Hi tom_in_az. Thanks for your help.
When the android is running, the memory usage is the following:
/ # free
total used free shared buffers
Mem: 121508 117184 4324 0 8
Swap: 0 0 0
Total: 121508 117184 4324
When the Linpack is running, the memory usage is the following:
total used free shared buffers
Mem: 121508 116188 5320 0 8
Swap: 0 0 0
Total: 121508 116188 5320
and I've run flops which is double precision and linpack which can run
single precision on my board.
I was using an arm926T at 266MHz with the CFLAGS and compiler shown below:
FLAGS = -O3
gcc version 4.4.1 (Sourcery G++ Lite 2009q3-67)
FLOPS C Program (Double Precision), V2.0 18 Dec 1992
Module Error RunTime MFLOPS
(usec)
1 4.4764e-13 6.4350 2.1756
2 -8.3933e-14 4.3825 1.5973
3 1.2879e-14 5.5400 3.0686
4 4.4187e-14 5.0775 2.9542
5 -3.9857e-14 11.5650 2.5076
6 1.2323e-14 9.6150 3.0161
7 7.9751e-11 9.8700 1.2158
8 -1.7431e-14 9.9500 3.0151
Iterations = 4000000
NullTime (usec) = 0.0000
MFLOPS(1) = *1.8942*
MFLOPS(2) = *1.9559*
MFLOPS(3) = *2.5150*
MFLOPS(4) = *3.0150*
For Linpack I was using the same platform and the same compiler and the
following CFLAGS:
CFLAGS=-O3
/sdmemory # ./linpack_test
Rolled Single Precision Linpack
Rolled Single Precision Linpack
norm. resid resid machep x[0]-1 x[n-1]-1
1.6 3.80277634e-05 1.19209290e-07 -1.38282776e-05
-7.51018524e-06
times are reported for matrices of order 100
dgefa dgesl total kflops unit ratio
times for array with leading dimension of 201
0.15 0.00 0.15 4578 0.44 2.68
0.15 0.01 0.16 4292 0.47 2.86
0.16 0.00 0.16 4292 0.47 2.86
0.15 0.00 0.16 4319 0.46 2.84
times for array with leading dimension of 200
0.16 0.00 0.16 4292 0.47 2.86
0.15 0.01 0.16 4292 0.47 2.86
0.16 0.00 0.16 4292 0.47 2.86
0.15 0.01 0.16 4402 0.45 2.79
Rolled Single Precision *4319* Kflops ; 10 Reps
And I ran the java version Linpack on android 2.2 , the benchmark
result is*1 MFLOPS
*, it's also *1 MFLOPS* on android 1.6.
Then I think the problem may be that my hardware is slow, so android 2.2
can't improve the Linpack, or should I enable the hardware float point
feature when I compile the benchmark? I'm trying it now.
On Fri, Jul 16, 2010 at 2:14 AM, tom_in_az <[email protected]> wrote:
> Did you try ADW Home? Also are you using any cache or JIT?
> One other question, how much free memory do you have once the system
> is running and fully loaded?
> Tom
>
> On Jul 14, 11:07 pm, Mali Laurent <[email protected]> wrote:
> > Yes it is. Although try installing another Launcher it will speed it up a
> > lot.
> >
> > On Thu, Jul 15, 2010 at 7:16 AM, jun ge <[email protected]> wrote:
> > > Hi all,
> >
> > > I’ve ported android 2.2 to my board, but I found that android 2.2 UI
> is
> > > slower than android 1.6. For example, android 2.2 UI keypad response
> time is
> > > longer than android 1.6. When I click the left button my board, it
> takes
> > > 500ms for android 2.2 to response it, but in android 1.6 it is
> responded
> > > immediately. I’ve printed log in my keypad driver, it seems that my
> keypad
> > > event is captured immediately in driver, but it takes 500ms for android
> UI
> > > to response it. I don't what android is doing in the 500ms. I think the
> > > problem should be android "fade in fade out" UI feature, but how can I
> > > disable the feature? Or does anybody have the same problem? Any help is
> > > appreciated, thanks.
> >
> > > ps: my board runs on arm 926t with 128M SDRAM.
> >
> > > Jun Ge
> >
> > > --
> > > unsubscribe:
> > > [email protected]<android-porting%[email protected]>
> <android-porting%[email protected]<android-porting%[email protected]>
> >
> > > website:http://groups.google.com/group/android-porting
>
> --
> unsubscribe:
> [email protected]<android-porting%[email protected]>
> website: http://groups.google.com/group/android-porting
>
--
unsubscribe: [email protected]
website: http://groups.google.com/group/android-porting
