Since I'm going to be talking about the megadrivers idea next week at
XDC, I thought I'd look at what the impact would be of an alternate
option, and to do that I wanted to first fix up any stupid wastes of
disk space I found. I found more than I expected.
The last commit is an important one. I think it accounts for most of
the performance change of the series (and the performance loss that we
got from dricore). I'm planning on comparing this series to
non-dricore and to megadrivers for the talk.
Size of mesa before:
text data bss dec hex filename
872896 19912 3392 896200 dacc8 lib/i965_dri.so
3387494 112060 95592 3595146 36db8a lib/libdricore9.3.0-devel.so.1
After:
872850 19912 3392 896154 dac9a lib/i965_dri.so
3184667 88524 95592 3368783 33674f lib/libdricore9.3.0-devel.so.1
(a little over 6% reduction)
Results of testing of CPU overhead in mesa (cairo-gl with
INTEL_NO_HW=1. Also had a patch for reducing _mesa_DummyProgram size
that I found was broken after I piglited):
x before
+ after
+------------------------------------------------------------------------------+
| + x |
| + x |
| + x |
| + xx |
| ++ + xx |
| ++ + xx |
| ++++xxxx |
| ++++xxxxx + |
| ++++xxxxx x+ x + + + x |
| ++++xxxxx++* +x++x+++ + x + + x |
| ++++*xxxx+** +x++x+++ + * +x + + + xx x |
| ++++**xxx+**+*x++x+** x ++x+ * * x+ +x ++x + x+ xx+ + ++* xx xx x |
|+++++***xx+**+*x*+x***x xxx**x*+*+*x**x*x+* ++xxx+ x** x* xx++++++***xx xxxxx|
|+++++********+*x******* **x****+***x**x****+******x*** **xx****+++***xxxxxxxx|
| |__|____________M_____AM___A________________|_______| |
+------------------------------------------------------------------------------+
N Min Max Median Avg Stddev
x 200 29.417143 47.735255 35.27498 36.471098 6.0928765
+ 200 28.254668 45.819881 33.445248 35.119495 5.5758312
Difference at 95.0% confidence
-1.3516 +/- 1.14466
-3.70596% +/- 3.13853%
(Student's t, pooled s = 5.84008)
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
