mirror of
https://igit.ific.uv.es/alramos/latticegpu.jl.git
synced 2025-05-14 19:23:42 +02:00
Twisted mass
This commit is contained in:
Fernando P. Panadero
parent
5bd1aadfd0
commit
6a90d74024
4 changed files with 47 additions and 45 deletions
|
|
@ -28,21 +28,22 @@ The parameters are:
|
|||
- `m0::T` : Mass of the fermion
|
||||
- `csw::T` : Improvement coefficient for the Csw term
|
||||
- `th{Ntuple{4,Complex{T}}}` : Phase for the fermions included in the boundary conditions, reabsorbed in the Dirac operator.
|
||||
- `tm` : Twisted mass paramete
|
||||
- `ct` : Boundary improvement term, only used for Schrödinger Funtional boundary conditions.
|
||||
"""
|
||||
struct DiracParam{T,R}
|
||||
m0::T
|
||||
csw::T
|
||||
th::NTuple{4,Complex{T}}
|
||||
tm::T
|
||||
ct::T
|
||||
|
||||
|
||||
function DiracParam{T}(::Type{R},m0,csw,th,ct) where {T,R}
|
||||
return new{T,R}(m0,csw,th,ct)
|
||||
function DiracParam{T}(::Type{R},m0,csw,th,tm,ct) where {T,R}
|
||||
return new{T,R}(m0,csw,th,tm,ct)
|
||||
end
|
||||
|
||||
function DiracParam{T}(m0,csw,th,ct) where {T}
|
||||
return new{T,SU3fund}(m0,csw,th,ct)
|
||||
function DiracParam{T}(m0,csw,th,tm,ct) where {T}
|
||||
return new{T,SU3fund}(m0,csw,th,tm,ct)
|
||||
end
|
||||
end
|
||||
function Base.show(io::IO, dpar::DiracParam{T,R}) where {T,R}
|
||||
|
|
@ -50,8 +51,9 @@ function Base.show(io::IO, dpar::DiracParam{T,R}) where {T,R}
|
|||
println(io, "Wilson fermions in the: ", R, " representation")
|
||||
println(io, " - Bare mass: ", dpar.m0," // Kappa = ",0.5/(dpar.m0+4))
|
||||
println(io, " - Csw : ", dpar.csw)
|
||||
println(io, " - c_t: ", dpar.ct)
|
||||
println(io, " - Theta: ", dpar.th)
|
||||
println(io, " - Twisted mass: ", dpar.tm)
|
||||
println(io, " - c_t: ", dpar.ct)
|
||||
return nothing
|
||||
end
|
||||
|
||||
|
|
@ -181,13 +183,13 @@ function Dw!(so, U, si, dpar::DiracParam, dws::DiracWorkspace, lp::SpaceParm{4,6
|
|||
if abs(dpar.csw) > 1.0E-10
|
||||
@timeit "Dw" begin
|
||||
CUDA.@sync begin
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_Dwimpr!(so, U, si, dws.csw, dpar.m0, dpar.th, dpar.csw, lp)
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_Dwimpr!(so, U, si, dws.csw, dpar.m0, dpar.tm, dpar.th, dpar.csw, lp)
|
||||
end
|
||||
end
|
||||
else
|
||||
@timeit "Dw" begin
|
||||
CUDA.@sync begin
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_Dw!(so, U, si, dpar.m0, dpar.th, lp)
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_Dw!(so, U, si, dpar.m0, dpar.tm, dpar.th, lp)
|
||||
end
|
||||
end
|
||||
end
|
||||
|
|
@ -195,7 +197,7 @@ function Dw!(so, U, si, dpar::DiracParam, dws::DiracWorkspace, lp::SpaceParm{4,6
|
|||
return nothing
|
||||
end
|
||||
|
||||
function krnl_Dwimpr!(so, U, si, Fcsw, m0, th, csw, lp::SpaceParm{4,6,B,D}) where {B,D}
|
||||
function krnl_Dwimpr!(so, U, si, Fcsw, m0, tm, th, csw, lp::SpaceParm{4,6,B,D}) where {B,D}
|
||||
|
||||
b = Int64(CUDA.threadIdx().x); r = Int64(CUDA.blockIdx().x)
|
||||
|
||||
|
|
@ -210,7 +212,7 @@ function krnl_Dwimpr!(so, U, si, Fcsw, m0, th, csw, lp::SpaceParm{4,6,B,D}) wher
|
|||
|
||||
@inbounds begin
|
||||
|
||||
so[b,r] = (4+m0)*si[b,r] + 0.5*csw*im*( Fcsw[b,1,r]*dmul(Gamma{10},si[b,r]) + Fcsw[b,2,r]*dmul(Gamma{11},si[b,r]) + Fcsw[b,3,r]*dmul(Gamma{12},si[b,r])
|
||||
so[b,r] = (4+m0)*si[b,r]+ im*tm*dmul(Gamma{5},si[b,r]) + 0.5*csw*im*( Fcsw[b,1,r]*dmul(Gamma{10},si[b,r]) + Fcsw[b,2,r]*dmul(Gamma{11},si[b,r]) + Fcsw[b,3,r]*dmul(Gamma{12},si[b,r])
|
||||
+Fcsw[b,4,r]*dmul(Gamma{15},si[b,r]) + Fcsw[b,5,r]*dmul(Gamma{14},si[b,r]) + Fcsw[b,6,r]*dmul(Gamma{13},si[b,r]))
|
||||
|
||||
so[b,r] -= 0.5*(th[1]*gpmul(Pgamma{1,-1},U[b,1,r],si[bu1,ru1]) +conj(th[1])*gdagpmul(Pgamma{1,+1},U[bd1,1,rd1],si[bd1,rd1]) +
|
||||
|
|
@ -223,7 +225,7 @@ function krnl_Dwimpr!(so, U, si, Fcsw, m0, th, csw, lp::SpaceParm{4,6,B,D}) wher
|
|||
return nothing
|
||||
end
|
||||
|
||||
function krnl_Dw!(so, U, si, m0, th, lp::SpaceParm{4,6,B,D}) where {B,D}
|
||||
function krnl_Dw!(so, U, si, m0, tm, th, lp::SpaceParm{4,6,B,D}) where {B,D}
|
||||
|
||||
b = Int64(CUDA.threadIdx().x); r = Int64(CUDA.blockIdx().x)
|
||||
|
||||
|
|
@ -238,7 +240,7 @@ function krnl_Dw!(so, U, si, m0, th, lp::SpaceParm{4,6,B,D}) where {B,D}
|
|||
|
||||
@inbounds begin
|
||||
|
||||
so[b,r] = (4+m0)*si[b,r]
|
||||
so[b,r] = (4+m0)*si[b,r] + im*tm*dmul(Gamma{5},si[b,r])
|
||||
|
||||
so[b,r] -= 0.5*(th[1]*gpmul(Pgamma{1,-1},U[b,1,r],si[bu1,ru1]) +conj(th[1])*gdagpmul(Pgamma{1,+1},U[bd1,1,rd1],si[bd1,rd1]) +
|
||||
th[2]*gpmul(Pgamma{2,-1},U[b,2,r],si[bu2,ru2]) +conj(th[2])*gdagpmul(Pgamma{2,+1},U[bd2,2,rd2],si[bd2,rd2]) +
|
||||
|
|
@ -255,13 +257,13 @@ function Dw!(so, U, si, dpar::DiracParam, dws::DiracWorkspace, lp::Union{SpacePa
|
|||
if abs(dpar.csw) > 1.0E-10
|
||||
@timeit "Dw" begin
|
||||
CUDA.@sync begin
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_Dwimpr!(so, U, si, dws.csw, dpar.m0, dpar.th, dpar.csw, dpar.ct, lp)
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_Dwimpr!(so, U, si, dws.csw, dpar.m0, dpar.tm, dpar.th, dpar.csw, dpar.ct, lp)
|
||||
end
|
||||
end
|
||||
else
|
||||
@timeit "Dw" begin
|
||||
CUDA.@sync begin
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_Dw!(so, U, si, dpar.m0, dpar.th, dpar.ct, lp)
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_Dw!(so, U, si, dpar.m0, dpar.tm, dpar.th, dpar.ct, lp)
|
||||
end
|
||||
end
|
||||
end
|
||||
|
|
@ -269,7 +271,7 @@ function Dw!(so, U, si, dpar::DiracParam, dws::DiracWorkspace, lp::Union{SpacePa
|
|||
return nothing
|
||||
end
|
||||
|
||||
function krnl_Dwimpr!(so, U, si, Fcsw, m0, th, csw, ct, lp::Union{SpaceParm{4,6,BC_SF_ORBI,D},SpaceParm{4,6,BC_SF_AFWB,D}}) where {D}
|
||||
function krnl_Dwimpr!(so, U, si, Fcsw, m0, tm, th, csw, ct, lp::Union{SpaceParm{4,6,BC_SF_ORBI,D},SpaceParm{4,6,BC_SF_AFWB,D}}) where {D}
|
||||
|
||||
# The field si is assumed to be zero at t = 0
|
||||
|
||||
|
|
@ -288,7 +290,7 @@ function krnl_Dwimpr!(so, U, si, Fcsw, m0, th, csw, ct, lp::Union{SpaceParm{4,6,
|
|||
|
||||
@inbounds begin
|
||||
|
||||
so[b,r] = (4+m0)*si[b,r] + 0.5*csw*im*( Fcsw[b,1,r]*dmul(Gamma{10},si[b,r]) + Fcsw[b,2,r]*dmul(Gamma{11},si[b,r]) + Fcsw[b,3,r]*dmul(Gamma{12},si[b,r])
|
||||
so[b,r] = (4+m0)*si[b,r] + im*tm*dmul(Gamma{5},si[b,r]) + 0.5*csw*im*( Fcsw[b,1,r]*dmul(Gamma{10},si[b,r]) + Fcsw[b,2,r]*dmul(Gamma{11},si[b,r]) + Fcsw[b,3,r]*dmul(Gamma{12},si[b,r])
|
||||
+Fcsw[b,4,r]*dmul(Gamma{15},si[b,r]) + Fcsw[b,5,r]*dmul(Gamma{14},si[b,r]) + Fcsw[b,6,r]*dmul(Gamma{13},si[b,r]))
|
||||
|
||||
|
||||
|
|
@ -306,7 +308,7 @@ function krnl_Dwimpr!(so, U, si, Fcsw, m0, th, csw, ct, lp::Union{SpaceParm{4,6,
|
|||
return nothing
|
||||
end
|
||||
|
||||
function krnl_Dw!(so, U, si, m0, th, ct, lp::Union{SpaceParm{4,6,BC_SF_ORBI,D},SpaceParm{4,6,BC_SF_AFWB,D}}) where {D}
|
||||
function krnl_Dw!(so, U, si, m0, tm, th, ct, lp::Union{SpaceParm{4,6,BC_SF_ORBI,D},SpaceParm{4,6,BC_SF_AFWB,D}}) where {D}
|
||||
|
||||
# The field si is assumed to be zero at t = 0
|
||||
|
||||
|
|
@ -325,7 +327,7 @@ function krnl_Dw!(so, U, si, m0, th, ct, lp::Union{SpaceParm{4,6,BC_SF_ORBI,D},S
|
|||
|
||||
@inbounds begin
|
||||
|
||||
so[b,r] = (4+m0)*si[b,r]
|
||||
so[b,r] = (4+m0)*si[b,r] + im*tm*dmul(Gamma{5},si[b,r])
|
||||
so[b,r] -= 0.5*(th[1]*gpmul(Pgamma{1,-1},U[b,1,r],si[bu1,ru1]) +conj(th[1])*gdagpmul(Pgamma{1,+1},U[bd1,1,rd1],si[bd1,rd1]) +
|
||||
th[2]*gpmul(Pgamma{2,-1},U[b,2,r],si[bu2,ru2]) +conj(th[2])*gdagpmul(Pgamma{2,+1},U[bd2,2,rd2],si[bd2,rd2]) +
|
||||
th[3]*gpmul(Pgamma{3,-1},U[b,3,r],si[bu3,ru3]) +conj(th[3])*gdagpmul(Pgamma{3,+1},U[bd3,3,rd3],si[bd3,rd3]) +
|
||||
|
|
@ -351,13 +353,13 @@ function g5Dw!(so, U, si, dpar::DiracParam, dws::DiracWorkspace, lp::SpaceParm{4
|
|||
if abs(dpar.csw) > 1.0E-10
|
||||
@timeit "g5Dw" begin
|
||||
CUDA.@sync begin
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_g5Dwimpr!(so, U, si, dws.csw, dpar.m0, dpar.th, dpar.csw, lp)
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_g5Dwimpr!(so, U, si, dws.csw, dpar.m0, dpar.tm, dpar.th, dpar.csw, lp)
|
||||
end
|
||||
end
|
||||
else
|
||||
@timeit "g5Dw" begin
|
||||
CUDA.@sync begin
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_g5Dw!(so, U, si, dpar.m0, dpar.th, lp)
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_g5Dw!(so, U, si, dpar.m0, dpar.tm, dpar.th, lp)
|
||||
end
|
||||
end
|
||||
end
|
||||
|
|
@ -365,7 +367,7 @@ function g5Dw!(so, U, si, dpar::DiracParam, dws::DiracWorkspace, lp::SpaceParm{4
|
|||
return nothing
|
||||
end
|
||||
|
||||
function krnl_g5Dwimpr!(so, U, si, Fcsw, m0, th, csw, lp::SpaceParm{4,6,B,D}) where {B,D}
|
||||
function krnl_g5Dwimpr!(so, U, si, Fcsw, m0, tm, th, csw, lp::SpaceParm{4,6,B,D}) where {B,D}
|
||||
|
||||
b = Int64(CUDA.threadIdx().x); r = Int64(CUDA.blockIdx().x)
|
||||
|
||||
|
|
@ -388,13 +390,13 @@ function krnl_g5Dwimpr!(so, U, si, Fcsw, m0, th, csw, lp::SpaceParm{4,6,B,D}) wh
|
|||
th[3]*gpmul(Pgamma{3,-1},U[b,3,r],si[bu3,ru3]) +conj(th[3])*gdagpmul(Pgamma{3,+1},U[bd3,3,rd3],si[bd3,rd3]) +
|
||||
th[4]*gpmul(Pgamma{4,-1},U[b,4,r],si[bu4,ru4]) +conj(th[4])*gdagpmul(Pgamma{4,+1},U[bd4,4,rd4],si[bd4,rd4]) )
|
||||
|
||||
so[b,r] = dmul(Gamma{5}, so[b,r])
|
||||
so[b,r] = dmul(Gamma{5}, so[b,r])+ im*tm*si[b,r]
|
||||
end
|
||||
|
||||
return nothing
|
||||
end
|
||||
|
||||
function krnl_g5Dw!(so, U, si, m0, th, lp::SpaceParm{4,6,B,D}) where {B,D}
|
||||
function krnl_g5Dw!(so, U, si, m0, tm, th, lp::SpaceParm{4,6,B,D}) where {B,D}
|
||||
|
||||
b = Int64(CUDA.threadIdx().x); r = Int64(CUDA.blockIdx().x)
|
||||
|
||||
|
|
@ -416,7 +418,7 @@ function krnl_g5Dw!(so, U, si, m0, th, lp::SpaceParm{4,6,B,D}) where {B,D}
|
|||
th[3]*gpmul(Pgamma{3,-1},U[b,3,r],si[bu3,ru3]) +conj(th[3])*gdagpmul(Pgamma{3,+1},U[bd3,3,rd3],si[bd3,rd3]) +
|
||||
th[4]*gpmul(Pgamma{4,-1},U[b,4,r],si[bu4,ru4]) +conj(th[4])*gdagpmul(Pgamma{4,+1},U[bd4,4,rd4],si[bd4,rd4]) )
|
||||
|
||||
so[b,r] = dmul(Gamma{5}, so[b,r])
|
||||
so[b,r] = dmul(Gamma{5}, so[b,r]) + im*tm*si[b,r]
|
||||
end
|
||||
|
||||
return nothing
|
||||
|
|
@ -427,13 +429,13 @@ function g5Dw!(so, U, si, dpar::DiracParam, dws::DiracWorkspace, lp::Union{Space
|
|||
if abs(dpar.csw) > 1.0E-10
|
||||
@timeit "g5Dw" begin
|
||||
CUDA.@sync begin
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_g5Dwimpr!(so, U, si, dws.csw, dpar.m0, dpar.th, dpar.csw, dpar.ct, lp)
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_g5Dwimpr!(so, U, si, dws.csw, dpar.m0, dpar.tm, dpar.th, dpar.csw, dpar.ct, lp)
|
||||
end
|
||||
end
|
||||
else
|
||||
@timeit "g5Dw" begin
|
||||
CUDA.@sync begin
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_g5Dw!(so, U, si, dpar.m0, dpar.th, dpar.ct, lp)
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_g5Dw!(so, U, si, dpar.m0, dpar.tm, dpar.th, dpar.ct, lp)
|
||||
end
|
||||
end
|
||||
end
|
||||
|
|
@ -441,7 +443,7 @@ function g5Dw!(so, U, si, dpar::DiracParam, dws::DiracWorkspace, lp::Union{Space
|
|||
return nothing
|
||||
end
|
||||
|
||||
function krnl_g5Dwimpr!(so, U, si, Fcsw, m0, th, csw, ct, lp::Union{SpaceParm{4,6,BC_SF_ORBI,D},SpaceParm{4,6,BC_SF_AFWB,D}}) where {D}
|
||||
function krnl_g5Dwimpr!(so, U, si, Fcsw, m0, tm, th, csw, ct, lp::Union{SpaceParm{4,6,BC_SF_ORBI,D},SpaceParm{4,6,BC_SF_AFWB,D}}) where {D}
|
||||
|
||||
# The field si is assumed to be zero at t = 0
|
||||
|
||||
|
|
@ -475,12 +477,12 @@ function krnl_g5Dwimpr!(so, U, si, Fcsw, m0, th, csw, ct, lp::Union{SpaceParm{4,
|
|||
end
|
||||
end
|
||||
|
||||
so[b,r] = dmul(Gamma{5}, so[b,r])
|
||||
so[b,r] = dmul(Gamma{5}, so[b,r])+ im*tm*si[b,r]
|
||||
|
||||
return nothing
|
||||
end
|
||||
|
||||
function krnl_g5Dw!(so, U, si, m0, th, ct, lp::Union{SpaceParm{4,6,BC_SF_ORBI,D},SpaceParm{4,6,BC_SF_AFWB,D}}) where {D}
|
||||
function krnl_g5Dw!(so, U, si, m0, tm, th, ct, lp::Union{SpaceParm{4,6,BC_SF_ORBI,D},SpaceParm{4,6,BC_SF_AFWB,D}}) where {D}
|
||||
|
||||
# The field si is assumed to be zero at t = 0
|
||||
|
||||
|
|
@ -511,7 +513,7 @@ function krnl_g5Dw!(so, U, si, m0, th, ct, lp::Union{SpaceParm{4,6,BC_SF_ORBI,D}
|
|||
end
|
||||
end
|
||||
|
||||
so[b,r] = dmul(Gamma{5}, so[b,r])
|
||||
so[b,r] = dmul(Gamma{5}, so[b,r]) + im*tm*si[b,r]
|
||||
|
||||
return nothing
|
||||
end
|
||||
|
|
@ -519,7 +521,7 @@ end
|
|||
"""
|
||||
function DwdagDw!(so, U, si, dpar::DiracParam, dws::DiracWorkspace, lp::SpaceParm{4,6,B,D})
|
||||
|
||||
Applies the operator \`\` \\gamma_5 D_w \`\` twice to `si` and stores the result in `so`. This is equivalent to appling the operator \`\` \`\`
|
||||
Applies the operator \`\` \\gamma_5 D_w \`\` twice to `si` and stores the result in `so`. This is equivalent to appling the operator \`\` D_w^\\dagger D_w \`\`
|
||||
The Dirac operator is the same as in the functions `Dw!` and `g5Dw!`
|
||||
"""
|
||||
function DwdagDw!(so, U, si, dpar::DiracParam, dws::DiracWorkspace, lp::Union{SpaceParm{4,6,BC_SF_ORBI,D},SpaceParm{4,6,BC_SF_AFWB,D}}) where {D}
|
||||
|
|
@ -529,13 +531,13 @@ function DwdagDw!(so, U, si, dpar::DiracParam, dws::DiracWorkspace, lp::Union{Sp
|
|||
|
||||
@timeit "g5Dw" begin
|
||||
CUDA.@sync begin
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_g5Dwimpr!(dws.st, U, si, dws.csw, dpar.m0, dpar.th, dpar.csw, dpar.ct, lp)
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_g5Dwimpr!(dws.st, U, si, dws.csw, dpar.m0, dpar.tm, dpar.th, dpar.csw, dpar.ct, lp)
|
||||
end
|
||||
end
|
||||
|
||||
@timeit "g5Dw" begin
|
||||
CUDA.@sync begin
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_g5Dwimpr!(so, U, dws.st, dws.csw, dpar.m0, dpar.th, dpar.csw, dpar.ct, lp)
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_g5Dwimpr!(so, U, dws.st, dws.csw, dpar.m0, dpar.tm, dpar.th, dpar.csw, dpar.ct, lp)
|
||||
end
|
||||
end
|
||||
end
|
||||
|
|
@ -544,13 +546,13 @@ function DwdagDw!(so, U, si, dpar::DiracParam, dws::DiracWorkspace, lp::Union{Sp
|
|||
|
||||
@timeit "g5Dw" begin
|
||||
CUDA.@sync begin
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_g5Dw!(dws.st, U, si, dpar.m0, dpar.th, dpar.ct, lp)
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_g5Dw!(dws.st, U, si, dpar.m0, dpar.tm, dpar.th, dpar.ct, lp)
|
||||
end
|
||||
end
|
||||
|
||||
@timeit "g5Dw" begin
|
||||
CUDA.@sync begin
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_g5Dw!(so, U, dws.st, dpar.m0, dpar.th, dpar.ct, lp)
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_g5Dw!(so, U, dws.st, dpar.m0, dpar.tm, dpar.th, dpar.ct, lp)
|
||||
end
|
||||
end
|
||||
end
|
||||
|
|
@ -566,13 +568,13 @@ function DwdagDw!(so, U, si, dpar::DiracParam, dws::DiracWorkspace, lp::SpacePar
|
|||
|
||||
@timeit "g5Dw" begin
|
||||
CUDA.@sync begin
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_g5Dwimpr!(dws.st, U, si, dws.csw, dpar.m0, dpar.th, dpar.csw, lp)
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_g5Dwimpr!(dws.st, U, si, dws.csw, dpar.m0, dpar.tm, dpar.th, dpar.csw, lp)
|
||||
end
|
||||
end
|
||||
|
||||
@timeit "g5Dw" begin
|
||||
CUDA.@sync begin
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_g5Dwimpr!(so, U, dws.st, dws.csw, dpar.m0, dpar.th, dpar.csw, lp)
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_g5Dwimpr!(so, U, dws.st, dws.csw, dpar.m0, dpar.tm, dpar.th, dpar.csw, lp)
|
||||
end
|
||||
end
|
||||
end
|
||||
|
|
@ -581,13 +583,13 @@ function DwdagDw!(so, U, si, dpar::DiracParam, dws::DiracWorkspace, lp::SpacePar
|
|||
|
||||
@timeit "g5Dw" begin
|
||||
CUDA.@sync begin
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_g5Dw!(dws.st, U, si, dpar.m0, dpar.th, lp)
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_g5Dw!(dws.st, U, si, dpar.m0, dpar.tm, dpar.th, lp)
|
||||
end
|
||||
end
|
||||
|
||||
@timeit "g5Dw" begin
|
||||
CUDA.@sync begin
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_g5Dw!(so, U, dws.st, dpar.m0, dpar.th, lp)
|
||||
CUDA.@cuda threads=lp.bsz blocks=lp.rsz krnl_g5Dw!(so, U, dws.st, dpar.m0, dpar.tm, dpar.th, lp)
|
||||
end
|
||||
end
|
||||
end
|
||||
|
|
|
|||
|
|
@ -13,7 +13,7 @@ function fP_test(;theta = (0.5,0.7,1.0,0.0), m = 1.3, size = (8,8,8,16),prec = 1
|
|||
lp = SpaceParm{4}(size,(4,4,4,4),1,(0,0,0,0,0,0));
|
||||
exptheta = exp.(im.*theta./lp.iL);
|
||||
|
||||
dpar = DiracParam{Float64}(SU3fund,m,0.0,exptheta,1.0);
|
||||
dpar = DiracParam{Float64}(SU3fund,m,0.0,exptheta,0.0,1.0);
|
||||
dws = DiracWorkspace(SU3fund,Float64,lp);
|
||||
|
||||
U = fill!(vector_field(SU3{Float64},lp),one(SU3{Float64}));
|
||||
|
|
@ -65,7 +65,7 @@ function fA_test(;theta = (0.5,0.7,1.0,0.0), m = 1.3, size = (8,8,8,16),prec = 1
|
|||
lp = SpaceParm{4}(size,(4,4,4,4),1,(0,0,0,0,0,0));
|
||||
exptheta = exp.(im.*theta./lp.iL);
|
||||
|
||||
dpar = DiracParam{Float64}(SU3fund,m,0.0,exptheta,1.0);
|
||||
dpar = DiracParam{Float64}(SU3fund,m,0.0,exptheta,0.0,1.0);
|
||||
dws = DiracWorkspace(SU3fund,Float64,lp);
|
||||
|
||||
U = fill!(vector_field(SU3{Float64},lp),one(SU3{Float64}));
|
||||
|
|
|
|||
|
|
@ -7,7 +7,7 @@ using LatticeGPU, CUDA, TimerOutputs
|
|||
function Dwpw_test(;p=0,s=1,c=1)
|
||||
lp = SpaceParm{4}((16,16,16,16), (4,4,4,4), 0, (0,0,0,0,0,0))
|
||||
gp = GaugeParm{Float64}(SU3{Float64}, 6.0, 1.0)
|
||||
dpar = DiracParam{Float64}(SU3fund,1.3,0.0,(1.0,1.0,1.0,1.0),0.0)
|
||||
dpar = DiracParam{Float64}(SU3fund,1.3,0.0,(1.0,1.0,1.0,1.0),0.0,0.0)
|
||||
dws = DiracWorkspace(SU3fund,Float64,lp);
|
||||
|
||||
p==0 ? p = Int.(round.(lp.iL.*rand(4),RoundUp)) : nothing
|
||||
|
|
|
|||
|
|
@ -9,7 +9,7 @@ using CUDA, LatticeGPU, TimerOutputs
|
|||
lp = SpaceParm{4}((16,16,16,16), (4,4,4,4), 0, (0,0,0,0,0,0))
|
||||
gp = GaugeParm{Float64}(SU3{Float64}, 6.0, 1.0)
|
||||
ymws = YMworkspace(SU3, Float64, lp)
|
||||
dpar = DiracParam{Float64}(SU3fund,2.3,0.0,(1.0,1.0,1.0,1.0),0.0)
|
||||
dpar = DiracParam{Float64}(SU3fund,2.3,0.0,(1.0,1.0,1.0,1.0),0.0,0.0)
|
||||
dws = DiracWorkspace(SU3fund,Float64,lp);
|
||||
|
||||
randomize!(ymws.mom, lp, ymws)
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue