Elliot/measurements

measurements/variant_avx512_1/x86/1.log

@@ -0,0 +1,42 @@
+rm -f main_* sizes run_tests *.o *.s variants/*.s variants/*.o headers/*.s headers/*.o
+/usr/bin/g++ main.cpp ./variants/variant_avx512_1.cpp -o main_avx512_1 -O3 -march=native -std=c++23 -Wall -Wextra -flto -lsleef -DFLOPCOUNT=0 -DRANDOM_SEED=0 -DROOFLINE=0 -DBENCHMARK=1
+=== N_P=50 N_ITER=1000 ===
+W(n) per iteration: 184500.00 flops
+Q(n) per iteration: 6512 bytes
+I(n): 28.33 flops/byte
+W(n): 184500000.00 flops
+T(n): 63060640.00 cycles
+P(n): 2.93 flops/cycle
+
+=== N_P=100 N_ITER=1000 ===
+W(n) per iteration: 729000.00 flops
+Q(n) per iteration: 12912 bytes
+I(n): 56.46 flops/byte
+W(n): 729000000.00 flops
+T(n): 215114100.00 cycles
+P(n): 3.39 flops/cycle
+
+=== N_P=200 N_ITER=1000 ===
+W(n) per iteration: 2898000.00 flops
+Q(n) per iteration: 25712 bytes
+I(n): 112.71 flops/byte
+W(n): 2898000000.00 flops
+T(n): 744500640.00 cycles
+P(n): 3.89 flops/cycle
+
+=== N_P=500 N_ITER=1000 ===
+W(n) per iteration: 18045000.00 flops
+Q(n) per iteration: 64112 bytes
+I(n): 281.46 flops/byte
+W(n): 18045000000.00 flops
+T(n): 4480783600.00 cycles
+P(n): 4.03 flops/cycle
+
+=== N_P=1000 N_ITER=1000 ===
+W(n) per iteration: 72090000.00 flops
+Q(n) per iteration: 128112 bytes
+I(n): 562.71 flops/byte
+W(n): 72090000000.00 flops
+T(n): 16918374740.00 cycles
+P(n): 4.26 flops/cycle
+

measurements/variant_avx512_1/x86/2.log

@@ -0,0 +1,42 @@
+rm -f main_* sizes run_tests *.o *.s variants/*.s variants/*.o headers/*.s headers/*.o
+/usr/bin/g++ main.cpp ./variants/variant_avx512_1.cpp -o main_avx512_1 -O3 -march=native -std=c++23 -Wall -Wextra -flto -lsleef -DFLOPCOUNT=0 -DRANDOM_SEED=0 -DROOFLINE=0 -DBENCHMARK=1
+=== N_P=50 N_ITER=1000 ===
+W(n) per iteration: 184500.00 flops
+Q(n) per iteration: 6512 bytes
+I(n): 28.33 flops/byte
+W(n): 184500000.00 flops
+T(n): 62465750.00 cycles
+P(n): 2.95 flops/cycle
+
+=== N_P=100 N_ITER=1000 ===
+W(n) per iteration: 729000.00 flops
+Q(n) per iteration: 12912 bytes
+I(n): 56.46 flops/byte
+W(n): 729000000.00 flops
+T(n): 213111220.00 cycles
+P(n): 3.42 flops/cycle
+
+=== N_P=200 N_ITER=1000 ===
+W(n) per iteration: 2898000.00 flops
+Q(n) per iteration: 25712 bytes
+I(n): 112.71 flops/byte
+W(n): 2898000000.00 flops
+T(n): 749193840.00 cycles
+P(n): 3.87 flops/cycle
+
+=== N_P=500 N_ITER=1000 ===
+W(n) per iteration: 18045000.00 flops
+Q(n) per iteration: 64112 bytes
+I(n): 281.46 flops/byte
+W(n): 18045000000.00 flops
+T(n): 4481202620.00 cycles
+P(n): 4.03 flops/cycle
+
+=== N_P=1000 N_ITER=1000 ===
+W(n) per iteration: 72090000.00 flops
+Q(n) per iteration: 128112 bytes
+I(n): 562.71 flops/byte
+W(n): 72090000000.00 flops
+T(n): 17063409500.00 cycles
+P(n): 4.22 flops/cycle
+

measurements/variant_avx512_1/x86/3.log

@@ -0,0 +1,42 @@
+rm -f main_* sizes run_tests *.o *.s variants/*.s variants/*.o headers/*.s headers/*.o
+/usr/bin/g++ main.cpp ./variants/variant_avx512_1.cpp -o main_avx512_1 -O3 -march=native -std=c++23 -Wall -Wextra -flto -lsleef -DFLOPCOUNT=0 -DRANDOM_SEED=0 -DROOFLINE=0 -DBENCHMARK=1
+=== N_P=50 N_ITER=1000 ===
+W(n) per iteration: 184500.00 flops
+Q(n) per iteration: 6512 bytes
+I(n): 28.33 flops/byte
+W(n): 184500000.00 flops
+T(n): 62186140.00 cycles
+P(n): 2.97 flops/cycle
+
+=== N_P=100 N_ITER=1000 ===
+W(n) per iteration: 729000.00 flops
+Q(n) per iteration: 12912 bytes
+I(n): 56.46 flops/byte
+W(n): 729000000.00 flops
+T(n): 214174540.00 cycles
+P(n): 3.40 flops/cycle
+
+=== N_P=200 N_ITER=1000 ===
+W(n) per iteration: 2898000.00 flops
+Q(n) per iteration: 25712 bytes
+I(n): 112.71 flops/byte
+W(n): 2898000000.00 flops
+T(n): 739779420.00 cycles
+P(n): 3.92 flops/cycle
+
+=== N_P=500 N_ITER=1000 ===
+W(n) per iteration: 18045000.00 flops
+Q(n) per iteration: 64112 bytes
+I(n): 281.46 flops/byte
+W(n): 18045000000.00 flops
+T(n): 4486436580.00 cycles
+P(n): 4.02 flops/cycle
+
+=== N_P=1000 N_ITER=1000 ===
+W(n) per iteration: 72090000.00 flops
+Q(n) per iteration: 128112 bytes
+I(n): 562.71 flops/byte
+W(n): 72090000000.00 flops
+T(n): 16950058060.00 cycles
+P(n): 4.25 flops/cycle
+

measurements/variant_baseline/x86/1.log

@@ -0,0 +1,41 @@
+rm -f main_* sizes run_tests *.o *.s
+/usr/bin/g++ -O3 -march=native -std=c++23 -Wall -Wextra -flto -DFLOPCOUNT=0 -DRANDOM_SEED=0 -DROOFLINE=0 -DBENCHMARK=1 main.cpp ./variants/variant_baseline.cpp -o main_baseline
+=== N_P=50 N_ITER=1000 ===
+W(n) per iteration: 184500.00 flops
+Q(n) per iteration: 6512 bytes
+I(n): 28.33 flops/byte
+W(n): 184500000.00 flops
+T(n): 149741420.00 cycles
+P(n): 1.23 flops/cycle
+
+=== N_P=100 N_ITER=1000 ===
+W(n) per iteration: 729000.00 flops
+Q(n) per iteration: 12912 bytes
+I(n): 56.46 flops/byte
+W(n): 729000000.00 flops
+T(n): 590173880.00 cycles
+P(n): 1.24 flops/cycle
+
+=== N_P=200 N_ITER=1000 ===
+W(n) per iteration: 2898000.00 flops
+Q(n) per iteration: 25712 bytes
+I(n): 112.71 flops/byte
+W(n): 2898000000.00 flops
+T(n): 2394306640.00 cycles
+P(n): 1.21 flops/cycle
+
+=== N_P=500 N_ITER=1000 ===
+W(n) per iteration: 18045000.00 flops
+Q(n) per iteration: 64112 bytes
+I(n): 281.46 flops/byte
+W(n): 18045000000.00 flops
+T(n): 15758554280.00 cycles
+P(n): 1.15 flops/cycle
+
+=== N_P=1000 N_ITER=1000 ===
+W(n) per iteration: 72090000.00 flops
+Q(n) per iteration: 128112 bytes
+I(n): 562.71 flops/byte
+W(n): 72090000000.00 flops
+T(n): 61911299420.00 cycles
+P(n): 1.16 flops/cycle

measurements/variant_baseline/x86/2.log

@@ -0,0 +1,41 @@
+rm -f main_* sizes run_tests *.o *.s
+/usr/bin/g++ -O3 -march=native -std=c++23 -Wall -Wextra -flto -DFLOPCOUNT=0 -DRANDOM_SEED=0 -DROOFLINE=0 -DBENCHMARK=1 main.cpp ./variants/variant_baseline.cpp -o main_baseline
+=== N_P=50 N_ITER=1000 ===
+W(n) per iteration: 184500.00 flops
+Q(n) per iteration: 6512 bytes
+I(n): 28.33 flops/byte
+W(n): 184500000.00 flops
+T(n): 145460860.00 cycles
+P(n): 1.27 flops/cycle
+
+=== N_P=100 N_ITER=1000 ===
+W(n) per iteration: 729000.00 flops
+Q(n) per iteration: 12912 bytes
+I(n): 56.46 flops/byte
+W(n): 729000000.00 flops
+T(n): 589429960.00 cycles
+P(n): 1.24 flops/cycle
+
+=== N_P=200 N_ITER=1000 ===
+W(n) per iteration: 2898000.00 flops
+Q(n) per iteration: 25712 bytes
+I(n): 112.71 flops/byte
+W(n): 2898000000.00 flops
+T(n): 2395474680.00 cycles
+P(n): 1.21 flops/cycle
+
+=== N_P=500 N_ITER=1000 ===
+W(n) per iteration: 18045000.00 flops
+Q(n) per iteration: 64112 bytes
+I(n): 281.46 flops/byte
+W(n): 18045000000.00 flops
+T(n): 15694979780.00 cycles
+P(n): 1.15 flops/cycle
+
+=== N_P=1000 N_ITER=1000 ===
+W(n) per iteration: 72090000.00 flops
+Q(n) per iteration: 128112 bytes
+I(n): 562.71 flops/byte
+W(n): 72090000000.00 flops
+T(n): 61883782540.00 cycles
+P(n): 1.16 flops/cycle

measurements/variant_baseline/x86/3.log

@@ -0,0 +1,41 @@
+rm -f main_* sizes run_tests *.o *.s
+/usr/bin/g++ -O3 -march=native -std=c++23 -Wall -Wextra -flto -DFLOPCOUNT=0 -DRANDOM_SEED=0 -DROOFLINE=0 -DBENCHMARK=1 main.cpp ./variants/variant_baseline.cpp -o main_baseline
+=== N_P=50 N_ITER=1000 ===
+W(n) per iteration: 184500.00 flops
+Q(n) per iteration: 6512 bytes
+I(n): 28.33 flops/byte
+W(n): 184500000.00 flops
+T(n): 147463240.00 cycles
+P(n): 1.25 flops/cycle
+
+=== N_P=100 N_ITER=1000 ===
+W(n) per iteration: 729000.00 flops
+Q(n) per iteration: 12912 bytes
+I(n): 56.46 flops/byte
+W(n): 729000000.00 flops
+T(n): 583811800.00 cycles
+P(n): 1.25 flops/cycle
+
+=== N_P=200 N_ITER=1000 ===
+W(n) per iteration: 2898000.00 flops
+Q(n) per iteration: 25712 bytes
+I(n): 112.71 flops/byte
+W(n): 2898000000.00 flops
+T(n): 2406765140.00 cycles
+P(n): 1.20 flops/cycle
+
+=== N_P=500 N_ITER=1000 ===
+W(n) per iteration: 18045000.00 flops
+Q(n) per iteration: 64112 bytes
+I(n): 281.46 flops/byte
+W(n): 18045000000.00 flops
+T(n): 15769461780.00 cycles
+P(n): 1.14 flops/cycle
+
+=== N_P=1000 N_ITER=1000 ===
+W(n) per iteration: 72090000.00 flops
+Q(n) per iteration: 128112 bytes
+I(n): 562.71 flops/byte
+W(n): 72090000000.00 flops
+T(n): 61870155280.00 cycles
+P(n): 1.17 flops/cycle

measurements/variant_branchless/x86/1.log

@@ -0,0 +1,41 @@
+rm -f main_* sizes run_tests *.o *.s
+/usr/bin/g++ -O3 -march=native -std=c++23 -Wall -Wextra -flto -DFLOPCOUNT=0 -DRANDOM_SEED=0 -DROOFLINE=0 -DBENCHMARK=1 main.cpp ./variants/variant_branchless.cpp -o main_branchless
+=== N_P=50 N_ITER=1000 ===
+W(n) per iteration: 184500.00 flops
+Q(n) per iteration: 6512 bytes
+I(n): 28.33 flops/byte
+W(n): 184500000.00 flops
+T(n): 164628520.00 cycles
+P(n): 1.12 flops/cycle
+
+=== N_P=100 N_ITER=1000 ===
+W(n) per iteration: 729000.00 flops
+Q(n) per iteration: 12912 bytes
+I(n): 56.46 flops/byte
+W(n): 729000000.00 flops
+T(n): 668671300.00 cycles
+P(n): 1.09 flops/cycle
+
+=== N_P=200 N_ITER=1000 ===
+W(n) per iteration: 2898000.00 flops
+Q(n) per iteration: 25712 bytes
+I(n): 112.71 flops/byte
+W(n): 2898000000.00 flops
+T(n): 2797909140.00 cycles
+P(n): 1.04 flops/cycle
+
+=== N_P=500 N_ITER=1000 ===
+W(n) per iteration: 18045000.00 flops
+Q(n) per iteration: 64112 bytes
+I(n): 281.46 flops/byte
+W(n): 18045000000.00 flops
+T(n): 18473174460.00 cycles
+P(n): 0.98 flops/cycle
+
+=== N_P=1000 N_ITER=1000 ===
+W(n) per iteration: 72090000.00 flops
+Q(n) per iteration: 128112 bytes
+I(n): 562.71 flops/byte
+W(n): 72090000000.00 flops
+T(n): 72557284580.00 cycles
+P(n): 0.99 flops/cycle

measurements/variant_branchless/x86/2.log

@@ -0,0 +1,41 @@
+rm -f main_* sizes run_tests *.o *.s
+/usr/bin/g++ -O3 -march=native -std=c++23 -Wall -Wextra -flto -DFLOPCOUNT=0 -DRANDOM_SEED=0 -DROOFLINE=0 -DBENCHMARK=1 main.cpp ./variants/variant_branchless.cpp -o main_branchless
+=== N_P=50 N_ITER=1000 ===
+W(n) per iteration: 184500.00 flops
+Q(n) per iteration: 6512 bytes
+I(n): 28.33 flops/byte
+W(n): 184500000.00 flops
+T(n): 163598600.00 cycles
+P(n): 1.13 flops/cycle
+
+=== N_P=100 N_ITER=1000 ===
+W(n) per iteration: 729000.00 flops
+Q(n) per iteration: 12912 bytes
+I(n): 56.46 flops/byte
+W(n): 729000000.00 flops
+T(n): 672653360.00 cycles
+P(n): 1.08 flops/cycle
+
+=== N_P=200 N_ITER=1000 ===
+W(n) per iteration: 2898000.00 flops
+Q(n) per iteration: 25712 bytes
+I(n): 112.71 flops/byte
+W(n): 2898000000.00 flops
+T(n): 2815032380.00 cycles
+P(n): 1.03 flops/cycle
+
+=== N_P=500 N_ITER=1000 ===
+W(n) per iteration: 18045000.00 flops
+Q(n) per iteration: 64112 bytes
+I(n): 281.46 flops/byte
+W(n): 18045000000.00 flops
+T(n): 18559971420.00 cycles
+P(n): 0.97 flops/cycle
+
+=== N_P=1000 N_ITER=1000 ===
+W(n) per iteration: 72090000.00 flops
+Q(n) per iteration: 128112 bytes
+I(n): 562.71 flops/byte
+W(n): 72090000000.00 flops
+T(n): 72285399260.00 cycles
+P(n): 1.00 flops/cycle

measurements/variant_branchless/x86/3.log

@@ -0,0 +1,41 @@
+rm -f main_* sizes run_tests *.o *.s
+/usr/bin/g++ -O3 -march=native -std=c++23 -Wall -Wextra -flto -DFLOPCOUNT=0 -DRANDOM_SEED=0 -DROOFLINE=0 -DBENCHMARK=1 main.cpp ./variants/variant_branchless.cpp -o main_branchless
+=== N_P=50 N_ITER=1000 ===
+W(n) per iteration: 184500.00 flops
+Q(n) per iteration: 6512 bytes
+I(n): 28.33 flops/byte
+W(n): 184500000.00 flops
+T(n): 165200000.00 cycles
+P(n): 1.12 flops/cycle
+
+=== N_P=100 N_ITER=1000 ===
+W(n) per iteration: 729000.00 flops
+Q(n) per iteration: 12912 bytes
+I(n): 56.46 flops/byte
+W(n): 729000000.00 flops
+T(n): 686277160.00 cycles
+P(n): 1.06 flops/cycle
+
+=== N_P=200 N_ITER=1000 ===
+W(n) per iteration: 2898000.00 flops
+Q(n) per iteration: 25712 bytes
+I(n): 112.71 flops/byte
+W(n): 2898000000.00 flops
+T(n): 2805241940.00 cycles
+P(n): 1.03 flops/cycle
+
+=== N_P=500 N_ITER=1000 ===
+W(n) per iteration: 18045000.00 flops
+Q(n) per iteration: 64112 bytes
+I(n): 281.46 flops/byte
+W(n): 18045000000.00 flops
+T(n): 18557266380.00 cycles
+P(n): 0.97 flops/cycle
+
+=== N_P=1000 N_ITER=1000 ===
+W(n) per iteration: 72090000.00 flops
+Q(n) per iteration: 128112 bytes
+I(n): 562.71 flops/byte
+W(n): 72090000000.00 flops
+T(n): 72552877660.00 cycles
+P(n): 0.99 flops/cycle

measurements/variant_branchless_sqrtfree/x86/1.log

@@ -0,0 +1,41 @@
+rm -f main_* sizes run_tests *.o *.s
+/usr/bin/g++ -O3 -march=native -std=c++23 -Wall -Wextra -flto -DFLOPCOUNT=0 -DRANDOM_SEED=0 -DROOFLINE=0 -DBENCHMARK=1 main.cpp ./variants/variant_branchless_sqrtfree.cpp -o main_branchless_sqrtfree
+=== N_P=50 N_ITER=1000 ===
+W(n) per iteration: 184500.00 flops
+Q(n) per iteration: 6512 bytes
+I(n): 28.33 flops/byte
+W(n): 184500000.00 flops
+T(n): 160027720.00 cycles
+P(n): 1.15 flops/cycle
+
+=== N_P=100 N_ITER=1000 ===
+W(n) per iteration: 729000.00 flops
+Q(n) per iteration: 12912 bytes
+I(n): 56.46 flops/byte
+W(n): 729000000.00 flops
+T(n): 632178740.00 cycles
+P(n): 1.15 flops/cycle
+
+=== N_P=200 N_ITER=1000 ===
+W(n) per iteration: 2898000.00 flops
+Q(n) per iteration: 25712 bytes
+I(n): 112.71 flops/byte
+W(n): 2898000000.00 flops
+T(n): 2477854340.00 cycles
+P(n): 1.17 flops/cycle
+
+=== N_P=500 N_ITER=1000 ===
+W(n) per iteration: 18045000.00 flops
+Q(n) per iteration: 64112 bytes
+I(n): 281.46 flops/byte
+W(n): 18045000000.00 flops
+T(n): 15380682320.00 cycles
+P(n): 1.17 flops/cycle
+
+=== N_P=1000 N_ITER=1000 ===
+W(n) per iteration: 72090000.00 flops
+Q(n) per iteration: 128112 bytes
+I(n): 562.71 flops/byte
+W(n): 72090000000.00 flops
+T(n): 61619267280.00 cycles
+P(n): 1.17 flops/cycle

measurements/variant_ilp/x86/1.log

@@ -0,0 +1,2 @@
+rm -f main_* sizes run_tests *.o *.s
+/usr/bin/g++ -O3 -march=native -std=c++23 -Wall -Wextra -flto -DFLOPCOUNT=0 -DRANDOM_SEED=0 -DBENCHMARK=1 main.cpp ./variants/variant_ilp.cpp -o main_ilp

measurements/variant_inlining_7/x86/1.log

@@ -0,0 +1,41 @@
+rm -f main_* sizes run_tests *.o *.s
+/usr/bin/g++ -O3 -march=native -std=c++23 -Wall -Wextra -flto -DFLOPCOUNT=0 -DRANDOM_SEED=0 -DROOFLINE=0 -DBENCHMARK=1 main.cpp ./variants/variant_inlining_7.cpp -o main_inlining_7
+=== N_P=50 N_ITER=1000 ===
+W(n) per iteration: 184500.00 flops
+Q(n) per iteration: 6512 bytes
+I(n): 28.33 flops/byte
+W(n): 184500000.00 flops
+T(n): 131794980.00 cycles
+P(n): 1.40 flops/cycle
+
+=== N_P=100 N_ITER=1000 ===
+W(n) per iteration: 729000.00 flops
+Q(n) per iteration: 12912 bytes
+I(n): 56.46 flops/byte
+W(n): 729000000.00 flops
+T(n): 527605420.00 cycles
+P(n): 1.38 flops/cycle
+
+=== N_P=200 N_ITER=1000 ===
+W(n) per iteration: 2898000.00 flops
+Q(n) per iteration: 25712 bytes
+I(n): 112.71 flops/byte
+W(n): 2898000000.00 flops
+T(n): 2166884360.00 cycles
+P(n): 1.34 flops/cycle
+
+=== N_P=500 N_ITER=1000 ===
+W(n) per iteration: 18045000.00 flops
+Q(n) per iteration: 64112 bytes
+I(n): 281.46 flops/byte
+W(n): 18045000000.00 flops
+T(n): 14615797280.00 cycles
+P(n): 1.23 flops/cycle
+
+=== N_P=1000 N_ITER=1000 ===
+W(n) per iteration: 72090000.00 flops
+Q(n) per iteration: 128112 bytes
+I(n): 562.71 flops/byte
+W(n): 72090000000.00 flops
+T(n): 57677070100.00 cycles
+P(n): 1.25 flops/cycle

measurements/variant_inlining_7/x86/2.log

@@ -0,0 +1,41 @@
+rm -f main_* sizes run_tests *.o *.s
+/usr/bin/g++ -O3 -march=native -std=c++23 -Wall -Wextra -flto -DFLOPCOUNT=0 -DRANDOM_SEED=0 -DROOFLINE=0 -DBENCHMARK=1 main.cpp ./variants/variant_inlining_7.cpp -o main_inlining_7
+=== N_P=50 N_ITER=1000 ===
+W(n) per iteration: 184500.00 flops
+Q(n) per iteration: 6512 bytes
+I(n): 28.33 flops/byte
+W(n): 184500000.00 flops
+T(n): 132693220.00 cycles
+P(n): 1.39 flops/cycle
+
+=== N_P=100 N_ITER=1000 ===
+W(n) per iteration: 729000.00 flops
+Q(n) per iteration: 12912 bytes
+I(n): 56.46 flops/byte
+W(n): 729000000.00 flops
+T(n): 527980800.00 cycles
+P(n): 1.38 flops/cycle
+
+=== N_P=200 N_ITER=1000 ===
+W(n) per iteration: 2898000.00 flops
+Q(n) per iteration: 25712 bytes
+I(n): 112.71 flops/byte
+W(n): 2898000000.00 flops
+T(n): 2161999920.00 cycles
+P(n): 1.34 flops/cycle
+
+=== N_P=500 N_ITER=1000 ===
+W(n) per iteration: 18045000.00 flops
+Q(n) per iteration: 64112 bytes
+I(n): 281.46 flops/byte
+W(n): 18045000000.00 flops
+T(n): 14567319680.00 cycles
+P(n): 1.24 flops/cycle
+
+=== N_P=1000 N_ITER=1000 ===
+W(n) per iteration: 72090000.00 flops
+Q(n) per iteration: 128112 bytes
+I(n): 562.71 flops/byte
+W(n): 72090000000.00 flops
+T(n): 57637351660.00 cycles
+P(n): 1.25 flops/cycle

measurements/variant_inlining_7/x86/3.log

@@ -0,0 +1,41 @@
+rm -f main_* sizes run_tests *.o *.s
+/usr/bin/g++ -O3 -march=native -std=c++23 -Wall -Wextra -flto -DFLOPCOUNT=0 -DRANDOM_SEED=0 -DROOFLINE=0 -DBENCHMARK=1 main.cpp ./variants/variant_inlining_7.cpp -o main_inlining_7
+=== N_P=50 N_ITER=1000 ===
+W(n) per iteration: 184500.00 flops
+Q(n) per iteration: 6512 bytes
+I(n): 28.33 flops/byte
+W(n): 184500000.00 flops
+T(n): 132083640.00 cycles
+P(n): 1.40 flops/cycle
+
+=== N_P=100 N_ITER=1000 ===
+W(n) per iteration: 729000.00 flops
+Q(n) per iteration: 12912 bytes
+I(n): 56.46 flops/byte
+W(n): 729000000.00 flops
+T(n): 526775360.00 cycles
+P(n): 1.38 flops/cycle
+
+=== N_P=200 N_ITER=1000 ===
+W(n) per iteration: 2898000.00 flops
+Q(n) per iteration: 25712 bytes
+I(n): 112.71 flops/byte
+W(n): 2898000000.00 flops
+T(n): 2160050540.00 cycles
+P(n): 1.34 flops/cycle
+
+=== N_P=500 N_ITER=1000 ===
+W(n) per iteration: 18045000.00 flops
+Q(n) per iteration: 64112 bytes
+I(n): 281.46 flops/byte
+W(n): 18045000000.00 flops
+T(n): 14605475200.00 cycles
+P(n): 1.24 flops/cycle
+
+=== N_P=1000 N_ITER=1000 ===
+W(n) per iteration: 72090000.00 flops
+Q(n) per iteration: 128112 bytes
+I(n): 562.71 flops/byte
+W(n): 72090000000.00 flops
+T(n): 57692748180.00 cycles
+P(n): 1.25 flops/cycle

measurements/variant_inlining_7_SoA/x86/1.log

@@ -0,0 +1,33 @@
+rm -f main_* sizes run_tests *.o *.s
+/usr/bin/g++ -O3 -march=native -std=c++23 -Wall -Wextra -flto -DFLOPCOUNT=0 -DRANDOM_SEED=0 -DROOFLINE=0 -DBENCHMARK=1 main.cpp ./variants/variant_inlining_7_SoA.cpp -o main_inlining_7_SoA
+=== N_P=50 N_ITER=1000 ===
+W(n) per iteration: 184500.00 flops
+Q(n) per iteration: 6512 bytes
+I(n): 28.33 flops/byte
+W(n): 184500000.00 flops
+T(n): 174279560.00 cycles
+P(n): 1.06 flops/cycle
+
+=== N_P=100 N_ITER=1000 ===
+W(n) per iteration: 729000.00 flops
+Q(n) per iteration: 12912 bytes
+I(n): 56.46 flops/byte
+W(n): 729000000.00 flops
+T(n): 771260820.00 cycles
+P(n): 0.95 flops/cycle
+
+=== N_P=200 N_ITER=1000 ===
+W(n) per iteration: 2898000.00 flops
+Q(n) per iteration: 25712 bytes
+I(n): 112.71 flops/byte
+W(n): 2898000000.00 flops
+T(n): 3263227240.00 cycles
+P(n): 0.89 flops/cycle
+
+=== N_P=500 N_ITER=1000 ===
+W(n) per iteration: 18045000.00 flops
+Q(n) per iteration: 64112 bytes
+I(n): 281.46 flops/byte
+W(n): 18045000000.00 flops
+T(n): 19848792000.00 cycles
+P(n): 0.91 flops/cycle

measurements/variant_inlining_7_SoA/x86/2.log

@@ -0,0 +1,33 @@
+rm -f main_* sizes run_tests *.o *.s
+/usr/bin/g++ -O3 -march=native -std=c++23 -Wall -Wextra -flto -DFLOPCOUNT=0 -DRANDOM_SEED=0 -DROOFLINE=0 -DBENCHMARK=1 main.cpp ./variants/variant_inlining_7_SoA.cpp -o main_inlining_7_SoA
+=== N_P=50 N_ITER=1000 ===
+W(n) per iteration: 184500.00 flops
+Q(n) per iteration: 6512 bytes
+I(n): 28.33 flops/byte
+W(n): 184500000.00 flops
+T(n): 174641860.00 cycles
+P(n): 1.06 flops/cycle
+
+=== N_P=100 N_ITER=1000 ===
+W(n) per iteration: 729000.00 flops
+Q(n) per iteration: 12912 bytes
+I(n): 56.46 flops/byte
+W(n): 729000000.00 flops
+T(n): 769502860.00 cycles
+P(n): 0.95 flops/cycle
+
+=== N_P=200 N_ITER=1000 ===
+W(n) per iteration: 2898000.00 flops
+Q(n) per iteration: 25712 bytes
+I(n): 112.71 flops/byte
+W(n): 2898000000.00 flops
+T(n): 3259749920.00 cycles
+P(n): 0.89 flops/cycle
+
+=== N_P=500 N_ITER=1000 ===
+W(n) per iteration: 18045000.00 flops
+Q(n) per iteration: 64112 bytes
+I(n): 281.46 flops/byte
+W(n): 18045000000.00 flops
+T(n): 19799476220.00 cycles
+P(n): 0.91 flops/cycle

measurements/variant_inlining_7_SoA/x86/3.log

@@ -0,0 +1,33 @@
+rm -f main_* sizes run_tests *.o *.s
+/usr/bin/g++ -O3 -march=native -std=c++23 -Wall -Wextra -flto -DFLOPCOUNT=0 -DRANDOM_SEED=0 -DROOFLINE=0 -DBENCHMARK=1 main.cpp ./variants/variant_inlining_7_SoA.cpp -o main_inlining_7_SoA
+=== N_P=50 N_ITER=1000 ===
+W(n) per iteration: 184500.00 flops
+Q(n) per iteration: 6512 bytes
+I(n): 28.33 flops/byte
+W(n): 184500000.00 flops
+T(n): 174646140.00 cycles
+P(n): 1.06 flops/cycle
+
+=== N_P=100 N_ITER=1000 ===
+W(n) per iteration: 729000.00 flops
+Q(n) per iteration: 12912 bytes
+I(n): 56.46 flops/byte
+W(n): 729000000.00 flops
+T(n): 771133380.00 cycles
+P(n): 0.95 flops/cycle
+
+=== N_P=200 N_ITER=1000 ===
+W(n) per iteration: 2898000.00 flops
+Q(n) per iteration: 25712 bytes
+I(n): 112.71 flops/byte
+W(n): 2898000000.00 flops
+T(n): 3262481280.00 cycles
+P(n): 0.89 flops/cycle
+
+=== N_P=500 N_ITER=1000 ===
+W(n) per iteration: 18045000.00 flops
+Q(n) per iteration: 64112 bytes
+I(n): 281.46 flops/byte
+W(n): 18045000000.00 flops
+T(n): 19817595500.00 cycles
+P(n): 0.91 flops/cycle

measurements/variant_inlining_7_symmetry_2/x86/1.log

@@ -0,0 +1,33 @@
+rm -f main_* sizes run_tests *.o *.s
+/usr/bin/g++ -O3 -march=native -std=c++23 -Wall -Wextra -flto -DFLOPCOUNT=0 -DRANDOM_SEED=0 -DROOFLINE=0 -DBENCHMARK=1 main.cpp ./variants/variant_inlining_7_symmetry_2.cpp -o main_inlining_7_symmetry_2
+=== N_P=50 N_ITER=1000 ===
+W(n) per iteration: 184500.00 flops
+Q(n) per iteration: 6512 bytes
+I(n): 28.33 flops/byte
+W(n): 184500000.00 flops
+T(n): 134217900.00 cycles
+P(n): 1.37 flops/cycle
+
+=== N_P=100 N_ITER=1000 ===
+W(n) per iteration: 729000.00 flops
+Q(n) per iteration: 12912 bytes
+I(n): 56.46 flops/byte
+W(n): 729000000.00 flops
+T(n): 539909700.00 cycles
+P(n): 1.35 flops/cycle
+
+=== N_P=200 N_ITER=1000 ===
+W(n) per iteration: 2898000.00 flops
+Q(n) per iteration: 25712 bytes
+I(n): 112.71 flops/byte
+W(n): 2898000000.00 flops
+T(n): 2204928560.00 cycles
+P(n): 1.31 flops/cycle
+
+=== N_P=500 N_ITER=1000 ===
+W(n) per iteration: 18045000.00 flops
+Q(n) per iteration: 64112 bytes
+I(n): 281.46 flops/byte
+W(n): 18045000000.00 flops
+T(n): 15229802740.00 cycles
+P(n): 1.18 flops/cycle

measurements/variant_inlining_7_symmetry_2/x86/2.log

@@ -0,0 +1,33 @@
+rm -f main_* sizes run_tests *.o *.s
+/usr/bin/g++ -O3 -march=native -std=c++23 -Wall -Wextra -flto -DFLOPCOUNT=0 -DRANDOM_SEED=0 -DROOFLINE=0 -DBENCHMARK=1 main.cpp ./variants/variant_inlining_7_symmetry_2.cpp -o main_inlining_7_symmetry_2
+=== N_P=50 N_ITER=1000 ===
+W(n) per iteration: 184500.00 flops
+Q(n) per iteration: 6512 bytes
+I(n): 28.33 flops/byte
+W(n): 184500000.00 flops
+T(n): 134590240.00 cycles
+P(n): 1.37 flops/cycle
+
+=== N_P=100 N_ITER=1000 ===
+W(n) per iteration: 729000.00 flops
+Q(n) per iteration: 12912 bytes
+I(n): 56.46 flops/byte
+W(n): 729000000.00 flops
+T(n): 537772020.00 cycles
+P(n): 1.36 flops/cycle
+
+=== N_P=200 N_ITER=1000 ===
+W(n) per iteration: 2898000.00 flops
+Q(n) per iteration: 25712 bytes
+I(n): 112.71 flops/byte
+W(n): 2898000000.00 flops
+T(n): 2216669180.00 cycles
+P(n): 1.31 flops/cycle
+
+=== N_P=500 N_ITER=1000 ===
+W(n) per iteration: 18045000.00 flops
+Q(n) per iteration: 64112 bytes
+I(n): 281.46 flops/byte
+W(n): 18045000000.00 flops
+T(n): 15269978860.00 cycles
+P(n): 1.18 flops/cycle

measurements/variant_inlining_7_symmetry_2/x86/3.log

@@ -0,0 +1,33 @@
+rm -f main_* sizes run_tests *.o *.s
+/usr/bin/g++ -O3 -march=native -std=c++23 -Wall -Wextra -flto -DFLOPCOUNT=0 -DRANDOM_SEED=0 -DROOFLINE=0 -DBENCHMARK=1 main.cpp ./variants/variant_inlining_7_symmetry_2.cpp -o main_inlining_7_symmetry_2
+=== N_P=50 N_ITER=1000 ===
+W(n) per iteration: 184500.00 flops
+Q(n) per iteration: 6512 bytes
+I(n): 28.33 flops/byte
+W(n): 184500000.00 flops
+T(n): 133252040.00 cycles
+P(n): 1.38 flops/cycle
+
+=== N_P=100 N_ITER=1000 ===
+W(n) per iteration: 729000.00 flops
+Q(n) per iteration: 12912 bytes
+I(n): 56.46 flops/byte
+W(n): 729000000.00 flops
+T(n): 533142280.00 cycles
+P(n): 1.37 flops/cycle
+
+=== N_P=200 N_ITER=1000 ===
+W(n) per iteration: 2898000.00 flops
+Q(n) per iteration: 25712 bytes
+I(n): 112.71 flops/byte
+W(n): 2898000000.00 flops
+T(n): 2200045220.00 cycles
+P(n): 1.32 flops/cycle
+
+=== N_P=500 N_ITER=1000 ===
+W(n) per iteration: 18045000.00 flops
+Q(n) per iteration: 64112 bytes
+I(n): 281.46 flops/byte
+W(n): 18045000000.00 flops
+T(n): 15171854800.00 cycles
+P(n): 1.19 flops/cycle

measurements/variant_inlining_7_symmetry_2_ilp_1/x86/1.log

@@ -0,0 +1,33 @@
+rm -f main_* sizes run_tests *.o *.s
+/usr/bin/g++ -O3 -march=native -std=c++23 -Wall -Wextra -flto -DFLOPCOUNT=0 -DRANDOM_SEED=0 -DROOFLINE=0 -DBENCHMARK=1 main.cpp ./variants/variant_inlining_7_symmetry_2_ilp_1.cpp -o main_inlining_7_symmetry_2_ilp_1
+=== N_P=50 N_ITER=1000 ===
+W(n) per iteration: 184500.00 flops
+Q(n) per iteration: 6512 bytes
+I(n): 28.33 flops/byte
+W(n): 184500000.00 flops
+T(n): 134117580.00 cycles
+P(n): 1.38 flops/cycle
+
+=== N_P=100 N_ITER=1000 ===
+W(n) per iteration: 729000.00 flops
+Q(n) per iteration: 12912 bytes
+I(n): 56.46 flops/byte
+W(n): 729000000.00 flops
+T(n): 529885740.00 cycles
+P(n): 1.38 flops/cycle
+
+=== N_P=200 N_ITER=1000 ===
+W(n) per iteration: 2898000.00 flops
+Q(n) per iteration: 25712 bytes
+I(n): 112.71 flops/byte
+W(n): 2898000000.00 flops
+T(n): 2158215760.00 cycles
+P(n): 1.34 flops/cycle
+
+=== N_P=500 N_ITER=1000 ===
+W(n) per iteration: 18045000.00 flops
+Q(n) per iteration: 64112 bytes
+I(n): 281.46 flops/byte
+W(n): 18045000000.00 flops
+T(n): 14658868260.00 cycles
+P(n): 1.23 flops/cycle

measurements/variant_inlining_7_symmetry_2_ilp_1/x86/2.log

@@ -0,0 +1,33 @@
+rm -f main_* sizes run_tests *.o *.s
+/usr/bin/g++ -O3 -march=native -std=c++23 -Wall -Wextra -flto -DFLOPCOUNT=0 -DRANDOM_SEED=0 -DROOFLINE=0 -DBENCHMARK=1 main.cpp ./variants/variant_inlining_7_symmetry_2_ilp_1.cpp -o main_inlining_7_symmetry_2_ilp_1
+=== N_P=50 N_ITER=1000 ===
+W(n) per iteration: 184500.00 flops
+Q(n) per iteration: 6512 bytes
+I(n): 28.33 flops/byte
+W(n): 184500000.00 flops
+T(n): 135847100.00 cycles
+P(n): 1.36 flops/cycle
+
+=== N_P=100 N_ITER=1000 ===
+W(n) per iteration: 729000.00 flops
+Q(n) per iteration: 12912 bytes
+I(n): 56.46 flops/byte
+W(n): 729000000.00 flops
+T(n): 531086160.00 cycles
+P(n): 1.37 flops/cycle
+
+=== N_P=200 N_ITER=1000 ===
+W(n) per iteration: 2898000.00 flops
+Q(n) per iteration: 25712 bytes
+I(n): 112.71 flops/byte
+W(n): 2898000000.00 flops
+T(n): 2149640340.00 cycles
+P(n): 1.35 flops/cycle
+
+=== N_P=500 N_ITER=1000 ===
+W(n) per iteration: 18045000.00 flops
+Q(n) per iteration: 64112 bytes
+I(n): 281.46 flops/byte
+W(n): 18045000000.00 flops
+T(n): 14662439140.00 cycles
+P(n): 1.23 flops/cycle

measurements/variant_inlining_7_symmetry_2_ilp_1/x86/3.log

@@ -0,0 +1,33 @@
+rm -f main_* sizes run_tests *.o *.s
+/usr/bin/g++ -O3 -march=native -std=c++23 -Wall -Wextra -flto -DFLOPCOUNT=0 -DRANDOM_SEED=0 -DROOFLINE=0 -DBENCHMARK=1 main.cpp ./variants/variant_inlining_7_symmetry_2_ilp_1.cpp -o main_inlining_7_symmetry_2_ilp_1
+=== N_P=50 N_ITER=1000 ===
+W(n) per iteration: 184500.00 flops
+Q(n) per iteration: 6512 bytes
+I(n): 28.33 flops/byte
+W(n): 184500000.00 flops
+T(n): 132982240.00 cycles
+P(n): 1.39 flops/cycle
+
+=== N_P=100 N_ITER=1000 ===
+W(n) per iteration: 729000.00 flops
+Q(n) per iteration: 12912 bytes
+I(n): 56.46 flops/byte
+W(n): 729000000.00 flops
+T(n): 530176140.00 cycles
+P(n): 1.38 flops/cycle
+
+=== N_P=200 N_ITER=1000 ===
+W(n) per iteration: 2898000.00 flops
+Q(n) per iteration: 25712 bytes
+I(n): 112.71 flops/byte
+W(n): 2898000000.00 flops
+T(n): 2162889920.00 cycles
+P(n): 1.34 flops/cycle
+
+=== N_P=500 N_ITER=1000 ===
+W(n) per iteration: 18045000.00 flops
+Q(n) per iteration: 64112 bytes
+I(n): 281.46 flops/byte
+W(n): 18045000000.00 flops
+T(n): 14697507480.00 cycles
+P(n): 1.23 flops/cycle

measurements/variant_inlining_7_symmetry_2_ilp_2/x86/1.log

@@ -0,0 +1,33 @@
+rm -f main_* sizes run_tests *.o *.s
+/usr/bin/g++ -O3 -march=native -std=c++23 -Wall -Wextra -flto -DFLOPCOUNT=0 -DRANDOM_SEED=0 -DROOFLINE=0 -DBENCHMARK=1 main.cpp ./variants/variant_inlining_7_symmetry_2_ilp_2.cpp -o main_inlining_7_symmetry_2_ilp_2
+=== N_P=50 N_ITER=1000 ===
+W(n) per iteration: 184500.00 flops
+Q(n) per iteration: 6512 bytes
+I(n): 28.33 flops/byte
+W(n): 184500000.00 flops
+T(n): 138339020.00 cycles
+P(n): 1.33 flops/cycle
+
+=== N_P=100 N_ITER=1000 ===
+W(n) per iteration: 729000.00 flops
+Q(n) per iteration: 12912 bytes
+I(n): 56.46 flops/byte
+W(n): 729000000.00 flops
+T(n): 534997480.00 cycles
+P(n): 1.36 flops/cycle
+
+=== N_P=200 N_ITER=1000 ===
+W(n) per iteration: 2898000.00 flops
+Q(n) per iteration: 25712 bytes
+I(n): 112.71 flops/byte
+W(n): 2898000000.00 flops
+T(n): 2168912380.00 cycles
+P(n): 1.34 flops/cycle
+
+=== N_P=500 N_ITER=1000 ===
+W(n) per iteration: 18045000.00 flops
+Q(n) per iteration: 64112 bytes
+I(n): 281.46 flops/byte
+W(n): 18045000000.00 flops
+T(n): 14986758980.00 cycles
+P(n): 1.20 flops/cycle

measurements/variant_inlining_7_symmetry_2_ilp_2/x86/2.log

@@ -0,0 +1,33 @@
+rm -f main_* sizes run_tests *.o *.s
+/usr/bin/g++ -O3 -march=native -std=c++23 -Wall -Wextra -flto -DFLOPCOUNT=0 -DRANDOM_SEED=0 -DROOFLINE=0 -DBENCHMARK=1 main.cpp ./variants/variant_inlining_7_symmetry_2_ilp_2.cpp -o main_inlining_7_symmetry_2_ilp_2
+=== N_P=50 N_ITER=1000 ===
+W(n) per iteration: 184500.00 flops
+Q(n) per iteration: 6512 bytes
+I(n): 28.33 flops/byte
+W(n): 184500000.00 flops
+T(n): 135244420.00 cycles
+P(n): 1.36 flops/cycle
+
+=== N_P=100 N_ITER=1000 ===
+W(n) per iteration: 729000.00 flops
+Q(n) per iteration: 12912 bytes
+I(n): 56.46 flops/byte
+W(n): 729000000.00 flops
+T(n): 537453160.00 cycles
+P(n): 1.36 flops/cycle
+
+=== N_P=200 N_ITER=1000 ===
+W(n) per iteration: 2898000.00 flops
+Q(n) per iteration: 25712 bytes
+I(n): 112.71 flops/byte
+W(n): 2898000000.00 flops
+T(n): 2175781640.00 cycles
+P(n): 1.33 flops/cycle
+
+=== N_P=500 N_ITER=1000 ===
+W(n) per iteration: 18045000.00 flops
+Q(n) per iteration: 64112 bytes
+I(n): 281.46 flops/byte
+W(n): 18045000000.00 flops
+T(n): 14932133120.00 cycles
+P(n): 1.21 flops/cycle

measurements/variant_inlining_7_symmetry_2_ilp_2/x86/3.log

@@ -0,0 +1,33 @@
+rm -f main_* sizes run_tests *.o *.s
+/usr/bin/g++ -O3 -march=native -std=c++23 -Wall -Wextra -flto -DFLOPCOUNT=0 -DRANDOM_SEED=0 -DROOFLINE=0 -DBENCHMARK=1 main.cpp ./variants/variant_inlining_7_symmetry_2_ilp_2.cpp -o main_inlining_7_symmetry_2_ilp_2
+=== N_P=50 N_ITER=1000 ===
+W(n) per iteration: 184500.00 flops
+Q(n) per iteration: 6512 bytes
+I(n): 28.33 flops/byte
+W(n): 184500000.00 flops
+T(n): 135881620.00 cycles
+P(n): 1.36 flops/cycle
+
+=== N_P=100 N_ITER=1000 ===
+W(n) per iteration: 729000.00 flops
+Q(n) per iteration: 12912 bytes
+I(n): 56.46 flops/byte
+W(n): 729000000.00 flops
+T(n): 534853360.00 cycles
+P(n): 1.36 flops/cycle
+
+=== N_P=200 N_ITER=1000 ===
+W(n) per iteration: 2898000.00 flops
+Q(n) per iteration: 25712 bytes
+I(n): 112.71 flops/byte
+W(n): 2898000000.00 flops
+T(n): 2166176600.00 cycles
+P(n): 1.34 flops/cycle
+
+=== N_P=500 N_ITER=1000 ===
+W(n) per iteration: 18045000.00 flops
+Q(n) per iteration: 64112 bytes
+I(n): 281.46 flops/byte
+W(n): 18045000000.00 flops
+T(n): 14773516340.00 cycles
+P(n): 1.22 flops/cycle

measurements/variant_lut_exp/x86/1.log

@@ -0,0 +1,33 @@
+rm -f main_* sizes run_tests *.o *.s
+/usr/bin/g++ -O3 -march=native -std=c++23 -Wall -Wextra -flto -DFLOPCOUNT=0 -DRANDOM_SEED=0 -DROOFLINE=0 -DBENCHMARK=1 main.cpp ./variants/variant_lut_exp.cpp -o main_lut_exp
+=== N_P=50 N_ITER=1000 ===
+W(n) per iteration: 184500.00 flops
+Q(n) per iteration: 6512 bytes
+I(n): 28.33 flops/byte
+W(n): 184500000.00 flops
+T(n): 145567620.00 cycles
+P(n): 1.27 flops/cycle
+
+=== N_P=100 N_ITER=1000 ===
+W(n) per iteration: 729000.00 flops
+Q(n) per iteration: 12912 bytes
+I(n): 56.46 flops/byte
+W(n): 729000000.00 flops
+T(n): 569315980.00 cycles
+P(n): 1.28 flops/cycle
+
+=== N_P=200 N_ITER=1000 ===
+W(n) per iteration: 2898000.00 flops
+Q(n) per iteration: 25712 bytes
+I(n): 112.71 flops/byte
+W(n): 2898000000.00 flops
+T(n): 2268145640.00 cycles
+P(n): 1.28 flops/cycle
+
+=== N_P=500 N_ITER=1000 ===
+W(n) per iteration: 18045000.00 flops
+Q(n) per iteration: 64112 bytes
+I(n): 281.46 flops/byte
+W(n): 18045000000.00 flops
+T(n): 15227306960.00 cycles
+P(n): 1.19 flops/cycle

measurements/variant_lut_exp/x86/2.log

@@ -0,0 +1,33 @@
+rm -f main_* sizes run_tests *.o *.s
+/usr/bin/g++ -O3 -march=native -std=c++23 -Wall -Wextra -flto -DFLOPCOUNT=0 -DRANDOM_SEED=0 -DROOFLINE=0 -DBENCHMARK=1 main.cpp ./variants/variant_lut_exp.cpp -o main_lut_exp
+=== N_P=50 N_ITER=1000 ===
+W(n) per iteration: 184500.00 flops
+Q(n) per iteration: 6512 bytes
+I(n): 28.33 flops/byte
+W(n): 184500000.00 flops
+T(n): 141259480.00 cycles
+P(n): 1.31 flops/cycle
+
+=== N_P=100 N_ITER=1000 ===
+W(n) per iteration: 729000.00 flops
+Q(n) per iteration: 12912 bytes
+I(n): 56.46 flops/byte
+W(n): 729000000.00 flops
+T(n): 570826340.00 cycles
+P(n): 1.28 flops/cycle
+
+=== N_P=200 N_ITER=1000 ===
+W(n) per iteration: 2898000.00 flops
+Q(n) per iteration: 25712 bytes
+I(n): 112.71 flops/byte
+W(n): 2898000000.00 flops
+T(n): 2281692040.00 cycles
+P(n): 1.27 flops/cycle
+
+=== N_P=500 N_ITER=1000 ===
+W(n) per iteration: 18045000.00 flops
+Q(n) per iteration: 64112 bytes
+I(n): 281.46 flops/byte
+W(n): 18045000000.00 flops
+T(n): 15242958120.00 cycles
+P(n): 1.18 flops/cycle

measurements/variant_lut_exp/x86/3.log

@@ -0,0 +1,33 @@
+rm -f main_* sizes run_tests *.o *.s
+/usr/bin/g++ -O3 -march=native -std=c++23 -Wall -Wextra -flto -DFLOPCOUNT=0 -DRANDOM_SEED=0 -DROOFLINE=0 -DBENCHMARK=1 main.cpp ./variants/variant_lut_exp.cpp -o main_lut_exp
+=== N_P=50 N_ITER=1000 ===
+W(n) per iteration: 184500.00 flops
+Q(n) per iteration: 6512 bytes
+I(n): 28.33 flops/byte
+W(n): 184500000.00 flops
+T(n): 140807980.00 cycles
+P(n): 1.31 flops/cycle
+
+=== N_P=100 N_ITER=1000 ===
+W(n) per iteration: 729000.00 flops
+Q(n) per iteration: 12912 bytes
+I(n): 56.46 flops/byte
+W(n): 729000000.00 flops
+T(n): 563063780.00 cycles
+P(n): 1.29 flops/cycle
+
+=== N_P=200 N_ITER=1000 ===
+W(n) per iteration: 2898000.00 flops
+Q(n) per iteration: 25712 bytes
+I(n): 112.71 flops/byte
+W(n): 2898000000.00 flops
+T(n): 2268612980.00 cycles
+P(n): 1.28 flops/cycle
+
+=== N_P=500 N_ITER=1000 ===
+W(n) per iteration: 18045000.00 flops
+Q(n) per iteration: 64112 bytes
+I(n): 281.46 flops/byte
+W(n): 18045000000.00 flops
+T(n): 15237244940.00 cycles
+P(n): 1.18 flops/cycle