what NVIDIA has to say about NVFP4 it sure sounds impressive [1]. But look closely that initial graph never compares fp8 and fp4 on the same hardware. They jump from H100 to B200 while implying a 5x jump of going with fp4 which it isn't. Accompanied with scary words like if you use MXFP4 "Risk of noticeable accuracy drop compared to FP8"…
Comment linking /software-tools-optimization/mxfp4-mxfp6-quantization/README.html
on “TPUs vs. GPUs and why Google is positioned to win AI race in the long term”Submitted Matrix Core Programming on AMD CDNA Architecture as a story · 61 points
HN discussionHere is a good article on it: https://rocm.blogs.amd.com/software-tools-optimization/compu...
Comment linking /software-tools-optimization/compute-memory-modes/README.html
on “Why DeepSeek is cheap at scale but expensive to run locally”Submitted Boosting Computational Fluid Dynamics Performance with AMD MI300X as a story · 47 points
HN discussion(Split into two parts due to comment length restrictions) > I work closely with OLCF and Frontier (I have a job running on Frontier right now). This is incorrect. The overwhelming majority of compute and resource allocation are not "nuclear stockpile modeling code" projects or anything close to it. AMD often gets directly involved with various issues (OLCF staff has plenty of stories about this). I know because I've…
Comment linking /artificial-intelligence/vllm/README.html
on “An interview with AMD CEO Lisa Su about solving hard problems”Submitted ROCm 10.0: A Decade of Open Compute, Built for the Age of Agentic AI as a story · 18 points
HN discussionI think a lot of stuff will get upstreamed eventually. PyTorch just moves slower and since it’s a stable library, I think it cannot rapidly adopt something like fused MoE until the dust has settled a little and it’s clear what the API would look like long-term. I think it’s ok that stuff is tried first in Torch extensions. That’s how Flash Attention started after all and the same is true for newer kernels in CUDA-la…
Comment linking /artificial-intelligence/DeepSeekR1-Part2/README.html
on “Aiter: AI Tensor Engine for ROCm”Submitted C++17 Parallel Algorithms and Hipstdpar as a story · 8 points
HN discussionOh, interesting! Thanks for the additional context. This is one area where it is clear that Nvidia is a leader by not only providing the underlying kernels, but also the overall product framework integrations. One would have to port that entire plugin project over, which would probably be a ton of work to maintain. For what it is worth, AMD just recently released two blog posts on hipDF, so at least they are putting…
Comment linking /artificial-intelligence/hipDF_pandas_accelerated/README.html
on “Launch HN: ParaQuery (YC X25) – GPU Accelerated Spark/SQL”Submitted ROCm 10.1: Breaking the Data-Movement Bottleneck – ROCm Blogs as a story · 3 points
HN discussionHaving looked briefly at the code I still think C++17 parallel algorithms are more ergonomic compared to OpenMP: https://rocm.blogs.amd.com/software-tools-optimization/hipst...
Comment linking /software-tools-optimization/hipstdpar/README.html
on “Porting HPC Applications to AMD Instinct MI300A Using Unified Memory and OpenMP”Submitted A Deep Dive into LDS Optimizations on AMD Instinct MI450 GPUs as a story · 3 points
HN discussionThe benchmark this article relies on is here[1], and is focused on a single-user case and uses QwQ-32B-AWQ with a context length of 4096 as far as I can see. Which, from what I can gather, is a quite unrealistic setup for someone seriously considering buying a H100 for the home lab. That said, the tensor parallelism[2] of vLLM is interesting. [1]: https://github.com/DeutscheKI/llm-performance-tests#vllm-per... [2]:…
Comment linking /artificial-intelligence/tensor-parallelism/README.html
on “Dual RTX 5090 Beats $25,000 H100 in Real-World LLM Performance”Submitted SPIR-V on ROCm: A Portable IR for AMD GPUs as a story · 3 points
HN discussionSpir-v is already happening: https://rocm.blogs.amd.com/software-tools-optimization/spir-...
Comment linking /software-tools-optimization/spir-v-rocm/README.html
on “AMD's Instinct MI455X: Aiming for the Sun”Submitted FP8 GEMM Optimization on AMD CDNA4 Architecture as a story · 3 points
HN discussionNot anymore: https://rocm.blogs.amd.com/software-tools-optimization/spir-...
Comment linking /software-tools-optimization/spir-v-rocm/README.html
on “AMD's Instinct MI455X: Aiming for the Sun”Submitted Matrix Core Programming on AMD CDNA4 Architecture as a story · 3 points
HN discussionhttps://rocm.blogs.amd.com/artificial-intelligence/rocjpeg-d... https://alax.info/blog/1992 https://en.wikipedia.org/wiki/Unified_Video_Decoder added around H.265 in UVD 6.0
Comment linking /artificial-intelligence/rocjpeg-decoding-performance-blog/README.html
on “FFmpeg 9.1's new AAC encoder”Submitted Installing ROCm from Source with Spack as a story · 3 points
HN discussionSubmitted Large language model inference optimizations on AMD GPUs as a story · 3 points
HN discussionSubmitted FP16 GEMM Optimization on AMD RDNA 3 and RDNA 4 Architectures as a story · 2 points
HN discussionSubmitted A Deep Dive into LDS Optimizations on AMD Instinct MI450 GPUs as a story · 2 points
HN discussionSubmitted Instella-Moe: An Open Mixture-of-Experts Language Model as a story · 2 points
HN discussionSubmitted SPIR-V on ROCm: A Portable IR for AMD GPUs as a story · 2 points
HN discussionSubmitted Occupancy Math on the AMD MI355X GPU (CDNA4): A From-First-Principles Guide as a story · 2 points
HN discussion
