quadrant benchmark: 0xdroid in DEVKIT8000 versus a custom rom in motorola DEFY

[航 劲]

Hi,

i did a quandrant bench mark test on 0xdroid with DEVKIT8000,
the results are somehow lower than my expected.

I am using the eclair with 3d support prebuild image. the overall mark
is 750 something.
this is much lower than a custom rom (MIUI) running on my motorola
defy with 2.3 sdk.
the custom rom on defy gives ~2000.

I can clearly see that the 3d test is broken on devkit8000. there is
one test with a rotating planet
but on 0xdroid in devkit8000, the planet becomes pieces (seriously).

I understand that the defy is using 2.3 android and 0xdroid is 2.1.
but the difference is still too much.
considering 0xdroid already enable the hardware 3d acceleration, the
results should be better.

of course, the devkit is running a 600Mhz 3530 but defy is said to be
with 3630. this could be the
most significant difference.

[Jim Huang]

2012/1/15 航 劲 <pixelse...@gmail.com>:

> i did a quandrant bench mark test on 0xdroid with DEVKIT8000,
> the results are somehow lower than my expected.

hi!

Thanks for sharing the comparison materials to us!

> I am using the eclair with 3d support prebuild image. the overall mark
> is 750 something.
> this is much lower than a custom rom (MIUI) running on my motorola
> defy with 2.3 sdk.
> the custom rom on defy gives ~2000.
>
> I can clearly see that the 3d test is broken on devkit8000. there is
> one test with a rotating planet
> but on 0xdroid in devkit8000, the planet becomes pieces (seriously).

Okay, can you post some pictures or scores regarding your experiments
over different software combination?

> I understand that the defy is using 2.3 android and 0xdroid is 2.1.
> but the difference is still too much.
> considering 0xdroid already enable the hardware 3d acceleration, the
> results should be better.
>
> of course, the devkit is running a 600Mhz 3530 but defy is said to be
> with 3630. this could be the
> most significant difference.

Yes, I think so. OMAP3430 and OMAP3630 have slight difference between
each other. Please share the information about CPU clocking and
configurations. Dumping /proc/cpuinfo should be enough.

Sincerely,
Jim Huang (jserv)
http://0xlab.org/