Palm to Shift to ARM Processor

krugdm writes "According to this article, Palm is expected to announce that it will be modifying its OS to run on faster ARM processors instead of the current Motorola Dragonball processor."

Comments

More CPU power can be useful for palm

[Lally Singh]

Palm's biggest advantage has always been their mission plan: be a good handheld organizer. Note that I didn't say 'handheld computer.'

Palm's single greatest asset has been a _WELL_DEFINED_ _PLAN_ and a _TARGET_MARKET_. CE's target market is "Palm's customers," without a second thought about _WHO_ those people are: salespeople, businesspeople, etc... you know, the _ordinary_user_, who doesn't really care that much about running a full web browser in their pocket (hence the philosophical difference between full internet connectivity and web clipping :full unadulterated access at disgustingly low speeds versus an efficient access point for the data most often accessed)... Don't let me go off on how little foresight or design MS actually puts into their products...

I don't see Palm turning their products into the ubergeek-toys that the CE crowd has been putting out: tiny handheld computers that are simply painful to use because they try to replace desktop computers with shitty, overpriced, small screened, keyboardless versions. Palm aims for the general public with a tool that lets you store your data, recall it, and otherwise, just gets out of your damn way.

CE's just turning into another Apple Newton: a handheld computer that's waaay too overpriced for too little functionality; useful only to those .1% of the computer buying population that can't possibly sit down and use a desktop for the majority of their real work.

Palm's probably going to use the new CPU power for things like a prettier GUI, voice recognition, MP3 playing, and perhaps some multimedia. At the end of the day, just improving what it does now: assists you with your life's tasks, not taking them over like CE.

But, let me end my random tirade on WinCE versus PalmOS... I'm afraid that the percentage of people here on /. that seek an ubergeek-toy is disproportionately high.

--Lally Singh, Palm Software Developer.

--

Re:More CPU power can be useful for palm

[MrBogus]

The Palm versus WinCE battle sorta reminds me of the old PC versus Mac versus Unix battle of the 1980s. Sure 90% of the people could get their work done fine and move along with an el cheapo DOS PC. However the other 10% couldn't shut up about their colorful graphical user interface with WYSIWYG printing, or their multi-user multi-tasking memory-protected super stable workstation.

Then one day, the Mac people and the Unix people woke up and figured that the PC could do almost exactly everything their specialized and more expensive systems could do.

It's the same thing in the handheld world, turned on it's head. Eventually WinCE-type handheld computers will have all the battery life, cost, and 'filofax' features of the Palm. BUT, if Palm gets the MPEG and Quake and Spreadsheet 'features' (which means a fast CPU and something resembling a real OS) before then without losing it's primary advantages, it will probably hold it's dominant marketshare lead forever.

Re:The obvious question:

[bcombee]

Moving to ARM does two important things:

1. Palm OS now can run on the same chipset used in many cell phones, meaning its easier to make Palm-based smartphones.
2. There are multiple companies making ARM-core based chips. The 68K is a dead end, while ARM allows a wide range of scalability, from cheap ARM 7 devices that are cheaper than Dragonballs to fast XScale processors that can power information appliances.

The obvious question:

[Accipiter]

WHY?

Tell me why an electronic datebook / addressbook / personal information unit needs a fast processor? Is it that important to display a datebook entry in .03 milliseconds rather than .1 milliseconds?

The Palm line was designed for a few specific tasks, and to perform those tasks well. Palm succeeded in doing this; the Palm is a wonderful organizer. But who needs an organizer that runs faster than 33 megahertz?

A switch to ARM-based chips means that Palm OS-based handhelds will run at significantly higher clock speeds than their current 33MHz. This means the devices will be able to handle more complex applications, such as video streaming and digital-audio playback, and to match the processing power of handhelds based on Microsoft's Pocket PC operating system.

Okay, hold up. Let's review this:

Palm has significantly more market share than Windows CE, because it does what it's designed to do a lot BETTER than CE.

So why are they changing their product to adapt to the weaker market segment? Who the fuck needs digital audio and streaming video embedded in their addressbook? Yes, it's badass whiz-bang stuff, but not practical. They throw in the excuse of "planning for the future", but that's marketspeak bullshit. They need to concentrate on making their products BETTER, and not necessarily COOLER.

-- Give him Head? Be a Beacon?

On schedule

[hatless]

For Palm to stay in their longstanding $200-$450 price range, this move is right on schedule. Palms have been shipping for about 5 years now, just over 3 generations of Moore's Law. Since the original Palms were 20 MHz 68030-class computers, it should be possible to produce something 8-10 times faster overall with 8-10 times the memory. This roughly translates to 150-200 MHz ARM-based systems (right on target) with 4MB RAM in the lowest-end devices (think m100) and several times that in the high-end ones.

Do all the math and you can easily come up with roughly the same hardware spec as a current iPAQ handheld, but about $200 cheaper since it will hit the market about a year later, which in 2002 puts a color 200 MHz ARM-based device with 16-32MB memory built in right in Palm's pricing sweet spot.

While Compaq will be targeting its $500-$600 price range with lovely 300 MHz 64 MB devices, this hypothetical $400 Palm device should be comparable performance-wise to the current iPAQ, but with longer battery life and what will probably be transparent support for all Dragonball Palm apps. They're not dead yet.

Random observations from a Palm user

[dennism]

This is great, in my opinion. Faster processor, with backward compatibility. And no sacrificing the easy to use PalmOS. At least if Palm does it right :)

The ARM has a number of advantages over the Dragonball/68k line. One of which is that it is designed to be used in battery powered devices such as the Palm and the GameBoy Advance. The current Palms run between 16-33mhz. You can even overclock them (in software!) to run faster. The ARM chip in the Palm should be running at 200mhz.

Why such a jump? Well, there are a couple of reasons:

- 68k emulation. To run the old dragonball apps, you'll need to emulate a 68k. I'm going to guess here, and say that instead of using something like POSE running under a new OS, they will instead go the same route Apple did. The core OS will be written in ARM, and the 68k code when calling a trap, will run code natively on the ARM. i.e, OS routines will run at full speed.

- Multimedia. I don't think that the Palm will be ready to run very much in the line of movies, but MP3 playback becomes very practical. Along with other things such as dictation, telephony, etc, etc. The extra boost of speed will definetly help here.

- OS functions. I currently have a Visor Deluxe, with about 800k free. And the thing is slow at times. If I go to check info, delete an app, whatever, it takes 10-15 seconds just to get the list. After that, it's cached so it appears instantly, but the initial time it's slow. And once another app is started, the cache is gone. In addition, the Find function isn't exactly the fastest thing in the world. The extra speed will help here.

As far as the PalmOS goes, I think that the ARM version will be an evolution of the current version. I'm sure we'll see Palm push different screen sizes, we already have some variations on the market now (Sony's Clie - 320x320, for one). 160x160 will probally become the minimum size.

I think another thing that will happen is that Palm tosses out the current kernel they use, and use another kernel that they can actually expose to the developer. This will allow things like MP3 playback to continue in the background while other applications continue to run -- a kernel thread would handle it. I don't see Palm getting rid of the single application running scheme -- it just works so well, and prevents things like out-of-memory errors from occuring because you have 20 programs open (WinCE/PocketPC).

Palm knows what they need to do. I think they want to make the transition to ARM as seemless as can be. This means making source code compatible (as much as possible) between the Dragonball and ARM version. If programming an ARM version of the Palm is completely different than the 68k version, developers may just get up and move to WinCE.

On the other hand, I would imagine that most developers could write 68k version and allow the emulation to handle running it under the ARM. For most apps, you don't need the power of the ARM. And on the ARM, those apps would benefit from the faster CPU (for OS functions).

Palm has a lot of ex-Apple employees (or at least they did :) working for them, so I'm sure they are studying the 68k->PPC transition that Apple pulled off. Don't expect the radical (completely new OS) on this, but expect a smooth transition... they have way too much investment built around the 68k Palm (both in binaries and source code) to just throw it away for a different CPU.

This is going to be an exciting time for both Palm users and developers. I personally can't wait until I can buy a ARM based Palm of the shelf at the local computer store.

Re:the world of incompatibility

[Moofie]

Keep in mind that despite the fact that Microsoft has utterly failed to make this sort of a transition work well, it has been done. Apple's emulation of 68k processors on PowerPC was pretty damn bulletproof. The overwhelming majority of software out at the time just kept on workin', with only modest speed penalties. After a recompile, it was off to the races.

This can be done, and done well. I just hope they don't spank their battery life (PalmOS's single hugest advantage) moving to this gee-whiz new processor.

Need for speed?

[GenetixSW]

This is an interesting, and mostly welcome development. I could argue that a palmtop platform really doesn't need to be terribly fast, since most people simply use them for keeping up with contacts and maintaining schedules, rather than watching video and streaming MP3s wirelessly...

What has me worried is this:

Although speeds have not been announced, the company has demonstrated the upcoming chips at 1GHz.

200MHz is a lot. 1GHz is utterly ridiculous. The biggest concern with having faster chips is reducing battery life. Most, if not all, WinCE devices have a max battery life in the hours. Granted, most such devices have colour screens, but it would be foolish to say the faster chips doesn't play a part in it. Even Palms are using more power nowadays. A year ago, it was common for a Palm (say IIIxe) to last a month or more on one set of batteries. Right now, I'm averaging 6 weeks between replacement, which is terrific. The newer m500 models are touted as lasting up to two weeks. That's nice, yes, but a disturbing fact of life is becoming increasingly obvious and disruptive: more features require more power and drain more energy.

Do we need the extra abilities? As I see it, we'll be losing sight of the original goal of a palmtop. Instead, what is now occurring is that the personal agenda of a year ago is being replaced with ever more powerful be-everything devices.

This has me worried. Perhaps my fears are misplaced?

Re:Need for speed?

[tjwhaynes]

200MHz is a lot. 1GHz is utterly ridiculous. The biggest concern with having faster chips is reducing battery life. Most, if not all, WinCE devices have a max battery life in the hours. Granted, most such devices have colour screens, but it would be foolish to say the faster chips doesn't play a part in it.

The Motorola MC68328 DragonBall microprocessor uses about 55mW. The power consumption for a modern colour LCD screen is about 25mW. So CPU power usage is an important factor in battery life. There is a paper on power consumption which is worth a read which shows that power consumption for the Palm Pro 16MHz uses between 26mW (sleeping) and 160mW for intensive tasks.

Now the ARM family has been designed for high MIPS/Watt ratios. Looking at the ARM10 tech specs the ARM chip uses 275mW when running, and can drop down to less than 5mW when idling, and in sleep mode uses 0mW with the option for fast wake up.

So what does this mean for the average Palm user? If the CPU goes to sleep in between uses, and idle mode is invoked during any period of inactivity, the new Palms should maintain good battery life while being a lot snappier to use. To me, it looks like you should get the same sort of battery life if you use it for the same sort of tasks. If you want to play MP3's all day long, then your Palm will be chewing around 350mW out of the batteries. Given that a single AAA battery gives roughly 1000mAH (milliAmp Hours) so that would imply a pair of AAAs would give you about 8 hours of continuous 100% CPU usage.

Cheers,

Toby Haynes

The Real Limitations - Not CPU, though it helps

[billstewart]

My Palm7 is dog-slow. My Psion 3A wasn't, in spite of having 3 times as much gorgeous screen area and a 7.8MHz 8086-clone processor (Of course, having an excellent keyboard instead of a touch-screen with handwriting recognition saves a lot of horsepower.) Quadrupling the CPU might make the handwriting recognition faster and more accurate, and make up for the clunky operating system. And somebody else already commented that using an ARM core makes it much easier to use already-written ASIC libraries, which means that Palm can do more integration and keep costs down in the future, which is important when you want your product to cost less than 50 shares of your stock. The other big win is that changing processors gives you an excuse to fix some of the appallingly small limits built into the OS, like the 4KB data structures used for memos and email. The Psion had a few 32KB or 64KB limits from 8086 small model, but the last system I used with a 4KB limit was IBM 360 Assembler Language - the PDP-11 had gotten away from the PDP-8's 12-bit limit....

Memory is important, and it's pretty close to free. Selling a 2MB machine is tacky enough if you're doing it to get people to buy the overpriced memory expansion card, but there's simply no excuse for doing so on a non-expandable machine just to create product differentiation, or to bait&switch people into buying the much more expensive model just to get $6 more RAM. For many applications it doesn't matter, but if you want people do buy the box for things beyond the basics - ebooks, or industry-specific applications that require more data, or reading some real fraction of their email (especially using the overpriced radio link), you need more memory, sometimes lots more. Even if the future Palms don't play MP3s, storing compressed speech uses about 1MB per 20 minutes, so it helps to have more memory.

Better Screens - the Psion 3a has 480x160 mono, and lasts about as long on batteries as the Palm; the WinCE and iPaQ machines really do have good-looking screens, more readable as well as flashy, but the battery life is too short for practicality.

Audio, especially speech recognition - that does need more horsepower, though some of it can be done with ASICs like cell-phone voice compression instead of the CPU if that makes sense. Microsoft is going after the MP3-player / Video Game / TV set in your pocket market, but for business users and other people who want organizers rather than toys, the two obvious directions to go are cellphone capability and speech recognition interfaces to the box, and those may be CPU burners.

SDK builds processors?

[molo]

Motorola, TI and Intel will use Palm's software developer kit to build processors using ARM technology, sources say.

Man, thats some SDK. Why don't my SDKs do that?

One chance for glory

[steveha]

This is a risky move, but if Palm does it right it will pay off. They only get one chance.

Apple showed it can be done. Apple really did a great job of smoothly transitioning their users over to the PowerPC.

As long as they are doing this, Palm needs to break out of the 160x160 display trap. I'm sure they are eager to do it. All the current Palm apps, when running under emulation, can have a 160x160 box to run in, but new software can have more room. And, Palm needs to release two different reference designs, with different screen sizes, just to make people get serious about testing their apps with different screen sizes.

I'd be thrilled if they used this as an opportunity to transition from PalmOS to Linux, with the emulator running PalmOS of course. If they did a good job on the SDK, people should be able to recompile PalmOS apps to run under the new Linux. I don't have a feel for how likely this is, but it would give Palm more control over their destiny, and save them paying licensing fees.

They should make sure the new Palm can play DTMF tones through its speaker; people have been crying for this since the first Palm. Look up a phone number, hold the Palm next to the phone, and click on "Dial" and the Palm makes DTMF tones and dials for you. I'd like that.

If they make it rock-solid reliable, and give it really long battery life, this transition could be a very good thing. Oh, and they need to keep making DragonBall Palms for a while: don't make people feel they are being forced at gunpoint to switch.

steveha

More difficult for Handspring

[steveha]

This will be more difficult for Handspring than for Palm. The Springboard is basically a direct connection to the DragonBall's bus, so I can't see how it would be very easy to do a compatible Springboard on an ARM chip. Also, Springboard modules have code on board for the cool plug-and-play, and that code is of course DragonBall code.

So, does Handspring stick with the DragonBall? Or do they try to emulate the Springboard bus using the ARM? Or do they do Springboard 2.0 using the ARM bus? Or can they do a new bus that isn't so tightly coupled to the CPU's bus? (How about making the plug-and-play use some sort of portable bytecode?)

By the way, I hope Handspring will adopt the "universal connector" idea Palm introduced with their latest devices. Having both a serial port and a USB port is a nice thing. Having both of them plus a Springboard would be even better.

steveha

Re:oh no!

[rjamestaylor]

Oh no the program i had to write for assembly class to draw rectangles will be usless now!

Now?

This has been kicking around for a while.

[suwalski]

I first heard of this last summer.

This brings along great opportunities for Linux. For example, the Rebel.Com NetWinder kernel could fairly easily be modified to run on this Palm system, with full X, etc. The news could impact the Palm and Linux even more than it did on the iPaq.

The ARM is a greatly versatile, very low power RISC chip. I'm happy they are finally pursuing this. Nokia is supposed to also take advantage of this in accordance with their deal with Palm to use the PalmOS. This is why the PalmOS isn't on Nokia phones yet, they are waiting for the ARM version...

This will be cool on two fronts. :)

They are going to be kicking themselves...

[Amigori]

"...Palm maintains a huge lead in market share over its rivals..." Then why would they screw themselves by making their products compete more directly with the products in the minority market share? People are buying Palms to replace paper organizers, not laptop computers. But people are people and they usually want they latest thing with the most power, can do many more things than they need, can fit the most stuff on it, and might be able to play quake on it.

"That's why they are revising their platform--to grow room for the future." A standard idea, but I for one, do not want streaming multimedia and MP3s on my handheld. My Palm m105 does exactly what I need it to do. If I want to watch a movie on a plane, $4 headphones from the stewardess or a laptop w/DVD will do the trick. If I want to read a book, I'll spend $6 and get a paperback from the store or check out one from a library. No batteries, backlighting, or internet connection needed.

The Palms were designed to replace the paper organizer that contained our calendars, to do lists, addresses, etc. Last I checked, my paper dayplanner did not play MP3s. It doesn't need to. My Palm doesn't play MP3s either. It doesn't need to. Unfortunately, too many people forget about this and want the latest and greatest. Maybe when they start burning holes in suits because they run so hot, people will realize that the current Palm line does more than they need it to.

Strike A New POSE

[FrankDrebin]

The shift to ARM will mean of course incompatible binaries for all those vendor-supplied and open-source projects. Generally not too big a deal since there are lots of tools for ARM (gcc supports cross-compiling to ARM of course).

But, perhaps more interestingly, a new version of POSE is required. Development of the original POSE (nee CoPilot) is a fascinating story of reverse-engineering, ingenuity, and personal persistence of Greg Hewgill. Do a Google search for 13hewgil.pdf for an interesting account.

Let's hope Palm publishes the data sheet for the chip they're using -- unlike the originally secret Dragonball -- so POSE can be implemented for the new architecture without too much headache.

Maybe they should have considered...

[Bonker]

Shogakukan's 'Ranma' processor. If you douse it in cold water, it transforms into an 'ARM', but if you pour hot water over it, it changes back into a 'Dragonball'.

"Oh sirs, very tragic story of microprocessor that drown in spring one thousand, two hundred year ago!"

Okay, enough with the stupid anime jokes already.

Need for speed

[spellcheckur]

I don't know about you, but my Palm is great for the limited things I call on it to do.

It essentially replaces my organizer, which was, as far as I can tell, it's intended purpose.

Why does it need to get faster. More memory I understand, color and backlighting and better batteries also, but are people really feeling the lag when looking up phone numbers or appointments?

How about making the things cheaper? I don't want a palm to replace a computer... if I did, I'd buy an iPaq (and put linux on it). I want it to take care of scheduling, take notes and remember phone numbers.
...oh, and I guess a few simple games never hurt anybody.

What I want to know is, why does my palm really need to be faster?

A slightly different version of the story

[stu42j]

From the August issue of MaximumPC:

A sequal to the DragonBall VZ, the Super VZ doubles clockspeed to 66MHz and adds on chip support for USB 2.0, on-chip SRAM cache , DMA memory transfers, and a frame buffer to speed up video. Its 0.18-micron shrink (down from 0.25-micron) should offer a two-to four-fold increase in speed and drink less power compared to the fastest Palm procs available today. The Super VZ will debut alongside the DragonBall MX1, which will be based on ARM technology and run at 200MHz.

No more Dragonball!?

[baumanj]

...instead of the current Motorola Dragonball processor

Oh great! Now how is Goku going to buy gohan for Gohan?

Those heartless bastards.