MixPert: Optimizing Mixed-Precision Floating-Point Emulation on GPU Integer Tensor Cores.
Zejia LinAoyuan SunXianwei ZhangYutong LuPublished in: LCTES (2024)
Keyphrases
- floating point
- graphics processing units
- fixed point
- square root
- memory bandwidth
- higher order
- parallel architectures
- real time
- sparse matrices
- graphics hardware
- fast fourier transform
- interval arithmetic
- gpu implementation
- instruction set
- parallel computation
- processing units
- commodity hardware
- graphical models
- bayesian networks