ZEN INVESTING
NVIDIA CUDA 13.2 Expands Tile Programming to Ampere and Ada GPUs
CUDA 13.2 extends tile-based GPU programming to older architectures, adds Python profiling tools, and delivers up to 5x speedups with new Top-K algorithms.
NVIDIA CCCL 3.1 Adds Floating-Point Determinism Controls for GPU Computing
NVIDIA's CCCL 3.1 introduces three determinism levels for parallel reductions, letting developers trade performance for reproducibility in GPU computations.
NVIDIA cuda.compute Brings C++ GPU Performance to Python Developers
NVIDIA's new cuda.compute library topped GPU MODE benchmarks, delivering CUDA C++ performance through pure Python with 2-4x speedups over custom kernels.
NVIDIA Launches GPU-Accelerated Endpoints for Moonshot AI's Kimi K2.5 Model
NVIDIA now offers free GPU-accelerated API access to Kimi K2.5, a 1T parameter multimodal AI model with 384 experts and 262K context length for developers.
NVIDIA Megatron Core Gets Dynamic-CP Update With 48% Training Speedups
NVIDIA releases Dynamic Context Parallelism for Megatron Core, achieving up to 1.48x faster LLM training and 35% gains in industrial deployments.
FlashAttention-4 Hits 1,605 TFLOPS on NVIDIA Blackwell GPUs
NVIDIA's FlashAttention-4 achieves 71% hardware efficiency on Blackwell chips, delivering 3.6x speedup over FA2 for AI training workloads.
NVIDIA cuOpt Solver Cracks Four Previously Unsolved Optimization Problems
NVIDIA's GPU-accelerated cuOpt engine discovers new solutions for four MIPLIB benchmark problems, outperforming CPU solvers with 22% lower objective gaps.
Enhancing CUDA Kernel Performance with Shared Memory Register Spilling
Discover how CUDA 13.0 optimizes kernel performance by using shared memory for register spilling, reducing latency and improving efficiency in GPU computations.
Decoding PTX: The Core of NVIDIA CUDA GPU Computing
Explore PTX, the assembly language for NVIDIA CUDA GPUs, its role in enabling forward compatibility, and its significance in the GPU computing landscape.