Benchmark for APM X-Gene
From Leo's Notes
Last edited on 10 August 2017, at 23:31.
APM X-Gene has a 64-bit ARM processor.
Original APM Linux[edit | edit source]
The following benchmark was done on the original linux kernel that was shipped with the box. This was able to see all the cores and was able to run slightly faster even on one CPU.
Results Overview[edit source]
| Hardware | APM X-Gene |
|---|---|
| CPU | APM X-Gene |
| Memory | 16 GB DDR3 EEC |
| Disk | 1x500 GB SATA |
| Operating System | Fedora 22 aarch64 |
| Score | 926.9 / 3402.7 |
Raw Output
========================================================================
BYTE UNIX Benchmarks (Version 5.1.3)
System: csa.cpsc.ucalgary.ca: GNU/Linux
OS: GNU/Linux -- 3.12.0-mustang_sw_1.12.09-beta -- #1 SMP Thu Jun 12 10:27:15 PDT 2014
Machine: aarch64 (aarch64)
Language: en_US.utf8 (charmap="UTF-8", collate="UTF-8")
CPU 0: APM X-Gene Mustang board @ 2.40GHz (0.0 bogomips)
CPU 1: APM X-Gene Mustang board @ 2.40GHz (0.0 bogomips)
CPU 2: APM X-Gene Mustang board @ 2.40GHz (0.0 bogomips)
CPU 3: APM X-Gene Mustang board @ 2.40GHz (0.0 bogomips)
CPU 4: APM X-Gene Mustang board @ 2.40GHz (0.0 bogomips)
CPU 5: APM X-Gene Mustang board @ 2.40GHz (0.0 bogomips)
CPU 6: APM X-Gene Mustang board @ 2.40GHz (0.0 bogomips)
CPU 7: APM X-Gene Mustang board @ 2.40GHz (0.0 bogomips)
17:28:54 up 27 min, 3 users, load average: 1.39, 0.87, 0.61; runlevel 3
------------------------------------------------------------------------
Benchmark Run: Fri Jun 26 2015 17:28:54 - 17:56:56
8 CPUs in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 17763947.5 lps (10.0 s, 7 samples)
Double-Precision Whetstone 2114.7 MWIPS (9.6 s, 7 samples)
Execl Throughput 2459.8 lps (29.9 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 468331.1 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 130247.8 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 1151363.8 KBps (30.0 s, 2 samples)
Pipe Throughput 1067484.4 lps (10.0 s, 7 samples)
Pipe-based Context Switching 192167.1 lps (10.0 s, 7 samples)
Process Creation 7660.0 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 4207.7 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 1634.0 lpm (60.0 s, 2 samples)
System Call Overhead 1438914.1 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 17763947.5 1522.2
Double-Precision Whetstone 55.0 2114.7 384.5
Execl Throughput 43.0 2459.8 572.0
File Copy 1024 bufsize 2000 maxblocks 3960.0 468331.1 1182.7
File Copy 256 bufsize 500 maxblocks 1655.0 130247.8 787.0
File Copy 4096 bufsize 8000 maxblocks 5800.0 1151363.8 1985.1
Pipe Throughput 12440.0 1067484.4 858.1
Pipe-based Context Switching 4000.0 192167.1 480.4
Process Creation 126.0 7660.0 607.9
Shell Scripts (1 concurrent) 42.4 4207.7 992.4
Shell Scripts (8 concurrent) 6.0 1634.0 2723.3
System Call Overhead 15000.0 1438914.1 959.3
========
System Benchmarks Index Score 926.9
------------------------------------------------------------------------
Benchmark Run: Fri Jun 26 2015 17:56:56 - 18:25:05
8 CPUs in system; running 8 parallel copies of tests
Dhrystone 2 using register variables 141540380.6 lps (10.0 s, 7 samples)
Double-Precision Whetstone 16919.0 MWIPS (9.6 s, 7 samples)
Execl Throughput 13012.0 lps (29.8 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 579580.8 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 155434.8 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 1673229.9 KBps (30.0 s, 2 samples)
Pipe Throughput 8441190.9 lps (10.0 s, 7 samples)
Pipe-based Context Switching 1463165.9 lps (10.0 s, 7 samples)
Process Creation 32854.4 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 16740.8 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 2477.4 lpm (60.1 s, 2 samples)
System Call Overhead 7650246.7 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 141540380.6 12128.6
Double-Precision Whetstone 55.0 16919.0 3076.2
Execl Throughput 43.0 13012.0 3026.1
File Copy 1024 bufsize 2000 maxblocks 3960.0 579580.8 1463.6
File Copy 256 bufsize 500 maxblocks 1655.0 155434.8 939.2
File Copy 4096 bufsize 8000 maxblocks 5800.0 1673229.9 2884.9
Pipe Throughput 12440.0 8441190.9 6785.5
Pipe-based Context Switching 4000.0 1463165.9 3657.9
Process Creation 126.0 32854.4 2607.5
Shell Scripts (1 concurrent) 42.4 16740.8 3948.3
Shell Scripts (8 concurrent) 6.0 2477.4 4128.9
System Call Overhead 15000.0 7650246.7 5100.2
========
System Benchmarks Index Score 3402.7
Fedora 23[edit | edit source]
Ran again on a Fedora 23 install. This time with the TianoCore UEFI bios installed.
Results Overview[edit source]
| Hardware | APM X-Gene |
|---|---|
| CPU | APM X-Gene |
| Memory | 16 GB DDR3 EEC |
| Disk | 1x500 GB SATA |
| Operating System | Fedora 23 aarch64 |
| Score | 890.0 / 3436.9 |
Raw Output
========================================================================
BYTE UNIX Benchmarks (Version 5.1.3)
System: csa2: GNU/Linux
OS: GNU/Linux -- 4.2.8-300.fc23.aarch64 -- #1 SMP Mon Dec 21 06:10:24 UTC 201 5
Machine: aarch64 (aarch64)
Language: en_US.utf8 (charmap="UTF-8", collate="UTF-8")
17:32:18 up 22:39, 1 user, load average: 0.13, 1.36, 1.31; runlevel 3
------------------------------------------------------------------------
Benchmark Run: Fri Jan 08 2016 17:32:18 - 18:00:14
8 CPUs in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 17767160.3 lps (10.0 s, 7 samples)
Double-Precision Whetstone 2177.1 MWIPS (9.6 s, 7 samples)
Execl Throughput 2373.3 lps (29.9 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 545523.3 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 160952.3 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 1286987.6 KBps (30.0 s, 2 samples)
Pipe Throughput 727293.2 lps (10.0 s, 7 samples)
Pipe-based Context Switching 117655.4 lps (10.0 s, 7 samples)
Process Creation 5668.4 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 4352.4 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 1841.7 lpm (60.0 s, 2 samples)
System Call Overhead 1537463.6 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 17767160.3 1522.5
Double-Precision Whetstone 55.0 2177.1 395.8
Execl Throughput 43.0 2373.3 551.9
File Copy 1024 bufsize 2000 maxblocks 3960.0 545523.3 1377.6
File Copy 256 bufsize 500 maxblocks 1655.0 160952.3 972.5
File Copy 4096 bufsize 8000 maxblocks 5800.0 1286987.6 2218.9
Pipe Throughput 12440.0 727293.2 584.6
Pipe-based Context Switching 4000.0 117655.4 294.1
Process Creation 126.0 5668.4 449.9
Shell Scripts (1 concurrent) 42.4 4352.4 1026.5
Shell Scripts (8 concurrent) 6.0 1841.7 3069.6
System Call Overhead 15000.0 1537463.6 1025.0
========
System Benchmarks Index Score 890.0
------------------------------------------------------------------------
Benchmark Run: Fri Jan 08 2016 18:00:14 - 18:28:09
8 CPUs in system; running 8 parallel copies of tests
Dhrystone 2 using register variables 142091681.3 lps (10.0 s, 7 samples)
Double-Precision Whetstone 17414.9 MWIPS (9.6 s, 7 samples)
Execl Throughput 11301.6 lps (29.9 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 1032947.0 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 286632.5 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 2569016.9 KBps (30.0 s, 2 samples)
Pipe Throughput 5662756.1 lps (10.0 s, 7 samples)
Pipe-based Context Switching 1090607.6 lps (10.0 s, 7 samples)
Process Creation 15654.0 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 17548.7 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 2488.6 lpm (60.0 s, 2 samples)
System Call Overhead 7592134.7 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 142091681.3 12175.8
Double-Precision Whetstone 55.0 17414.9 3166.4
Execl Throughput 43.0 11301.6 2628.3
File Copy 1024 bufsize 2000 maxblocks 3960.0 1032947.0 2608.5
File Copy 256 bufsize 500 maxblocks 1655.0 286632.5 1731.9
File Copy 4096 bufsize 8000 maxblocks 5800.0 2569016.9 4429.3
Pipe Throughput 12440.0 5662756.1 4552.1
Pipe-based Context Switching 4000.0 1090607.6 2726.5
Process Creation 126.0 15654.0 1242.4
Shell Scripts (1 concurrent) 42.4 17548.7 4138.9
Shell Scripts (8 concurrent) 6.0 2488.6 4147.6
System Call Overhead 15000.0 7592134.7 5061.4
========
System Benchmarks Index Score 3436.9
Fedora 26[edit | edit source]
Fedora 26 with the most recent firmware and BIOS (Tianocore / SlimPro 3.06.25).
The decrease in performance could be related to the BIOS update or Fedora 26.
Results Overview[edit source]
| Hardware | APM X-Gene |
|---|---|
| CPU | APM X-Gene |
| Memory | 16 GB DDR3 EEC |
| Disk | 1x500 GB SATA |
| Operating System | Fedora 26 aarch64 |
| Score | 723.0 / 2765.1 |
Raw Output
========================================================================
BYTE UNIX Benchmarks (Version 5.1.3)
System: csa2: GNU/Linux
OS: GNU/Linux -- 4.11.9-300.fc26.aarch64 -- #1 SMP Wed Jul 5 16:15:00 UTC 2017
Machine: aarch64 (aarch64)
Language: en_US.utf8 (charmap="UTF-8", collate="UTF-8")
14:45:47 up 17 min, 1 user, load average: 0.46, 0.18, 0.12; runlevel 2017-07-17
------------------------------------------------------------------------
Benchmark Run: Mon Jul 17 2017 14:45:47 - 15:13:43
8 CPUs in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 18892194.7 lps (10.0 s, 7 samples)
Double-Precision Whetstone 2471.9 MWIPS (9.6 s, 7 samples)
Execl Throughput 1755.7 lps (29.9 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 374673.9 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 102829.9 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 963569.3 KBps (30.0 s, 2 samples)
Pipe Throughput 676573.4 lps (10.0 s, 7 samples)
Pipe-based Context Switching 94482.3 lps (10.0 s, 7 samples)
Process Creation 5147.9 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 3200.9 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 1478.3 lpm (60.0 s, 2 samples)
System Call Overhead 1080252.6 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 18892194.7 1618.9
Double-Precision Whetstone 55.0 2471.9 449.4
Execl Throughput 43.0 1755.7 408.3
File Copy 1024 bufsize 2000 maxblocks 3960.0 374673.9 946.1
File Copy 256 bufsize 500 maxblocks 1655.0 102829.9 621.3
File Copy 4096 bufsize 8000 maxblocks 5800.0 963569.3 1661.3
Pipe Throughput 12440.0 676573.4 543.9
Pipe-based Context Switching 4000.0 94482.3 236.2
Process Creation 126.0 5147.9 408.6
Shell Scripts (1 concurrent) 42.4 3200.9 754.9
Shell Scripts (8 concurrent) 6.0 1478.3 2463.9
System Call Overhead 15000.0 1080252.6 720.2
========
System Benchmarks Index Score 723.0
------------------------------------------------------------------------
Benchmark Run: Mon Jul 17 2017 15:13:43 - 15:41:42
8 CPUs in system; running 8 parallel copies of tests
Dhrystone 2 using register variables 150729262.4 lps (10.0 s, 7 samples)
Double-Precision Whetstone 19767.7 MWIPS (9.6 s, 7 samples)
Execl Throughput 12385.7 lps (29.9 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 531394.2 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 143088.9 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 1530740.2 KBps (30.0 s, 2 samples)
Pipe Throughput 5364483.4 lps (10.0 s, 7 samples)
Pipe-based Context Switching 821279.1 lps (10.0 s, 7 samples)
Process Creation 26902.7 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 15055.4 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 2095.9 lpm (60.1 s, 2 samples)
System Call Overhead 3121065.5 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 150729262.4 12916.0
Double-Precision Whetstone 55.0 19767.7 3594.1
Execl Throughput 43.0 12385.7 2880.4
File Copy 1024 bufsize 2000 maxblocks 3960.0 531394.2 1341.9
File Copy 256 bufsize 500 maxblocks 1655.0 143088.9 864.6
File Copy 4096 bufsize 8000 maxblocks 5800.0 1530740.2 2639.2
Pipe Throughput 12440.0 5364483.4 4312.3
Pipe-based Context Switching 4000.0 821279.1 2053.2
Process Creation 126.0 26902.7 2135.1
Shell Scripts (1 concurrent) 42.4 15055.4 3550.8
Shell Scripts (8 concurrent) 6.0 2095.9 3493.2
System Call Overhead 15000.0 3121065.5 2080.7
========
System Benchmarks Index Score 2765.1