Love the N900, but most make it sound like its dwarfed under Snapdragon processors. Question is targeting experts not *people that know everything*

Specifically, how does N900 processor hold up against an underclocked snapdragon.
For eg, N900 (600Mhz, OMAP 3430, 65nm A8) vs Acer Liquid A1 (768MHz, QSD 8250, 65nm custom A8).

Power difference:
Energy difference:
Effeciency difference:
*Figures can be relative (ie 130% instead of 128%)

But I am strictly talking about the processors; I want to know which has the better processor not which is the better device.
If their are differences in RAM/battery capacity, please guestimate to bring both devices to same RAM/battery capacity!
(for instance/ A1 has 512Mb and N900 has 256Mb/1GB RAM, this might give the A1 unfair advantage in a benchmark)


Working Example: (Assuming are equal effeciency)
Power difference= 768/600 = 128% (ie the A1 is 28% more powerful than N900)
Energy difference= 768/600 = 128% (ie the A1 uses 28% more energy than N900)
Effeciency= Power/Energy difference = 128/128 = 1 (same effeciency)

Dont say "this is a pointless post" and scram = troll.
Point of this is, how powerful and effecient is QSD?
And today where theres so many options, this helps clarify what-is-what.

I think the 1ghz snapdragon will probably do more calulations per second than a 600mhz arm cortex a8.

Just like a desk fan will spin faster than a ceiling fan.

Which one moves more air is specifically what you said you did not want to know.

edit: I think this post may get a different response at anandtech forums. Maybe ask their experts to humor you and do a bit of extrapolating comparisons as an exercise. or whatever.

This was a simple question, but it has a complex answer.

Both processors are ARM-based, meaning they have a common set of instructions (the language processors understand). ARM language is relatively simple, which is reflected on the architecture of their processors, which make them very power efficient, a plus on the mobile market.

That being said, comparing the OMAP 3430 and the QSD 8250, the first difference in their architecture is that the 3430 off loads some CPU intensive tasks (video, audio, digital signal processing) to dedicated co-processors (GPU, DPS, etc) while the Snapdragon processes all of it in their CPU.

So, if you are watching a video in the 3430, the main 600 Mhz CPU is doing nothing, letting the GPU work on it; while the 8250 main 1 Ghz CPU is busy decoding the video, which probably takes a fair chunk of that 1 Ghz.

It is a different approach to the solve the same basic problem.

From some random Google searches I found that the 3430 can do 2000 Millions of Instructions per Second, while the 8250 can do 2100, but I can't say that is official information.

So, in my opinion, the only thing Snapdragon has over OMAP3 is a fancier name.

Now, on the other hand, you have to consider the OS running the processor. If you want those Mhz to truly reflect in the applications you run, you need those applications to be as closer as the hardware as possible. If you are thinking Android, then there is a big bottleneck because most Android apps have to run through a Virtual Machine (Dalvik) and few are written in native code (that gives them direct access to the hardware). Maemo gives apps direct access to the hardware, and Symbian too (Snapdragon is said to be coming to Symbian^3 and 4 devices); so they can better use the hardware available. Can't speak about iPhone OS, RIM or webOS because I don't know much about them.

Edit: Speaking about AnandTech, this is a good read: http://www.anandtech.com/gadgets/sho...spx?i=3595&p=1

thanks mrojas,

that was insightful. I knew OMAP3x had this co-processor concept but was unsure of QSD. Looks like all QSD has going for it on top of OMAP3 is its speedier processor.

Usually a speedier processor without increasing effeciency (eg downsize to 45nm) correlates to more energy expenditure. Plus the lack of hardware support for it really takes its toll. So perhaps the underclocked (768MHz) performs equal to a 600Mhz OMAP3 ... in video playback? On another note, adding co-processors would mean you would need to supply each unit (unless Tegra genious) with power. This might decrease the energy effeciency of OMAP3 to level of underclocked QSD.
So both processors have different design/start but achieve roughly same qualities. That's what I've been pondering, anyone care to elaborate your thoughts?

I realize Android is a linux kernel heavily modified with java vs maemo which is native. As a rule of thumb:
Java=slower+cross compatability+hard ports+easy dev.
Native=faster+less cross compatibility+easy ports+hard dev.

So for ultimate experience, native is your plan.
But because we live in an imperfect world, we need cross-compatability.
The success of iPhone is that is has processors better than your standard PDA (in 2007 era) and runs in native so you get maximum benefits.
How? There's only 3 variants, with 4 (iPad) coming soon (all similar hardware) so cross-compatability in native is possible/easy.

So iPhone is quality over quantity.
Whereas Android is quantity over quality.

Android=devices can have a large option of hardware, cross-compataility is still possible on java, applications run slower.

re power, the omap can power down the different co-processors entirely, so if they're not used they don't use power.. I guess this is the same "Tegra genious".