Phil,
Phil wrote:
Thanks for your efforts on this, working quite nicely (at least for the image I can get to run :-) on my Honeycomb tablet. One question I do have: are you aware of any changes that need to be made to or things to watch out for when running other images? I've tried about half a dozen images (pre- and post-Cog) but they all hang while loading resulting in a force close. However, the image you provided based on the earlier work of Andreas, works as expected.
Thanks for the update. Can you post the tiny benchmark result as in http://code.google.com/p/squeakvm-tablet/wiki/TinyBenchmark so I could add it to the table?
It is good to hear that Android 3 can run it. I am not sure what can be done with newer images: I tried some as well and got same results (program hangs). Keep in mind that this is 3.9 VM.
Also an FYI if you weren't already aware: the recently released NDKr5c appears to have finally addressed the issue that prevented building with the latest NDK.
I am trying to find time to play with recent Pharo, VMmaker and Cog (Mariano Peck has a good review of the VM internals in his blog which I am going to use as reference material) to do the same trick of "turning the event loop inside out" as Andreas did to the classic VM used in this port, and then to use ndk5, newer images, better platform code management, etc. This may result in a more universal embeddable Squeak VM.
I think that whatever was done with the original port has shown that Squeak may be used on Android tablets pretty well. Next step is to get in sync with the current VM sources.
Thanks.
On Jun 28, 2011, at 1:35 PM, Dimitry Golubovsky wrote:
Phil,
Phil wrote:
Thanks for your efforts on this, working quite nicely (at least for the image I can get to run :-) on my Honeycomb tablet. One question I do have: are you aware of any changes that need to be made to or things to watch out for when running other images? I've tried about half a dozen images (pre- and post-Cog) but they all hang while loading resulting in a force close. However, the image you provided based on the earlier work of Andreas, works as expected.
Thanks for the update. Can you post the tiny benchmark result as in http://code.google.com/p/squeakvm-tablet/wiki/TinyBenchmark so I could add it to the table?
Sure... 37,058,482 bytecodes/sec, 1,205,854 sends/sec on an Acer Iconia (Android 3.0.1, dual core 1GHz Tegra 2 though obviously only one core is being used by the VM)
It is good to hear that Android 3 can run it. I am not sure what can be done with newer images: I tried some as well and got same results (program hangs). Keep in mind that this is 3.9 VM.
Bummer, but that makes sense.
Also an FYI if you weren't already aware: the recently released NDKr5c appears to have finally addressed the issue that prevented building with the latest NDK.
I am trying to find time to play with recent Pharo, VMmaker and Cog (Mariano Peck has a good review of the VM internals in his blog which I am going to use as reference material) to do the same trick of "turning the event loop inside out" as Andreas did to the classic VM used in this port, and then to use ndk5, newer images, better platform code management, etc. This may result in a more universal embeddable Squeak VM.
That's good to know. I'll keep an eye on the mailing list for any new developments as well as doing a little poking around and post anything I come up with as well (though I'm not likely to be of much help on the VM side as my focus is more on dealing with Android these days)
I think that whatever was done with the original port has shown that Squeak may be used on Android tablets pretty well. Next step is to get in sync with the current VM sources.
I agree, a common and current VM codebase should be the goal.
Thanks.
-- Dimitry Golubovsky
Anywhere on the Web
Thanks again, Phil
Thanks, added benchmark results to the Wiki.
On Thu, Jun 30, 2011 at 3:17 PM, Phil (list) pbpublist@gmail.com wrote:
Sure... 37,058,482 bytecodes/sec, 1,205,854 sends/sec on an Acer Iconia (Android 3.0.1, dual core 1GHz Tegra 2 though obviously only one core is being used by the VM)
On 30.06.2011, at 21:17, Phil (list) wrote:
On Jun 28, 2011, at 1:35 PM, Dimitry Golubovsky wrote:
Thanks for the update. Can you post the tiny benchmark result as in http://code.google.com/p/squeakvm-tablet/wiki/TinyBenchmark so I could add it to the table?
Sure... 37,058,482 bytecodes/sec, 1,205,854 sends/sec on an Acer Iconia (Android 3.0.1, dual core 1GHz Tegra 2 though obviously only one core is being used by the VM)
Interesting. On the old iPad (single-code 1 GHz) I get 59,479,553 bytecodes/sec and 1,884,210 sends/sec. Have not seen number for iPad2 yet.
And this is not even the Stack VM, just the plain old interpreter.
- Bert -
vm-dev@lists.squeakfoundation.org
-
Bert Freudenberg -
Dimitry Golubovsky -
Phil (list)