История изменений
Исправление arax, (текущая версия) :
Потестил на 6.1.38, в unigine-superposition разницы между rt и не rt нет, а вот в byte-unixbench на многопоточной нагрузке rt проигрывает не rt почти 60%.
Кому интересно детали:
не rt:
32 CPUs in system; running 32 parallel copies of tests
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 1292392460.3 110744.9
Double-Precision Whetstone 55.0 281852.3 51245.9
Execl Throughput 43.0 54018.1 12562.4
File Copy 1024 bufsize 2000 maxblocks 3960.0 2324349.0 5869.6
File Copy 256 bufsize 500 maxblocks 1655.0 599601.0 3623.0
File Copy 4096 bufsize 8000 maxblocks 5800.0 6629064.0 11429.4
Pipe Throughput 12440.0 72469091.3 58254.9
Pipe-based Context Switching 4000.0 8595323.6 21488.3
Process Creation 126.0 222388.0 17649.8
Shell Scripts (1 concurrent) 42.4 128404.3 30284.0
Shell Scripts (8 concurrent) 6.0 18119.1 30198.5
System Call Overhead 15000.0 81529002.9 54352.7
========
System Benchmarks Index Score 22731.1
rt:
32 CPUs in system; running 32 parallel copies of tests
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 1277060335.7 109431.0
Double-Precision Whetstone 55.0 279420.5 50803.7
Execl Throughput 43.0 14701.3 3418.9
File Copy 1024 bufsize 2000 maxblocks 3960.0 406645.0 1026.9
File Copy 256 bufsize 500 maxblocks 1655.0 94881.0 573.3
File Copy 4096 bufsize 8000 maxblocks 5800.0 1497178.0 2581.3
Pipe Throughput 12440.0 67906249.3 54587.0
Pipe-based Context Switching 4000.0 7187764.7 17969.4
Process Creation 126.0 40188.5 3189.6
Shell Scripts (1 concurrent) 42.4 59245.7 13973.0
Shell Scripts (8 concurrent) 6.0 6230.3 10383.8
System Call Overhead 15000.0 75191437.3 50127.6
========
System Benchmarks Index Score 9656.3
Исправление arax, :
Потестил на 6.1.38, в unigine-superposition разницы между rt и не rt нет, а вот в byte-unixbench на многопоточной нагрузке rt проигрывает не rt почти 60%.
Кому интересно детали:
не rt:
32 CPUs in system; running 32 parallel copies of tests
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 1292392460.3 110744.9
Double-Precision Whetstone 55.0 281852.3 51245.9
Execl Throughput 43.0 54018.1 12562.4
File Copy 1024 bufsize 2000 maxblocks 3960.0 2324349.0 5869.6
File Copy 256 bufsize 500 maxblocks 1655.0 599601.0 3623.0
File Copy 4096 bufsize 8000 maxblocks 5800.0 6629064.0 11429.4
Pipe Throughput 12440.0 72469091.3 58254.9
Pipe-based Context Switching 4000.0 8595323.6 21488.3
Process Creation 126.0 222388.0 17649.8
Shell Scripts (1 concurrent) 42.4 128404.3 30284.0
Shell Scripts (8 concurrent) 6.0 18119.1 30198.5
System Call Overhead 15000.0 81529002.9 54352.7
========
System Benchmarks Index Score 22731.1
rt:
32 CPUs in system; running 32 parallel copies of tests
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 1277060335.7 109431.0
Double-Precision Whetstone 55.0 279420.5 50803.7
Execl Throughput 43.0 14701.3 3418.9
File Copy 1024 bufsize 2000 maxblocks 3960.0 406645.0 1026.9
File Copy 256 bufsize 500 maxblocks 1655.0 94881.0 573.3
File Copy 4096 bufsize 8000 maxblocks 5800.0 1497178.0 2581.3
Pipe Throughput 12440.0 67906249.3 54587.0
Pipe-based Context Switching 4000.0 7187764.7 17969.4
Process Creation 126.0 40188.5 3189.6
Shell Scripts (1 concurrent) 42.4 59245.7 13973.0
Shell Scripts (8 concurrent) 6.0 6230.3 10383.8
System Call Overhead 15000.0 75191437.3 50127.6
========
System Benchmarks Index Score 9656.3
Исходная версия arax, :
Потестил на 6.1.38, в unigine-superposition разницы между rt и не rt нет, а вот в byte-unixbench на многопоточной нагрузке rt проигрывает не rt почти 60%.
Кому интересно детали:
не rt:
32 CPUs in system; running 32 parallel copies of tests
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 1292392460.3 110744.9
Double-Precision Whetstone 55.0 281852.3 51245.9
Execl Throughput 43.0 54018.1 12562.4
File Copy 1024 bufsize 2000 maxblocks 3960.0 2324349.0 5869.6
File Copy 256 bufsize 500 maxblocks 1655.0 599601.0 3623.0
File Copy 4096 bufsize 8000 maxblocks 5800.0 6629064.0 11429.4
Pipe Throughput 12440.0 72469091.3 58254.9
Pipe-based Context Switching 4000.0 8595323.6 21488.3
Process Creation 126.0 222388.0 17649.8
Shell Scripts (1 concurrent) 42.4 128404.3 30284.0
Shell Scripts (8 concurrent) 6.0 18119.1 30198.5
System Call Overhead 15000.0 81529002.9 54352.7
========
System Benchmarks Index Score 22731.1
rt:
32 CPUs in system; running 32 parallel copies of tests
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 1277060335.7 109431.0
Double-Precision Whetstone 55.0 279420.5 50803.7
Execl Throughput 43.0 14701.3 3418.9
File Copy 1024 bufsize 2000 maxblocks 3960.0 406645.0 1026.9
File Copy 256 bufsize 500 maxblocks 1655.0 94881.0 573.3
File Copy 4096 bufsize 8000 maxblocks 5800.0 1497178.0 2581.3
Pipe Throughput 12440.0 67906249.3 54587.0
Pipe-based Context Switching 4000.0 7187764.7 17969.4
Process Creation 126.0 40188.5 3189.6
Shell Scripts (1 concurrent) 42.4 59245.7 13973.0
Shell Scripts (8 concurrent) 6.0 6230.3 10383.8
System Call Overhead 15000.0 75191437.3 50127.6
========
System Benchmarks Index Score 9656.3