Sharp's Upcoming Linux PDA

Bill Kendrick writes: "ZDNet reports that Sharp is getting ready to make its Linux-based PDA available to developers in the next few weeks. They'll include a 206MHz StrongARM, 32MB (in the cheaper, developer edition), a JVM, the Opera web browser, and a slide-out keyboard. A profile of the device is available at LinuxDevices.com." We've mentioned this before, but it looks like it'll be here soon.

Comments

Any more information?

[PenguinX]

I would love to get ahold of this little box - I currently have a Visor Prisim and the best thing they have for it is the VisorPhone. Does anyone know of a CF variant of a GSM "visorphone" device? If anyone has anymore details on this device I would love to hear about it.

Sharp Zaurus PDA

[Spootnik]

I just returned from Java One and Sharp had a booth there. They showed off this PDA that looked very cool. I think it takes Palm attachments. The bottom slides down to reveal a tiny keyboard. But the cool thing is that the PDA is a Java app thing that runs under Linux. It was running a 2.4 kernel and it just looked friggin' cool. I don't know what kind of development environment they've got (does gcc have a StrongARM backend?), but I got the feeling that they were looking for people to develop apps for it. I suppose that's because no one will buy it without apps. I signed up to get an early development release, but I don't really know what that means. Does anyone have any more information on this? All the web pages I find are in Japanese.

I'll definitely buy it...

[UserChrisCanter4]

if the ad campaign here is as funny as the japanese one .

Looks good.

[proxima]

Sharp looks like it is actually trying to be a bit innovative with a PDA, unlike many manufacturers. First of all, the reflective TFT color display is good choice - consumers and business users seem to have this desire for color (I personally own a Visor Platinum with a grayscale screen, I love the battery life).

I don't really see what Java and Linux bring to a handheld device. Development isn't that difficult for the Palm OS, even Pocket PC, which have each picked a niche in the handheld market (the Palm OS for basic PIM functions with lots of little add-on software, Pocket PC for built-in support of Office documents and multimedia). I have spent some time thinking about it, and the advantages of Linux (multitasking, different processor support, open source) don't seem as important in the handheld market. At least not yet. If Palm OS and the Pocket PC platforms weren't mature, I would definately think that using Linux would be a much better choice. Unfortunately, it is still quite immature, as one can quickly tell from reading through the Linux development mailing lists of the Agenda. Not to say it isn't useful, but on the same hardware it seems to be slower than the Palm equivalents, from the reports I have read.

Moving on, the choice of compact flash and lithium ion battery was very wise. Better than a proprietary expansion slot, in my opinion, but somewhat more limited. Handspring's sprinboards are capable of doing so much more than memory expansion and modem/ethernet devices - like a remote module, GPS, cell phone, wireless internet, etc. I am not sure how many of these things the compact flash design on this palmtop could support - with something sticking out the top. Seeing as this has a 206 Mhz processor and a color screen, the good rechargable battery will be quite needed. It would be nice if these are easily removable, so that those who don't get a chance to charge for quite some time will be able to pop in a second battery.

The sliding keyboard seems nice, but obviously useful mostly for "thumb-typing". Handspring just announced a clip-on sort of keyboard for their devices that does a similar thing - SnapNType. One thing that I wonder about this Sharp device - will it support handwriting recognition? The site claims the color screen has "touch panel support". Handwriting recognition is fairly difficult to code, as the Agenda creators have found. Grafiti is nice, especially for those that have learned it, but there is some sort of licensing with it.

All in all, this looks like a promising Linux handheld. They learned from the Agenda's mistakes, by including USB connectivity, a rechargable battery, and compact flash slot. With all these features it will definately be in the price range of the already-mature color Compaq's, which means a limited consumer base. I look forward to hearing how well the developer models work.

Re:Looks good.

[jelle]

"I don't really see what Java and Linux bring to a handheld device. "

"One thing that I wonder about this Sharp device - will it support handwriting recognition?"

I can help you a bit with those two questions with one url: xscribble

Re:Looks good.

[proxima]

Perhaps it was a little strongly worded. I can see some advantages of using Linux and Java - the incredible ease of software development and the ability to create programs that run on both the PC and a handheld identically (for business use). However, there are some disadvantages to having a full-featured multitasking OS with a programming language that is notoriously slower than good C/C++ code - speed. Speed means a lot in the handheld world, and the harder a handheld has to use the processor, the shorter the battery life is. This is where the Palm apps shine. They aren't complicated, but they do the job and they are written to be extremely small and optimized. Thus Palms (and Handsprings, and TRGs, and Sonys) need the smallest amount of RAM and processor compared to the other handhelds available on the market. They can manufacture the things cheaper and with longer battery life.

Obviously this Sharp is aimed for the high end market. My main point was emphasized in my last paragraph - this high end market is already cluttered with mature devices like Compaq's and HP's. This is not to say the new Sharp handheld won't meet with some success, but that it will need to mature much more quickly in order to be successful, because too many high-end mature options exist now. Like I said, a few years ago this type of software would have a much better chance before Pocket PC got off the ground, and while Palm was even more primitive (not to say that it is a bad thing, like I mentioned, I am a very happy Visor Platinum user).

It requires more than just software to support good handwriting recognition - the touch screen must be sensitive enough to work well. I had an old Tandy handheld quite some time ago that "supported" handwriting recognition. The software was bad, but the touch-screen really was not designed to handle handwriting well. One could tap through things well, but the handwriting was quite inaccurate, caused by both hardware and software.

I cannot speak for xscribble, but I do know that the Agenda team had to revamp the handwriting recognition recently because it didn't work well (perhaps someone closer to the project could elaborate). I don't know if they based it off of xscribble code or not.

love the input options.

[motherhead]

Wow, touch screen support for future implementations of graffiti or handwriting recognition software and a dropout keyboard. That is just plain polite. As opposed to say... this.

Re:206 mhz Strongarm VS 200 mhz pentium?

[Waffle Iron]

I loaded Linux on my 200MHz iPaq, and it ran vastly slower than a 200MHz Pentium Pro machine that I have. I didn't do any official benchmarks, but it felt like it ran on par with a 66MHz 486.

I would guess that the performance difference has to do with issues such as tiny cache, lack of parallelism and pipelining in the CPU, slow narrow memory, software framebuffer, etc.

StrongARM comments

[horza]

1) No, because it runs at 206MHz does not mean it comsumes a lot of power. It draws 0.7W.
2) It is RISC rather than CISC, and having used a 200MHz StrongARM desktop I can tell you it FLIES. Much faster than a P2-266
3) You use gcc to compile on StrongARM because Linux runs on StrongARM (well obviously). ARMLinux has been around for years running on Acorn machines. You can also cross-compile to StrongARM using a x86 box - just ./configure --target=arm-linux when compiling GCC.
4) You can even use them for Beowolf ;-)

Phillip.

Re:StrongARM comments

[A Commentor]

2) It is RISC rather than CISC, and having used a 200MHz StrongARM desktop I can tell you it FLIES. Much faster than a P2-266

Do you have any hard numbers to confirm this?? RISC processors at the same speed as a CISC processor are typically SLOWER because they do LESS work per instruction than a CISC processor.

Are you really comparing the speed of the processors or how the overall systems 'feels'... It's an invalid comparision, if you are comparing a fully installed Linux system on a P2-266, with all the extra processes and X, etc), to a stripped down StrongARM system with minimum processes and much smaller graphic display.

Re:StrongARM comments

[spiro_killglance]

"Do you have any hard numbers to confirm this?? RISC processors at the same speed as a CISC processor are typically SLOWER because they do LESS work per instruction than a CISC processor. "

This would be true if RISC and CISC today ment
what they used to me. In fact many of todays so
called RISC machines have more powerful instruction sets, with for example three operand
instructions with multiple addressing modes. Mean while the
major architectual inivations from risc processors
like pipe-lining and superscalar are on all modern
microprocessors. For more info see this ars-technica article.

All this, plus the AMD vs INTEL megahertz wars, leads to a curious roll reversal where so called
RISC chips do more work per MHz, while so called
CISC chips (actually only the x86 is called CISC
these days), have the highest clock rates.

Re:StrongARM comments

[horza]

RISC processors at the same speed as a CISC processor are typically SLOWER because they do LESS work per instruction than a CISC processor.

That would be true if CISC instructions all executed on one clock cycle (as RISC instructions do) but that isn't true. CISC processory do MORE work per instruction. In fact, some (little used) CISC instructions can take hundreds of clock cycles. Advantages of having one cycle per instruction include efficient pipelining. The Intel Pentium is a strange hybrid. It has a RISC core which works on its own microcode, and 90% of the silicon is actually a hardware translator which converts the x86 CISC instructions to the Intel RISC microcode.

My basis was using a 200MHz RiscPC running RiscOS. From turning the machine on to the desktop running takes less than one second.

Phillip.

Re:StrongARM comments

[ReelOddeeo]

"Do you have any hard numbers to confirm this?? RISC processors at the same speed as a CISC processor are typically SLOWER because they do LESS work per instruction than a CISC processor. "

Less work per instruction, not less work per clock cycle.

RISC, with simpler design, are easier to put more functional units onto a chip of the same size with the same technology to deposit the same number of transistors per chip. Therefore RISC can do multiple instructions per clock with proper instruction scheduling. That is, as long as you can keep all the functional units busy. Thus, the compiler's instruction scheduling can make the difference between lousy and excellent RISC performance.

Re:StrongARM comments

[ReelOddeeo]

many of todays so called RISC machines have more powerful instruction sets, with for example three operand instructions with multiple addressing modes.

Are you sure about the addressing modes?

I have not surveyed a lot of RISC designs. But as I recall, one basic feature was that there were very few instructions that could load and store to memory, and a limited number of memory addressing modes.

The three operands make sense. But aren't the risc instructions predominantly all register-to-register with a very large number of registers? A few register-to-memory instructions, and then you optimally schedule in your register-to-memory instructions at cricual points, rather than having to take where they fall "inside" of the CISC instruction? Basically you write several ops to "build" a CISC instruction. Then you take the overall stream of instructions and optimally schedule operations to keep all functional units busy. This is done in the compiler statically, rather than by trying to rearrange instruction execution dynamically in hardware, thus needing even more hardware. I've only read up on all of Apple's PPC propaganda during the early 90's, so I'm no expert here.

Back to addressing. Yes, I don't expect to see RISC instructions with an addressing mode such as: preindexed, double indirect, added to a register, indirected, then postindexed by another register and added to the phase of the moon.

Sharp sign Amiga as content provider

[Twig]

The Sharp Zaurus PDA is also the first mainstream device to see the return of old friend Amiga to modern computing.

Amiga have been signed by Sharp as a content provider for its new Zuarus platform. The Zaurus ships with Amiga's "AmigaDE", a platform agnostic digital environment which is hosted by the Linux OS.

Sharp demonstrated the Zaurus running AmigaDE applications a while back. Here's the link.

Amiga have also been signed by Psion to provide its AmigaDE system for their NetBook products.

--
Ben.

There's no room in the marketplace

[Kris_J]

Sure, it's got some interesting specs, but there simply isn't room in the marketplace for this PDA. It's not a gaurantee of failure, but it's close enough -- this device would have to unseat MS' PocketPC or the Palm to really be worth bothering about. And by the sounds of it a colour Handera 330 would remove most of its market. (CF II and SD slots, QVGA.) The keyboard's cute, but it looks like it wastes a lot of space that coulb be better used for Flash RAM or battery life. If only it had Bluetooth built-in or some other particularly interesting technology. Just having a Linux core really isn't enough of a hook.

Anyway, this is not the time, economy wise, to be trying to introduce a completely new product in a genre of questionable usefulness. My TRGpro spends only about one in five days out of its drawer, and I really like it, I just can't find a use for it that justifies lugging it around. (Particularly now summer is on it's way.)

Re:There's no room in the marketplace

[macpeep]

I don't know.. I just came back from Japan and the Sharp Zaurus is the PDA that had sold the most there. I personally didn't like the slide out keyboard all that much, but I must say that on a whole, it's a very nice device. Tao's Java VM and media API's are very good indeed, so this really has some potential!

On a slightly unrelated note, bluetooth was *everywhere*. This truly amazed me. Just about every device and manufacturer had shitloads of bluetooth stuff!! It seems the reports of its death are greatly exaggerated, at least over there!

Needs better connectivity

[hey!]

I think this thing needs either a serial port or a second CF slot.

The USB is a nice touch, but it looks like it might get in the way of the CF slot.

I see the real possibilities in a Linux powered device like this is in integration into larger system and field based data collection. There's no way anybody is going to break into the PalmOS/WinCE dominated world.

The problem is when you start assembling systems to do things like field surveying systems, the features you get don't add up (e.g. you need a huge CF card to hold your maps files, but then yo have no way to connect your GPS). I do a lot of (simple) stuff with GPS hand PDAs -- I think every PDA should have a serial port!

Re:Needs better connectivity

[alhaz]

I think this thing needs either a serial port or a second CF slot.

There's no reason to believe you couldn't get a serial port dongle that clips onto the docking port as are available for palm pilots.

Two CF slots would make it rather large, but the prototype I've played with does have an MMC slot in addition to the CF slot.

The USB is a nice touch, but it looks like it might get in the way of the CF slot.

What the heck are you looking at? There's no USB port on the device except for the USB what runs through the docking connector.

Regardless, like most (all?) strongarm handhelds it's probably using the built-in USB on the strongarm chip itself. That means that it's target-only. The SA1110 is a USB device, not a USB host. You can't attach USB peripherals to it as it is a USB peripheral.

SA1110 usb also isn't very fast, it's a design limitation. But how fast does it have to be to copy a few megs of data to the host computer?

There are usb chips available that can be interfaced to the SA1110, but I haven't heard of a PDA that has one, given the voltage requirements. A USB hub has to be able provide 500mA of +5v to each device. Not practical for something that fits in your pocket.

The problem is when you start assembling systems to do things like field surveying systems, the features you get don't add up (e.g. you need a huge CF card to hold your maps files, but then yo have no way to connect your GPS). I do a lot of (simple) stuff with GPS hand PDAs -- I think every PDA should have a serial port!

It does have a serial port. The docking interface may provide the serial buffer chip but that's no big deal to build into a dongle. It just doesn't have a DB9 right on the case, which is perfectly reasonable.

Like i pointed out before, it does have an MMC slot on the side. MMC cards are not as cheap as CF or SmartMedia but they do exist, and if push came to shove you could put a 64 meg MMC card in the side and stick something else in the CF slot.

Keep in mind that while CF can be implemented as a storage-only interface, in order to be capable of hot-swap it is generally implemented as sortof a small formfactor PCMCIA. There exist LAN cards, modems, video confrencing cameras, and all manner of peripherals available in the CF formfactor. It's just like pcmcia, but smaller.

Re:pull-out keyboard

[biglig2]

Eh? As a long-time Psion user, I have to say that thumb-typing is perfect on the 5MX.

The 5's have the best PDA keyboards there are. (Of course, I can't get one with color, or a big fat fast processor, etc.etc., hence my current bout of "new PDA lust".)

Ruby on PDAs (was:maybe someone can answer this)

[Bill Kendrick]

I haven't played with it myself, but 'miniRuby' is available for the AgendaVR3 PDA (which also runs Linux).

Then again, NetHack and Apache are available on the Agenda, too. ;)

Re:MPEG4 movies listed under features?

[Jagasian]

Well, considering that IBM and others make 1GB Microdrive CF cards, you could store at least a few high quality MPEG4 movie such as the Matrix or Star Wars Ep1. Considering that you could recompress the MPEGs for such a low resolution (of the LCD screen), you could probably fit allot more.

Thumb Keyboards

[FallLine]

Such as those on the RIM (Blackberry) pagers and Motorolla P900 (though those are much more cramped) are actually quite cramped. In case you're unfamiliar with them, you don't type on them like you do with a traditional keyboard, you just use your thumbs, as in the edge of your thumb. I've got fairly big hands and I can comfortably type ~30wpm on the RIM pagers. Ignoring space issues, the thumb keyboards beat the pants off Palm and PocketPC's handwriting recognition and other common forms.

That said, I agree with you, there should be better solutions out there.... Who ever invents one that:
A) can be implimented without taking up a great deal of space (at least when compacted)
B) can be LEARNED relatively quickly
C) allow proficient users to type comfortably upwards of 40wpm [especially in PDA/road-type situations]

will be in a real position to dominate the PDA market....