View Single Post
  #4 (permalink)  
Old 04-20-2009, 05:21 PM
Tenny's Avatar
Tenny
Halfway to VIP Status
Offline
Location: Hayward, CA
 
Join Date: Jul 2007
Posts: 516
Reputation: 250
Tenny is becoming a PPCGeeks regularTenny is becoming a PPCGeeks regularTenny is becoming a PPCGeeks regular
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Send a message via AIM to Tenny
Re: Omnia vs OEM CDMA d3d drivers

Alright, so i decided to do a bit of testing. I tested the Omnia vs. OEM CDMA D3D vs OEM GSM D3D.

So I built a rom off Calk's kitchen (21042). They all had the same settings, etc. I used both SPB Benchmark 1.6 (Freeware) and Dynamixsoft 3D Benchmark (Freeware).

With the SPB Benchmark Results, FS means File System, GRF means Graphics. I included both just incase anyone wanted to see the results.

Each rom was built with the following selected in Calcs 4.6.1 Kitchen:
Code:
System
-Games
-Internet Explorer
-Office Mobile
-Remote Desktop
-SqlCeMobile
-Rhodium Ringtones
Carrier
-Generic
D3D
-Omnia First
Icons
-3D
-HTC Logo
Apps
-Dynamix 3D Benchmark
-SPB Benchmark
Tweaks
-3 Row
-Battery
-Preload
-Registry

Also had Topaz Notification Enhancement. (forgot to remove it from OEM folder).


Problems with this test:

1. I realize that after a few hard resets the device becomes faster, so I tested them w/o any hard resets, to try and mimic the results as much as possible. To combat this problem, I included FS tests in SPB Benchmark, as the D3D drivers should not affect the FS.
2. I don't know if either of the graphics tests utilize D3D. (big problem...)

Anyways some results:
Omnia
SPB Benchmark - (Results are in MS (miliseconds). Lower numbers = Better)
Quote:
fs-largefilewrite 637.300000
fs-largefileread 39.500000
fs-largefilecopy 771.000000
fs-manyfileswrite 4201.800000
fs-manyfilesread 247.200000
fs-manyfilescopy 3604.000000
fs-dirlisting 1377.600000
fs-db 735.800000
grf-ddb 8.800000
grf-dib 34.170000
grf-gapi 28.135000
Dynamix 3D Benchmark - (Results are in Frames Per Second, Higher numbers are better)
Quote:
Average FPS: 31.66
Min FPS: 25
Max FPS: 33

OEM GSM D3D

SPB Benchmark - (Results are in MS (miliseconds). Lower numbers = Better)
Quote:
fs-largefilewrite 620.100000
fs-largefileread 39.100000
fs-largefilecopy 765.300000
fs-manyfileswrite 3680.500000
fs-manyfilesread 242.500000
fs-manyfilescopy 3429.600000
fs-dirlisting 1281.600000
fs-db 677.500000
grf-ddb 8.810000
grf-dib 34.345000
grf-gapi 28.955000
Dynamix 3D Benchmark - (Results are in Frames Per Second, Higher numbers are better)
Quote:
Average FPS: 31.80
Min FPS: 19
Max FPS: 33
OEM CDMA D3D
SPB Benchmark - (Results are in MS (miliseconds). Lower numbers = Better)
Quote:
fs-largefilewrite 613.200000
fs-largefileread 35.500000
fs-largefilecopy 753.800000
fs-manyfileswrite 3604.600000
fs-manyfilesread 238.500000
fs-manyfilescopy 3399.200000
fs-dirlisting 1282.100000
fs-db 672.400000
grf-ddb 6.165000
grf-dib 24.060000
grf-gapi 5.760000
Dynamix 3D Benchmark - (Results are in Frames Per Second, Higher numbers are better)
Quote:
Average FPS: 57.16
Min FPS: 30
Max FPS: 67
Conclusion:
No it's not the most scientific test that can be done. I didn't think it through when I started. Despite the problems, there appears to be a clear winner: OEM CDMA D3D. For all future roms that I cook, they will all be with that until a new driver comes out, or another test the proves my test results with an actual 100% D3D benchmark.
__________________
Treo 600 > Treo 650 > Motorola Q > Sprint Mogul > Sprint Touch > Sprint Touch Diamond > Sprint Hero > HTC EVO 4G

Last edited by Tenny; 04-20-2009 at 05:45 PM.
Reply With Quote
This post has been thanked 9 times.