The evaluation edition of MonoDroid is emulator-only, so you can't do any 
benchmarking or even verify their benchmark results without buying a very 
expensive license. I have a ton of ideas for benchmarking this, and 
installed MonoDroid and did some coding, but the MonoDroid eval edition 
wouldn't let me build a binary to run on an Android device. The regular 
license is $999 and while there may be some cheaper academic/hobbyist 
options, I'm not giving them large sums of money to verify their claims. 
BTW, the regular Eclipse Android development stack is 100% free. And that's 
not some limited starter edition, you get the full toolset for free.

The benchmark results posted on the Mono blog aren't using one of the many 
standard benchmarking suites, they weren't run by an impartial entity, and 
they aren't even making a reasonable pretense of being fair. The Mono guys 
developed their own internal custom benchmark designed to make Mono look 
good and Dalvik look bad and they cherry pick the results accordingly. 
Their Git repo also has the SciMark benchmark source code running on 
Dalvik/Mono, but they didn't choose to publish those results. If you look 
around for Android benchmarks, there are a ton of them. If I could without 
buying a $999 MonoDroid license, I'd port them to C# and run them, and see 
how they stack up. I'd imagine if these results were favorable to Mono, 
they'd be publishing them on their blog themselves. If some big company 
like Intel/AMD/Nvidia published these types of heavily customized 
"benchmarks" like this, they would be laughed at.

Google has the Renderscript and the C/C++ NDK which I suspect are the more 
high performance low level SDK options. Dalvik's design goals weren't 
primarily centered around runtime performance, but memory considerations, 
battery life issues, UI responsivenes, app startup/switching issues.

Dalvik Turbo, is a drop in replacement for Dalvik that supposedly delivers 
2.8x faster runtime performance on benchmarks. What are the ups and downs, 
between this, native Dalvik, the Mono VM, and others.

Dalvik Turbo seems to have the simple goal of providing a faster VM without 
any other associated agenda. Many alternate programming languages like 
Scala/Kotlin/Clojure/Fantom aim to provide a variety of innovations at the 
programming language level but really strive to acommodate multiple VM 
ecosytems without playing a heavily partisan role. Mono isn't like that. 
They have a history of evagenglizing everything Microsoft and bashing 
opposing infrastructure like Java, using technical bullet points like stack 
value types or reified generics as ammo fodder for that contentious debate, 
and being absolutely obnoxious about the whole thing.

Casper, you say the JDK ecosystem is uninspiring. To each his own, but I 
disagree. I love the culture of innovation that has happened outside of 
Sun/Oracle within the JDK community. All kinds of alternate languages 
(Scala/Kotlin/etc) alternate build tools and paradigms (Maven/Gradle), web 
frameworks, configuration frameworks, competing IDEs, all innovate and 
suceed or fail based on largely merit. In the .NET world, the community at 
large is very resistant to using anything that isn't officially Microsoft. 
It's extremely rare to see a .NET shop use a non-Microsoft IDE, development 
OS (there is a very small minority doing .NET/Mono on Linux/Mac), 
programming language, build tool, web framework, testing framework, or 
anything. Even when there are high quality alternatives out there; for 
example, Scala/Clojure/Fantom have been on .NET for a while, but there is 
nearly zero adoption. The .NET world is less a culture of open innovation 
and more a culture of Microsoft and that isn't appealing to me.

-- 
You received this message because you are subscribed to the Google Groups "Java 
Posse" group.
To view this discussion on the web visit 
https://groups.google.com/d/msg/javaposse/-/wuXtXdPOc-YJ.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/javaposse?hl=en.

Reply via email to