1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
TEST_HEADER
 id = $Id: //info.ravenbrook.com/project/mps/master/test/function/500.txt#3 $
 summary = compare MM and boehm performance in Dylan
 language = english
END_HEADER

1. Get a recent, stable release of Dylan. If you're not sure what release would
be appropriate, ask Tony.

2. Create a version of hqn-dylan.dll that uses the memory manager to be tested.
Do this by compiling the dylan library, then copying the mmdw.lib you want to
the dylan\lib directory, then from the dylan\build\dylan directory run
"build microsoft force dll". This is assuming you have not already done
step 3, which creates a new pentium-run-time.lib.

3. Create a version of hqn-dylan.dll that uses the boehm collector.
Do this by running "nmake install-boehm" from from the Sources\lib\pentium-run-time
directory.

4. Run the following tests on each version:

dylan-compile
compile-library -force-parse -force-compile harlequin-extensions
compile-library -force-parse -force-compile harlequin-extensions
compile-library -force-compile dylan
quit

dylan-compile
update-libraries -force -save -flush dfmc-modeling
quit

Or make up some other similar tests.
Compare CPU times and lines per minute as reported by Dylan.

The test is passed if the MM being tested is as fast as or faster than the
boehm collector (in all cases). The two may have different memory requirements,
so make sure to note how much memory your machine has, as well as the version of
dylan used, and to check whether cpu time and clock-on-the-wall time are noticeably
different.

rit 1998-03-13