Port-m68k archive
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]
bytebench on 4KB/8KB pagesize kernels
Hi,
After cleanup of pmap_bootstrap.c on all hp300 derived ports,
I also made some changes in my local tree which make
NetBSD/news68k kernel also work with 8KB/page settings.
I've tried pkgsrc/benchmark/bytebench on both
4KB/page and 8KB/page kernels on the same machine
(NWS-1750D, 68030/25MHz with 16KB L2 cache), and
I'm surprised because 8KB/page kernel is
significantly faster than 4KB/page kernel.
Are these results really expected?
On 4KB/page kernel:
------------------------------------------------------------
BYTE UNIX Benchmarks (Version 4.1.0)
System -- libero
Start Benchmark Run: Tue Dec 8 22:11:49 JST 2009
1 interactive users.
10:11PM up 10 mins, 1 user, load averages: 0.47, 0.79, 0.54
-r-xr-xr-x 1 root wheel 135783 Apr 27 2009 /bin/sh
/bin/sh: ELF 32-bit MSB executable, Motorola 68020, version 1 (SYSV), \
dynamically linked (uses shared libs), for NetBSD 5.0, not stripped
/dev/sd0a 314015 44559 253756 14% /
Dhrystone 2 using register variables 7924.6 lps (10.4 secs, 10 samples)
Double-Precision Whetstone 1.7 MWIPS (6.0 secs, 10 samples)
System Call Overhead 1473.7 lps (10.4 secs, 10 samples)
Pipe Throughput 1136.2 lps (10.4 secs, 10 samples)
Pipe-based Context Switching 442.5 lps (10.4 secs, 10 samples)
Process Creation 5.1 lps (30.4 secs, 3 samples)
Execl Throughput 4.0 lps (29.9 secs, 3 samples)
File Read 1024 bufsize 2000 maxblocks 704.0 KBps (30.0 secs, 3 samples)
File Write 1024 bufsize 2000 maxblocks 133.0 KBps (30.0 secs, 3 samples)
File Copy 1024 bufsize 2000 maxblocks 60.0 KBps (30.0 secs, 3 samples)
File Read 256 bufsize 500 maxblocks 380.0 KBps (30.0 secs, 3 samples)
File Write 256 bufsize 500 maxblocks 33.0 KBps (30.0 secs, 3 samples)
File Copy 256 bufsize 500 maxblocks 18.0 KBps (30.0 secs, 3 samples)
File Read 4096 bufsize 8000 maxblocks 349.0 KBps (30.0 secs, 3 samples)
File Write 4096 bufsize 8000 maxblocks 266.0 KBps (30.0 secs, 3 samples)
File Copy 4096 bufsize 8000 maxblocks 98.0 KBps (30.0 secs, 3 samples)
Shell Scripts (1 concurrent) 6.3 lpm (60.5 secs, 3 samples)
Shell Scripts (8 concurrent) 1.0 lpm (0.0 secs, 3 samples)
Shell Scripts (16 concurrent) 1.0 lpm (0.0 secs, 3 samples)
Arithmetic Test (type = short) 410.7 lps (13.5 secs, 3 samples)
Arithmetic Test (type = int) 1450.7 lps (10.4 secs, 3 samples)
Arithmetic Test (type = long) 1428.2 lps (10.4 secs, 3 samples)
Arithmetic Test (type = float) 755.6 lps (10.4 secs, 3 samples)
Arithmetic Test (type = double) 632.8 lps (10.4 secs, 3 samples)
Arithoh 624111.6 lps (10.4 secs, 3 samples)
C Compiler Throughput 1.6 lpm (60.6 secs, 3 samples)
Dc: sqrt(2) to 99 decimal places 126.4 lpm (30.4 secs, 3 samples)
Recursion Test--Tower of Hanoi 122.8 lps (20.4 secs, 3 samples)
INDEX VALUES
TEST BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 7924.6 0.7
Double-Precision Whetstone 55.0 1.7 0.3
Execl Throughput 43.0 4.0 0.9
File Copy 1024 bufsize 2000 maxblocks 3960.0 60.0 0.2
File Copy 256 bufsize 500 maxblocks 1655.0 18.0 0.1
File Copy 4096 bufsize 8000 maxblocks 5800.0 98.0 0.2
Pipe Throughput 12440.0 1136.2 0.9
Pipe-based Context Switching 4000.0 442.5 1.1
Process Creation 126.0 5.1 0.4
Shell Scripts (8 concurrent) 6.0 1.0 1.7
System Call Overhead 15000.0 1473.7 1.0
=========
FINAL SCORE 0.5
------------------------------------------------------------
On 8KB/page kernel:
------------------------------------------------------------
BYTE UNIX Benchmarks (Version 4.1.0)
System -- libero
Start Benchmark Run: Tue Dec 8 23:10:45 JST 2009
1 interactive users.
11:10PM up 1:14, 1 user, load averages: 0.57, 0.15, 0.05
-r-xr-xr-x 1 root wheel 135783 Apr 27 2009 /bin/sh
/bin/sh: ELF 32-bit MSB executable, Motorola 68020, version 1 (SYSV), \
dynamically linked (uses shared libs), for NetBSD 5.0, not stripped
/dev/sd0a 314015 44593 253722 14% /
Dhrystone 2 using register variables 8137.5 lps (10.3 secs, 10 samples)
Double-Precision Whetstone 1.7 MWIPS (6.0 secs, 10 samples)
System Call Overhead 1656.1 lps (10.3 secs, 10 samples)
Pipe Throughput 1188.3 lps (10.3 secs, 10 samples)
Pipe-based Context Switching 466.7 lps (10.3 secs, 10 samples)
Process Creation 10.5 lps (30.3 secs, 3 samples)
Execl Throughput 4.9 lps (30.1 secs, 3 samples)
File Read 1024 bufsize 2000 maxblocks 1015.0 KBps (30.0 secs, 3 samples)
File Write 1024 bufsize 2000 maxblocks 133.0 KBps (30.0 secs, 3 samples)
File Copy 1024 bufsize 2000 maxblocks 63.0 KBps (30.0 secs, 3 samples)
File Read 256 bufsize 500 maxblocks 394.0 KBps (30.0 secs, 3 samples)
File Write 256 bufsize 500 maxblocks 33.0 KBps (30.0 secs, 3 samples)
File Copy 256 bufsize 500 maxblocks 20.0 KBps (30.0 secs, 3 samples)
File Read 4096 bufsize 8000 maxblocks 549.0 KBps (30.0 secs, 3 samples)
File Write 4096 bufsize 8000 maxblocks 266.0 KBps (30.0 secs, 3 samples)
File Copy 4096 bufsize 8000 maxblocks 123.0 KBps (30.0 secs, 3 samples)
Shell Scripts (1 concurrent) 8.3 lpm (60.4 secs, 3 samples)
Shell Scripts (8 concurrent) 1.0 lpm (20.2 secs, 3 samples)
Shell Scripts (16 concurrent) 1.0 lpm (0.0 secs, 3 samples)
Arithmetic Test (type = short) 1441.5 lps (10.6 secs, 3 samples)
Arithmetic Test (type = int) 1482.8 lps (10.3 secs, 3 samples)
Arithmetic Test (type = long) 1485.0 lps (10.3 secs, 3 samples)
Arithmetic Test (type = float) 683.5 lps (10.3 secs, 3 samples)
Arithmetic Test (type = double) 647.5 lps (10.3 secs, 3 samples)
Arithoh 629653.3 lps (10.2 secs, 3 samples)
C Compiler Throughput 1.3 lpm (60.4 secs, 3 samples)
Dc: sqrt(2) to 99 decimal places 152.4 lpm (30.4 secs, 3 samples)
Recursion Test--Tower of Hanoi 131.7 lps (20.3 secs, 3 samples)
INDEX VALUES
TEST BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 8137.5 0.7
Double-Precision Whetstone 55.0 1.7 0.3
Execl Throughput 43.0 4.9 1.1
File Copy 1024 bufsize 2000 maxblocks 3960.0 63.0 0.2
File Copy 256 bufsize 500 maxblocks 1655.0 20.0 0.1
File Copy 4096 bufsize 8000 maxblocks 5800.0 123.0 0.2
Pipe Throughput 12440.0 1188.3 1.0
Pipe-based Context Switching 4000.0 466.7 1.2
Process Creation 126.0 10.5 0.8
Shell Scripts (8 concurrent) 6.0 1.0 1.7
System Call Overhead 15000.0 1656.1 1.1
=========
FINAL SCORE 0.6
------------------------------------------------------------
---
Izumi Tsutsui
Home |
Main Index |
Thread Index |
Old Index