Closed kibaekkim closed 1 year ago
I think the CPU and GPU codes are not using the same norm.
https://github.com/exanauts/ExaTron.jl/blob/3ebc6bc7f73c1e4af5e93d3540785975addb0742/src/CPU/dgpnorm.jl#L1-L20 https://github.com/exanauts/ExaTron.jl/blob/3ebc6bc7f73c1e4af5e93d3540785975addb0742/src/CUDA/dgpnorm.jl#L1-L32 https://github.com/exanauts/ExaTron.jl/blob/3ebc6bc7f73c1e4af5e93d3540785975addb0742/src/KA/dgpnorm.jl#L1-L39 https://github.com/exanauts/ExaTron.jl/blob/3ebc6bc7f73c1e4af5e93d3540785975addb0742/src/driver.jl#L42-L67
I am not sure if this is intended. We need @youngdae to confirm this.
Never mind.. they are same.. but computing in different ways.
I think the CPU and GPU codes are not using the same norm.
https://github.com/exanauts/ExaTron.jl/blob/3ebc6bc7f73c1e4af5e93d3540785975addb0742/src/CPU/dgpnorm.jl#L1-L20 https://github.com/exanauts/ExaTron.jl/blob/3ebc6bc7f73c1e4af5e93d3540785975addb0742/src/CUDA/dgpnorm.jl#L1-L32 https://github.com/exanauts/ExaTron.jl/blob/3ebc6bc7f73c1e4af5e93d3540785975addb0742/src/KA/dgpnorm.jl#L1-L39 https://github.com/exanauts/ExaTron.jl/blob/3ebc6bc7f73c1e4af5e93d3540785975addb0742/src/driver.jl#L42-L67
I am not sure if this is intended. We need @youngdae to confirm this.