╠2️⃣│rocm

80 threads · Page 1 of 2

ROCm Windows · gfx1100 · BEATEK Holdings 3 messages
vLLM + Step-3.7-Flash-FP8 R9700 seeking optimization 5 messages
HIP AI Setup
Qwen3.6 on AMD GPUs: what changed since Qwen3.5 (with R9700 benchmarks) 4 messages
vLLM setup with 8xR9700? 8 messages
HIP AI Compiler Setup
hipDeviceGetUuid still doesn't work on windows in 2026 5 messages
HIP Tools AI
Amdgpu-dkms crashes on new Ubuntu 26.04 LTS 3 messages
Running Therock on comfy 3 messages
Sharing Tunable Ops results 14 messages
hipEngine: ROCm-native local LLM inference for RDNA3/3.5 3 messages
ComfyUI on AMD GPUs (Linux): the pitfalls that freeze your whole box 9 messages
Solving the issue of dual gpu not working on Ubuntu/Cachyos 3 messages
Games are slow after ROCm install 6 messages
VLLM and Qwen3.6-27B-FP8 on Radeon R9700 5 messages
probs offloading tensors & layers from Mac M4pro over thunderbolt 5 to usb4 rocm gpu 3 messages
ROCm 7.2.2 is out — and it finally has an official RDNA 3.5 optimization guide 2 messages
Severe UI Latency with Blender Built from Source on MI300X (HIP/PTX Translation bottlenecks? 19 messages
HIP Performance Compiler Frameworks HIPThreads
ROCm vs Vulkan for LLM inference on AMD: what I've found across different hardware 4 messages
My AMD Radeon RX 7600M XT was brought two years ago,but I want to run Rocm on my laptop. 2 messages
Tools AI Frameworks Setup
Qwen3.5 on AMD — what's working, what's not, and why 11 messages
hipfire — from-scratch LLM inference for RDNA GPUs in Rust + HIP 18 messages
HIP AI Performance Tools
Qwen3.5‑2B Runs at 26 tok/s on a 2019 MacBook Pro (No ROCm, No CUDA) 18 messages
Resurrecting Legacy AMD: From 2 tok/s to 37 tok/s on a 7-year-old Radeon GPU without ROCm 4 messages
Stop blaming AMD silicon: Ollama / Standard Stacks vs. Custom Bare-Metal (45+ tok/s on a 2019 GPU) 2 messages
3x HP Z2 Mini G1a (Ryzen AI Max+ Pro 395, 128GB) — What's the supported ROCm stack for Strix ? 10 messages
[rocprofv3] torch.cuda.nvtx support and marker tracing in ROCm 6.4.4 (MI250X) 14 messages
Dual gpu not happening 15 messages
Difference between hipModule and hipLibrary 3 messages
HIP
Does rocm aiter has fused kernels for FP4 quantization and gemm? 4 messages
Show: Bypassing the 𝑂(𝑁3) Matrix Inversion Wall – 67.7x Faster in Pure PyTorch on Ryzen 9950X 2 messages
Anyone running QWEN 3.5 yet? Seems there are llama issues with it? 37 messages
How does Windows compare to Linux in performance right now? 28 messages
Building a Custom PyTorch Backend with Rust and Vulkan on Windows 5 messages
Tools HPC AI
HIP Threads: GPU power for teams without GPU experts 12 messages
HIP Compiler HIPThreads
Does SCALE support the MI300X? 3 messages
Compiler Setup
Installing ROCM 7.2 on Linux Mint 22 20 messages
amdrocm-core-sdk-gfx1150 apt package doesn't install rocminfo or amdsmi? 4 messages
Using gplat on ROCM 7.2 9 messages
BS Roformer optimisation 14 messages
Rocm with lmstudio in windows 5 messages
Ollama 0.15.4 with ROCm 7.2 and gfx1201 94 messages
HIP AI
GPU Reset 890m/gfx1150 6 messages
AI Frameworks Setup
Observations on ROCm 7.1, gfx1030 and hope for the future 38 messages
ROCm 7.2 vs 7.10 on 395+ 4 messages
ROCm on 395+ (Ubuntu 24.04) 70 messages
ADTOF running really slow 9 messages
[Benchmark] qingming-engine Vector Search Performance: RX 7900 XTX 24G Shows Excellent Results 6 messages
** :rocket: [Benchmark] AOTriton Speed Mystery: Nightly Build vs. Staging/Dev Builds on 7900 X 45 messages
Performance
Amd 2 messages
Building FlashAttention with debug flags 3 messages
Compiler
Omega2.0 just changed the game. 2 messages