I am working with the devg-chips_hiqv.so library on an Arcom SBC-Geode
development board. (x86, Chips & Technologies 69000 video hardware) I am
wondering why this video hardware is so slow. I ran the bench utility on
the Arcom board (233 MHz) and on my development machine (x86, approx. 500
MHz, devg-rage.so graphics library). When I compare the results of the
individual tests, the Arcom is much slower than my development machine.
Even accounting for the difference in CPU speed, there is still a
significant difference between the video hardware. Anyone have any ideas?
The test results are below.
thanks
Charlie
ARCOM Results:
*** Core Tests ***
*** Context Tests (opaque) ***
context_draw_span: 674233 ops (44948/sec)
context_draw_rect 89773 ops (5984/sec)
*** Context Tests (trans pattern) ***
context_draw_span: 503495 ops (33566/sec)
context_draw_rect 74643 ops (4976/sec)
*** Context Tests (mono pattern) ***
context_draw_span: 491858 ops (32790/sec)
context_draw_rect 74729 ops (4981/sec)
*** Context Tests (source alpha global) ***
context_draw_span: 217617 ops (14507/sec)
context_draw_rect 4362 ops (290/sec)
*** Context Tests (source alpha map) ***
context_draw_span: 3551 ops (236/sec)
context_draw_rect 15 ops (1/sec)
Development Machine Results
*** Core Tests ***
*** Context Tests (opaque) ***
context_draw_span: 9978218 ops (665214/sec)
context_draw_rect 56632 ops (3775/sec)
*** Context Tests (trans pattern) ***
context_draw_span: 1821080 ops (121405/sec)
context_draw_rect 56339 ops (3755/sec)
*** Context Tests (mono pattern) ***
context_draw_span: 1649830 ops (109988/sec)
context_draw_rect 56907 ops (3793/sec)
*** Context Tests (source alpha global) ***
context_draw_span: 8013948 ops (534263/sec)
context_draw_rect 311092 ops (20739/sec)
*** Context Tests (source alpha map) ***
context_draw_span: 93904 ops (6260/sec)
context_draw_rect 290 ops (19/sec)