diff options
Diffstat (limited to 'TAO/performance-tests/Cubit/COOL/MT_Cubit/README')
| -rw-r--r-- | TAO/performance-tests/Cubit/COOL/MT_Cubit/README | 166 |
1 files changed, 0 insertions, 166 deletions
diff --git a/TAO/performance-tests/Cubit/COOL/MT_Cubit/README b/TAO/performance-tests/Cubit/COOL/MT_Cubit/README deleted file mode 100644 index f8e7941cbdf..00000000000 --- a/TAO/performance-tests/Cubit/COOL/MT_Cubit/README +++ /dev/null @@ -1,166 +0,0 @@ -The MT-Cubit example for the CHORUS COOL ORB --------------------------------------------- - -Compiling ---------- - -Setup the CHORUS COOL environment variables before compiling, the -following is an example. You would typically want this in the shells' -resource file (e.g. .cshrc.mine). The example provided is for a C -shell. - - setenv CHORUS_ROOT /project/doc/miniCOOL/minicool.r4.3/solaris2.5-CC-mt - set path = ($path /project/doc/miniCOOL/minicool.r4.3/solaris2.5-CC-mt/bin) - setenv LD_LIBRARY_PATH /project/doc/miniCOOL/minicool.r4.3/solaris2.5-CC-mt/lib:$LD_LIBRARY_PATH - setenv MANPATH /project/doc/miniCOOL/minicool.r4.3/solaris2.5-CC-mt/man:$MANPATH - -Makefiles are provided. Use make to compile. - -You can either run the server in the background in the same window as -the client or open a separate window for the client and server. - -server: -------- - -./server <tpr | tpool> - <cool-tcp://THIS_HOSTNAME:THIS_HOST_PORT | cool-chorus://:THIS_HOST_PORT> - -v - <-cool-tcp -cool-iiop | -cool-chorus> - -tpr : this indicates to the server to use the Thread Per Request concurrency model. - -tpool : this indicates to the server to use the Thread Pool concurrency model. - -THIS_HOSTNAME : the name of the host that the server will be running on. This is useful - if you want to specifya different interface (e.g. ATM, 100Mbit Ethernet) - -THIS_HOST_PORT: is the port number at which the server will start the high - priority servant. The low priority servant will be created at - (THIS_HOST_PORT+1). - -The server will print the IOR of the high and low priority servants, -but it is not required by the client. - -client: -------- - -./client [-n <LOOP_COUNT>] - [-t <NUM_THREADS>] - [-h SERVER_HOSTNAME] - [-p <SERVER_PORT_NUM>] - [-cool-tcp -cool-iiop | -cool-chorus] - [-I] - -[-I] indicates to the client to use Chorus IPC as the underlying transport - instead of TCP - -<SERVER_HOSTNAME>: is the hostname of the server. - -<SERVER_PORT_NUM>: is the port number of the servant. - -Based on the SERVER_HOSTNAME and the SERVER_PORT_NUM, the client constructs -the IOR's for the high priority servant and the low priority servant. - -<LOOP_COUNT> : is the number of times the group of CORBA twoway calls are to - be made. The group consists of the following calls: cube_octet(), cube_short(), - cube_long() and cube_struct(). The default is 100 loops (i.e. 400 calls are made). - -<NUM_THREADS> : is the total number of concurrent clients to be - created. One of these clients will be a high priority client (and thus - send requests to the high priority servant), and the rest will have - low priority. - -The high priority client makes requests at 20Hz, and the low priority -clients make requests at 10Hz. - -Results -------- - -The client prints the latency observed by the high priority client, -and the average latency observed by the low priority client. - -Examples --------- - -(1) An Example on a Sparc running SunOS 5.5.1 using TCP as the underlying transport - - - On the host lambada.cs.wustl.edu : - - % ./server tpool cool-tcp://lambada.cs.wustl.edu:3000 -v -cool-tcp -cool-iiop - - On the host merengue.cs.wustl.edu : - - % ./client -t 2 -h lambada.cs.wustl.edu -p 3000 -cool-tcp -cool-iiop - - The output of the client program may look like this: - - (5) Everyone's done, here I go!! - (4) Threads have bound, utilization test started - (9) Everyone's done, here I go!! - (8) Everyone's done, here I go!! - (6) Everyone's done, here I go!! - (7) Everyone's done, here I go!! - cube average call ACE_OS::time = 20.917016 msec, 47.807966 calls/second - (5) My latency was 20.917016 - 5 calls, 0 errors - cube average call ACE_OS::time = 19.299501 msec, 51.814811 calls/second - (8) My latency was 19.299501 - cube average call ACE_OS::time = 19.262436 msec, 51.914514 calls/second - 5 calls, 0 errors - (9) My latency was 19.262436 - 5 calls, 0 errors - cube average call ACE_OS::time = 15.101540 msec, 66.218412 calls/second - (6) My latency was 15.101540 - 5 calls, 0 errors - cube average call ACE_OS::time = 10.320359 msec, 96.895858 calls/second - (7) My latency was 10.320359 - 5 calls, 0 errors - Test done. - High priority client latency : 20.917016 msec, jitter: 36.582330 msec - Low priority client latency : 15.995959 msec, jitter: 527.515277 msec - utilization task performed 435 computations - -(2) An Example on a MVME177 backplane using 68k processors running ClassiX using - Chorus IPC as the underlying transport - - - On the host tambora2.cs.wustl.edu : - - % rsh tambora2 arun /tests/Cubit/COOL/MT_Cubit/server tpool cool-chorus://:1023 -v -cool-chorus - - On the host tambora1.cs.wustl.edu : - - % rsh tambora1 arun /tests/Cubit/COOL/MT_Cubit/client -n 400 -I -t 2 -p 1023 -cool-chorus - - The output of the client program might look like this: - - started aid = 12 - COOL:0000000001000000010000001C00750AA60000000200750AA65555555500000001429A00000000000A(13) Thread created - (13) Im the high priority client, my id is 0. - Using ior = cool-chorus://:1023 - Creating 1 clients with low priority of 125 - (13) Binding succeeded - (13) Cb == 2c6144 - Object Created at: '2908484l'connected to object 'COOL:0000000001000000010000001C00750AA60000000200750AA65555555500000001429A00000000000A' - COOL:0000000001000000010000001C00750AA60000000200750AA65555555500000001429A00000000000A - (13) Waiting for other threads to finish binding.. - (12) Thread created - (12) Im a low priority client, my id is 1 - Using ior = cool-chorus://:1023 - (12) Binding succeeded - (12) Cb == 2c6164 - Object Created at: '2908516l'connected to object 'COOL:0000000001000000010000001C00750AA60000000200750AA65555555500000001429A00000000000A'(12) Waiting for other threads to finish binding.. - (13) Everyone's done, here I go!! - (12) Everyone's done, here I go!! - (12|13) cube average call ACE_OS::time = 12835 usec, 77 calls/second - (13) My latency was 12835 - 400 calls, 0 errors - (12|12) cube average call ACE_OS::time = 12931 usec, 77 calls/second - (12) My latency was 12931 - Test done. - High priority client latency : 12835 usec - Low priority client latency : 12931 usec - 400 calls, 0 errors - --->Output file for latency data is "cb12__2.txt" - (8) utilization task performed 0 computations |