Hi All:
   I'm developing wince 6 OS on my FPGA board, which has a CPU clock
with 24 MHZ. Of course, only run simple applications in the OS. Now I
encountered a problem:
   When I add display driver and shell components into OS design, it
works fine, but the OEMIdle() function is never been called even there
is no operation with the OS. when remove display driver or when
gwes.exe process is Data abort, the OEMIdle() function will be called
frequently.
  My question is: Is the above situation normal or not? if it is
normal, why the gwes.exe and display driver make the CPU so busy?

Which ARM core do you use?
24 MHz should be more than enough for running the core OS (including GWES) 
of CE, we have run FPGA with ARM9 core before and without any problem. The 
display driver sounds more suspected. Have you tried to do a profiling to see 
how CPU time get used? Another suggestion is to use dummy display driver 
(ddi_nop) to reduce the factor.

"XuanKe Liu" wrote:

> Hi All:
>    I'm developing wince 6 OS on my FPGA board, which has a CPU clock
> with 24 MHZ. Of course, only run simple applications in the OS. Now I
> encountered a problem:
>    When I add display driver and shell components into OS design, it
> works fine, but the OEMIdle() function is never been called even there
> is no operation with the OS. when remove display driver or when
> gwes.exe process is Data abort, the OEMIdle() function will be called
> frequently.
>   My question is: Is the above situation normal or not? if it is
> normal, why the gwes.exe and display driver make the CPU so busy?
> .
> 

Hi KMOS:
   Thanks for your reply.
   I=E2=80=99m using ARM cortex-A8. It's should be more powerful than ARM9:=
)

   The ddi.dll seems to be the most suspector, but ActiveSync is not
ok currently, so the profiling
method can't be used.

   Use dummy display driver sounds like a reasonable suggestion, I
will have a try.

   Do you know what the main tasks when display driver works?

On 6=E6=9C=884=E6=97=A5, =E4=B8=8B=E5=8D=884=E6=97=B649=E5=88=86, KMOS <K..=
..@discussions.microsoft.com> wrote:
> Which ARM core do you use?
> 24 MHz should be more than enough for running the core OS (including GWES=
)
> of CE, we have run FPGA with ARM9 core before and without any problem. Th=
e
> display driver sounds more suspected. Have you tried to do a profiling to=
 see
> how CPU time get used? Another suggestion is to use dummy display driver
> (ddi_nop) to reduce the factor.
>
>
>
> "XuanKe Liu" wrote:
> > Hi All:
> > =C2=A0 =C2=A0I'm developing wince 6 OS on my FPGA board, which has a CP=
U clock
> > with 24 MHZ. Of course, only run simple applications in the OS. Now I
> > encountered a problem:
> > =C2=A0 =C2=A0When I add display driver and shell components into OS des=
ign, it
> > works fine, but the OEMIdle() function is never been called even there
> > is no operation with the OS. when remove display driver or when
> > gwes.exe process is Data abort, the OEMIdle() function will be called
> > frequently.
> > =C2=A0 My question is: Is the above situation normal or not? if it is
> > normal, why the gwes.exe and display driver make the CPU so busy?
> > .

On Jun 4, 10:45=C2=A0am, XuanKe Liu <pilixua...@gmail.com> wrote:
> Hi KMOS:
> =C2=A0 =C2=A0Thanks for your reply.
> =C2=A0 =C2=A0I=E2=80=99m using ARM cortex-A8. It's should be more powerfu=
l than ARM9:)
>
> =C2=A0 =C2=A0The ddi.dll seems to be the most suspector, but ActiveSync i=
s not
> ok currently, so the profiling
> method can't be used.
>
> =C2=A0 =C2=A0Use dummy display driver sounds like a reasonable suggestion=
, I
> will have a try.
>
> =C2=A0 =C2=A0Do you know what the main tasks when display driver works?
>
> On 6=E6=9C=884=E6=97=A5, =E4=B8=8B=E5=8D=884=E6=97=B649=E5=88=86, KMOS <K=
....@discussions.microsoft.com> wrote:
>
>
>
> > Which ARM core do you use?
> > 24 MHz should be more than enough for running the core OS (including GW=
ES)
> > of CE, we have run FPGA with ARM9 core before and without any problem. =
The
> > display driver sounds more suspected. Have you tried to do a profiling =
to see
> > how CPU time get used? Another suggestion is to use dummy display drive=
r
> > (ddi_nop) to reduce the factor.
>
> > "XuanKe Liu" wrote:
> > > Hi All:
> > > =C2=A0 =C2=A0I'm developing wince 6 OS on my FPGA board, which has a =
CPU clock
> > > with 24 MHZ. Of course, only run simple applications in the OS. Now I
> > > encountered a problem:
> > > =C2=A0 =C2=A0When I add display driver and shell components into OS d=
esign, it
> > > works fine, but the OEMIdle() function is never been called even ther=
e
> > > is no operation with the OS. when remove display driver or when
> > > gwes.exe process is Data abort, the OEMIdle() function will be called
> > > frequently.
> > > =C2=A0 My question is: Is the above situation normal or not? if it is
> > > normal, why the gwes.exe and display driver make the CPU so busy?
> > > .- Hide quoted text -
>
> - Show quoted text -

Are you really using a Cortex-A8 at 24MHz... it doesn't sound correct
to me? Perhaps you mean that crystal oscillator input is 24MHz. There
is usually a PLL clock generator in the device to boost this up to
much higher frequencies. Some of the Cortex-A8's can run faster than
1GHz CPU clock! You need to find out how the clock tree in the device
is configured... this is usually done in the initial boot code used by
EBOOT.
Andrew.

Hi Andrew:
   It's just at FPGA stage, the real chip is not ok now:)

On 6=E6=9C=884=E6=97=A5, =E4=B8=8B=E5=8D=886=E6=97=B625=E5=88=86, "AndrewSc=
holan[MCTS]" <a...@plextek.co.uk> wrote:
> On Jun 4, 10:45=C2=A0am, XuanKe Liu <pilixua...@gmail.com> wrote:
>

> > - Show quoted text -
>
> Are you really using a Cortex-A8 at 24MHz... it doesn't sound correct
> to me? Perhaps you mean that crystal oscillator input is 24MHz. There
> is usually a PLL clock generator in the device to boost this up to
> much higher frequencies. Some of the Cortex-A8's can run faster than
> 1GHz CPU clock! You need to find out how the clock tree in the device
> is configured... this is usually done in the initial boot code used by
> EBOOT.
> Andrew.

So you don't have KITL for your platform?
KITL connection is strongly recommended, especially for developing a new ASIC.
Besides to KITL, assume the platform does have JTAG port for HW debugger, 
you may want to use a HW debugger supporting PB to make your work more 
efficiency.

"XuanKe Liu" wrote:

> Hi KMOS:
>    Thanks for your reply.
>    I’m using ARM cortex-A8. It's should be more powerful than ARM9:)
> 
>    The ddi.dll seems to be the most suspector, but ActiveSync is not
> ok currently, so the profiling
> method can't be used.
> 
>    Use dummy display driver sounds like a reasonable suggestion, I
> will have a try.
> 
>    Do you know what the main tasks when display driver works?
> 
> On 6月4日, 下午4时49分, KMOS <K....@discussions.microsoft.com> wrote:
> > Which ARM core do you use?
> > 24 MHz should be more than enough for running the core OS (including GWES)
> > of CE, we have run FPGA with ARM9 core before and without any problem. The
> > display driver sounds more suspected. Have you tried to do a profiling to see
> > how CPU time get used? Another suggestion is to use dummy display driver
> > (ddi_nop) to reduce the factor.
> >
> >
> >
> > "XuanKe Liu" wrote:
> > > Hi All:
> > >    I'm developing wince 6 OS on my FPGA board, which has a CPU clock
> > > with 24 MHZ. Of course, only run simple applications in the OS. Now I
> > > encountered a problem:
> > >    When I add display driver and shell components into OS design, it
> > > works fine, but the OEMIdle() function is never been called even there
> > > is no operation with the OS. when remove display driver or when
> > > gwes.exe process is Data abort, the OEMIdle() function will be called
> > > frequently.
> > >   My question is: Is the above situation normal or not? if it is
> > > normal, why the gwes.exe and display driver make the CPU so busy?
> > > .
> 
> .
> 

Hi KMOS:
   The problem has been solved.
   The reason is that OEM RTC related functions implemented
improperly.

   Thanks for you all.

On 6=E6=9C=884=E6=97=A5, =E4=B8=8B=E5=8D=8811=E6=97=B656=E5=88=86, KMOS <K.=
...@discussions.microsoft.com> wrote:
> So you don't have KITL for your platform?
> KITL connection is strongly recommended, especially for developing a new =
ASIC.
> Besides to KITL, assume the platform does have JTAG port for HW debugger,
> you may want to use a HW debugger supporting PB to make your work more
> efficiency.
>
>
>
> "XuanKe Liu" wrote:
> > Hi KMOS:
> > =C2=A0 =C2=A0Thanks for your reply.
> > =C2=A0 =C2=A0I=E2=80=99m using ARM cortex-A8. It's should be more power=
ful than ARM9:)
>
> > =C2=A0 =C2=A0The ddi.dll seems to be the most suspector, but ActiveSync=
 is not
> > ok currently, so the profiling
> > method can't be used.
>
> > =C2=A0 =C2=A0Use dummy display driver sounds like a reasonable suggesti=
on, I
> > will have a try.
>
> > =C2=A0 =C2=A0Do you know what the main tasks when display driver works?
>
> > On 6=E6=9C=884=E6=97=A5, =E4=B8=8B=E5=8D=884=E6=97=B649=E5=88=86, KMOS =
<K....@discussions.microsoft.com> wrote:
> > > Which ARM core do you use?
> > > 24 MHz should be more than enough for running the core OS (including =
GWES)
> > > of CE, we have run FPGA with ARM9 core before and without any problem=
.. The
> > > display driver sounds more suspected. Have you tried to do a profilin=
g to see
> > > how CPU time get used? Another suggestion is to use dummy display dri=
ver
> > > (ddi_nop) to reduce the factor.
>
> > > "XuanKe Liu" wrote:
> > > > Hi All:
> > > > =C2=A0 =C2=A0I'm developing wince 6 OS on my FPGA board, which has =
a CPU clock
> > > > with 24 MHZ. Of course, only run simple applications in the OS. Now=
 I
> > > > encountered a problem:
> > > > =C2=A0 =C2=A0When I add display driver and shell components into OS=
 design, it
> > > > works fine, but the OEMIdle() function is never been called even th=
ere
> > > > is no operation with the OS. when remove display driver or when
> > > > gwes.exe process is Data abort, the OEMIdle() function will be call=
ed
> > > > frequently.
> > > > =C2=A0 My question is: Is the above situation normal or not? if it =
is
> > > > normal, why the gwes.exe and display driver make the CPU so busy?
> > > > .
>
> > .