Science Daily

Breakthrough optical processor lets AI compute at the speed of light

Researchers at Tsinghua University developed the Optical Feature Extraction Engine (OFE2), an optical engine that processes data at 12.5 GHz using light rather than electricity. Its integrated ...
IEEE

Optimizing Sparse Matrix-Vector Multiplication on GPUs using the Mathematics of Arrays

Abstract: We present a Mathematics of Arrays (MoA) and ψ-calculus derivation of the memory-optimal operational normal form for ELLPACK sparse matrix-vector multiplication (SpMV) on GPUs. Under the ...
GitHub

Basic Matrix/Vector Operations Using SIMD

Dot Product calculation between two vectors calculated using v1.x * v2.x + v1.y * v2.y + v1.z * v2.z + v1.2 + v2.2 To do this with Simd Dot product loads vectors into float arrays for simd use, then ...
IEEE

SSpMV: A Sparsity-aware SpMV Framework Empowered by Multimodal Machine Learning

Abstract: Sparse Matrix-Vector Multiplication (SpMV) is an essential sparse operation in scientific computing and artificial intelligence. Efficiently adapting SpMV algorithms to diverse matrices and ...
GitHub

Fused Quantized GEMV Kernels for LLM Inference

Quill is a CUDA library that accelerates the decode phase of large language model inference by fusing weight dequantization directly into matrix-vector multiplication. This eliminates intermediate ...