459-GemsFDTD Report

**459-GemsFDTD**
Exec Time (%)	Codelet Name	Error (%)
100.3	leapfrog___leapfrog_mod_MOD_leapfrog_200	8.0
2.9	UPML_fppized___upml_mod_MOD_upmlupdatee...	5.0
3.0	excite___excite_mod_MOD_excitation_arra...	0.9
2.8	UPML_fppized___upml_mod_MOD_upmlupdateh...	8.3
6.4	UPML_fppized___upml_mod_MOD_upmlupdatee...	12.0
6.8	UPML_fppized___upml_mod_MOD_upmlupdatee...	8.6
3.2	UPML_fppized___upml_mod_MOD_upmlupdateh...	6.6
2.8	UPML_fppized___upml_mod_MOD_upmlupdateh...	6.8
7.1	UPML_fppized___upml_mod_MOD_upmlupdateh...	14.8
2.7	UPML_fppized___upml_mod_MOD_upmlupdatee...	3.6
2.0	huygens_fppized___huygens_mod_MOD_huyge...	100.0
3.0	excite___excite_mod_MOD_excitation_arra...	0.9
2.0	huygens_fppized___huygens_mod_MOD_huyge...	100.0
3.0	excite___excite_mod_MOD_excitation_arra...	0.9
3.2	UPML_fppized___upml_mod_MOD_upmlupdateh...	7.4
3.2	UPML_fppized___upml_mod_MOD_upmlupdatee...	6.7
10.4	update_fppized___update_mod_MOD_updateh...	30.4
6.2	UPML_fppized___upml_mod_MOD_upmlupdateh...	26.7
2.9	UPML_fppized___upml_mod_MOD_upmlupdatee...	3.5
11.6	update_fppized___update_mod_MOD_updatee...	19.5
18.9	NFT_fppized___nft_mod_MOD_nft_store_1065	14.4

Exec Time : 1.458696e+12 cycles

Matching : 100.3%

                CODE NOT FOUND -> THIS CODELET NOT IN regions.csv?

invivoUPML_fppized___upml_mod_MOD_upmlupdateh_904

Real time: 9.160525e+10 -- Predicted time: 1.250746e+11

Callcount: 1000

Invocation	Cluster	Part	Invitro (cycles)	Invivo (cycles)	Error (%)
410	### 1	202.000644902754	1.252626e+08	1.052376e+08	15.9
361	### 2	247.000027550059	1.273118e+08	1.052629e+08	17.3
132	### 3	230.000461417679	1.247310e+08	1.052929e+08	15.5
14	### 4	112.006712673046	1.207503e+08	1.053303e+08	12.7
158	### 5	193.988643819768	1.248593e+08	1.052093e+08	15.7
834	### 6	15.0108026856178	1.259864e+08	1.055030e+08	16.2

                !+
! NAME
!       leapfrog_mod - Leap-frog module
!
! DESCRIPTION
!       Contains the Subroutine leapfrog       
!
! PUBLIC
!       SUBROUTINE leapfrog
!
! HISTORY
!       Version       Date                 Name
!       Comments
!       -------------------------------------------
!	$Log: leapfrog.f90,v $
!	Revision 1.4  2003/09/26 22:05:21  ulfa
!	Removed a missplaced comment.
!	
!	Revision 1.3  2003/09/26 21:50:38  ulfa
!	Improved sdout info.
!	
!	Revision 1.2  2003/09/26 20:15:18  ulfa
!	Added global computation of #bytes allocated.
!	
!	Revision 1.1  2003/09/23 14:06:51  ulfa
!	Initial version.
!-
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

MODULE leapfrog_mod

IMPLICIT NONE

PUBLIC leapfrog

PRIVATE 

CONTAINS

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!+
! NAME
!       leapfrog - Leap-frog scheme
!
! DESCRIPTION
!       1) calls init routines for the modules used
!       2) allocate the field variables
!       3) performs explicit timestepping for nts time steps
!       4) writes desired output
!       5) calls end routines for the modules used
!
! METHOD
!       During timestepping these SUBROUTINES are called
!          timer                          if (ts==2)
!          updateH_homo                   (always)
!          UPMLupdateH                    if (OBC_Type==OBC_UPML)
!          HuygensH                       if (HuyPulsetype>0)
!          ------------- Formally, a half time step is taken here -------------
!          updateE_homo                   (always)
!          UPMLupdateE                    if (OBC_Type==OBC_UPML)
!          HuygensE                       if (HuyPulsetype>0)
!          PECApply                       if PEC_found
!          NFT_Store                      if (NF_Type==2)
!          progress                       (always)
!
! SYNOPSIS
!       CALL leapfrog(nx,ny,nz)
!         integer, intent(in) :: nx, ny, nz
!
! ERRORS
!       Uses the errorcheck module
!
! HISTORY
!       Written by Ulf Andersson
!-
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

SUBROUTINE leapfrog(nx,ny,nz)
USE errorcheck_mod, ONLY : check_allocate, check_deallocate, AllocStatus
USE errorcheck_mod, ONLY : fatal, warning
USE parameter_mod, ONLY  : rfp, sfp, dfp, bytes_per_float
USE huygens_mod, ONLY    : HuyPulsetype, HuygensH, HuygensE, Huygens_end
USE update_mod, ONLY     : Update_init, UpdateH_homo, UpdateE_homo
USE UPML_mod, ONLY       : UPMLinit, UPMLend, UPMLupdateH, UPMLupdateE
USE PEC_mod, ONLY        : PEC_found, PECinit, PECend, PECapply
USE NFT_mod, ONLY        : NFT_Init, NFT_Store, NFT_Print
USE calcflops_mod, ONLY  : calcflops
USE globalvar_mod, ONLY  : dx, dy, dz, nts, dt,                               &
                           xstop, xstart, ystop, ystart, zstop, zstart,       &
                           OBC_Type, OBC_UPML, NF_Type, bytes_allocated
USE progress_mod, ONLY   : progress, progress_awtic, progress_awtwb

!------------------------------------------------------------------------------
!                     A r g u m e n t s             
!------------------------------------------------------------------------------

integer, intent(in) :: nx, ny, nz

!------------------------------------------------------------------------------
!                     L o c a l  v a r i a b l e st
!------------------------------------------------------------------------------

integer        :: ts  ! time step
integer        :: allocstat
integer        :: nf_Err ! Near-to-far-field error variable

real(kind=sfp) :: t_start, t_final

! Field variables
real(kind=rfp), dimension(:,:,:), allocatable :: Ex, Ey, Ez, Hx, Hy, Hz
integer :: bytes, bytes2

!------------------------------------------------------------------------------
!                     E x e c u t a b l e   c o d e
!------------------------------------------------------------------------------
!
! Initialize leap-frog update 
!
CALL Update_init()
!
! Initialize PEC
!
if (PEC_found) then
  CALL PECinit(nx,ny,nz)
end if
!
! Initialize ABC
if (OBC_Type==OBC_UPML) then
  CALL UPMLinit(nx,ny,nz)
end if
!
! Initialize near-to-far-field transforms
!
if (NF_Type==2) then
  CALL NFT_Init(nx,ny,nz,nf_Err)
end if
if (nf_Err/=0) then
  write(*,*) 'FATAL ERROR in SUBROUTINE NFT_Init, nf_Err = ', nf_Err
  write(*,*) '(see comments in NFT_init for an explanation)'
  stop
end if
!
! Allocate memory for the electromagnetic fields. Do this last so that other
! initializations may use temporary memory. Note that we waste some memory
! when allocating the E fields. The following fields are never used:
!   Ex(   xstop+1    ,ystart:ystop+1,zstart:zstop+1)
!   Ey(xstart:xstop+1,   ystop+1    ,zstart:zstop+1)
!   Ez(xstart:xstop+1,ystart:ystop+1,   zstop+1    )
! By doing this we get the same size for all E fields.
!
write(*,'(A,I3)') ' Size of leading dimension for H-fields = ', xstop+1-xstart

allocate( Hx(xstart:xstop  ,ystart:ystop  ,zstart:zstop  ), STAT=allocstat )
call check_allocate(allocstat, 'Hx', fatal)

allocate( Hy(xstart:xstop  ,ystart:ystop  ,zstart:zstop  ), STAT=allocstat )
call check_allocate(allocstat, 'Hy', fatal)

allocate( Hz(xstart:xstop  ,ystart:ystop  ,zstart:zstop  ), STAT=allocstat )
call check_allocate(allocstat, 'Hz', fatal)

write(*,'(A,I3)') ' Size of leading dimension for E-fields = ', xstop+2-xstart

allocate( Ex(xstart:xstop+1,ystart:ystop+1,zstart:zstop+1), STAT=allocstat )
call check_allocate(allocstat, 'Ex', fatal)

allocate( Ey(xstart:xstop+1,ystart:ystop+1,zstart:zstop+1), STAT=allocstat )
call check_allocate(allocstat, 'Ey', fatal)

allocate( Ez(xstart:xstop+1,ystart:ystop+1,zstart:zstop+1), STAT=allocstat )
call check_allocate(allocstat, 'Ez', fatal)

bytes = (xstop-xstart+1)*(ystop-ystart+1)*(zstop-zstart+1)*3 + & ! H-fields   &
        (xstop-xstart+2)*(ystop-ystart+2)*(zstop-zstart+2)*3     ! E-fields 
bytes = bytes*bytes_per_float
bytes_allocated = bytes_allocated + bytes
write(*,*) 'Fields allocated in leapfrog_mod, bytes used = ', bytes
bytes2 = (nx*ny*nz*3 + (nx+1)*(ny+1)*(nz+1)*3)*bytes_per_float
write(*,*) '(', bytes-bytes2, ' bytes of this is due to UPML)'
write(*,*)
write(*,'(A,I5)') ' Total amount of Mbytes allocated = ',                     &
           bytes_allocated/(1024*1024)
write(*,*)
!
! Initialize the fields 
!
Ex = 0.0_rfp ; Ey = 0.0_rfp ; Ez = 0.0_rfp
Hx = 0.0_rfp ; Hy = 0.0_rfp ; Hz = 0.0_rfp

! Only perform initialization. This is useful if a flop counter tool is used.
if (progress_awtwb) then                                             ! (ulfa)
  write(*,*) 'TERMINATING, because PROGRESS/Awtwb was set'
  stop
end if

!-------------- TIMESTEPPING -----------------

do ts = 1,nts

  if (ts==1) then
    write(*,*) '============================================================'
    write(*,*) 'Entering timestepping loop'
    write(*,*) 'Total number of iterations are ', nts
  end if

! Start timer. The first time step may differ from the others in performance,
! since it does not have any preceding time step. We do not want to pollute
! the timing by including it.
  if (ts==2) then                                    
    write(*,*) 'Starting timer in the beginning of the second time step'
!   !SPEC: no, don't; SPEC prefers to do its own timing jh/12/feb/04
!   !CALL timer(t_start,0.0_sfp)
  end if

  CALL updateH_homo(nx,ny,nz,Hx,Hy,Hz,Ex,Ey,Ez)

  if (OBC_Type==OBC_UPML) then
    CALL UPMLupdateH(nx,ny,nz,Hx,Hy,Hz,Ex,Ey,Ez)
  end if
  
  if (HuyPulsetype>0) then
    CALL HuygensH(ts,Hx,Hy,Hz)
  end if

!
!-------------- Formally, a half time step is taken here -----------------
!

  CALL updateE_homo(nx,ny,nz,Hx,Hy,Hz,Ex,Ey,Ez)

  if (OBC_Type==OBC_UPML) then
    CALL UPMLupdateE(nx,ny,nz,Hx,Hy,Hz,Ex,Ey,Ez)
  end if

  if (HuyPulsetype>0) then
    CALL HuygensE(ts,Ex,Ey,Ez)
  end if

  if (PEC_found) then
    CALL PECapply(Ex,Ey,Ez)
  end if

  if (NF_Type==2) then
    CALL NFT_Store(Ex,Ey,Ez,Hx,Hy,Hz,(ts-1)*dt,ts)
  end if

  CALL progress(ts)

end do ! Main loop ends here
! =+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
! Do not insert anything here! It would disrupt the timing.
if (nts>1) then              ! Timer not started if we only made one time step!
!  !SPEC  CALL timer(t_final,t_start)! Stop the timer 
end if

write(*,*) 'Ending timestepping loop' 

if (nts>1) then  ! Calculate the performance
!  !SPEC CALL calcflops(nx,ny,nz,t_final)
end if

!--------------------- Output and deallocation --------------------------------

! Excluding post-processing. This is useful if a flop counter tool is used.
if (progress_awtic) then                                           ! (ulfa)
  write(*,*) 'TERMINATING, because PROGRESS/Awtic was set'
  stop
end if

! Output and deallocation for near-to-far-field transforms.
if (NF_Type==2) then
  CALL NFT_Print()
end if

!--------------------- Deallocate memory --------------------------------------


deallocate( Ex, STAT=allocstat )
call check_deallocate(allocstat, 'Ex', warning)
deallocate( Ey, STAT=allocstat )
call check_deallocate(allocstat, 'Ey', warning)
deallocate( Ez, STAT=allocstat )
call check_deallocate(allocstat, 'Ez', warning)

deallocate( Hx, STAT=allocstat )
call check_deallocate(allocstat, 'Hx', warning)
deallocate( Hy, STAT=allocstat )
call check_deallocate(allocstat, 'Hy', warning)
deallocate( Hz, STAT=allocstat )
call check_deallocate(allocstat, 'Hz', warning)

call UPMLend

if (HuyPulsetype>0) then
  call Huygens_end
end if

if (PEC_found) then
  call PECend
end if

CALL AllocStatus

END SUBROUTINE leapfrog

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

END MODULE leapfrog_mod

invivoleapfrog___leapfrog_mod_MOD_leapfrog_200

Real time: 1.463251e+12 -- Predicted time: 1.592044e+12

Callcount: 1

Invocation	Cluster	Part	Invitro (cycles)	Invivo (cycles)	Error (%)
1	### 1	1	1.592044e+12	1.452532e+12	8.7

                CODE NOT FOUND -> THIS CODELET NOT IN regions.csv?

invivoNFT_fppized___nft_mod_MOD_nft_store_1065

Real time: 2.764544e+11 -- Predicted time: 3.231779e+11

Callcount: 1000

Invocation	Cluster	Part	Invitro (cycles)	Invivo (cycles)	Error (%)
97	### 1	162.991972419295	3.217508e+08	2.881620e+08	10.4
296	### 2	168.990581623588	3.282020e+08	2.885321e+08	12.0
737	### 3	273.996528336234	3.208363e+08	2.887422e+08	10.0
491	### 4	139.012788826891	3.216610e+08	2.892184e+08	10.0
863	### 5	208.000278255338	3.247551e+08	2.889432e+08	11.0
881	### 6	46.02419787597	3.210381e+08	2.897319e+08	9.7
635	### 7	1	3.243582e+08	3.394253e+08	4.4

                CODE NOT FOUND -> THIS CODELET NOT IN regions.csv?

invivoUPML_fppized___upml_mod_MOD_upmlupdatee_1202

Real time: 9.463068e+10 -- Predicted time: 1.076333e+11

Callcount: 1000

Invocation	Cluster	Part	Invitro (cycles)	Invivo (cycles)	Error (%)
326	### 1	999.056777057681	1.076275e+08	9.448112e+07	12.2
338	### 2	1	1.073322e+08	1.212337e+08	11.4

                CODE NOT FOUND -> THIS CODELET NOT IN regions.csv?

invivoUPML_fppized___upml_mod_MOD_upmlupdateh_846

Real time: 1.036339e+11 -- Predicted time: 1.217484e+11

Callcount: 1000

Invocation	Cluster	Part	Invitro (cycles)	Invivo (cycles)	Error (%)
230	### 1	140.000969588382	1.214804e+08	1.175468e+08	3.2
241	### 2	160.000002893582	1.215090e+08	1.175014e+08	3.2
847	### 3	65.0003306928029	1.213123e+08	1.175834e+08	3.0
838	### 4	153.999972829898	1.211096e+08	1.174821e+08	2.9
5	### 5	36.005501337285	1.215508e+08	1.176369e+08	3.2
440	### 6	161.001956643428	1.217502e+08	1.175196e+08	3.4
184	### 7	143.998741421788	1.215275e+08	1.174627e+08	3.3
411	### 8	136.991474246471	1.234452e+08	1.174343e+08	4.8
392	### 9	2.99972632612737	1.247042e+08	1.182283e+08	5.1

                CODE NOT FOUND -> THIS CODELET NOT IN regions.csv?

invivoUPML_fppized___upml_mod_MOD_upmlupdatee_1401

Real time: 4.358890e+10 -- Predicted time: 4.588921e+10

Callcount: 1000

Invocation	Cluster	Part	Invitro (cycles)	Invivo (cycles)	Error (%)
380	### 1	294.003678011386	4.599283e+07	4.453385e+07	3.1
43	### 2	252.989885556307	4.588507e+07	4.451377e+07	2.9
768	### 3	258.006282376291	4.581531e+07	4.455575e+07	2.7
331	### 4	14.0023652889093	4.580287e+07	4.475563e+07	2.2
371	### 5	179.023382873426	4.583155e+07	4.458853e+07	2.7
30	### 6	2.0058815366706	4.575633e+07	4.557312e+07	0.4

                CODE NOT FOUND -> THIS CODELET NOT IN regions.csv?

invivoUPML_fppized___upml_mod_MOD_upmlupdatee_1569

Real time: 3.944061e+10 -- Predicted time: 4.095546e+10

Callcount: 1000

Invocation	Cluster	Part	Invitro (cycles)	Invivo (cycles)	Error (%)
600	### 1	114.019486232794	4.159093e+07	3.992952e+07	3.9
2	### 2	10.0120490901413	4.155185e+07	4.011722e+07	3.4
46	### 3	342.003080722444	4.090478e+07	3.991272e+07	2.4
258	### 4	194.98957902845	4.085293e+07	3.988688e+07	2.3
194	### 5	338.998680419655	4.083145e+07	3.990056e+07	2.2

                CODE NOT FOUND -> THIS CODELET NOT IN regions.csv?

invivoUPML_fppized___upml_mod_MOD_upmlupdateh_1091

Real time: 4.203746e+10 -- Predicted time: 4.585702e+10

Callcount: 1000

Invocation	Cluster	Part	Invitro (cycles)	Invivo (cycles)	Error (%)
91	### 1	243.952544200492	4.551457e+07	4.423299e+07	2.8
56	### 2	249.995353433269	4.619196e+07	4.429679e+07	4.1
641	### 3	203.042172653893	4.572570e+07	4.441779e+07	2.8
564	### 4	273.992155388604	4.594441e+07	4.436217e+07	3.4
755	### 5	29.019421941131	4.594155e+07	4.462355e+07	2.8

                ERROR UNICODE -> FILE:excite.f90

invivoexcite___excite_mod_MOD_excitation_array_236

Real time: 4.393474e+10 -- Predicted time: 4.350182e+10

Callcount: 25001

Invocation	Cluster	Part	Invitro (cycles)	Invivo (cycles)	Error (%)
31	### 1	4973.60718699496	8.491600e+04	8.415200e+04	0.8
3435	### 2	5699.11418231807	2.732674e+06	2.723504e+06	0.3
3096	### 3	1635.76844993753	9.068170e+05	8.964800e+05	1.1
10686	### 4	9057.30578593585	1.579919e+06	1.628708e+06	2.9
7955	### 5	2464.28538034035	2.385186e+06	2.366512e+06	0.7
16806	### 6	2437.04844424197	1.970360e+06	2.010724e+06	2.0
18074	### 7	92.8427175682174	1.112489e+07	1.123864e+07	1.0

                CODE NOT FOUND -> THIS CODELET NOT IN regions.csv?

invivohuygens_fppized___huygens_mod_MOD_huygensh_693

Real time: 3.020855e+10 -- Predicted time: 0.000000e+00

Callcount: 1000

Invocation	Cluster	Part	Invitro (cycles)	Invivo (cycles)	Error (%)
80	### 1	152.158078830455	0.000000e+00	1.818526e+07	100.0
907	### 2	424.089458436582	0.000000e+00	2.766125e+07	100.0
697	### 3	439.569225696292	0.000000e+00	3.634367e+07	100.0

                CODE NOT FOUND -> THIS CODELET NOT IN regions.csv?

459-GemsFDTD

Exec Time : 1.458696e+12 cycles

Matching : 100.3%

__invivo__UPML_fppized___upml_mod_MOD_upmlupdateh_904

Real time: 9.160525e+10 -- Predicted time: 1.250746e+11

Callcount: 1000

__invivo__leapfrog___leapfrog_mod_MOD_leapfrog_200

Real time: 1.463251e+12 -- Predicted time: 1.592044e+12

Callcount: 1

__invivo__NFT_fppized___nft_mod_MOD_nft_store_1065

Real time: 2.764544e+11 -- Predicted time: 3.231779e+11

Callcount: 1000

__invivo__UPML_fppized___upml_mod_MOD_upmlupdatee_1202

Real time: 9.463068e+10 -- Predicted time: 1.076333e+11

Callcount: 1000

__invivo__UPML_fppized___upml_mod_MOD_upmlupdateh_846

Real time: 1.036339e+11 -- Predicted time: 1.217484e+11

Callcount: 1000

__invivo__UPML_fppized___upml_mod_MOD_upmlupdatee_1401

Real time: 4.358890e+10 -- Predicted time: 4.588921e+10

Callcount: 1000

__invivo__UPML_fppized___upml_mod_MOD_upmlupdatee_1569

Real time: 3.944061e+10 -- Predicted time: 4.095546e+10

Callcount: 1000

__invivo__UPML_fppized___upml_mod_MOD_upmlupdateh_1091

Real time: 4.203746e+10 -- Predicted time: 4.585702e+10

Callcount: 1000

__invivo__excite___excite_mod_MOD_excitation_array_236

Real time: 4.393474e+10 -- Predicted time: 4.350182e+10

Callcount: 25001

__invivo__huygens_fppized___huygens_mod_MOD_huygensh_693

Real time: 3.020855e+10 -- Predicted time: 0.000000e+00

Callcount: 1000

invivoUPML_fppized___upml_mod_MOD_upmlupdateh_904

invivoleapfrog___leapfrog_mod_MOD_leapfrog_200

invivoNFT_fppized___nft_mod_MOD_nft_store_1065

invivoUPML_fppized___upml_mod_MOD_upmlupdatee_1202

invivoUPML_fppized___upml_mod_MOD_upmlupdateh_846

invivoUPML_fppized___upml_mod_MOD_upmlupdatee_1401

invivoUPML_fppized___upml_mod_MOD_upmlupdatee_1569

invivoUPML_fppized___upml_mod_MOD_upmlupdateh_1091

invivoexcite___excite_mod_MOD_excitation_array_236

invivohuygens_fppized___huygens_mod_MOD_huygensh_693