Difference between revisions of "Benchmark DirectFB"

From eLinux.org
Jump to: navigation, search
(Benchmarks)
(added page to SuperH category)
 
(One intermediate revision by one other user not shown)
Line 24: Line 24:
 
|-
 
|-
 
|E||Intel Pentium4||2.4GHz||AGP||1GB||Matrox G450||2.4.20
 
|E||Intel Pentium4||2.4GHz||AGP||1GB||Matrox G450||2.4.20
 +
|-
 +
|G||rowspan="2"|Intel CE2110||rowspan="2"|1Ghz||rowspan="2"|CPU||rowspan="2"|100Mb||software drawing||2.6.16
 +
|-
 +
|G*||Intel GDL accelerated||2.6.16
 
|}
 
|}
  
Line 29: Line 33:
 
{| border="1" cellpadding="5" cellspacing="0" align="center"
 
{| border="1" cellpadding="5" cellspacing="0" align="center"
 
! style="background:#80FF80;" rowspan="2"|'''Benchmarks'''
 
! style="background:#80FF80;" rowspan="2"|'''Benchmarks'''
! style="background:#80FF80;" colspan="7"|'''Platform'''
+
! style="background:#80FF80;" colspan="9"|'''Platform'''
 
|-
 
|-
 
! style="background:#80FF80;"|'''A'''
 
! style="background:#80FF80;"|'''A'''
Line 38: Line 42:
 
! style="background:#80FF80;"|'''E'''
 
! style="background:#80FF80;"|'''E'''
 
! style="background:#80FF80;"|'''F'''
 
! style="background:#80FF80;"|'''F'''
 +
! style="background:#80FF80;"|'''G'''
 +
! style="background:#80FF80;"|'''G*'''
 
|-
 
|-
|Anti-aliased Text [KChars/sec]||29.22||29.41||20.40||24.83||23.96||607.39||750.00
+
|Anti-aliased Text [KChars/sec]||29.22||29.41||20.40||24.83||23.96||607.39||750.00||84||45
 
|-
 
|-
|Anti-aliased Text (blend)  [KChars/sec]||7.52||7.47||6.12||16.52||16.66||613.00||752.85
+
|Anti-aliased Text (blend)  [KChars/sec]||7.52||7.47||6.12||16.52||16.66||613.00||752.85||25||19
 
|-
 
|-
|Fill Rectangles [MPixel/sec]||style="background:#80FFFF;"|14.07||style="background:#80FFFF;"|221.49||63.63||116.37||53.25||892.88||849.22
+
|Fill Rectangles [MPixel/sec]||style="background:#80FFFF;"|14.07||style="background:#80FFFF;"|221.49||63.63||116.37||53.25||892.88||849.22||style="background:#80FFFF;"|34||style="background:#80FFFF;"|105
 
|-
 
|-
|Fill Rectangles (blend) [MPixel/sec]||1.64||1.67||1.20||3.18||3.26||236.38||225.84
+
|Fill Rectangles (blend) [MPixel/sec]||1.64||1.67||1.20||3.18||3.26||236.38||225.84||style="background:#80FFFF;"|3||style="background:#80FFFF;"|6
 
|-
 
|-
|Fill Triangles [MPixel/sec]||style="background:#80FFFF;"|12.25||style="background:#80FFFF;"|93.87||93.87||108.79||50.51||748.55||730.24
+
|Fill Triangles [MPixel/sec]||style="background:#80FFFF;"|12.25||style="background:#80FFFF;"|93.87||93.87||108.79||50.51||748.55||730.24||style="background:#80FFFF;"|32||style="background:#80FFFF;"|1
 
|-
 
|-
|Fill Triangles (blend) [MPixel/sec]||1.63||1.61||1.17||3.13||3.17||223.32||218.24
+
|Fill Triangles (blend) [MPixel/sec]||1.63||1.61||1.17||3.13||3.17||223.32||218.24||style="background:#80FFFF;"|4||style="background:#80FFFF;"|0.026
 
|-
 
|-
|Draw Rectangles [KRects/sec]||style="background:#80FFFF;"|1.81||style="background:#80FFFF;"|16.01||10.67||12.95||8.57||59.53||36.27
+
|Draw Rectangles [KRects/sec]||style="background:#80FFFF;"|1.81||style="background:#80FFFF;"|16.01||10.67||12.95||8.57||59.53||36.27||style="background:#80FFFF;"|11||style="background:#80FFFF;"|3.6
 
|-
 
|-
|Draw Rectangles (blend) [KRects/sec]||0.52||0.57||0.43||0.83||0.84||25.40||17.09
+
|Draw Rectangles (blend) [KRects/sec]||0.52||0.57||0.43||0.83||0.84||25.40||17.09||style="background:#80FFFF;"|2.142||style="background:#80FFFF;"|0.194
 
|-
 
|-
|Draw Lines [KLines/sec]||style="background:#80FFFF;"|7.10||style="background:#80FFFF;"|66.98||61.33||62.60||48.84||281.86||162.40
+
|Draw Lines [KLines/sec]||style="background:#80FFFF;"|7.10||style="background:#80FFFF;"|66.98||61.33||62.60||48.84||281.86||162.40||style="background:#80FFFF;"|41||style="background:#80FFFF;"|16
 
|-
 
|-
|Draw Lines (blend) [KLines/sec]||2.33||2.43||1.94||3.69||3.70||127.29||80.04
+
|Draw Lines (blend) [KLines/sec]||2.33||2.43||1.94||3.69||3.70||127.29||80.04||style="background:#80FFFF;"|10||style="background:#80FFFF;"|7
 
|-
 
|-
|Blit [MPixel/sec]||style="background:#80FFFF;"|8.12||style="background:#80FFFF;"|100.49||38.68||53.75||32.56||435.02||398.84
+
|Blit [MPixel/sec]||style="background:#80FFFF;"|8.12||style="background:#80FFFF;"|100.49||38.68||53.75||32.56||435.02||398.84||style="background:#80FFFF;"|25||style="background:#80FFFF;"|64
 
|-
 
|-
|Blit colorkeyed [MPixel/sec]||style="background:#80FFFF;"|1.63||style="background:#80FFFF;"|102.16||39.19||58.69||32.54||445.20||421.97
+
|Blit colorkeyed [MPixel/sec]||style="background:#80FFFF;"|1.63||style="background:#80FFFF;"|102.16||39.19||58.69||32.54||445.20||421.97||style="background:#80FFFF;"|28||style="background:#80FFFF;"|36
 
|-
 
|-
|Blit with format conversion [MPixel/sec]||4.04||4.13||3.59||18.11||17.79||203.38||193.26
+
|Blit with format conversion [MPixel/sec]||4.04||4.13||3.59||18.11||17.79||203.38||193.26||style="background:#80FFFF;"|13||style="background:#80FFFF;"|38
 
|-
 
|-
|Blit from 32bit (alphachannel blend) [MPixel/sec]||1.05||1.05||0.82||2.71||2.71||171.32||158.10
+
|Blit from 32bit (alphachannel blend) [MPixel/sec]||1.05||1.05||0.82||2.71||2.71||171.32||158.10||3||36
 
|-
 
|-
|Blit from 8bit palette [MPixel/sec]||3.54||2.35||3.20||17.40||17.38||17.28||95.17
+
|Blit from 8bit palette [MPixel/sec]||3.54||2.35||3.20||17.40||17.38||17.28||95.17||-||-
 
|-
 
|-
|Blit from 8bit palette (alphachannel blend) [MPixel/sec]||1.02||0.90||0.81||2.67||2.71||3.55||5.53
+
|Blit from 8bit palette (alphachannel blend) [MPixel/sec]||1.02||0.90||0.81||2.67||2.71||3.55||5.53||-||-
 
|-
 
|-
|Stretch Blit [MPixel/sec]||style="background:#80FFFF;"|12.17||style="background:#80FFFF;"|83.50||7.06||83.50||47.61||210.32||220.77
+
|Stretch Blit [MPixel/sec]||style="background:#80FFFF;"|12.17||style="background:#80FFFF;"|83.50||7.06||83.50||47.61||210.32||220.77||25||21
 
|-
 
|-
|Stretch Blit colorkeyed [MPixel/sec]||7.46||7.43||4.20||46.17||46.30||211.97||221.64
+
|Stretch Blit colorkeyed [MPixel/sec]||7.46||7.43||4.20||46.17||46.30||211.97||221.64||26||20
 
|}
 
|}
  
  
 
After enabling 2D acceleration in A* environment, the graphics benchmark score improves. It looks like even embedded processor like SH-4 can be used when DirectFB can use hardware acceleration. So a graphics library would be essential in a graphical system.
 
After enabling 2D acceleration in A* environment, the graphics benchmark score improves. It looks like even embedded processor like SH-4 can be used when DirectFB can use hardware acceleration. So a graphics library would be essential in a graphical system.
 +
 +
For G* (Intel CE2110) environment, just basic operations like opaque rectangle and blits are really optimized, some operations are much slower and we can notice rendering artifacts/bugs. So take care, possible you'll have to fallback to software rendering on such platform. Cyan-colored cells are supposed to be accelerated.
  
 
[[Category:DirectFB]]
 
[[Category:DirectFB]]
 
[[Category:Multimedia]]
 
[[Category:Multimedia]]
 +
[[Category:SuperH]]

Latest revision as of 01:26, 12 May 2010

Benchmarks

The DirectFB example suites include benchmark 'df_dok'. We have ran this benchmark on the following platform:

Platform CPU Clock I/F System RAM Graphics Card Kernel Version
A Renesas SH-4 240MHz CPU 64MB SMI SM501 software drawing 2.4.20 CELF
A* SMI SM501 enable 2D acceleration 2.4.20 CELF
B Renesas SH-4 240MHz PCI 64MB Matrox Millenium 2.4.19
C Intel Celeron 450MHz PCI 128MB Matrox Mystique 2.4.20
D Intel Celeron 450MHz PCI 128MB Matrox Millenium 2.4.20
F Intel Celeron 450MHz AGP 128MB Matrox G450 2.4.20
E Intel Pentium4 2.4GHz AGP 1GB Matrox G450 2.4.20
G Intel CE2110 1Ghz CPU 100Mb software drawing 2.6.16
G* Intel GDL accelerated 2.6.16


Benchmarks Platform
A A* B C D E F G G*
Anti-aliased Text [KChars/sec] 29.22 29.41 20.40 24.83 23.96 607.39 750.00 84 45
Anti-aliased Text (blend) [KChars/sec] 7.52 7.47 6.12 16.52 16.66 613.00 752.85 25 19
Fill Rectangles [MPixel/sec] 14.07 221.49 63.63 116.37 53.25 892.88 849.22 34 105
Fill Rectangles (blend) [MPixel/sec] 1.64 1.67 1.20 3.18 3.26 236.38 225.84 3 6
Fill Triangles [MPixel/sec] 12.25 93.87 93.87 108.79 50.51 748.55 730.24 32 1
Fill Triangles (blend) [MPixel/sec] 1.63 1.61 1.17 3.13 3.17 223.32 218.24 4 0.026
Draw Rectangles [KRects/sec] 1.81 16.01 10.67 12.95 8.57 59.53 36.27 11 3.6
Draw Rectangles (blend) [KRects/sec] 0.52 0.57 0.43 0.83 0.84 25.40 17.09 2.142 0.194
Draw Lines [KLines/sec] 7.10 66.98 61.33 62.60 48.84 281.86 162.40 41 16
Draw Lines (blend) [KLines/sec] 2.33 2.43 1.94 3.69 3.70 127.29 80.04 10 7
Blit [MPixel/sec] 8.12 100.49 38.68 53.75 32.56 435.02 398.84 25 64
Blit colorkeyed [MPixel/sec] 1.63 102.16 39.19 58.69 32.54 445.20 421.97 28 36
Blit with format conversion [MPixel/sec] 4.04 4.13 3.59 18.11 17.79 203.38 193.26 13 38
Blit from 32bit (alphachannel blend) [MPixel/sec] 1.05 1.05 0.82 2.71 2.71 171.32 158.10 3 36
Blit from 8bit palette [MPixel/sec] 3.54 2.35 3.20 17.40 17.38 17.28 95.17 - -
Blit from 8bit palette (alphachannel blend) [MPixel/sec] 1.02 0.90 0.81 2.67 2.71 3.55 5.53 - -
Stretch Blit [MPixel/sec] 12.17 83.50 7.06 83.50 47.61 210.32 220.77 25 21
Stretch Blit colorkeyed [MPixel/sec] 7.46 7.43 4.20 46.17 46.30 211.97 221.64 26 20


After enabling 2D acceleration in A* environment, the graphics benchmark score improves. It looks like even embedded processor like SH-4 can be used when DirectFB can use hardware acceleration. So a graphics library would be essential in a graphical system.

For G* (Intel CE2110) environment, just basic operations like opaque rectangle and blits are really optimized, some operations are much slower and we can notice rendering artifacts/bugs. So take care, possible you'll have to fallback to software rendering on such platform. Cyan-colored cells are supposed to be accelerated.