I've been playing with multimedia & A2DP and found out that...
...A2DP is f*ckingly well implemented (speaking of N95-1)! With every audio formats except WMA, the additional CPU usage it causes is around 2% (with WMA, slightly more - but playing back WMA in the factory 'Music Player' itself requires definitely more CPU time than other formats). Does anyone know whether A2DP encoding is done by hardware? It MUST be something like that, given that the software-only solutions take, in general, an order of magnitude more CPU time on Windows Mobile (see my WM CPU usage results if interested in the linked articles).
Strangely, the same isn't the case with playing back MPEG-4 videos. Then, A2DP may result in an additional 20...50% CPU usage, depending on the player (with the factory RealPlayer, the former; with the latest CorePlayer, the latter). Do you know what the problem can be caused by?
Finally, a chart, showing the CPU usage of the factory "Music Player" (with the video, RealPlayer instead) and the latest CorePlayer. Excellent results - I wish Windows Mobile had similarly CPU-friendly players (in there, the CPU usage is at least five times higher). Again, I need to ask whether decoding is done by the hardware - it must be, unlike on current WM devices.
I'll, tomorrow, publish some cool battery life tests, both with and without A2DP, showing playing back MP3 and WMA.
The built-in Music Player: 5:00 with WMA + A2DP (~13.5% CPU usage); 5:26 with the same file set and the factory wired earbuds (~7% CPU usage; volume continuusly at moderately loud – 70% - level) with both dis- and enabled equalizer and stereo widening. I’ve run this test twice to really see the effects of the qualizer / widening. Fortunately, unlike with (non-hardware equalizers only effective with wired headphones and nothing else) Windows Mobile, these (with widening, essential) goodies don’t cause any additional battery drain.
MP3 playback (<1% CPU usage) with exactly the same settings (wired headset) flattened the battery in less than 6:20 (dunno exactly when – I’d bet around 6:00? Must be considerably more than 5:26 because of the much less CPU usage but can’t be more than 6:20). This is really bad news.
As far as using the CPU at 100% is concerned, on the other hand, I’ve got much better results than on Windows Mobile. With CorePlayer 1.1.2 beta2, H.264 file playback with maximal backlight level and A2DP (therefore, at ~100% CPU usage): 2:52. This is WAY better than anything on Windows Mobile, assuming maximal backlight and 100% CPU usage.
The test conditions were all the same: disabled 3G (and, with non-A2DP tests, BT too), disabled Wi-Fi scanning, starting with two times fully recharged battery, after a reset, not having anything else in the memory and waiting until the phone shut down. All tests have been made during exactly the same circumstances (same fixed, static position; same external temperature etc). It had very good GSM signal.
All in all,
1. Audio files requiring low CPU time (particularly the MP3 playback test), where I’ve expected MUCH better results, has turned out to be much worse than I’ve expected (and what some people either here or AAS have reported): you can’t expect more than six hours of (continuous) playback of MP3’s, and WMA’s are even worse. Frankly, I don’t know why some people state the phone is able to play back music for more than ten hours. I’m absolutely sure they aren’t right.
2. Video playback (or, in general, running any app requiring (almost) 100% CPU time, with the highest backlight level), on the other hand, fared much better than I’ve thought. The almost three hours I’ve measured is FAR better than with a Windows Mobile device with comparable CPU raw power AND battery. Interestingly, the battery didn’t heat up either (it “only” felt a bit warm) – as opposed to the Windows Mobile case, where 100% CPU usage considerably heats up the battery.
What were the music files encoded at (128,256,320,etc)? Do you think different encoding rates would make much of a difference? Were the WMA's protected?
N95's tend to lose connection to A2DP headphones pretty easily. Like when my arm is in between the phone on my desk and the headset, the music would start skipping. This happened with all 4 N95's that I've owned.
I used the Nokia BH-501 headset all day today for about 2-3 hours of music, in car, in room, in school, walking, everywhere...
the n95 would be far from me sometimes..in my pocket, in my cup holder...various places
no skipping at all with Mp3's...or WMAs either...or anything...even voice calls were awesome, nobody knew i was on a headset..
I think the Nokia BH-501's are by far the best in terms of picking up my voice and still sounding perfect for the other caller....the BASS and the "boom" effects in them are LOUD and sometimes i would have to lower the volume..
and again i just got them in the mail yesterday ...
the only thing they are missing..is a carrying case...and...they need to have more music controls like pause, next, and maybe a mute button...or maybe i just havent opened the manuals yet haha
I used the Nokia BH-501 headset all day today for about 2-3 hours of music, in car, in room, in school, walking, everywhere...
the n95 would be far from me sometimes..in my pocket, in my cup holder...various places
no skipping at all with Mp3's...or WMAs either...or anything...even voice calls were awesome, nobody knew i was on a headset..
I think the Nokia BH-501's are by far the best in terms of picking up my voice and still sounding perfect for the other caller....the BASS and the "boom" effects in them are LOUD and sometimes i would have to lower the volume..
and again i just got them in the mail yesterday ...
the only thing they are missing..is a carrying case...and...they need to have more music controls like pause, next, and maybe a mute button...or maybe i just havent opened the manuals yet haha
What were the music files encoded at (128,256,320,etc)? Do you think different encoding rates would make much of a difference? Were the WMA's protected?
Everything encoded with the default settings of iTunes (AAC), CDex (FLAC, MP3, Ogg) and WMP (WMA 10 and 11 - I've tested two sets of WMA's). For example, MP3's used 160 kbps.
N95's tend to lose connection to A2DP headphones pretty easily. Like when my arm is in between the phone on my desk and the headset, the music would start skipping. This happened with all 4 N95's that I've owned.
My Sony K790a was never half as sensitive.
This is definitely a headphones problem. I've published (in the Windows Mobile forums here at HoFo) several A2DP headphones reviews; several headphones models (for example, the Jabra BT620s) indeed suffer from this problem. See for example http://www.howardforums.com/showthre...readid=1025471
With my 590A, the A2DP connection is very stable and, in general, I can listen to music even in another room (with non-thick walls). I've found it even slightly better than my previous phone used for A2DP purposes (the HTC Vox / s710).
Do you think different encoding rates would make much of a difference?
Based on the two different sets of WMA's, nope. Frankly, I don't think anyone can produce better results - therefore, I don't think the N95 can be at all used as a long-time (over 10 hours on one charge - most TI OMAP-based (but not with other CPU types) Windows Mobile models are capable of that feat, even with enabled (!) A2DP) music player.
This is certainly bad news as, otherwise, it's a great device.
Guys and gals, I’ve just finished my multimedia playback battery life tests.
Exactly the same tests, conducted under exactly the same circumstances (static and same placement, disabled 3G, no Net connections / programs running in the background, no Bluetooth, lowest backlight; no incoming / outgoing phone (call, SMS etc.) activity; factory earbuds with 70% volume level and exactly the same set of WMA’s, in the same order) that yielded in 5:26 total playback time (from being topped up battery to the handset entirely shutting down) with the v12 ROM, now, resulted in 7:55 total playback time with the new, v20 ROM.
That is, the low-CPU multimedia playback battery life has been increased by 45% !!!! This means it’s true what the Nokia folks have claimed: that is, they have implemented far better scaling (underclocking) than in previous firmware versions.
I’ll repeat the tests with A2DP to see whether the improvements are similar, or, does A2DP result in a significantly less battery life gain.
Reply With Quote
Originally Posted by Blue13x
...
Bookmarks