diff --git a/.gitattributes b/.gitattributes
index cb6389607cec7ec700eda30eab1cf3c5be602ad6..95194ed9a9e03fcec68412e7c8a99953e592302e 100644
--- a/.gitattributes
+++ b/.gitattributes
@@ -3594,6 +3594,7 @@ RTCP/IONProc/test/tDelayCompensation.sh -text
RTCP/IONProc/test/tSSH.cc -text
RTCP/IONProc/test/tSSH.sh -text
RTCP/IONProc/test/tSSH.stdout -text
+RTCP/Interface/include/Interface/BGPAsm.h -text
RTCP/Interface/include/Interface/BeamCoordinates.h -text
RTCP/Interface/include/Interface/BeamFormedData.h -text
RTCP/Interface/include/Interface/DataFactory.h -text
@@ -3606,9 +3607,11 @@ RTCP/Interface/include/Interface/Stream.h -text
RTCP/Interface/include/Interface/StreamableData.h -text
RTCP/Interface/include/Interface/TransposedBeamFormedData.h -text
RTCP/Interface/include/Interface/TriggerData.h -text
+RTCP/Interface/src/BGPAsm.S -text
RTCP/Interface/src/BeamCoordinates.cc -text
RTCP/Interface/src/DataFactory.cc -text
RTCP/Interface/src/Stream.cc -text
+RTCP/Interface/test/tCorrelatedData.cc -text
RTCP/LofarStMan/CMakeLists.txt -text
RTCP/LofarStMan/include/LofarStMan/CMakeLists.txt -text
RTCP/LofarStMan/src/CMakeLists.txt -text
diff --git a/RTCP/CNProc/src/BeamFormerAsm.S b/RTCP/CNProc/src/BeamFormerAsm.S
index 9b4eadf8ecb79f34f7af25a8eed045a44506b9a6..680e37ebf8b0322aad7b79aa09df167f1b1236b3 100644
--- a/RTCP/CNProc/src/BeamFormerAsm.S
+++ b/RTCP/CNProc/src/BeamFormerAsm.S
@@ -1,765 +1,5 @@
#if defined HAVE_BGP
-.global _add_2_single_precision_vectors
-_add_2_single_precision_vectors:
-
- li 0,-16
- stfpdux 14,1,0
- stfpdux 15,1,0
-
- srwi 6,6,4
- li 11,8
- subi 6,6,1
- mtctr 6
-
- cmpwi 0,6,0
- lfpsx 0,0,4
- lfpsx 1,0,5
- lfpsux 2,4,11
- lfpsux 3,5,11
- lfpsux 4,4,11
- lfpsux 5,5,11
- lfpsux 6,4,11
- lfpsux 7,5,11
- lfpsux 8,4,11
- lfpsux 9,5,11
- lfpsux 10,4,11
- lfpsux 11,5,11
- fpadd 0,0,1
- lfpsux 12,4,11
- lfpsux 13,5,11
- fpadd 2,2,3
- lfpsux 14,4,11
- lfpsux 15,5,11
- stfpsx 0,0,3
- fpadd 4,4,5
- beq- 0,1f
-
-0: lfpsux 0,4,11
- lfpsux 1,5,11
- stfpsux 2,3,11
- fpadd 6,6,7
- lfpsux 2,4,11
- lfpsux 3,5,11
- stfpsux 4,3,11
- fpadd 8,8,9
- lfpsux 4,4,11
- lfpsux 5,5,11
- stfpsux 6,3,11
- fpadd 10,10,11
- lfpsux 6,4,11
- lfpsux 7,5,11
- stfpsux 8,3,11
- fpadd 12,12,13
- lfpsux 8,4,11
- lfpsux 9,5,11
- stfpsux 10,3,11
- fpadd 14,14,15
- lfpsux 10,4,11
- lfpsux 11,5,11
- stfpsux 12,3,11
- fpadd 0,0,1
- lfpsux 12,4,11
- lfpsux 13,5,11
- stfpsux 14,3,11
- fpadd 2,2,3
- lfpsux 14,4,11
- lfpsux 15,5,11
- stfpsux 0,3,11
- fpadd 4,4,5
- bdnz 0b
-
-1: stfpsux 2,3,11
- fpadd 6,6,7
- stfpsux 4,3,11
- fpadd 8,8,9
- stfpsux 6,3,11
- fpadd 10,10,11
- stfpsux 8,3,11
- fpadd 12,12,13
- stfpsux 10,3,11
- fpadd 14,14,15
- stfpsux 12,3,11
- stfpsux 14,3,11
-
- li 0,16 # restore call-saved registers
- lfpdx 15,0,1
- lfpdux 14,1,0
- addi 1,1,16
-
- blr
-
-
-.global _add_3_single_precision_vectors
-_add_3_single_precision_vectors:
-
- li 0,-16
- stfpdux 14,1,0
- stfpdux 15,1,0
- stfpdux 16,1,0
- stfpdux 17,1,0
- stfpdux 18,1,0
- stfpdux 19,1,0
- stfpdux 20,1,0
- stfpdux 21,1,0
- stfpdux 22,1,0
- stfpdux 23,1,0
-
- srwi 7,7,4
- li 11,8
- subi 7,7,1
- mtctr 7
-
- cmpwi 0,7,0
- lfpsx 0,0,4
- lfpsx 1,0,5
- lfpsx 2,0,6
- lfpsux 3,4,11
- lfpsux 4,5,11
- lfpsux 5,6,11
- lfpsux 6,4,11
- lfpsux 7,5,11
- lfpsux 8,6,11
- lfpsux 9,4,11
- lfpsux 10,5,11
- lfpsux 11,6,11
- lfpsux 12,4,11
- lfpsux 13,5,11
- lfpsux 14,6,11
- lfpsux 15,4,11
- lfpsux 16,5,11
- fpadd 0,0,1
- lfpsux 17,6,11
- lfpsux 18,4,11
- lfpsux 19,5,11
- fpadd 3,3,4
- lfpsux 20,6,11
- fpadd 0,0,2
- lfpsux 21,4,11
- lfpsux 22,5,11
- fpadd 6,6,7
- lfpsux 23,6,11
- fpadd 3,3,5
- stfpsx 0,0,3
- beq- 0,1f
- li 10,32
-
-0: lfpsux 0,4,11
- lfpsux 1,5,11
- dcbt 4,10
- fpadd 9,9,10
- lfpsux 2,6,11
- fpadd 6,6,8
- stfpsux 3,3,11
- lfpsux 3,4,11
- lfpsux 4,5,11
- fpadd 12,12,13
- lfpsux 5,6,11
- fpadd 9,9,11
- stfpsux 6,3,11
- lfpsux 6,4,11
- lfpsux 7,5,11
- fpadd 15,15,16
- lfpsux 8,6,11
- fpadd 12,12,14
- stfpsux 9,3,11
- lfpsux 9,4,11
- lfpsux 10,5,11
- fpadd 18,18,19
- lfpsux 11,6,11
- fpadd 15,15,17
- stfpsux 12,3,11
- lfpsux 12,4,11
- lfpsux 13,5,11
- fpadd 21,21,22
- lfpsux 14,6,11
- fpadd 18,18,20
- stfpsux 15,3,11
- lfpsux 15,4,11
- lfpsux 16,5,11
- fpadd 0,0,1
- lfpsux 17,6,11
- fpadd 21,21,23
- stfpsux 18,3,11
- lfpsux 18,4,11
- lfpsux 19,5,11
- fpadd 3,3,4
- lfpsux 20,6,11
- fpadd 0,0,2
- stfpsux 21,3,11
- lfpsux 21,4,11
- lfpsux 22,5,11
- fpadd 6,6,7
- lfpsux 23,6,11
- fpadd 3,3,5
- stfpsux 0,3,11
- bdnz 0b
-
-1: fpadd 9,9,10
- fpadd 6,6,8
- stfpsux 3,3,11
- fpadd 12,12,13
- fpadd 9,9,11
- stfpsux 6,3,11
- fpadd 15,15,16
- fpadd 12,12,14
- stfpsux 9,3,11
- fpadd 18,18,19
- fpadd 15,15,17
- stfpsux 12,3,11
- fpadd 21,21,22
- fpadd 18,18,20
- stfpsux 15,3,11
- fpadd 21,21,23
- stfpsux 18,3,11
- stfpsux 21,3,11
-
- li 0,16 # restore call-saved registers
- lfpdx 23,0,1
- lfpdux 22,1,0
- lfpdux 21,1,0
- lfpdux 20,1,0
- lfpdux 19,1,0
- lfpdux 18,1,0
- lfpdux 17,1,0
- lfpdux 16,1,0
- lfpdux 15,1,0
- lfpdux 14,1,0
- addi 1,1,16
-
- blr
-
-
-.global _add_4_single_precision_vectors
-_add_4_single_precision_vectors:
-
- li 0,-16
- stfpdux 14,1,0
- stfpdux 15,1,0
- stfpdux 16,1,0
- stfpdux 17,1,0
- stfpdux 18,1,0
- stfpdux 19,1,0
- stfpdux 20,1,0
- stfpdux 21,1,0
- stfpdux 22,1,0
- stfpdux 23,1,0
- stfpdux 24,1,0
- stfpdux 25,1,0
- stfpdux 26,1,0
- stfpdux 27,1,0
- stfpdux 28,1,0
- stfpdux 29,1,0
- stfpdux 30,1,0
- stfpdux 31,1,0
-
- srwi 8,8,4
- li 11,8
- subi 8,8,1
- mtctr 8
-
- cmpwi 0,8,0
- lfpsx 0,0,4
- lfpsx 1,0,5
- lfpsx 2,0,6
- lfpsx 3,0,7
- lfpsux 4,4,11
- lfpsux 5,5,11
- lfpsux 6,6,11
- lfpsux 7,7,11
- lfpsux 8,4,11
- lfpsux 9,5,11
- lfpsux 10,6,11
- lfpsux 11,7,11
- lfpsux 12,4,11
- lfpsux 13,5,11
- lfpsux 14,6,11
- lfpsux 15,7,11
- lfpsux 16,4,11
- lfpsux 17,5,11
- lfpsux 18,6,11
- lfpsux 19,7,11
- lfpsux 20,4,11
- lfpsux 21,5,11
- fpadd 0,0,1
- fpadd 2,2,3
- lfpsux 22,6,11
- lfpsux 23,7,11
- lfpsux 24,4,11
- lfpsux 25,5,11
- fpadd 4,4,5
- fpadd 6,6,7
- lfpsux 26,6,11
- lfpsux 27,7,11
- fpadd 0,0,2
- lfpsux 28,4,11
- lfpsux 29,5,11
- fpadd 8,8,9
- fpadd 10,10,11
- lfpsux 30,6,11
- lfpsux 31,7,11
-
- stfpsx 0,0,3
- fpadd 4,4,6
- beq- 0,1f
-
-0: lfpsux 0,4,11
- lfpsux 1,5,11
- fpadd 12,12,13
- fpadd 14,14,15
- lfpsux 2,6,11
- lfpsux 3,7,11
-
- stfpsux 4,3,11
- fpadd 8,8,10
-
- lfpsux 4,4,11
- lfpsux 5,5,11
- fpadd 16,16,17
- fpadd 18,18,19
- lfpsux 6,6,11
- lfpsux 7,7,11
-
- stfpsux 8,3,11
- fpadd 12,12,14
-
- lfpsux 8,4,11
- lfpsux 9,5,11
- fpadd 20,20,21
- fpadd 22,22,23
- lfpsux 10,6,11
- lfpsux 11,7,11
-
- stfpsux 12,3,11
- fpadd 16,16,18
-
- lfpsux 12,4,11
- lfpsux 13,5,11
- fpadd 24,24,25
- fpadd 26,26,27
- lfpsux 14,6,11
- lfpsux 15,7,11
-
- stfpsux 16,3,11
- fpadd 20,20,22
-
- lfpsux 16,4,11
- lfpsux 17,5,11
- fpadd 28,28,29
- fpadd 30,30,31
- lfpsux 18,6,11
- lfpsux 19,7,11
-
- stfpsux 20,3,11
- fpadd 24,24,26
-
- lfpsux 20,4,11
- lfpsux 21,5,11
- fpadd 0,0,1
- fpadd 2,2,3
- lfpsux 22,6,11
- lfpsux 23,7,11
-
- stfpsux 24,3,11
- fpadd 28,28,30
-
- lfpsux 24,4,11
- lfpsux 25,5,11
- fpadd 4,4,5
- fpadd 6,6,7
- lfpsux 26,6,11
- lfpsux 27,7,11
-
- stfpsux 28,3,11
- fpadd 0,0,2
-
- lfpsux 28,4,11
- lfpsux 29,5,11
- fpadd 8,8,9
- fpadd 10,10,11
- lfpsux 30,6,11
- lfpsux 31,7,11
-
- stfpsux 0,3,11
- fpadd 4,4,6
-
- bdnz 0b
-
-1: fpadd 12,12,13
- fpadd 14,14,15
-
- stfpsux 4,3,11
- fpadd 8,8,10
-
- fpadd 16,16,17
- fpadd 18,18,19
-
- stfpsux 8,3,11
- fpadd 12,12,14
-
- fpadd 20,20,21
- fpadd 22,22,23
-
- stfpsux 12,3,11
- fpadd 16,16,18
-
- fpadd 24,24,25
- fpadd 26,26,27
-
- stfpsux 16,3,11
- fpadd 20,20,22
-
- lfpsux 16,4,11
- lfpsux 17,5,11
- fpadd 28,28,29
- fpadd 30,30,31
- stfpsux 20,3,11
- fpadd 24,24,26
- stfpsux 24,3,11
- fpadd 28,28,30
- stfpsux 28,3,11
-
- li 0,16 # restore call-saved registers
- lfpdx 31,0,1
- lfpdux 30,1,0
- lfpdux 29,1,0
- lfpdux 28,1,0
- lfpdux 27,1,0
- lfpdux 26,1,0
- lfpdux 25,1,0
- lfpdux 24,1,0
- lfpdux 23,1,0
- lfpdux 22,1,0
- lfpdux 21,1,0
- lfpdux 20,1,0
- lfpdux 19,1,0
- lfpdux 18,1,0
- lfpdux 17,1,0
- lfpdux 16,1,0
- lfpdux 15,1,0
- lfpdux 14,1,0
- addi 1,1,16
-
- blr
-
-
-.global _add_5_single_precision_vectors
-_add_5_single_precision_vectors:
-
- li 0,-16
- stfpdux 14,1,0
- stfpdux 15,1,0
- stfpdux 16,1,0
- stfpdux 17,1,0
- stfpdux 18,1,0
- stfpdux 19,1,0
- stfpdux 24,1,0
- stfpdux 25,1,0
- stfpdux 26,1,0
- stfpdux 27,1,0
- stfpdux 28,1,0
- stfpdux 29,1,0
-
- srwi 9,9,3
- li 11,8
- subi 9,9,1
- mtctr 9
- subi 3,3,8
-
- cmpwi 0,9,0
-
- lfpsx 0,0,4
- lfpsx 1,0,5
- lfpsx 2,0,6
- lfpsx 3,0,7
- lfpsx 4,0,8
-
- lfpsux 5,4,11
- lfpsux 6,5,11
- lfpsux 7,6,11
- lfpsux 8,7,11
- lfpsux 9,8,11
-
- lfpsux 10,4,11
- lfpsux 11,5,11
- lfpsux 12,6,11
- lfpsux 13,7,11
- lfpsux 14,8,11
- fpadd 24,0,1
- fpadd 25,2,3
-
- lfpsux 15,4,11
- lfpsux 16,5,11
- lfpsux 17,6,11
- fpadd 24,24,25
- lfpsux 18,7,11
- lfpsux 19,8,11
- fpadd 27,5,6
- fpadd 28,7,8
-
- beq- 0,1f
- li 10,48
-
-0:
- dcbt 4,10
- lfpsux 0,4,11
- fpadd 24,24,4
- lfpsux 1,5,11
- lfpsux 2,6,11
- fpadd 27,27,28
- lfpsux 3,7,11
- lfpsux 4,8,11
- stfpsux 24,3,11
- fpadd 24,10,11
- dcbt 5,10
- fpadd 25,12,13
-
- lfpsux 5,4,11
- fpadd 27,27,9
- lfpsux 6,5,11
- lfpsux 7,6,11
- fpadd 24,24,25
- lfpsux 8,7,11
- lfpsux 9,8,11
- stfpsux 27,3,11
- fpadd 27,15,16
- dcbt 6,10
- fpadd 28,17,18
-
- lfpsux 10,4,11
- fpadd 24,24,14
- lfpsux 11,5,11
- lfpsux 12,6,11
- fpadd 27,27,28
- lfpsux 13,7,11
- lfpsux 14,8,11
- stfpsux 24,3,11
- fpadd 24,0,1
- dcbt 7,10
- fpadd 25,2,3
-
- lfpsux 15,4,11
- fpadd 27,27,19
- lfpsux 16,5,11
- lfpsux 17,6,11
- fpadd 24,24,25
- dcbt 8,10
- lfpsux 18,7,11
- lfpsux 19,8,11
- stfpsux 27,3,11
- fpadd 27,5,6
- fpadd 28,7,8
-
- bdnz 0b
-
-1:
- fpadd 24,24,4
- fpadd 27,27,28
- stfpsux 24,3,11
- fpadd 24,10,11
- fpadd 25,12,13
-
- fpadd 27,27,9
- fpadd 24,24,25
- stfpsux 27,3,11
- fpadd 27,15,16
- fpadd 28,17,18
-
- fpadd 24,24,14
- fpadd 27,27,28
- stfpsux 24,3,11
-
- fpadd 27,27,19
- stfpsux 27,3,11
-
- li 0,16 # restore call-saved registers
- lfpdx 29,0,1
- lfpdux 28,1,0
- lfpdux 27,1,0
- lfpdux 26,1,0
- lfpdux 25,1,0
- lfpdux 24,1,0
- lfpdux 19,1,0
- lfpdux 18,1,0
- lfpdux 17,1,0
- lfpdux 16,1,0
- lfpdux 15,1,0
- lfpdux 14,1,0
- addi 1,1,16
- blr
-
-
-.global _add_6_single_precision_vectors
-_add_6_single_precision_vectors:
-
- li 0,-16
- stfpdux 14,1,0
- stfpdux 15,1,0
- stfpdux 16,1,0
- stfpdux 17,1,0
- stfpdux 18,1,0
- stfpdux 19,1,0
- stfpdux 20,1,0
- stfpdux 21,1,0
- stfpdux 22,1,0
- stfpdux 23,1,0
- stfpdux 24,1,0
- stfpdux 25,1,0
- stfpdux 26,1,0
- stfpdux 27,1,0
- stfpdux 28,1,0
- stfpdux 29,1,0
-
- srwi 10,10,3
- li 11,8
- subi 10,10,1
- mtctr 10
- subi 3,3,8
-
- cmpwi 0,10,0
-
- lfpsx 0,0,4
- lfpsx 1,0,5
- lfpsx 2,0,6
- lfpsx 3,0,7
- lfpsx 4,0,8
- lfpsx 5,0,9
-
- lfpsux 6,4,11
- lfpsux 7,5,11
- lfpsux 8,6,11
- lfpsux 9,7,11
- lfpsux 10,8,11
- lfpsux 11,9,11
-
- lfpsux 12,4,11
- lfpsux 13,5,11
- lfpsux 14,6,11
- lfpsux 15,7,11
- lfpsux 16,8,11
- lfpsux 17,9,11
- fpadd 24,0,1
- fpadd 25,2,3
-
- lfpsux 18,4,11
- lfpsux 19,5,11
- lfpsux 20,6,11
- fpadd 26,4,5
- fpadd 24,24,25
- lfpsux 21,7,11
- lfpsux 22,8,11
- lfpsux 23,9,11
- fpadd 27,6,7
- fpadd 28,8,9
-
- beq- 0,1f
- li 10,48
-
-0:
- dcbt 4,10
- lfpsux 0,4,11
- fpadd 24,24,26
- lfpsux 1,5,11
- lfpsux 2,6,11
- fpadd 29,10,11
- fpadd 27,27,28
- lfpsux 3,7,11
- lfpsux 4,8,11
- lfpsux 5,9,11
- dcbt 5,10
- stfpsux 24,3,11
- fpadd 24,12,13
- fpadd 25,14,15
-
- lfpsux 6,4,11
- fpadd 27,27,29
- lfpsux 7,5,11
- lfpsux 8,6,11
- dcbt 6,10
- fpadd 26,16,17
- fpadd 24,24,25
- lfpsux 9,7,11
- lfpsux 10,8,11
- lfpsux 11,9,11
- stfpsux 27,3,11
- fpadd 27,18,19
- fpadd 28,20,21
-
- dcbt 7,10
- lfpsux 12,4,11
- fpadd 24,24,26
- lfpsux 13,5,11
- lfpsux 14,6,11
- fpadd 29,22,23
- fpadd 27,27,28
- lfpsux 15,7,11
- lfpsux 16,8,11
- lfpsux 17,9,11
- dcbt 8,10
- stfpsux 24,3,11
- fpadd 24,0,1
- fpadd 25,2,3
-
- lfpsux 18,4,11
- fpadd 27,27,29
- lfpsux 19,5,11
- lfpsux 20,6,11
- dcbt 9,10
- fpadd 26,4,5
- fpadd 24,24,25
- lfpsux 21,7,11
- lfpsux 22,8,11
- lfpsux 23,9,11
- stfpsux 27,3,11
- fpadd 27,6,7
- fpadd 28,8,9
-
- bdnz 0b
-
-1:
- fpadd 24,24,26
- fpadd 29,10,11
- fpadd 27,27,28
- stfpsux 24,3,11
- fpadd 24,12,13
- fpadd 25,14,15
-
- fpadd 27,27,29
- fpadd 26,16,17
- fpadd 24,24,25
- stfpsux 27,3,11
- fpadd 27,18,19
- fpadd 28,20,21
-
- fpadd 24,24,26
- fpadd 29,22,23
- fpadd 27,27,28
- stfpsux 24,3,11
-
- fpadd 27,27,29
- stfpsux 27,3,11
-
- li 0,16 # restore call-saved registers
- lfpdx 29,0,1
- lfpdux 28,1,0
- lfpdux 27,1,0
- lfpdux 26,1,0
- lfpdux 25,1,0
- lfpdux 24,1,0
- lfpdux 23,1,0
- lfpdux 22,1,0
- lfpdux 21,1,0
- lfpdux 20,1,0
- lfpdux 19,1,0
- lfpdux 18,1,0
- lfpdux 17,1,0
- lfpdux 16,1,0
- lfpdux 15,1,0
- lfpdux 14,1,0
- addi 1,1,16
- blr
-
-
.align 5
.global _beamform_3stations_6beams
diff --git a/RTCP/CNProc/src/BeamFormerAsm.h b/RTCP/CNProc/src/BeamFormerAsm.h
index 88f8c136c0dfe0fff1ff3095125128e9cf4e9bde..2a8399e10dcd7e39d0f33354be9b8c19dd8afc44 100644
--- a/RTCP/CNProc/src/BeamFormerAsm.h
+++ b/RTCP/CNProc/src/BeamFormerAsm.h
@@ -3,6 +3,7 @@
#if defined HAVE_BGP
+#include <Interface/BGPAsm.h>
#include <cstring>
namespace LOFAR {
@@ -10,53 +11,6 @@ namespace RTCP {
extern "C" {
-// all float * must be aligned to 8 bytes
-
-void _add_2_single_precision_vectors(
- /* r3 */ float *dst,
- /* r4 */ const float *src1,
- /* r5 */ const float *src2,
- /* r6 */ unsigned count /* non-zero; multiple of 16 */
-);
-
-void _add_3_single_precision_vectors(
- /* r3 */ float *dst,
- /* r4 */ const float *src1,
- /* r5 */ const float *src2,
- /* r6 */ const float *src3,
- /* r7 */ unsigned count /* non-zero; multiple of 16 */
-);
-
-void _add_4_single_precision_vectors(
- /* r3 */ float *dst,
- /* r4 */ const float *src1,
- /* r5 */ const float *src2,
- /* r6 */ const float *src3,
- /* r7 */ const float *src4,
- /* r8 */ unsigned count /* non-zero; multiple of 16 */
-);
-
-void _add_5_single_precision_vectors(
- /* r3 */ float *dst,
- /* r4 */ const float *src1,
- /* r5 */ const float *src2,
- /* r6 */ const float *src3,
- /* r7 */ const float *src4,
- /* r8 */ const float *src5,
- /* r9 */ unsigned count /* non-zero; multiple of 16 */
-);
-
-void _add_6_single_precision_vectors(
- /* r3 */ float *dst,
- /* r4 */ const float *src1,
- /* r5 */ const float *src2,
- /* r6 */ const float *src3,
- /* r7 */ const float *src4,
- /* r8 */ const float *src5,
- /* r9 */ const float *src6,
- /* r10 */ unsigned count /* non-zero; multiple of 16 */
-);
-
#if 0
void _beamform_3beams(
/* r3 */ fcomplex *dst,
@@ -134,33 +88,6 @@ void _beamform_6beams_2times(
} // extern "C"
-// Similar functions that do not need or have an ASM version
-
-// defined just to aid the use of macros
-static inline void _add_1_single_precision_vectors(
- float *dst,
- const float *src1,
- unsigned count /* non-zero; multiple of 16 */
-) {
- // nothing to add, so just copy the values
- memcpy( dst, src1, count * sizeof(float) );
-}
-
-static inline void _add_7_single_precision_vectors(
- float *dst,
- const float *src1,
- const float *src2,
- const float *src3,
- const float *src4,
- const float *src5,
- const float *src6,
- const float *src7,
- unsigned count /* non-zero; multiple of 16 */
-) {
- _add_4_single_precision_vectors( dst, src1, src2, src3, src4, count );
- _add_4_single_precision_vectors( dst, dst, src5, src6, src7, count );
-}
-
} // namespace LOFAR::RTCP
} // namespace LOFAR
diff --git a/RTCP/Interface/CMakeLists.txt b/RTCP/Interface/CMakeLists.txt
index 93c6b119ba8eb0ea13d3d27763f959b20293f5f9..2b21d242021b5022ae698ce1ad7c4ce81ddc1d38 100644
--- a/RTCP/Interface/CMakeLists.txt
+++ b/RTCP/Interface/CMakeLists.txt
@@ -10,6 +10,20 @@ if(USE_VALGRIND)
add_definitions(-DUSE_VALGRIND)
endif(USE_VALGRIND)
+## ---------------------------------------------------------------------------
+## Enable BGP specific assembler.
+## Use the BGP assembler also for linking C/C++ programs.
+##
+## NOTE: Maybe this should be moved to the variants file. That requires that
+## CN and ION are no longer variants; we then need two compiler definitions:
+## BGPCN and BGPION, which is a cleaner solution anyway.
+## ---------------------------------------------------------------------------
+enable_language(ASM-BGP)
+if(LOFAR_COMPILER_SUITE STREQUAL "BGPCN")
+ set(CMAKE_C_LINK_EXECUTABLE ${CMAKE_ASM-BGP_LINK_EXECUTABLE})
+ set(CMAKE_CXX_LINK_EXECUTABLE ${CMAKE_ASM-BGP_LINK_EXECUTABLE})
+endif(LOFAR_COMPILER_SUITE STREQUAL "BGPCN")
+
add_subdirectory(include/Interface)
add_subdirectory(src)
add_subdirectory(test)
diff --git a/RTCP/Interface/include/Interface/BGPAsm.h b/RTCP/Interface/include/Interface/BGPAsm.h
new file mode 100644
index 0000000000000000000000000000000000000000..8a1c9a2faae1cd9fc318ae5aa670d8680cc26094
--- /dev/null
+++ b/RTCP/Interface/include/Interface/BGPAsm.h
@@ -0,0 +1,94 @@
+#ifndef LOFAR_INTERFACE_BGPASM_H
+#define LOFAR_INTERFACE_BGPASM_H
+
+#if defined HAVE_BGP
+
+#include <cstring>
+
+namespace LOFAR {
+namespace RTCP {
+
+extern "C" {
+
+// all float * must be aligned to 8 bytes
+
+void _add_2_single_precision_vectors(
+ /* r3 */ float *dst,
+ /* r4 */ const float *src1,
+ /* r5 */ const float *src2,
+ /* r6 */ unsigned count /* non-zero; multiple of 16 */
+);
+
+void _add_3_single_precision_vectors(
+ /* r3 */ float *dst,
+ /* r4 */ const float *src1,
+ /* r5 */ const float *src2,
+ /* r6 */ const float *src3,
+ /* r7 */ unsigned count /* non-zero; multiple of 16 */
+);
+
+void _add_4_single_precision_vectors(
+ /* r3 */ float *dst,
+ /* r4 */ const float *src1,
+ /* r5 */ const float *src2,
+ /* r6 */ const float *src3,
+ /* r7 */ const float *src4,
+ /* r8 */ unsigned count /* non-zero; multiple of 16 */
+);
+
+void _add_5_single_precision_vectors(
+ /* r3 */ float *dst,
+ /* r4 */ const float *src1,
+ /* r5 */ const float *src2,
+ /* r6 */ const float *src3,
+ /* r7 */ const float *src4,
+ /* r8 */ const float *src5,
+ /* r9 */ unsigned count /* non-zero; multiple of 16 */
+);
+
+void _add_6_single_precision_vectors(
+ /* r3 */ float *dst,
+ /* r4 */ const float *src1,
+ /* r5 */ const float *src2,
+ /* r6 */ const float *src3,
+ /* r7 */ const float *src4,
+ /* r8 */ const float *src5,
+ /* r9 */ const float *src6,
+ /* r10 */ unsigned count /* non-zero; multiple of 16 */
+);
+
+} // extern "C"
+
+// Similar functions that do not need or have an ASM version
+
+// defined just to aid the use of macros
+static inline void _add_1_single_precision_vectors(
+ float *dst,
+ const float *src1,
+ unsigned count /* non-zero; multiple of 16 */
+) {
+ // nothing to add, so just copy the values
+ memcpy( dst, src1, count * sizeof(float) );
+}
+
+static inline void _add_7_single_precision_vectors(
+ float *dst,
+ const float *src1,
+ const float *src2,
+ const float *src3,
+ const float *src4,
+ const float *src5,
+ const float *src6,
+ const float *src7,
+ unsigned count /* non-zero; multiple of 16 */
+) {
+ _add_4_single_precision_vectors( dst, src1, src2, src3, src4, count );
+ _add_4_single_precision_vectors( dst, dst, src5, src6, src7, count );
+}
+
+} // namespace LOFAR::RTCP
+} // namespace LOFAR
+
+#endif
+
+#endif
diff --git a/RTCP/Interface/include/Interface/CMakeLists.txt b/RTCP/Interface/include/Interface/CMakeLists.txt
index 655c0f5711b431f1af8ed204f7789392f93fdbe0..abec53f584b7e2d5d3159c0f5f945a672ce47b74 100644
--- a/RTCP/Interface/include/Interface/CMakeLists.txt
+++ b/RTCP/Interface/include/Interface/CMakeLists.txt
@@ -5,6 +5,7 @@ set(inst_HEADERS
Allocator.h
BeamCoordinates.h
BeamFormedData.h
+ BGPAsm.h
CN_Command.h
CN_Mapping.h
Stream.h
diff --git a/RTCP/Interface/include/Interface/CorrelatedData.h b/RTCP/Interface/include/Interface/CorrelatedData.h
index 381b629927ee9c8504de79b32bb637eb66839055..2c7df06ba211e11c776da6ec6bd04ddf190c2edb 100644
--- a/RTCP/Interface/include/Interface/CorrelatedData.h
+++ b/RTCP/Interface/include/Interface/CorrelatedData.h
@@ -7,6 +7,7 @@
#include <Interface/Allocator.h>
#include <Interface/Config.h>
#include <Interface/StreamableData.h>
+#include <Interface/BGPAsm.h>
#include <Interface/MultiDimArray.h>
#include <Stream/Stream.h>
@@ -122,13 +123,31 @@ inline void CorrelatedData::writeData(Stream *str)
}
-template <typename T> inline void addNrValidSamples(T *dst, const T *src, unsigned count)
+template <typename T> inline void addNrValidSamples(T * __restrict__ dst, const T * __restrict__ src, unsigned count)
{
for (unsigned i = 0; i < count; i ++)
dst[i] += src[i];
}
+template<> inline void addNrValidSamples<uint16_t>(uint16_t * __restrict__ dst, const uint16_t * __restrict__ src, unsigned count)
+{
+ addNrValidSamples<uint32_t>(reinterpret_cast<uint32_t*>(dst), reinterpret_cast<const uint32_t*>(src), count / 2);
+
+ if (count & 1)
+ dst[count - 1] += src[count - 1];
+}
+
+
+template<> inline void addNrValidSamples<uint8_t>(uint8_t * __restrict__ dst, const uint8_t * __restrict__ src, unsigned count)
+{
+ addNrValidSamples<uint16_t>(reinterpret_cast<uint16_t*>(dst), reinterpret_cast<const uint16_t*>(src), count / 2);
+
+ if (count & 1)
+ dst[count - 1] += src[count - 1];
+}
+
+
inline IntegratableData &CorrelatedData::operator += (const IntegratableData &other_)
{
const CorrelatedData &other = static_cast<const CorrelatedData &>(other_);
@@ -138,9 +157,19 @@ inline IntegratableData &CorrelatedData::operator += (const IntegratableData &ot
fcomplex *dst = visibilities.origin();
const fcomplex *src = other.visibilities.origin();
unsigned count = visibilities.num_elements();
+#ifdef HAVE_BGP
+ unsigned fastcopyfloats = (count * 2) & ~0xF;
+ unsigned remainder = count % 8;
+ for (unsigned i = 0; i < remainder; i ++)
+ dst[i] += src[i];
+
+ if (fastcopyfloats > 0)
+ _add_2_single_precision_vectors( reinterpret_cast<float*>(dst + remainder), reinterpret_cast<float*>(dst + remainder), reinterpret_cast<const float*>(src + remainder), fastcopyfloats );
+#else
for (unsigned i = 0; i < count; i ++)
dst[i] += src[i];
+#endif
}
// add nr. valid samples
diff --git a/RTCP/Interface/src/BGPAsm.S b/RTCP/Interface/src/BGPAsm.S
new file mode 100644
index 0000000000000000000000000000000000000000..db7a73cbd830a19320174eabf1b63954fc07cdfe
--- /dev/null
+++ b/RTCP/Interface/src/BGPAsm.S
@@ -0,0 +1,762 @@
+#if defined HAVE_BGP
+
+.global _add_2_single_precision_vectors
+_add_2_single_precision_vectors:
+
+ li 0,-16
+ stfpdux 14,1,0
+ stfpdux 15,1,0
+
+ srwi 6,6,4
+ li 11,8
+ subi 6,6,1
+ mtctr 6
+
+ cmpwi 0,6,0
+ lfpsx 0,0,4
+ lfpsx 1,0,5
+ lfpsux 2,4,11
+ lfpsux 3,5,11
+ lfpsux 4,4,11
+ lfpsux 5,5,11
+ lfpsux 6,4,11
+ lfpsux 7,5,11
+ lfpsux 8,4,11
+ lfpsux 9,5,11
+ lfpsux 10,4,11
+ lfpsux 11,5,11
+ fpadd 0,0,1
+ lfpsux 12,4,11
+ lfpsux 13,5,11
+ fpadd 2,2,3
+ lfpsux 14,4,11
+ lfpsux 15,5,11
+ stfpsx 0,0,3
+ fpadd 4,4,5
+ beq- 0,1f
+
+0: lfpsux 0,4,11
+ lfpsux 1,5,11
+ stfpsux 2,3,11
+ fpadd 6,6,7
+ lfpsux 2,4,11
+ lfpsux 3,5,11
+ stfpsux 4,3,11
+ fpadd 8,8,9
+ lfpsux 4,4,11
+ lfpsux 5,5,11
+ stfpsux 6,3,11
+ fpadd 10,10,11
+ lfpsux 6,4,11
+ lfpsux 7,5,11
+ stfpsux 8,3,11
+ fpadd 12,12,13
+ lfpsux 8,4,11
+ lfpsux 9,5,11
+ stfpsux 10,3,11
+ fpadd 14,14,15
+ lfpsux 10,4,11
+ lfpsux 11,5,11
+ stfpsux 12,3,11
+ fpadd 0,0,1
+ lfpsux 12,4,11
+ lfpsux 13,5,11
+ stfpsux 14,3,11
+ fpadd 2,2,3
+ lfpsux 14,4,11
+ lfpsux 15,5,11
+ stfpsux 0,3,11
+ fpadd 4,4,5
+ bdnz 0b
+
+1: stfpsux 2,3,11
+ fpadd 6,6,7
+ stfpsux 4,3,11
+ fpadd 8,8,9
+ stfpsux 6,3,11
+ fpadd 10,10,11
+ stfpsux 8,3,11
+ fpadd 12,12,13
+ stfpsux 10,3,11
+ fpadd 14,14,15
+ stfpsux 12,3,11
+ stfpsux 14,3,11
+
+ li 0,16 # restore call-saved registers
+ lfpdx 15,0,1
+ lfpdux 14,1,0
+ addi 1,1,16
+
+ blr
+
+
+.global _add_3_single_precision_vectors
+_add_3_single_precision_vectors:
+
+ li 0,-16
+ stfpdux 14,1,0
+ stfpdux 15,1,0
+ stfpdux 16,1,0
+ stfpdux 17,1,0
+ stfpdux 18,1,0
+ stfpdux 19,1,0
+ stfpdux 20,1,0
+ stfpdux 21,1,0
+ stfpdux 22,1,0
+ stfpdux 23,1,0
+
+ srwi 7,7,4
+ li 11,8
+ subi 7,7,1
+ mtctr 7
+
+ cmpwi 0,7,0
+ lfpsx 0,0,4
+ lfpsx 1,0,5
+ lfpsx 2,0,6
+ lfpsux 3,4,11
+ lfpsux 4,5,11
+ lfpsux 5,6,11
+ lfpsux 6,4,11
+ lfpsux 7,5,11
+ lfpsux 8,6,11
+ lfpsux 9,4,11
+ lfpsux 10,5,11
+ lfpsux 11,6,11
+ lfpsux 12,4,11
+ lfpsux 13,5,11
+ lfpsux 14,6,11
+ lfpsux 15,4,11
+ lfpsux 16,5,11
+ fpadd 0,0,1
+ lfpsux 17,6,11
+ lfpsux 18,4,11
+ lfpsux 19,5,11
+ fpadd 3,3,4
+ lfpsux 20,6,11
+ fpadd 0,0,2
+ lfpsux 21,4,11
+ lfpsux 22,5,11
+ fpadd 6,6,7
+ lfpsux 23,6,11
+ fpadd 3,3,5
+ stfpsx 0,0,3
+ beq- 0,1f
+ li 10,32
+
+0: lfpsux 0,4,11
+ lfpsux 1,5,11
+ dcbt 4,10
+ fpadd 9,9,10
+ lfpsux 2,6,11
+ fpadd 6,6,8
+ stfpsux 3,3,11
+ lfpsux 3,4,11
+ lfpsux 4,5,11
+ fpadd 12,12,13
+ lfpsux 5,6,11
+ fpadd 9,9,11
+ stfpsux 6,3,11
+ lfpsux 6,4,11
+ lfpsux 7,5,11
+ fpadd 15,15,16
+ lfpsux 8,6,11
+ fpadd 12,12,14
+ stfpsux 9,3,11
+ lfpsux 9,4,11
+ lfpsux 10,5,11
+ fpadd 18,18,19
+ lfpsux 11,6,11
+ fpadd 15,15,17
+ stfpsux 12,3,11
+ lfpsux 12,4,11
+ lfpsux 13,5,11
+ fpadd 21,21,22
+ lfpsux 14,6,11
+ fpadd 18,18,20
+ stfpsux 15,3,11
+ lfpsux 15,4,11
+ lfpsux 16,5,11
+ fpadd 0,0,1
+ lfpsux 17,6,11
+ fpadd 21,21,23
+ stfpsux 18,3,11
+ lfpsux 18,4,11
+ lfpsux 19,5,11
+ fpadd 3,3,4
+ lfpsux 20,6,11
+ fpadd 0,0,2
+ stfpsux 21,3,11
+ lfpsux 21,4,11
+ lfpsux 22,5,11
+ fpadd 6,6,7
+ lfpsux 23,6,11
+ fpadd 3,3,5
+ stfpsux 0,3,11
+ bdnz 0b
+
+1: fpadd 9,9,10
+ fpadd 6,6,8
+ stfpsux 3,3,11
+ fpadd 12,12,13
+ fpadd 9,9,11
+ stfpsux 6,3,11
+ fpadd 15,15,16
+ fpadd 12,12,14
+ stfpsux 9,3,11
+ fpadd 18,18,19
+ fpadd 15,15,17
+ stfpsux 12,3,11
+ fpadd 21,21,22
+ fpadd 18,18,20
+ stfpsux 15,3,11
+ fpadd 21,21,23
+ stfpsux 18,3,11
+ stfpsux 21,3,11
+
+ li 0,16 # restore call-saved registers
+ lfpdx 23,0,1
+ lfpdux 22,1,0
+ lfpdux 21,1,0
+ lfpdux 20,1,0
+ lfpdux 19,1,0
+ lfpdux 18,1,0
+ lfpdux 17,1,0
+ lfpdux 16,1,0
+ lfpdux 15,1,0
+ lfpdux 14,1,0
+ addi 1,1,16
+
+ blr
+
+
+.global _add_4_single_precision_vectors
+_add_4_single_precision_vectors:
+
+ li 0,-16
+ stfpdux 14,1,0
+ stfpdux 15,1,0
+ stfpdux 16,1,0
+ stfpdux 17,1,0
+ stfpdux 18,1,0
+ stfpdux 19,1,0
+ stfpdux 20,1,0
+ stfpdux 21,1,0
+ stfpdux 22,1,0
+ stfpdux 23,1,0
+ stfpdux 24,1,0
+ stfpdux 25,1,0
+ stfpdux 26,1,0
+ stfpdux 27,1,0
+ stfpdux 28,1,0
+ stfpdux 29,1,0
+ stfpdux 30,1,0
+ stfpdux 31,1,0
+
+ srwi 8,8,4
+ li 11,8
+ subi 8,8,1
+ mtctr 8
+
+ cmpwi 0,8,0
+ lfpsx 0,0,4
+ lfpsx 1,0,5
+ lfpsx 2,0,6
+ lfpsx 3,0,7
+ lfpsux 4,4,11
+ lfpsux 5,5,11
+ lfpsux 6,6,11
+ lfpsux 7,7,11
+ lfpsux 8,4,11
+ lfpsux 9,5,11
+ lfpsux 10,6,11
+ lfpsux 11,7,11
+ lfpsux 12,4,11
+ lfpsux 13,5,11
+ lfpsux 14,6,11
+ lfpsux 15,7,11
+ lfpsux 16,4,11
+ lfpsux 17,5,11
+ lfpsux 18,6,11
+ lfpsux 19,7,11
+ lfpsux 20,4,11
+ lfpsux 21,5,11
+ fpadd 0,0,1
+ fpadd 2,2,3
+ lfpsux 22,6,11
+ lfpsux 23,7,11
+ lfpsux 24,4,11
+ lfpsux 25,5,11
+ fpadd 4,4,5
+ fpadd 6,6,7
+ lfpsux 26,6,11
+ lfpsux 27,7,11
+ fpadd 0,0,2
+ lfpsux 28,4,11
+ lfpsux 29,5,11
+ fpadd 8,8,9
+ fpadd 10,10,11
+ lfpsux 30,6,11
+ lfpsux 31,7,11
+
+ stfpsx 0,0,3
+ fpadd 4,4,6
+ beq- 0,1f
+
+0: lfpsux 0,4,11
+ lfpsux 1,5,11
+ fpadd 12,12,13
+ fpadd 14,14,15
+ lfpsux 2,6,11
+ lfpsux 3,7,11
+
+ stfpsux 4,3,11
+ fpadd 8,8,10
+
+ lfpsux 4,4,11
+ lfpsux 5,5,11
+ fpadd 16,16,17
+ fpadd 18,18,19
+ lfpsux 6,6,11
+ lfpsux 7,7,11
+
+ stfpsux 8,3,11
+ fpadd 12,12,14
+
+ lfpsux 8,4,11
+ lfpsux 9,5,11
+ fpadd 20,20,21
+ fpadd 22,22,23
+ lfpsux 10,6,11
+ lfpsux 11,7,11
+
+ stfpsux 12,3,11
+ fpadd 16,16,18
+
+ lfpsux 12,4,11
+ lfpsux 13,5,11
+ fpadd 24,24,25
+ fpadd 26,26,27
+ lfpsux 14,6,11
+ lfpsux 15,7,11
+
+ stfpsux 16,3,11
+ fpadd 20,20,22
+
+ lfpsux 16,4,11
+ lfpsux 17,5,11
+ fpadd 28,28,29
+ fpadd 30,30,31
+ lfpsux 18,6,11
+ lfpsux 19,7,11
+
+ stfpsux 20,3,11
+ fpadd 24,24,26
+
+ lfpsux 20,4,11
+ lfpsux 21,5,11
+ fpadd 0,0,1
+ fpadd 2,2,3
+ lfpsux 22,6,11
+ lfpsux 23,7,11
+
+ stfpsux 24,3,11
+ fpadd 28,28,30
+
+ lfpsux 24,4,11
+ lfpsux 25,5,11
+ fpadd 4,4,5
+ fpadd 6,6,7
+ lfpsux 26,6,11
+ lfpsux 27,7,11
+
+ stfpsux 28,3,11
+ fpadd 0,0,2
+
+ lfpsux 28,4,11
+ lfpsux 29,5,11
+ fpadd 8,8,9
+ fpadd 10,10,11
+ lfpsux 30,6,11
+ lfpsux 31,7,11
+
+ stfpsux 0,3,11
+ fpadd 4,4,6
+
+ bdnz 0b
+
+1: fpadd 12,12,13
+ fpadd 14,14,15
+
+ stfpsux 4,3,11
+ fpadd 8,8,10
+
+ fpadd 16,16,17
+ fpadd 18,18,19
+
+ stfpsux 8,3,11
+ fpadd 12,12,14
+
+ fpadd 20,20,21
+ fpadd 22,22,23
+
+ stfpsux 12,3,11
+ fpadd 16,16,18
+
+ fpadd 24,24,25
+ fpadd 26,26,27
+
+ stfpsux 16,3,11
+ fpadd 20,20,22
+
+ lfpsux 16,4,11
+ lfpsux 17,5,11
+ fpadd 28,28,29
+ fpadd 30,30,31
+ stfpsux 20,3,11
+ fpadd 24,24,26
+ stfpsux 24,3,11
+ fpadd 28,28,30
+ stfpsux 28,3,11
+
+ li 0,16 # restore call-saved registers
+ lfpdx 31,0,1
+ lfpdux 30,1,0
+ lfpdux 29,1,0
+ lfpdux 28,1,0
+ lfpdux 27,1,0
+ lfpdux 26,1,0
+ lfpdux 25,1,0
+ lfpdux 24,1,0
+ lfpdux 23,1,0
+ lfpdux 22,1,0
+ lfpdux 21,1,0
+ lfpdux 20,1,0
+ lfpdux 19,1,0
+ lfpdux 18,1,0
+ lfpdux 17,1,0
+ lfpdux 16,1,0
+ lfpdux 15,1,0
+ lfpdux 14,1,0
+ addi 1,1,16
+
+ blr
+
+
+.global _add_5_single_precision_vectors
+_add_5_single_precision_vectors:
+
+ li 0,-16
+ stfpdux 14,1,0
+ stfpdux 15,1,0
+ stfpdux 16,1,0
+ stfpdux 17,1,0
+ stfpdux 18,1,0
+ stfpdux 19,1,0
+ stfpdux 24,1,0
+ stfpdux 25,1,0
+ stfpdux 26,1,0
+ stfpdux 27,1,0
+ stfpdux 28,1,0
+ stfpdux 29,1,0
+
+ srwi 9,9,3
+ li 11,8
+ subi 9,9,1
+ mtctr 9
+ subi 3,3,8
+
+ cmpwi 0,9,0
+
+ lfpsx 0,0,4
+ lfpsx 1,0,5
+ lfpsx 2,0,6
+ lfpsx 3,0,7
+ lfpsx 4,0,8
+
+ lfpsux 5,4,11
+ lfpsux 6,5,11
+ lfpsux 7,6,11
+ lfpsux 8,7,11
+ lfpsux 9,8,11
+
+ lfpsux 10,4,11
+ lfpsux 11,5,11
+ lfpsux 12,6,11
+ lfpsux 13,7,11
+ lfpsux 14,8,11
+ fpadd 24,0,1
+ fpadd 25,2,3
+
+ lfpsux 15,4,11
+ lfpsux 16,5,11
+ lfpsux 17,6,11
+ fpadd 24,24,25
+ lfpsux 18,7,11
+ lfpsux 19,8,11
+ fpadd 27,5,6
+ fpadd 28,7,8
+
+ beq- 0,1f
+ li 10,48
+
+0:
+ dcbt 4,10
+ lfpsux 0,4,11
+ fpadd 24,24,4
+ lfpsux 1,5,11
+ lfpsux 2,6,11
+ fpadd 27,27,28
+ lfpsux 3,7,11
+ lfpsux 4,8,11
+ stfpsux 24,3,11
+ fpadd 24,10,11
+ dcbt 5,10
+ fpadd 25,12,13
+
+ lfpsux 5,4,11
+ fpadd 27,27,9
+ lfpsux 6,5,11
+ lfpsux 7,6,11
+ fpadd 24,24,25
+ lfpsux 8,7,11
+ lfpsux 9,8,11
+ stfpsux 27,3,11
+ fpadd 27,15,16
+ dcbt 6,10
+ fpadd 28,17,18
+
+ lfpsux 10,4,11
+ fpadd 24,24,14
+ lfpsux 11,5,11
+ lfpsux 12,6,11
+ fpadd 27,27,28
+ lfpsux 13,7,11
+ lfpsux 14,8,11
+ stfpsux 24,3,11
+ fpadd 24,0,1
+ dcbt 7,10
+ fpadd 25,2,3
+
+ lfpsux 15,4,11
+ fpadd 27,27,19
+ lfpsux 16,5,11
+ lfpsux 17,6,11
+ fpadd 24,24,25
+ dcbt 8,10
+ lfpsux 18,7,11
+ lfpsux 19,8,11
+ stfpsux 27,3,11
+ fpadd 27,5,6
+ fpadd 28,7,8
+
+ bdnz 0b
+
+1:
+ fpadd 24,24,4
+ fpadd 27,27,28
+ stfpsux 24,3,11
+ fpadd 24,10,11
+ fpadd 25,12,13
+
+ fpadd 27,27,9
+ fpadd 24,24,25
+ stfpsux 27,3,11
+ fpadd 27,15,16
+ fpadd 28,17,18
+
+ fpadd 24,24,14
+ fpadd 27,27,28
+ stfpsux 24,3,11
+
+ fpadd 27,27,19
+ stfpsux 27,3,11
+
+ li 0,16 # restore call-saved registers
+ lfpdx 29,0,1
+ lfpdux 28,1,0
+ lfpdux 27,1,0
+ lfpdux 26,1,0
+ lfpdux 25,1,0
+ lfpdux 24,1,0
+ lfpdux 19,1,0
+ lfpdux 18,1,0
+ lfpdux 17,1,0
+ lfpdux 16,1,0
+ lfpdux 15,1,0
+ lfpdux 14,1,0
+ addi 1,1,16
+ blr
+
+
+.global _add_6_single_precision_vectors
+_add_6_single_precision_vectors:
+
+ li 0,-16
+ stfpdux 14,1,0
+ stfpdux 15,1,0
+ stfpdux 16,1,0
+ stfpdux 17,1,0
+ stfpdux 18,1,0
+ stfpdux 19,1,0
+ stfpdux 20,1,0
+ stfpdux 21,1,0
+ stfpdux 22,1,0
+ stfpdux 23,1,0
+ stfpdux 24,1,0
+ stfpdux 25,1,0
+ stfpdux 26,1,0
+ stfpdux 27,1,0
+ stfpdux 28,1,0
+ stfpdux 29,1,0
+
+ srwi 10,10,3
+ li 11,8
+ subi 10,10,1
+ mtctr 10
+ subi 3,3,8
+
+ cmpwi 0,10,0
+
+ lfpsx 0,0,4
+ lfpsx 1,0,5
+ lfpsx 2,0,6
+ lfpsx 3,0,7
+ lfpsx 4,0,8
+ lfpsx 5,0,9
+
+ lfpsux 6,4,11
+ lfpsux 7,5,11
+ lfpsux 8,6,11
+ lfpsux 9,7,11
+ lfpsux 10,8,11
+ lfpsux 11,9,11
+
+ lfpsux 12,4,11
+ lfpsux 13,5,11
+ lfpsux 14,6,11
+ lfpsux 15,7,11
+ lfpsux 16,8,11
+ lfpsux 17,9,11
+ fpadd 24,0,1
+ fpadd 25,2,3
+
+ lfpsux 18,4,11
+ lfpsux 19,5,11
+ lfpsux 20,6,11
+ fpadd 26,4,5
+ fpadd 24,24,25
+ lfpsux 21,7,11
+ lfpsux 22,8,11
+ lfpsux 23,9,11
+ fpadd 27,6,7
+ fpadd 28,8,9
+
+ beq- 0,1f
+ li 10,48
+
+0:
+ dcbt 4,10
+ lfpsux 0,4,11
+ fpadd 24,24,26
+ lfpsux 1,5,11
+ lfpsux 2,6,11
+ fpadd 29,10,11
+ fpadd 27,27,28
+ lfpsux 3,7,11
+ lfpsux 4,8,11
+ lfpsux 5,9,11
+ dcbt 5,10
+ stfpsux 24,3,11
+ fpadd 24,12,13
+ fpadd 25,14,15
+
+ lfpsux 6,4,11
+ fpadd 27,27,29
+ lfpsux 7,5,11
+ lfpsux 8,6,11
+ dcbt 6,10
+ fpadd 26,16,17
+ fpadd 24,24,25
+ lfpsux 9,7,11
+ lfpsux 10,8,11
+ lfpsux 11,9,11
+ stfpsux 27,3,11
+ fpadd 27,18,19
+ fpadd 28,20,21
+
+ dcbt 7,10
+ lfpsux 12,4,11
+ fpadd 24,24,26
+ lfpsux 13,5,11
+ lfpsux 14,6,11
+ fpadd 29,22,23
+ fpadd 27,27,28
+ lfpsux 15,7,11
+ lfpsux 16,8,11
+ lfpsux 17,9,11
+ dcbt 8,10
+ stfpsux 24,3,11
+ fpadd 24,0,1
+ fpadd 25,2,3
+
+ lfpsux 18,4,11
+ fpadd 27,27,29
+ lfpsux 19,5,11
+ lfpsux 20,6,11
+ dcbt 9,10
+ fpadd 26,4,5
+ fpadd 24,24,25
+ lfpsux 21,7,11
+ lfpsux 22,8,11
+ lfpsux 23,9,11
+ stfpsux 27,3,11
+ fpadd 27,6,7
+ fpadd 28,8,9
+
+ bdnz 0b
+
+1:
+ fpadd 24,24,26
+ fpadd 29,10,11
+ fpadd 27,27,28
+ stfpsux 24,3,11
+ fpadd 24,12,13
+ fpadd 25,14,15
+
+ fpadd 27,27,29
+ fpadd 26,16,17
+ fpadd 24,24,25
+ stfpsux 27,3,11
+ fpadd 27,18,19
+ fpadd 28,20,21
+
+ fpadd 24,24,26
+ fpadd 29,22,23
+ fpadd 27,27,28
+ stfpsux 24,3,11
+
+ fpadd 27,27,29
+ stfpsux 27,3,11
+
+ li 0,16 # restore call-saved registers
+ lfpdx 29,0,1
+ lfpdux 28,1,0
+ lfpdux 27,1,0
+ lfpdux 26,1,0
+ lfpdux 25,1,0
+ lfpdux 24,1,0
+ lfpdux 23,1,0
+ lfpdux 22,1,0
+ lfpdux 21,1,0
+ lfpdux 20,1,0
+ lfpdux 19,1,0
+ lfpdux 18,1,0
+ lfpdux 17,1,0
+ lfpdux 16,1,0
+ lfpdux 15,1,0
+ lfpdux 14,1,0
+ addi 1,1,16
+ blr
+
+#endif
diff --git a/RTCP/Interface/src/CMakeLists.txt b/RTCP/Interface/src/CMakeLists.txt
index cb2cf20ff67b02c8d0c6f0ff2ad075aa54795708..40f0e69fd95c1b36ba982509c03308941ba063e5 100644
--- a/RTCP/Interface/src/CMakeLists.txt
+++ b/RTCP/Interface/src/CMakeLists.txt
@@ -2,7 +2,7 @@
include(LofarPackageVersion)
-lofar_add_library(interface
+set(interface_LIB_SRCS
Package__Version.cc
Allocator.cc
BeamCoordinates.cc
@@ -12,4 +12,16 @@ lofar_add_library(interface
Parset.cc
RSPTimeStamp.cc)
+# Maybe we shoud use LOFAR_COMPILER_SUITE, because ASM-BGP_COMPILER_WORKS
+# is not reliable: i.e.
+#if(LOFAR_COMPILER_SUITE STREQUAL BGPCN)
+if(CMAKE_ASM-BGP_COMPILER_WORKS)
+ list(APPEND interface_LIB_SRCS
+ BGPAsm.S)
+endif(CMAKE_ASM-BGP_COMPILER_WORKS)
+#endif(LOFAR_COMPILER_SUITE STREQUAL BGPCN)
+
+lofar_add_library(interface ${interface_LIB_SRCS})
+
lofar_add_bin_program(versioninterface versioninterface.cc)
+
diff --git a/RTCP/Interface/test/CMakeLists.txt b/RTCP/Interface/test/CMakeLists.txt
index 391e1a16d323a55db0ecee4751f9b3fa7735eef0..53e188beba7a6064ed65110ddc9f91027b7696d2 100644
--- a/RTCP/Interface/test/CMakeLists.txt
+++ b/RTCP/Interface/test/CMakeLists.txt
@@ -2,5 +2,6 @@
include(LofarCTest)
+lofar_add_test(tCorrelatedData tCorrelatedData.cc)
lofar_add_test(tSparseSet tSparseSet.cc)
lofar_add_test(tRSPTimeStamp tRSPTimeStamp.cc)
diff --git a/RTCP/Interface/test/tCorrelatedData.cc b/RTCP/Interface/test/tCorrelatedData.cc
new file mode 100644
index 0000000000000000000000000000000000000000..ded003c78f00b668efc71043d6c96214a9ef5a0a
--- /dev/null
+++ b/RTCP/Interface/test/tCorrelatedData.cc
@@ -0,0 +1,86 @@
+#include <lofar_config.h>
+
+#include <Common/Timer.h>
+
+#include <Interface/CorrelatedData.h>
+
+#include <cassert>
+#include <iostream>
+
+
+using namespace LOFAR;
+using namespace LOFAR::RTCP;
+using namespace std;
+
+int main(void)
+{
+ NSTimer timer("addition", true, false);
+
+ unsigned nr_maxsamples[] = { 255, 65535, 1000000 }; // encode using 1, 2, 4 bytes, respectively
+ unsigned nr_channels[] = { 1, 16, 64, 256 };
+ unsigned nr_stations[] = { 1, 2, 3, 4, 5, 24 };
+
+ for( unsigned s = 0; s < sizeof nr_maxsamples / sizeof nr_maxsamples[0]; ++s )
+ for( unsigned ch = 0; ch < sizeof nr_channels / sizeof nr_channels[0]; ++ch )
+ for( unsigned st = 0; st < sizeof nr_stations / sizeof nr_stations[0]; ++st ) {
+ unsigned ns = nr_maxsamples[s];
+ unsigned nch = nr_channels[ch];
+ unsigned nst = nr_stations[st];
+ unsigned nbl = nst * (nst + 1) / 2;
+
+ cout << nst << " stations (= " << nbl << " baselines), " << nch << " channels, " << ns << " samples" << endl;
+
+ // we will test whether data1 + data2 = data3
+ CorrelatedData data1(nst, nch, ns), data2(nst, nch, ns), data3(nst, nch, ns);
+
+ // initialise data
+ cout << "init" << endl;
+ unsigned n = 0;
+
+ for( unsigned i = 0; i < nbl; i++ ) {
+ for( unsigned j = 0; j < nch; j++ ) {
+ n++;
+
+ data1.setNrValidSamples(i, j, (n*1) % (ns/2));
+ data2.setNrValidSamples(i, j, (n*2) % (ns/2));
+ data3.setNrValidSamples(i, j, ((n*1) % (ns/2)) + ((n*2) % (ns/2)));
+
+ for( unsigned k = 0; k < 2; k++ ) {
+ for( unsigned l = 0; l < 2; l++ ) {
+ data1.visibilities[i][j][k][l] = 1 * ((i + j) * 10 + k * 2 + l);
+ data2.visibilities[i][j][k][l] = 1000 * ((i + j) * 10 + k * 2 + l);
+ data3.visibilities[i][j][k][l] = 1001 * ((i + j) * 10 + k * 2 + l);
+ }
+ }
+ }
+ }
+
+ // add
+ cout << "add" << endl;
+ timer.start();
+ data1 += data2;
+ timer.stop();
+
+ // verify
+ cout << "verify" << endl;
+ for( unsigned i = 0; i < nbl; i++ ) {
+ for( unsigned j = 0; j < nch; j++ ) {
+ //cout << data1.nrValidSamples(i, j) << " == " << data3.nrValidSamples(i, j) << endl;
+ assert(data1.nrValidSamples(i, j) == data3.nrValidSamples(i, j));
+
+ for( unsigned k = 0; k < 2; k++ ) {
+ for( unsigned l = 0; l < 2; l++ ) {
+ assert(
+ data1.visibilities[i][j][k][l] ==
+ data3.visibilities[i][j][k][l]
+ );
+ }
+ }
+ }
+ }
+
+ cout << "ok" << endl;
+ }
+
+ return 0;
+}
diff --git a/RTCP/Run/src/OLAP.parset b/RTCP/Run/src/OLAP.parset
index 2d586feaf842887921859515f83ca795fb8c427b..7e5e4860669831978de29881c93a57723180ad5a 100644
--- a/RTCP/Run/src/OLAP.parset
+++ b/RTCP/Run/src/OLAP.parset
@@ -37,10 +37,10 @@ PIC.Core.CS007HBA0.clockCorrectionTime = 7.913020e-06
PIC.Core.CS007HBA1.clockCorrectionTime = 7.913260e-06
PIC.Core.CS007HBA.clockCorrectionTime = 7.913140e-06
-PIC.Core.CS001LBA.clockCorrectionTime = 4.309154e-06
-PIC.Core.CS001HBA0.clockCorrectionTime = 4.309154e-06
-PIC.Core.CS001HBA1.clockCorrectionTime = 4.309154e-06
-PIC.Core.CS001HBA.clockCorrectionTime = 4.309154e-06
+PIC.Core.CS001LBA.clockCorrectionTime = 4.759754e-06
+PIC.Core.CS001HBA0.clockCorrectionTime = 4.759754e-06
+PIC.Core.CS001HBA1.clockCorrectionTime = 4.759754e-06
+PIC.Core.CS001HBA.clockCorrectionTime = 4.759754e-06
PIC.Core.CS011LBA.clockCorrectionTime = 7.55795e-06
PIC.Core.CS011HBA0.clockCorrectionTime = 7.55795e-06
@@ -52,80 +52,80 @@ PIC.Core.CS013HBA0.clockCorrectionTime = 1.639118e-05
PIC.Core.CS013HBA1.clockCorrectionTime = 1.639118e-05
PIC.Core.CS013HBA.clockCorrectionTime = 1.639118e-05
-PIC.Core.CS017LBA.clockCorrectionTime = 1.540943e-05
-PIC.Core.CS017HBA0.clockCorrectionTime = 1.540943e-05
-PIC.Core.CS017HBA1.clockCorrectionTime = 1.540943e-05
-PIC.Core.CS017HBA.clockCorrectionTime = 1.540943e-05
+PIC.Core.CS017LBA.clockCorrectionTime = 1.541095e-05
+PIC.Core.CS017HBA0.clockCorrectionTime = 1.541095e-05
+PIC.Core.CS017HBA1.clockCorrectionTime = 1.541095e-05
+PIC.Core.CS017HBA.clockCorrectionTime = 1.541095e-05
-PIC.Core.CS021LBA.clockCorrectionTime = 1.829614e-05
-PIC.Core.CS021HBA0.clockCorrectionTime = 1.829614e-05
-PIC.Core.CS021HBA1.clockCorrectionTime = 1.829614e-05
-PIC.Core.CS021HBA.clockCorrectionTime = 1.829614e-05
+PIC.Core.CS021LBA.clockCorrectionTime = 6.04963e-06
+PIC.Core.CS021HBA0.clockCorrectionTime = 6.04963e-06
+PIC.Core.CS021HBA1.clockCorrectionTime = 6.04963e-06
+PIC.Core.CS021HBA.clockCorrectionTime = 6.04963e-06
-PIC.Core.CS024LBA.clockCorrectionTime = 4.7597e-06
-PIC.Core.CS024HBA0.clockCorrectionTime = 4.7597e-06
-PIC.Core.CS024HBA1.clockCorrectionTime = 4.7597e-06
-PIC.Core.CS024HBA.clockCorrectionTime = 4.7597e-06
+PIC.Core.CS024LBA.clockCorrectionTime = 4.65818e-06
+PIC.Core.CS024HBA0.clockCorrectionTime = 4.65818e-06
+PIC.Core.CS024HBA1.clockCorrectionTime = 4.65818e-06
+PIC.Core.CS024HBA.clockCorrectionTime = 4.65818e-06
-PIC.Core.CS026LBA.clockCorrectionTime = 1.619986e-05
-PIC.Core.CS026HBA0.clockCorrectionTime = 1.619986e-05
-PIC.Core.CS026HBA1.clockCorrectionTime = 1.619986e-05
-PIC.Core.CS026HBA.clockCorrectionTime = 1.619986e-05
+PIC.Core.CS026LBA.clockCorrectionTime = 1.619876e-05
+PIC.Core.CS026HBA0.clockCorrectionTime = 1.619876e-05
+PIC.Core.CS026HBA1.clockCorrectionTime = 1.619876e-05
+PIC.Core.CS026HBA.clockCorrectionTime = 1.619876e-05
-PIC.Core.CS028LBA.clockCorrectionTime = 1.70614e-06
-PIC.Core.CS028HBA0.clockCorrectionTime = 1.70614e-06
-PIC.Core.CS028HBA1.clockCorrectionTime = 1.70614e-06
-PIC.Core.CS028HBA.clockCorrectionTime = 1.70614e-06
+PIC.Core.CS028LBA.clockCorrectionTime = 1.6962571e-05
+PIC.Core.CS028HBA0.clockCorrectionTime = 1.6962571e-05
+PIC.Core.CS028HBA1.clockCorrectionTime = 1.6962571e-05
+PIC.Core.CS028HBA.clockCorrectionTime = 1.6962571e-05
PIC.Core.CS030LBA.clockCorrectionTime = 9.7160576e-06
PIC.Core.CS030HBA0.clockCorrectionTime = 9.7160576e-06
PIC.Core.CS030HBA1.clockCorrectionTime = 9.7160576e-06
PIC.Core.CS030HBA.clockCorrectionTime = 9.7160576e-06
-PIC.Core.CS031LBA.clockCorrectionTime = 6.471391e-06
-PIC.Core.CS031HBA0.clockCorrectionTime = 6.471391e-06
-PIC.Core.CS031HBA1.clockCorrectionTime = 6.471391e-06
-PIC.Core.CS031HBA.clockCorrectionTime = 6.471391e-06
+PIC.Core.CS031LBA.clockCorrectionTime = 6.370090e-06
+PIC.Core.CS031HBA0.clockCorrectionTime = 6.370090e-06
+PIC.Core.CS031HBA1.clockCorrectionTime = 6.370090e-06
+PIC.Core.CS031HBA.clockCorrectionTime = 6.370090e-06
PIC.Core.CS032LBA.clockCorrectionTime = 8.546815e-06
PIC.Core.CS032HBA0.clockCorrectionTime = 8.546815e-06
PIC.Core.CS032HBA1.clockCorrectionTime = 8.546815e-06
PIC.Core.CS032HBA.clockCorrectionTime = 8.546815e-06
-PIC.Core.CS101LBA.clockCorrectionTime = 1.525453e-05
-PIC.Core.CS101HBA0.clockCorrectionTime = 1.525453e-05
-PIC.Core.CS101HBA1.clockCorrectionTime = 1.525453e-05
-PIC.Core.CS101HBA.clockCorrectionTime = 1.525453e-05
+PIC.Core.CS101LBA.clockCorrectionTime = 1.5165101e-05
+PIC.Core.CS101HBA0.clockCorrectionTime = 1.5165101e-05
+PIC.Core.CS101HBA1.clockCorrectionTime = 1.5165101e-05
+PIC.Core.CS101HBA.clockCorrectionTime = 1.5165101e-05
-PIC.Core.CS103LBA.clockCorrectionTime = 3.560284e-05
-PIC.Core.CS103HBA0.clockCorrectionTime = 3.560284e-05
-PIC.Core.CS103HBA1.clockCorrectionTime = 3.560284e-05
-PIC.Core.CS103HBA.clockCorrectionTime = 3.560284e-05
+PIC.Core.CS103LBA.clockCorrectionTime = 3.5500922e-05
+PIC.Core.CS103HBA0.clockCorrectionTime = 3.5500922e-05
+PIC.Core.CS103HBA1.clockCorrectionTime = 3.5500922e-05
+PIC.Core.CS103HBA.clockCorrectionTime = 3.5500922e-05
PIC.Core.CS201LBA.clockCorrectionTime = 1.744858e-05
PIC.Core.CS201HBA0.clockCorrectionTime = 1.744858e-05
PIC.Core.CS201HBA1.clockCorrectionTime = 1.744858e-05
PIC.Core.CS201HBA.clockCorrectionTime = 1.744858e-05
-PIC.Core.CS301LBA.clockCorrectionTime = 7.687733e-06
-PIC.Core.CS301HBA0.clockCorrectionTime = 7.687733e-06
-PIC.Core.CS301HBA1.clockCorrectionTime = 7.687733e-06
-PIC.Core.CS301HBA.clockCorrectionTime = 7.687733e-06
+PIC.Core.CS301LBA.clockCorrectionTime = 7.692091e-06
+PIC.Core.CS301HBA0.clockCorrectionTime = 7.692091e-06
+PIC.Core.CS301HBA1.clockCorrectionTime = 7.692091e-06
+PIC.Core.CS301HBA.clockCorrectionTime = 7.692091e-06
-PIC.Core.CS302LBA.clockCorrectionTime = 1.494415e-05
-PIC.Core.CS302HBA0.clockCorrectionTime = 1.494415e-05
-PIC.Core.CS302HBA1.clockCorrectionTime = 1.494415e-05
-PIC.Core.CS302HBA.clockCorrectionTime = 1.494415e-05
+PIC.Core.CS302LBA.clockCorrectionTime = 1.5062785e-05
+PIC.Core.CS302HBA0.clockCorrectionTime = 1.5062785e-05
+PIC.Core.CS302HBA1.clockCorrectionTime = 1.5062785e-05
+PIC.Core.CS302HBA.clockCorrectionTime = 1.5062785e-05
PIC.Core.CS401LBA.clockCorrectionTime = 8.051870e-06
PIC.Core.CS401HBA0.clockCorrectionTime = 8.057504e-06
PIC.Core.CS401HBA1.clockCorrectionTime = 8.057770e-06
PIC.Core.CS401HBA.clockCorrectionTime = 8.057637e-06
-PIC.Core.CS501LBA.clockCorrectionTime = 1.65833e-05
-PIC.Core.CS501HBA0.clockCorrectionTime = 1.65833e-05
-PIC.Core.CS501HBA1.clockCorrectionTime = 1.65833e-05
-PIC.Core.CS501HBA.clockCorrectionTime = 1.65833e-05
+PIC.Core.CS501LBA.clockCorrectionTime = 1.65842e-05
+PIC.Core.CS501HBA0.clockCorrectionTime = 1.65842e-05
+PIC.Core.CS501HBA1.clockCorrectionTime = 1.65842e-05
+PIC.Core.CS501HBA.clockCorrectionTime = 1.65842e-05
#
# Stations outside of the core (no correction needed)