test: benchmarking pass

code/jasmin/mlkem_avx2/Makefile

@@ -49,7 +49,7 @@ TESTS := \
 
 test: $(TESTS)
 
-speed: test/speed_indcpa test/speed_mlkem
+speed: test/speed_mlkem
 
 # --
 
@@ -161,16 +161,12 @@ run-tests: compile-tests
 
 # --
 
-test/speed_indcpa: test/speed_indcpa.c $(HEADERS) $(C_SOURCES) $(S_INC) jspeed.s
+test/speed_mlkem: test/speed_mlkem.c $(HEADERS) $(C_SOURCES) $(S_INC) $(RANDOMBYTES) jspeed.s
 	$(CC) $(CFLAGS) -o $@ $(C_SOURCES) $(RANDOMBYTES) jspeed.s $<
 
-test/speed_mlkem: test/speed_mlkem.c $(HEADERS) $(C_SOURCES) $(S_INC) jspeed.s
-	$(CC) $(CFLAGS) -o $@ $(C_SOURCES) $(RANDOMBYTES) jspeed.s $<
-
-compile-speed: test/speed_indcpa test/speed_mlkem
+compile-speed: test/speed_mlkem
 
 run-speed: compile-speed
-	./test/speed_indcpa
 	./test/speed_mlkem
 
 # --
@@ -191,7 +187,6 @@ clean:
 	-rm -f test/test_fips202
 	-rm -f test/test_indcpa
 	-rm -f test/test_kem
-	-rm -f test/speed_indcpa
 	-rm -f test/speed_mlkem
 ifeq ($(OS),Darwin)
 	-rm -r -f test/*.dSYM

code/jasmin/mlkem_avx2/cycles.jinc

@@ -0,0 +1,23 @@
+inline fn tsc() -> stack u32[2]
+{
+  reg u64 l h;
+  stack u32[2] t;
+
+  h, l = #RDTSC();
+
+  t[0] = (32u) l;
+  t[1] = (32u) h; 
+
+  return t;
+}
+
+inline fn cycles(stack u32[2] start end) -> reg u64
+{
+  reg u64 t;
+
+  t = end[u64 0];
+  t -= start[u64 0];
+
+  return t;
+}
+