Commit 6be78e2a authored by Simon Peyton Jones's avatar Simon Peyton Jones
Browse files

Much better perf on 5030, higher peak on 3064

The peak-megabytes is vulnerable to when GC runs, but
the bytes-allocated on 5030 shows a big reduction from
466Mbytes to 367Mbytes.

This follows the arity improvement and CSE patches
parent c6ccdbbe
......@@ -173,7 +173,7 @@ test('T3064',
compiler_stats_num_field('peak_megabytes_allocated',
[(wordsize(32), 14, 1),
# expected value: 14 (x86/Linux 28-06-2012):
(wordsize(64), 30, 20)]),
(wordsize(64), 37, 20)]),
# (amd64/Linux): 18
# (amd64/Linux) 2012-02-07: 26
# (amd64/Linux) 2013-02-12: 23; increased range to 10%
......@@ -181,6 +181,7 @@ test('T3064',
# (amd64/Linux) 2013-09-11: 30; result of AMP patch
# Increased range to 20%. peak-usage varies from 22 to 26,
# depending on whether the old .hi file exists
# (amd64/Linux) 2013-09-11: 37; better arity analysis (weird)
compiler_stats_num_field('bytes allocated',
[(wordsize(32), 111189536, 10),
......@@ -226,6 +227,8 @@ test('T5030',
# ^ reason unknown
# 2013-08-02 454498592 (amd64/Linux)
# decrease from more aggressive coercion optimisations from roles
# 2013-11-12 397672152 (amd64/Linux)
# big decrease following better CSE and arity
only_ways(['normal'])
],
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment