Comments on: Audio Patch Released for the Tungsten T

Cheers,
Tony Cheung

--
With great power comes great responsiblity.

That's what happens when you multitask with only 1 CPU. If the Real player uses the OMAP DSP instead of the ARM9, then the slowdown of the foreground task should be less. (Almost) 2 CPU's is supposed to be the big advantage of the TI OMAP over the XScale or MXL.

How much was the slowdown? In percent.

That's not what should happen. Neither an MP3 player nor Bejewelled should take far less than 50% of a 175MHz RISC chip. And that means that if you run them simultaneously, neither of them should experience any slowdown.

If there is a slowdown, it is really a software architecture problem in Palm OS 5: some combination of poor multitasking support, busy waiting, etc. That's not surprising, given the history of Palm OS. Let's hope it gets fixed for good in Palm OS 6.

There is nothing wrong with wasting CPU cycles when there are CPU cycles to waste. But let's not pretend that something is a CPU limitation when it is pretty clearly due to some combination of emulation, limitations in multitasking, and busy-waiting applications. This is fixable, and software authors and Palm should fix it.

Good. So, that means that 2/3's of the TT CPU are left for playing Bejewelled. You just made my point for me.

"b) Bejeweled is running in emulation."

Indeed. Which just goes to show that Palm's claims that emulation doesn't impose a lot of overhead are wrong. Because 2/3 of an ARM processor should be plenty for a game that mostly sits around and waits, with some simple sprite animations on a tiny bitmap every now and then.

"The other person was right - a program utilizing DSP for MP3 decoding would get around this nicely."

It would. But that's not the point. The point is that these kinds of applications should not require special hacks: they should just run efficiently, out of the box.

Palm needs to fix Palm OS and come out with a native version quickly. No emulation. Proper multitasking. No ARMlets. Anything else is just rearranging deck chairs on the Titanic.

Huh? Apples 68k to PPC transition worked very well, and OS X does not require people to rewrite software. Apple has lots of problems, but they have managed their platform transitions very well.

Oh, that's a GREAT idea. Instead of Bejewelled running slow, then it wouldn't run at ALL. It's a 68k application, remember?

There is nothing wrong with providing emulation, but there is everything wrong with not providing a fully native API and good multitasking. You may say that this is a tradeoff Palm had to make.

But no matter what the reason, the fact remains that you can't run an MP3 player and Bejewelled simultaneously on hardware that is perfectly capable of doing that with ease (in fact, frankly, it should be able to do that in emulation--the problem must be deeper).

Palm OS 5 just isn't a very good system. Stick with Palm OS 4 until Palm OS 6 comes out. A 68k-based Sony Clie is perfectly capable of running Bejewelled and an MP3 player simultaneously, and it's cheaper and lighter than a TT as well.

---"Because 2/3 of an ARM processor should be plenty for a game that mostly sits around and waits, with some simple sprite animations on a tiny bitmap every now and then."---

I hate to break this to you, but Bejeweled is NOT as simple as you make it. It's not just a "a game that mostly sits around and waits, with some simple sprite animations on a tiny bitmap every now and then"; it's much more than that. Remember, Bejeweled's addictiveness stems partially from the fact that it is somewhat hypnotic. Therefore, Astraware couldn't run any risks of having it run slow and jerky (as this would ruin the "hypnosis"), and I, as a beta tester of Bejeweled, can testify that this was no simple matter on Clies - HiRes graphics with a slow Dragonball CPU?! Forget about it!

At first, Howard Tomlinson even refused to make a HiRes version of Bejeweled at all, until they would be able to find a way around this. Fortunately for us, he and David Oakley managed to develop some sort of assembly-level trick for getting the job done, and we now have a HiRes version of Bejeweled.

Thus, you are right about the "waiting around" part - in fact there is so little happening then, that Bejeweled lowers the frame rate to conserve battery life. HOWEVER, the "simple sprite animations", I repeat, are NOT simple - you have 64 gems that have to glitter, roll, slide, sparkle, fall, appear on the board, be taken off the board, and more, and combinations of these actions are happening all the time.

Therefore, naturally you wont notice the lag when sitting and waiting - there's nothing happening then anyway! The "lag" being referred to here refers to when the jewels are moving, right? Well, of course that's when you'll notice the lag - even a powerful ARM chip can't decode MP3's and animate Bejewelled at the same time - after all, both of these require timely processing, (you can't wait for one to finish before working on the other) and the interspersal that is acheived through multitasking is not nearly as good as multiprocessing - thus resulting in "lag", even on a fast processor - remember, this is multitasking, NOT multiprocessing.

If the MP3 player were to use the DSP on the other hand, LIKE IT IS SUPPOSED TO (!), then it would be like multiprocessing, so there would be no problems.

---"'The other person was right - a program utilizing DSP for MP3 decoding would get around this nicely.'

It would. But that's not the point. The point is that these kinds of applications should not require special hacks: they should just run efficiently, out of the box."---

EXCUSE ME, but using the DSP is NOT A SPECIAL HACK - what the heck else did TI implement a "multimedia DSP" into the OMAP for??!! You're not making any sense, buddy! The TI OMAP's DSP is there for a reason - to be used for MP3 decoding!

The patch is not related to the Palm OS itself, it's only for the Tungsten T. So it won't be part of any OS update.

Tony

--
With great power comes great responsiblity.

What's there to be disappointed about? I don't think there's any need to jump to conclusions so quickly. I don't recall Palm even releasing a 5.2 update yet for Tungsten owners (as it's not offered on the software download page of their site). For all we know, it could already be included in the next update, but they made it available for us now. I, for one, am glad they didn't have us wait longer for an update to enjoy better sound quality today.

And if Palm OS software history is any indication, updates (x.2, x.2.1, etc) are free for download. Upgrades (v6.0, v7.0, etc) are what we might have to pay for.

Jim

Seeing that the TT has 5.0, 5.2 should be a free update..... but you never know.....

Bumbleluck

Please reread my comment, as not to confuse 'upgrades' with 'updates.' When upgrading, the OS is moving up a full version (i.e. v2.0 > v3.0, v.3.2 > v4.1, etc). When you are updating, they're only incremental moves (i.e. v3.1 > v.3.2, v4.0 > v4.1, etc).

: History rarely repeats itself, if it is observed! :-)

Only if observed more carefully. ;)

Jim

The first digit is normally a major upgrade while the second is minor. In unix a 3rd is patch update. The jump from 3 to 4 was done because they felt that it was different enough to warrant it. 4.0 to 4.1 is a free patch because they contain minor changes and didnt warant a purchase.

When iSilo went from version 2 to version 3 they made you pay again. When eWallet went from 2 to 3 they made everyone purchase again.

And yes, I do use the system sounds. Deal with it.

Do you think this is a bug in the patch? Or are we just now hearing high frequences on the "tick" sample (?) that always existed but were being filtered out before the patch?

Zilch

1. Use on other than Windows PCs. Real usually just created PC based supporting software, conduits etc.

2. The (more than likey required) use of their "supporting" software.

I want an MP3 player (read media as well), runs on my Palm, and accesses the media files that I choose. I don't need nor want to be tied down to supporting software to get the features stated available when I purchased the TT. Since MP3 capabilities were never listed as requiring a Windows based PC, I hope Palm doesn't change that non-requirement.

I'm sure many others feel the same way. I purchased Aeroplayer, some use Pocket Tunes. We are getting what we were told from Palm was or would soon be available.

On another note, I just checked Real's mobile player website and saw the following:

SUPPORTED DEVICES

Supported Devices include:
Nokia 9210 Series Communicator
Nokia 7650
Palm OS5 Based Handheld
Audiovox's Maestro Pocket PC
Casio Cassiopeia E-200 Series
Compaq iPAQ Pocket PC H3600, H3700, and H3800 Series
HP Jornada 565, 567, and 568
O2 xda from mmO2
NEC MobilePro P300
Sagem WA3050 (audio only)

This is a static list on the site. You then enter in a pull down, for which device you want to download for. Well guess which device is not in the pulldown list?

The Palm one...

Just to reiterate some points, it should be free and due for a release mid-Februrary (registered Tungsten owners should receive official word via e-mail when available) according to the RealNetworks reps at LinuxWorld Expo in NYC 2 weeks ago.

But I haven't read/heard anything since then.

Jim

Tony

--
With great power comes great responsiblity.

-Ben [:)]

Compared to the NX70v (which I have handy to compare), they are comparable in quality. Compared to the Creative Zen and iPod (which I also have handy), the amps seem of similar quality, though the Zen and iPod have more headroom (more power), and they also have customizable EQs to iron out some of the deficits. It's most evident when you want to crank the music - the T|T (and most other PDAs) will start clipping violently.

Now for most people, this will be fine. If you're picky about audio, you usually don't puch much faith in portables anyway, and those that do almost always get external can amps (headroom.headphone.com is a good place to go to learn about that stuff).

If you play an MP3 (128kbps) without any EQ settings with mediocre headphones at about 50% volume on the iPod normalized to that scale, you will get practically identical results - very tinny, harsh sounds. I used Aeroplayer for testing. When you get into those headphone amps that are included in portables such as these, they're all so low end that there's not a significant difference among them; the better sounding ones (iPod, etc) are purely a function of a slight increase in power, and not real circuitry (where they're all very similar). In fact, try listening to one of the original iPods - the ones with the scroll wheel and lower powered head amp - you'll be surprised at how (when flat) the two will sound very much alike.

The reason for asking is that I am frequent train/plane traveller, and am looking into getting some er4 or er6's headphones. These are earplugs that are also amongst the very best headphones available - ie they will likely make a crap source very evident.

I also don't buy the palm philosophy of carrying separate dedicated devices around (and their chargers, sync cables, etc, etc) - I want as many functions in a single (small) device as possible - I wish palm 'got it' in this respect, since many other companies do (eg nokia, with build in cameras, mp3 players, web, etc).

Finally I wanted to experiment with lossless encoding, and add dolby headphone type stuff, etc to the output, - which the stand-alone players don't do.

Agreed - though the patch does make quite a difference. I'll wait to see how the RealPlayer sounds - maybe this is just a matter of making proper use of the OMAP DSP - before i pass judgement.