BC3 seems to be much slower (44%) than BC2 doing binary comparisons of my data. Is everyone seeing a similar slowdown? Is it expected or a bug?
Here's my setup so you an judge relative speeds.
BC 2.5.1 (build 251)
BC 3.0.1 (build 8602)
Windows 2000 Pro. SP5
Intel P4 1.7 GHz
1.256 GB
All source is on a separate server over a 100Mbit Ethernet connection.
The server has enough memory that this data is probably always cached
and the server is pretty much dedicated to serving only me when I run
the tests. So, variance in network and server delays is not significant.
Binary comparison, all settings as similar as I can make them.
Comparing two identical directory trees on the same server.
177 MB total data.
2800 C, C++ source files.
2000 directories in the tree - most are empty - about 200 contain files.
Tests repeated several times on the same box,
in quick succession but varying order.
Sometimes 3.0.1 goes first, sometimes 2.5.1
Nothing else running. Before the tests task manager performance
reports 0% to 2% utilization and the System Idle Process is getting
99% cpu. No other process shows anything but 0% cpu utilization.
Results are consistent within one or two seconds across all runs.
BC 2.5.1 takes 45 seconds to do a full refresh.
BC 3.0.1 takes 65 seconds to do a full refresh.
(65-44)/44 is approximately 0.44, so BC3 is 44% slower than BC2.
By the way, a timestamp only comparison takes 5 seconds for BC2
and 7 seconds for BC3. (7-5)/5 is 0.4 so even that is 40% slower!
Here's my setup so you an judge relative speeds.
BC 2.5.1 (build 251)
BC 3.0.1 (build 8602)
Windows 2000 Pro. SP5
Intel P4 1.7 GHz
1.256 GB
All source is on a separate server over a 100Mbit Ethernet connection.
The server has enough memory that this data is probably always cached
and the server is pretty much dedicated to serving only me when I run
the tests. So, variance in network and server delays is not significant.
Binary comparison, all settings as similar as I can make them.
Comparing two identical directory trees on the same server.
177 MB total data.
2800 C, C++ source files.
2000 directories in the tree - most are empty - about 200 contain files.
Tests repeated several times on the same box,
in quick succession but varying order.
Sometimes 3.0.1 goes first, sometimes 2.5.1
Nothing else running. Before the tests task manager performance
reports 0% to 2% utilization and the System Idle Process is getting
99% cpu. No other process shows anything but 0% cpu utilization.
Results are consistent within one or two seconds across all runs.
BC 2.5.1 takes 45 seconds to do a full refresh.
BC 3.0.1 takes 65 seconds to do a full refresh.
(65-44)/44 is approximately 0.44, so BC3 is 44% slower than BC2.
By the way, a timestamp only comparison takes 5 seconds for BC2
and 7 seconds for BC3. (7-5)/5 is 0.4 so even that is 40% slower!
Comment