https://pengchengneo.github.io/posts/scaling-book-cn/part0_introduction/2026-04-29T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/pallas-jax-kernel-language-cn/pallas_quickstart_cn/2026-04-19T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/scaling-book-cn/part1_roofline/2026-04-29T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/pallas-jax-kernel-language-cn/pallas_grid_blockspec_cn/2026-04-19T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/scaling-book-cn/part2_tpus/2026-04-29T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/pallas-jax-kernel-language-cn/pallas_pipelining_cn/2026-04-19T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/scaling-book-cn/part3_sharding/2026-04-29T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/pallas-jax-kernel-language-cn/pallas_tpu_details_cn/2026-04-19T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/scaling-book-cn/part4_transformers/2026-04-29T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/pallas-jax-kernel-language-cn/pallas_tpu_pipelining_cn/2026-04-19T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/scaling-book-cn/part5_training/2026-04-29T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/pallas-jax-kernel-language-cn/pallas_tpu_matmul_cn/2026-04-19T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/scaling-book-cn/part6_applied_training/2026-04-29T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/pallas-jax-kernel-language-cn/pallas_tpu_sparse_cn/2026-04-19T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/scaling-book-cn/part7_inference/2026-04-29T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/pallas-jax-kernel-language-cn/pallas_tpu_distributed_cn/2026-04-19T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/scaling-book-cn/part8_applied_inference/2026-04-29T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/pallas-jax-kernel-language-cn/pallas_tpu_coremap_cn/2026-04-19T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/scaling-book-cn/part9_profiling/2026-04-29T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/pallas-jax-kernel-language-cn/pallas_tpu_sparsecore_cn/2026-04-19T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/scaling-book-cn/part10_jax/2026-04-29T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/pallas-jax-kernel-language-cn/pallas_tpu_prng_cn/2026-04-19T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/scaling-book-cn/part11_conclusion/2026-04-29T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/pallas-jax-kernel-language-cn/pallas_gpu_reference_cn/2026-04-19T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/scaling-book-cn/part12_gpus/2026-04-29T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/pallas-jax-kernel-language-cn/pallas_gpu_pipelining_cn/2026-04-19T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/pallas-jax-kernel-language-cn/pallas_gpu_blackwell_matmul_cn/2026-04-19T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/pallas-jax-kernel-language-cn/pallas_gpu_collective_matmul_cn/2026-04-19T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/pallas-jax-kernel-language-cn/pallas_design_cn/2026-04-19T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/pallas-jax-kernel-language-cn/pallas_design_async_cn/2026-04-19T00:00:00+00:00weekly0.5https://pengchengneo.github.io/2026-10-06T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/fineweb-cn/2026-10-06T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/2026-10-06T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/ultrascale-playbook-cn/2026-10-06T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/ai-chip-architectures-zh/2026-08-24T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/anthropic-vliw-simd-kernel-optimization/2026-08-12T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/pathways-distributed-runtime/2026-05-28T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/reiner-pope-blackboard-lectures/2026-05-25T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/designing-ai-chip-cn/2026-05-22T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/scaling-book-cn/2026-04-29T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/pallas-jax-kernel-language-cn/2026-04-19T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/tpu-gpu-interconnect-topology/2026-04-15T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/tpu-v6e-ici-bandwidth-test/2026-04-15T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/nvidia-gpu-comparison/2026-04-14T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/tpu-mxu-systolic-array/2026-04-12T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/tpu-compiler/2026-04-06T00:00:00+00:00weekly0.5https://pengchengneo.github.io/posts/gemm-kernel-optimization-10-questions/2026-04-04T00:00:00+00:00weekly0.5https://pengchengneo.github.io/projects/2026-03-20T00:00:00+00:00weekly0.5