OpenAI hardware team member and XLA creator Sanjoy Das described how AI is being used as a 'stochastic optimizer' to write and optimize kernels, such as the Jalapeño MLA kernel presented at HotChips, …
A developer's hand-written Gemma 4 port in pure JAX runs on both Turing and Ada NVIDIA GPUs with identical weights, but performance differs drastically due to hardware-specific compute dtype handling …
Google Cloud has introduced native vLLM TPU support for embedding inference, targeting production long-context retrieval with Qwen3-Embedding-8B and Qwen3-VL-Embedding-8B. The engineering effort addre…
Google Cloud has integrated native TPU support into vLLM, the open-source LLM serving engine, to enable enterprise-grade precision for long-context multimodal embedding inference, targeting the Qwen3 …
OpenAI is hiring a Software Engineer, Trainium in San Francisco with a salary range of $295k–380k/yr, to build and optimize its inference stack for AWS Trainium, developing high-performance kernels an…
A GitHub repository introduces an experimental Proof of Concept for an Adiabatic Silicon Aging & Thermal Degradation Failure-Fencing Engine, designed to isolate NaN and infinity propagation in hypersc…
HeyGen, an AI video generation platform, announced that its Avatar IV diffusion model, which runs on more than 18B parameters, is now 1.86× faster on Google Cloud's eight-chip Trillium (v6e) TPU host …
TensorFlow's XLA compilation path silently converts tf.debugging.Assert and tf.Assert operations into no-ops, disabling runtime checks in jit_compile=True functions. This deliberate behavior, implemen…
Xiaomi has opened pre-orders for the N90 Max extended-range SUV at RMB299,900, with formal release scheduled for September. The vehicle integrates Xiaomi's HAD assisted-driving system and XLA cognitiv…
Xanadu's quantum compiler Catalyst, built for the PennyLane library, accidentally became an LLVM compiler for JAX when users fed it pure JAX NumPy code without quantum instructions, bypassing XLA to l…
JAX's Pallas kernel system, which lowers through Triton on GPU and Mosaic on TPU, lets developers write custom kernels when XLA's automatic fusion is insufficient for operations like flash attention, …
Google's open-source TPU compiler, shipped as the Mosaic TPU dialect inside JAX, documents eight explicit memory spaces that programmers must manage, including vector memory (VMEM), scalar memory (SME…
Token dropping in mixture-of-experts (MoE) layers occurs when a capacity factor limits each expert's buffer, causing excess tokens to bypass the expert MLP and degrade model quality under production l…
XLA, the compiler behind JAX, TensorFlow, and PyTorch/XLA, optimizes array programs by freezing shapes, statically allocating buffers, and fusing operations against a global cost model, which makes it…
XLA, the compiler under JAX, TensorFlow, and PyTorch/XLA, uses two intermediate representations: classic HLO (a hand-built C++ IR) and MLIR dialects such as StableHLO and CHLO, with a translation laye…
A hardware-software co-design prototype called PIM-HBM explores zero-copy memory view fusion and fault telemetry for LLM training, using JAX/XLA integration and branchless CUDA kernels to mitigate com…
A developer ported Karpathy's nanochat to run on a TPU v6e-8 using JAX, achieving a CORE score that reproduces the original quality but with a model FLOPs utilization (MFU) of about 24%, roughly half …
NVIDIA and Google researchers have developed a host offloading technique in JAX that reduces high-bandwidth memory bottlenecks during LLM training, achieving up to 57% throughput improvements on NVIDI…
Google Cloud demonstrated elastic training on TPUs, where a worker failure during multi-node LLM training was recovered in under two minutes without restarting the job. Using the JAX AI stack (MaxText…
A developer built a hybrid GNN and decision-tree architecture for predicting XLA compiler runtime, achieving high speed and low memory footprint. The solution outperforms standard deep GNNs by shiftin…