Use this branch to profile

.gitignore

@@ -6,3 +6,17 @@
 *.dSYM
 minimap2
 mappy.c
+data
+.vscode/**
+test.sam
+*.sam
+Log/**
+debug/**
+verf
+trace
+ncu
+nsys
+*_output*
+workloads
+.cmake/**
+.depend

.gitmodules

@@ -1,3 +1,6 @@
 [submodule "lib/simde"]
 	path = lib/simde
 	url = https://github.com/nemequ/simde.git
+[submodule "cJSON"]
+	path = cJSON
+	url = https://github.com/DaveGamble/cJSON.git

LICENSE.txt

@@ -2,6 +2,7 @@ The MIT License
 
 Copyright (c) 2018-     Dana-Farber Cancer Institute
               2017-2018 Broad Institute, Inc.
+              2022      Advanced Micro Devices, Inc.
 
 Permission is hereby granted, free of charge, to any person obtaining
 a copy of this software and associated documentation files (the

Makefile

@@ -1,12 +1,16 @@
-CFLAGS=		-g -Wall -O2 -Wc++-compat #-Wextra
-CPPFLAGS=	-DHAVE_KALLOC
-INCLUDES=
+CFLAGS=		 -O2 -g -DNDEBUG
+CDEBUG_FLAGS= -g -O2 #-Wall -Wextra -Wno-unused-parameter -Wno-unused-variable -Wno-sign-compare -Wno-unused-function -Wno-c++17-extensions -Wno-\#warnings #-O0 -DNDEBUG
+CPPFLAGS=	-DHAVE_KALLOC -D__AMD_SPLIT_KERNELS__ # -Wno-unused-but-set-variable -Wno-unused-variable
+CPPFLAGS+= 	$(if $(MAX_MICRO_BATCH),-DMAX_MICRO_BATCH=\($(MAX_MICRO_BATCH)\))
+INCLUDES=	-I .
 OBJS=		kthread.o kalloc.o misc.o bseq.o sketch.o sdust.o options.o index.o \
 			lchain.o align.o hit.o seed.o map.o format.o pe.o esterr.o splitidx.o \
 			ksw2_ll_sse.o
-PROG=		minimap2
+# PROG=		minimap2-zerobranch-debug
+# PROG=		minimap2-nobalance-debug
+PROG=		minimap2$(SUFFIX)
 PROG_EXTRA=	sdust minimap2-lite
-LIBS=		-lm -lz -lpthread
+LIBS=		-lm -lz -lpthread 
 
 ifeq ($(arm_neon),) # if arm_neon is not defined
 ifeq ($(sse2only),) # if sse2only is not defined
@@ -34,7 +38,21 @@ ifneq ($(tsan),)
 	LIBS+=-fsanitize=thread
 endif
 
-.PHONY:all extra clean depend
+
+# turn on debug flags 
+ifeq ($(DEBUG),info) 
+	CFLAGS += -DDEBUG_PRINT
+endif
+ifeq ($(DEBUG), analyze) 
+	CFLAGS += $(CDEBUG_FLAGS) 
+	CFLAGS += -DDEBUG_CHECK -DDEBUG_PRINT
+endif
+ifeq ($(DEBUG), verbose)
+	CFLAGS += $(CDEBUG_FLAGS) 
+	CFLAGS += -DDEBUG_CHECK -DDEBUG_PRINT -DDEBUG_VERBOSE
+endif
+
+.PHONY:all extra clean depend # profile
 .SUFFIXES:.c .o
 
 .c.o:
@@ -44,14 +62,25 @@ all:$(PROG)
 
 extra:all $(PROG_EXTRA)
 
-minimap2:main.o libminimap2.a
-		$(CC) $(CFLAGS) main.o -o $@ -L. -lminimap2 $(LIBS)
+# build cJSON
+CJSON_OBJ= 	cJSON/cJSON.o
+INCLUDES += -I cJSON
+$(CJSON_OBJ): 
+	make -C cJSON
+
+# build kernel objs
+include gpu/gpu.mk
+
+
+# compile with nvcc/hipcc
+$(PROG):main.o libminimap2.a
+		$(GPU_CC) $(CFLAGS) $(GPU_FLAGS) main.o -o $@ -L. -lminimap2 $(LIBS)
 
 minimap2-lite:example.o libminimap2.a
-		$(CC) $(CFLAGS) $< -o $@ -L. -lminimap2 $(LIBS)
+		$(GPU_CC) $(CFLAGS)  $(GPU_FLAGS) $< -o $@ -L. -lminimap2 $(LIBS)
 
-libminimap2.a:$(OBJS)
-		$(AR) -csru $@ $(OBJS)
+libminimap2.a:$(OBJS) $(CU_OBJS) $(CJSON_OBJ)
+		$(AR) -csru $@ $^
 
 sdust:sdust.c kalloc.o kalloc.h kdq.h kvec.h kseq.h ketopt.h sdust.h
 		$(CC) -D_SDUST_MAIN $(CFLAGS) $< kalloc.o -o $@ -lz
@@ -97,7 +126,7 @@ ksw2_exts2_neon.o:ksw2_exts2_sse.c ksw2.h kalloc.h
 
 # other non-file targets
 
-clean:
+clean: cleangpu
 		rm -fr gmon.out *.o a.out $(PROG) $(PROG_EXTRA) *~ *.a *.dSYM build dist mappy*.so mappy.c python/mappy.c mappy.egg*
 
 depend:
@@ -129,4 +158,4 @@ pe.o: mmpriv.h minimap.h bseq.h kseq.h kvec.h kalloc.h ksort.h
 sdust.o: kalloc.h kdq.h kvec.h sdust.h
 seed.o: mmpriv.h minimap.h bseq.h kseq.h kalloc.h ksort.h
 sketch.o: kvec.h kalloc.h mmpriv.h minimap.h bseq.h kseq.h
-splitidx.o: mmpriv.h minimap.h bseq.h kseq.h
+splitidx.o: mmpriv.h minimap.h bseq.h kseq.h

a6000_config.json

@@ -0,0 +1,27 @@
+{
+    "//config is for": "a6000. Fits one batch + 5% x 4 long buffer avg_read_n 10k",
+    "num_streams": 1,
+    "min_n": 512, 
+    "//min_n": "queries with less anchors will be handled on cpu",
+    "long_seg_buffer_size": 258880000,
+    "max_total_n": 893440000, 
+    "max_read": 893440,
+    "avg_read_n": 20000,
+    "//avg_read_n": "expect average number of anchors per read, not used if max_total_n and max_read are specified",
+    "range_kernel": {
+        "blockdim": 512,
+        "cut_check_anchors": 10,
+        "//cut_check_anchors": "Number of anchors to check to attemp a cut",
+        "anchor_per_block": 32768,
+        "//anchor_per_block": "Number of anchors each block handle. Must be int * blockdim"
+    },
+    "score_kernel": {
+        "short_blockdim": 64,
+        "long_blockdim": 64,
+        "mid_blockdim": 64,
+        "//blockdim config": "options are not used: static config specified at compile time (make ... LONG_BLOCK_SIZE=1024)",
+        "short_griddim": 2688,
+        "long_griddim": 1024,
+        "mid_griddim": 2688
+    }
+}

aac_config.json

@@ -0,0 +1,27 @@
+{
+    "//config is for": "aac cloud. Fits one batch + 5% x 4 long buffer avg_read_n 10k",
+    "num_streams": 1,
+    "min_n": 512, 
+    "//min_n": "queries with less anchors will be handled on cpu",
+    "long_seg_buffer_size": 1117376000,
+    "max_total_n": 2036880000, 
+    "max_read": 2036880,
+    "avg_read_n": 20000,
+    "//avg_read_n": "expect average number of anchors per read, not used if max_total_n and max_read are specified",
+    "range_kernel": {
+        "blockdim": 512,
+        "cut_check_anchors": 10,
+        "//cut_check_anchors": "Number of anchors to check to attemp a cut",
+        "anchor_per_block": 32768,
+        "//anchor_per_block": "Number of anchors each block handle. Must be int * blockdim"
+    },
+    "score_kernel": {
+        "short_blockdim": 64,
+        "long_blockdim": 64,
+        "mid_blockdim": 64,
+        "//blockdim config": "options are not used: static config specified at compile time (make ... LONG_BLOCK_SIZE=1024)",
+        "short_griddim": 16128,
+        "long_griddim": 208,
+        "mid_griddim": 16128
+    }
+}

gfx1030.json

@@ -0,0 +1,28 @@
+{
+    "//config is for": "AMD Radeon RX 6800 XT on amdxfx. Fits one batch + 5% x 4 long buffer avg_read_n 10k",
+    "num_streams": 1,
+    "min_n": 512, 
+    "//min_n": "queries with less anchors will be handled on cpu",
+    "long_seg_buffer_size": 100000000,
+    "max_total_n": 493440000, 
+    "max_read": 493440,
+    "avg_read_n": 20000,
+    "//avg_read_n": "expect average number of anchors per read, not used if max_total_n and max_read are specified",
+    "range_kernel": {
+        "blockdim": 512,
+        "cut_check_anchors": 10,
+        "//cut_check_anchors": "Number of anchors to check to attemp a cut",
+        "anchor_per_block": 32768,
+        "//anchor_per_block": "Number of anchors each block handle. Must be int * blockdim"
+    },
+    "score_kernel": {
+        "micro_batch": 4,
+        "mid_blockdim": 512,
+        "//blockdim config": "options are not used: static config specified at compile time (make ... LONG_BLOCK_SIZE=1024)",
+        "short_griddim": 2688,
+        "long_griddim": 144,
+        "mid_griddim": 2688,
+        "long_seg_cutoff": 20,
+        "mid_seg_cutoff": 3
+    }
+}