4 results
JULY 14, 2026
To serve the 397B-parameter Qwen 3.5 Mixture-of-Experts (MoE) model on Ironwood TPUs, engineers developed a modular JAX/Pallas optimization stack that achieved up to a 4.7x inference speedup for prefill-heavy workloads. The team bypassed severe hardware sharding constraints by deploying a hybrid Data Parallelism and Expert Parallelism (DP+EP) topology, paired with custom low-level communication fusions like a hierarchical reduce-scatter to optimize cross-device token routing. Finally, by executing hardware-aware custom kernels—such as Batched Ragged Page Attention and a fully-fused Gated DeltaNet (GDN) block—they successfully saturated HBM bandwidth and TensorCore MXUs to push system throughput near its theoretical roofline limits.
AUG. 8, 2024 / Flutter
Purrfect Code is a sokoban style box pushing puzzler Purrfect based on Google tech (Flutter, IDX, FLAME, Firebase) and designed to be playable on web browsers. Open the game in Project IDX, earn badges for each level, and showcase your badges on your Developer Profile.
OCT. 30, 2023 / Best Practices
For many, Halloween is the perfect excuse to dress up and celebrate the things that haunt us. Google...
SEPT. 27, 2023 / Cloud
Whether you’re stuck on a problem, reading documentation, or figuring out the best tech stack for your project, Google Search has been a helpful tool.
