Problem
The leading positions of Palm on the market of handhelds and, in particular,
PDAs are getting somewhat doubtful. The reason is continuously growing
competition from Pocket PCs. Let's leave aside the Symbian EPOC which sank
down in other market sectors (in industrial PDAs and in smartphones, for
example, from Nokia) and still popular but aged monochrome incarnations
from PSION, and come up to cumbersome and weighty Windows CE devices with
short run-down time (Windows CE can be called also Windows for Pocket
PC or even .NET - the idea is the same). So, what can we see today?
1. These devices are not bulky anymore - the most compact models
come very close to Palm OS from Sony. Such a rapid pace can soon make them
equal in dimensions with the Palm computers of the 500 series.
2. The weight has fallen down as well; a typical difference is a couple
of dozens of grams - say, 120 g of Palm OS against 170 g of Pocket PC.
3. The running time is still a sensitive point, though it makes 4-8
hours which seems sufficient for a daily-charged device. At the same time
a color Palm OS swallow batteries at 6-12 hours, that is why the advantage
is not great.
4. A higher resolution of the screen and its rectangular shape which
is better for reading play into the hands of PocketPC. Although there are
Palm OS models of 320x320, this resolution is not official for Palm OS
4.x, and its support requires special modified OS versions, for example,
from Sony. That is why it's still unsupported by most programs, and the
screen remains square. There are just two models with a non-square screen
and an active graffiti area, but only few programs are able to work with
such screen.
5. Palm has nothing to oppose to wider multimedia capabilities of the
Pocket PC. The system has no multimedia support, any operations with sound
are non-standard and supported only in few models. At the same time, any
modern Pocket PC is able to play mp3 and work as a dictaphone. The Pocket
PC 400 MHz versions will probably bring playback of video clips in MPEG*.
Taking into account moderate prices for Compact Flash cards (below $100
for 256 MB) audio and video records on PDAs do not look extraordinary.
6. A higher level of games for Pocket PC - they benefit from sound and
higher processor's performance.
7. A wider range of network programs, a normal browser supporting most
network technologies typical of PCs.
8. Fast porting of the Windows code - I was surprised how easily and
swiftly I transferred to my pocket PC video and speech codecs first developed
for a desktop version of software for video conferences. As a result, coupled
with a cell phone a Pocket PC turns into a videophone.
Well, the situation doesn't look very pleasant for Palm OS. Here are their
advantages for today:
1. Unsophisticated user interface. But each new Pocket PC version
gets more friendly toward users, it becomes more convenient and gets rid
of flaws in UI.
2. Lower time of response to user actions. As the frequency of the Pocket
PC grows up the advantage vanishes away, but it's still noticeable in some
applications.
3. High stability of operation - the system hangs up and reboots quite
seldom.
4. Compact, strong and stylish metallic cases of senior models.
5. Color models (with backlight used sometimes) has a run-down time
greater by 1.5-2 times and monochrome models live tenfold longer.
6. Miniature dimensions and small weight
7. A great batch of accumulated software
8. Lower price. But some color Pocket PCs are already available at moderate
prices.
Well, if Palm doesn't want to lose its share (the current situation is
more or less balanced but possible changes would bring no good to Palm)
it must take certain measures.
Solution
So, they made a decision on a new hardware platform and development of
a new operating system - Palm OS 5.
Hardware platform
The platform replacement is necessary because of too slow and unsatisfactory
development of the Motorola Dragon Ball processors which earlier were used
for Palm OS. Motorola lost its customers in its time as now it can offer
only a 66MHz processor bundled with various built-in controllers! The aged
software architecture of the Dragon Ball (32bits, though with a 16-bit
data bus) makes the things even worse. Today one should go with 100 or
200 MHz and a normal 32-bit multitask processor. The developers from Palm
took notice of ARM - a good popular RISC architecture first developed for
desktop workstations and then succeeded on the market of mobile and built-in
solutions. This architecture, or rather its superset from Intel, is used
in all modern Pocket PCs. Here are strong points of this solution:
1. High clock speeds and performance, well-scalable 32-bit
RISC architecture
2. Processors compatible with ARM can be chosen from different manufacturers
3. High-quality optimizing compilers and a great heap of software for
this architecture
4. Inexpensive processors with integrated "all-in-one" peripheral controllers
And here are its weak points:
1. Potentially higher power consumption is arguable as it depends
on many factors and processors
2. Incompatibility with old software
The second item can cause headache. There are two wayouts - either using
a second processor (probably located on one die with the main one) compatible
with the Dragon Ball for old applications, or software emulation of the
Dragon Ball. The second option looks more suitable. The technology of emulation
is well developed for today, i.e. there are speedy algorithms that translate
codes before fulfilling them. But all of them require extra memory usage
and other expenses. Besides, such emulators don't cope well with a self-modifying
code which can be used for copy protection of old applications of Palm
OS. Reportedly, the ARM 200 MHz processor copes perfectly with emulation
of Dragon Ball (like modern 33 MHz models). Well, we'll see it soon.
One more advantage of the emulation is that typical non-game applications
spend most part of time in the OS code waiting for user's moves or working
with a user interface. Certainly, the whole OS will be rewritten for the
new processor and no emulation will be needed - in contrast to Apple: during
the similar platform replacement (it was also a Motorola's fault :-)) they
first changed the core and most important subsystems and then, in new versions,
they were getting rid of the remains of the old code.
It's also known that hardware solutions for the Palm OS 5 will be based
on the OMAP suite from TI (Texas Instruments), which together with the
ARM RISC core houses the DSP which releases the main processor from such
heavy multimedia tasks as decoding and encoding of multimedia data or,
maybe, handwritten text and voice recognition.
Besides, the suite comes with a screen digitizer interface, a color
LCD controller, USB and serial interfaces, controllers of memory cards
such as CF, SD and MS and other interfaces. It also provides support for
wireless radio modules: Bluetooth and 802.11b
Software platform
The key new features of the system are modularity and multitasking. This
will bring in dynamic-link libraries (DLL) and determine a clear concept
of a device driver. Multitasking will be implemented on the thread level
only, which is enough for pocket PCs where all programs coexist in the
shared memory. There is no file system as PC or Pocket PC sees it - all
information will be stored in bases of RAM. File systems of external storage
media such as memory cards will be supported via a special API like in
the Palm OS 4.
There isn't much to be changed from the standpoint of the base API.
Most new features will arrive in the form of new APIs without losses of
compatibility at the program code level with the old one. Theoretically,
adaptation of programs must be extremely simple, - it's enough to recompile
them for the new platform. But porting onto the new architecture can be
a possible source of problems. Resolution of 320x320 will finally be supported
at the level of the base API. However, non-square screens will still be
lacking in original Palm models and, therefore, won't be supported by the
base Palm OS. Palm makes just one basic resolution as it's easier to develop
software and standardize interfaces of programs. (will modularity of the
system let it to get adapted to non-standard resolutions? or Sony will
have to deal with the original system code again creating its own special
versions?).
The sound support will be standardized. Both digital sound and playback
of any PCM flow (16bit, stereo) will be possible. It seems that a microphone
is going to become an essential part of new Palm models. I don't know whether
a speaker is going to be integrated but a headphone jack will certainly
be provided. Network and cryptographic capabilities and API of the new
system will mushroom up. The wireless connection support (Bluetooth and
802.11b) is incorporated, as well as the cryptography API (RSA RC4 128
bit); and all critical data will be stored encrypted.
At the moment at http://www.palm.com/
in the section for developers you can download for free a Palm OS 5 emulator
which, first of all, is interested for the developers themselves - the
standard pack of programs hasn't changed much, and first models on the
Palm OS 5 will be of the most interest for us.
Summary
1. This fall will give us the new soft hardware platform PalmOS
5
2. The platform replacement will be crucial for Palm - they will either
try to maintain its market share or be swept away with new waves of Pocket
PCs and smartphones.
3. New, mostly multimedia, capabilities might bring not only pleasure
for users but also lower operating time of future Palm computers. In this
case Palm can become equal in dimensions and size with Pocket PCs, thus,
losing its best advantage.
Well, this "revolution of hope" will bring certain results quite soon,
and now I hope Palm will climb the mountain (ARM architecture and multimedia
capabilities) without getting new bumps (dimensions and weight).
PS At IDF Fall 2002 in San Jose it was mentioned that Intel XScale
based devices running under the PalmOS 5 would soon arrive in the market!
