30
30
30
NVIDIA Announces CUDA Rust with cuda-oxide (SIMT) and cutile-rs (Tile) for Compile-Time-Safe GPU Kernels
NVIDIA launches CUDA Rust via open-source cuda-oxide (SIMT) and cutile-rs (Tile), bringing compile-time-safe Rust GPU kernels.
NVIDIA announced CUDA Rust, making Rust a first-class language for GPU kernels through two NVlabs open-source projects: cuda-oxide for the SIMT model and cutile-rs for the Tile model. Both use Rust's ownership and borrow checker to catch buffer aliasing bugs at compile time. cutile-rs is published on crates.io, runs on stable Rust 1.89+ with CUDA 13.3, and is already used in Hugging Face's Grout inference engine and mistral.rs; cuda-oxide is early alpha requiring nightly Rust, CUDA 12.x, and compute capability 8.0+. cuda-oxide compiles Rust MIR through the community Pliron IR framework and LLVM to PTX, while cutile-rs JIT-compiles kernels via CUDA Tile IR.
45
AIR raises $50M to help companies vet the skills and add-ons AI agents use
AI security startup AIR exits stealth with $50M across two seed rounds to continuously vet skills, plugins, and MCP servers used by enterprise AI agents.
AIR, founded by Unit 8200 veterans Yair Saban and Niv Hoffman, raised $10 million led by Sequoia and $40 million led by Greenoaks to secure the emerging AI agent software supply chain. Its platform discovers agents in enterprise environments, intercepts their actions, and continuously vets skills, add-ons, and MCP servers against a maintained whitelist, filtering out about 27% of found add-ons. The company claims over 20 customers, roughly a quarter large enterprises, with strong demand in financial services and pharma, and competes with Noma Security, Zenity, Astrix Security, and Operant AI.
46
Retrospectively Reverse-Engineering Apple's Neural Engine
A developer reverse-engineers Apple's M1 Neural Engine architecture, mapping compute cores, MAC datapaths, and schedulers to explain the NPU's decline as transformers displaced CNN workloads.
A developer who previously maintained a reverse-engineered Linux driver for Apple's Neural Engine (ANE) published a retrospective deep dive mapping the M1 ANE's full internal architecture: compute, datapath, scheduler, memory, and execution model. The M1 ANE has 16 compute cores with 128 FP16 (or 256 INT8) MAC lanes each, totaling 2048 parallel MAC lanes, using 32-bit Q16.16 fixed-point accumulation with FP16 readout and an accumulator that saturates at 2^15. The author argues the ANE's dataflow was architected around the predictable reuse patterns of 2017-era CNN workloads (dating to the A11 Bionic), which autoregressive transformer decode broke, limiting its usefulness for general ML. With Apple's M5 folding ANE cores into GPU cores to tout LLM performance, the post frames this as the beginning of the end for the standalone NPU.
28
FlashVector: Agent for Hierarchical Model Serving Stack Optimization
FlashVector agent optimizes all layers of Unity's ad-serving stack, delivering up to 2x model-server throughput and 1.98x latency speedup in production.
FlashVector is an agentic system that optimizes performance across GPU kernels, ML framework computation graphs, model servers, and on-demand feature processing. Deployed in Unity's Vector advertising platform, it achieved up to 2x model-server throughput increase, 1.98x latency speedup, and 1.6x feature-store throughput gain. Optimizations spanned NVIDIA Triton's C++ codebase and the Python feature transformation service, demonstrating extensibility beyond single-kernel tuning.
30
Perplexity Details Its GPU Embedding Stack: How Ivy, Tulip and ROSE Serve pplx-embed
Perplexity details its GPU embedding serving stack (Ivy, Tulip, ROSE), which reuses LLM prefill/decode kernels, CUDA graphs, and LazyTensors to cut launch overhead.
Perplexity engineers published a deep dive on the serving infrastructure behind pplx-embed, used across Perplexity Search and its API platform. The stack comprises Ivy (Rust HTTP gateway), Tulip (gRPC scheduling and batching), and ROSE (Runtime-Optimized Serving Engine), which reuses LLM prefill and decode kernels rather than running a separate embedding engine. Optimizations include whole-model CUDA graphs with lazy capture and a LazyTensor abstraction that overlaps CPU batch preparation with in-flight GPU work. Benchmarks are reported against vLLM v0.22.0 in BF16, with FlashAttention 4 generally fastest but FlashInfer 3 winning on Qwen-based models at very long sequence lengths.
33
30
Dream-RSI: Recursive Self-Improvement through Evolving Worlds
Dream-RSI refines exploration policies by dreaming in replay simulators built from discovery history, cutting discovery costs across coding tasks.
Dream-RSI is a framework for scalable recursive self-improvement in autonomous coding agents, where a lightweight orchestration layer makes exploration explicit and programmable while leaving the underlying agent unchanged. Its core insight is that accumulated discovery history can serve as a replay simulator over the realized search space, providing immediate, low-cost off-policy feedback to evaluate and refine exploration policies without expensive online evaluations. Across algorithm engineering, mathematical optimization, and GPU kernel engineering, Dream-RSI achieves competitive or improved discovery quality at substantially reduced cost.
32
NVIDIA Details BioNeMo Inference Runtime (BioIR): 2.90x Higher Boltz-2 Folding Throughput and 58.5K Residues per GPU-Hour on 8xH100
NVIDIA released BioNeMo Inference Runtime (BioIR), an open-source PyTorch-compatible library delivering 2.90x higher Boltz-2 protein-folding throughput on 8xH100 GPUs.
NVIDIA detailed BioIR, a Python library that accelerates Boltz-2, OpenFold2, and OpenFold3 structure-prediction inference on NVIDIA GPUs while preserving standard PyTorch workflows. On a matched benchmark of 1,000 human dimers on 8xH100 80GB GPUs, BioIR delivered 58.5K folded residues per GPU-hour versus 20.2K for a torch.compile baseline, a 2.90x throughput gain. BioIR already powered the AlphaFold Database expansion, generating about 31 million candidate complexes across 4,777 proteomes, with 1.81 million released as high-confidence predictions. Extrapolated to 1 million targets, estimated folding energy drops from 35 MWh to 11 MWh at 8-GPU TDP equivalents.
25
30