Comments on: Palm OS 5 Can Take Full Advantage of Intel's XScale

It was recently revealed in a silicon.com article that the soon to be released handhelds that run the Pocket PC OS on the PXA250, Intel's new XScale processor, won't see an increase in speed when compared to the the company's older StrongARM processors. As Palm OS 5 can run on these same chips, some people have wondered if it will also be limited. Fortunately, this turns out to not be the case.
Return to Story - Permalink

Article Comments

 (74 comments)

The following comments are owned by whoever posted them. PalmInfocenter is not responsible for them in any way.
Please Login or register here to add your comments.

Comments Closed Comments Closed
This article is no longer accepting new comments.

Down

Good to go!

I.M. Anonymous @ 6/18/2002 8:52:31 AM #
I'm sure there are a few dozen engineers at Intel cheering at this one. Time and time again, Intel released documents on how to optimize for the x86 CPUs only to have Microsoft just ignore them. Now, Intel gets to write the docs and code for it too!
RE: Good to go!
I.M. Anonymous @ 6/18/2002 9:39:02 AM #
Ah, one up for Palm. One less for PPC fans (not so much PPC itself... I think... those companies are only out of short term money anyway. No one in the right mind colludes with M$ thinking of long-term survival... Which partner of M$ lasts long enough, other than Intel? And even then, PocketPCs/WinCE was originally written to undermine Intel's dominance [hence ARM, MIPS, etc. non-Intel platforms originally]).
MicroSoft addresses the issue
Palm_Otaku @ 6/20/2002 8:42:01 PM #
After getting the word directly from MicroSoft, there are some very disillusioned PPC fans out there right now:

http://www.pocketpcthoughts.com/forums/viewtopic.php?t=1772


RE: MicroSoft addresses the issue
I.M. Anonymous @ 6/20/2002 10:52:28 PM #
Here's my favorite part of this interview. Ed Suwanjindar from the Microsoft Mobile Devices group says "In the mobile device space, we don’t think that MHz is what ultimately matters to customers. What matters the most in this market is whether customers can do what they want to do with devices quickly and easily. "

Isn't that what we Palm users have been saying for years?

Oh, the pain...

sandbuck @ 6/18/2002 9:45:42 AM #
>*gloating laughter*<

This is just priceless! I wonder if the PPC hardware makers will design their units to be flashable. If so, I predict a lot of dual-boot PPC's that boot up with PPC less and less as owners convert to Palm OS.

RE: Oh, the pain...
sub_tex @ 6/18/2002 9:52:46 AM #
Why is it so important to everyone that people only use one OS over the other?

Who cares!

People will use what they're comfortable with. Period.

RE: Oh, the pain...
I.M. Anonymous @ 6/18/2002 10:17:57 AM #
Actually, we will not see any end users loading PalmOS on their ppc hardware.

They will run into the same problem they would run into right now if they tried to load an m130 OS onto an m515. The OS must be customized at the HAL and (for ARM) DAL level for each specific hardware set, and this can only be done by someone with intimate knowledge of the actual hardware, the source code to the reference HAL, and the source code to the DAL for that particular processor. You would also need to write custom code to handle the slots, infrared, and serial ports. Palm doesn’t exactly provide these resources to anyone who asks.

Good luck booting PalmOS on your iPaq. You bought it, live with it.

RE: Oh, the pain...
I.M. Anonymous @ 6/18/2002 12:53:26 PM #
People need to get a real life. I frankly don't care about either PPC or PALM "winning." I own both types of machines and frankly I would rather have consumers win.

To all you people who gloat and have nothing better to do: Get a life. A real life. Go on a date. Read a book. And think outside of the box.

You people are really, really sad specimens of humanity or is it in-humanity.

RE: Oh, the pain...
I.M. Anonymous @ 6/18/2002 1:36:18 PM #
My Toshiba is flashable now, so I presume that the Xscale machines will be too. Then you can expect MS to release a patch to PPC2002 or include it in PPC2003 or whatever.
RE: Oh, the pain...
sandbuck @ 6/18/2002 6:02:08 PM #
>>I frankly don't care about either PPC or PALM "winning." I own both types of machines .....

This is a signature "closet troll" statement. The bitterness in your post is also typical. The idea of Palm OS eradicating your precious high-end market, while maintaing its massive base on the low-end must make your PPC gall bladder do summersaults. Lose graciously. Be one with the Palm. You'll feel much better!

RE: Oh, the pain...
I.M. Anonymous @ 6/18/2002 8:55:24 PM #
Why is everyone on this site not picking apart PPC OS and hardware, get called a troll? True, false, or exagurrated the Palm OS zealots here are ok, but play devils advocate and you get your ass kicked.

I don't mind this site, and believe it or not, I am not a PPC zealot (although I know I will be called one). I use both OS's happily.

Palm Zealots on this site act like members of a cult, it drives me crazy.

RE: Oh, the pain...
I.M. Anonymous @ 6/19/2002 11:34:16 PM #
I think it's amusing how the trolls always go on about how Palm is going to die, and now they say it doesn't matter which wins.

Temporary advantage

sub_tex @ 6/18/2002 9:49:43 AM #
I think it's great that OS 5 will take to Xscale with no problem. However, i don't think it will be long before the PPC does as well.

Personally i don't know why people care anyhow. It's not like Palm OS ever had to contend Mhz wise with PPC before. We all know Mhz has nothing to do with performance.

The same applies to how 1.6Ghz Athlons are running faster than 2.2Ghz P4s.

RE: Temporary advantage
JimBob @ 6/18/2002 9:56:39 AM #
> Personally i don't know why people cazre anyhow. It's not like
> Palm OS ever had to contend Mhz wise with PPC before. We all know
> Mhz has nothing to do with performance.

While we have always known MHz isn't the best way to judge the performance of a handheld, we've had to put up with years of PPC fans dismissing our handhelds because theirs had a faster processor, without caring about actual performance. It is only human nature that we'd do a bit of gloating.

While I agree that this is temporary, Microsoft doesn't seem to be in all that big a hurry to do anything about it. The silicon.com article said a version of the PPC that takes full advantage of XScale may not be out until 2004. That is a long time for the Palm OS to take back any temporay gains the PPC has made in marketshare.

RE: Temporary advantage
mtg101 @ 6/18/2002 10:01:09 AM #
The next generation of PPC - CE.net - will also not take advantage of the Xscale processor. It won't be until the generation of PPC after the next until we might see Xscale support...

As for the MHz arguments - it's a valid point. I prefer the "I can run the Xscale at 200MHz with half the power consumption" view of things.

---
russ@russb.fsnet.co.uk

RE: Temporary advantage
I.M. Anonymous @ 6/18/2002 10:07:37 AM #
I don't think it's a "temporary advantage." As an engineer it's a strong indication to me that thte Palm OS is superior in terms of design to the Pocket PC. I really appreciate good design, and I think that this demonstrates exactly why.

Internally the Palm OS has ALWAYS been a superior PDA operating system... It is my hope that we'll see a slugfest between symbian (which is interesting in its own right) and the Palm OS in the coming years.. Its frusturating to me that people take to a product just because its a Microsoft OS, and everyone knows that Microsoft makes the best products.

RE: Temporary advantage
I.M. Anonymous @ 6/18/2002 10:39:42 AM #
"I don't think it's a "temporary advantage." As an engineer it's a strong indication to me that thte Palm OS is superior in terms of design to the Pocket PC. "

How hard it is to port a "Hello World" Program, and how had it is to port a program like Word, Excel?

it doesn't show Palm's superiority in design, it only shows Palm is much much less complex than Win CE.

RE: Temporary advantage
mtg101 @ 6/18/2002 11:34:26 AM #
>it doesn't show Palm's superiority in design, it
>only shows Palm is much much less complex than Win CE.

I think having to include Win3.1 parameters in startup code is complexity developers could do without!

---
russ@russb.fsnet.co.uk

RE: Temporary advantage
I.M. Anonymous @ 6/18/2002 11:52:04 AM #
>>I prefer the "I can run the Xscale at 200MHz with half the power consumption" view of things. <<

Exactly. This is the main reason why PPC fans are excited about Xscale. I've owned Palms for years, and now I use a Toshiba e570. Whereas my Palm V would run for weeks on a single charge, my e570 runs out of gas in several hours. Of course, there are reasons for that: power-hungry processor, color screen, frontlight, multi-tasking code, etc. But poor battery life is the bane of a PPC user's existence. Underclocking the Xscale can extend the battery life dramatically.

RE: Temporary advantage
I.M. Anonymous @ 6/18/2002 12:50:46 PM #
>>Exactly. This is the main reason why PPC fans are excited about Xscale. I've owned Palms for years, and now I use a Toshiba e570.

Yeah - but sorry guy: underclocking yr PPC on a 400mhz x-scale (down to 200mhz) will actually result in slower performance then a 200mhz StrongArm - but you will save power. Again, this is due to MS not optimizing for x-scale. To get performance equivalant to current 206mhx StrongArm processors, you have to run the x-scale at 400mhz - at which point yr back to the same power consumption.

RE: Temporary advantage
I.M. Anonymous @ 6/18/2002 12:55:54 PM #
> Internally the Palm OS has ALWAYS been a superior PDA operating system

You must be a pretty lousy engineer. Can you claim the PalmOS is superior when it doesn't even have a native filing system? It has to use a hack for files on memory cards.

Gimme a break and go back to engineering 101 or in your case basket weaving 101.

RE: Temporary advantage
I.M. Anonymous @ 6/18/2002 4:06:17 PM #
>> Internally the Palm OS has ALWAYS been a superior
>> PDA operating system
>
>Gimme a break and go back to engineering 101 or in
>your case basket weaving 101.

the poster said "superior PDA operating system" that makes the question very complex, what is a "PDA operating system", and what is considered as "Superior".

your "Superior" might be different from mine
your definition of "PDA operating system" might be different from mine.

Long term advantage
I.M. Anonymous @ 6/20/2002 10:47:43 PM #
> I think it's great that OS 5 will take to Xscale with no problem. However,
> i don't think it will be long before the PPC does as well.

According to Ed Suwanjindar from the Microsoft Mobile Devices group, Microsoft has already done everything it is going to do to support XScale for a long time. Turns out PPC2002 can't support both XScale chips and the SA1110 procesessors in iPaqs.
www.pocketpcthoughts.com/forums/viewtopic.php?t=1772

interesting

I.M. Anonymous @ 6/18/2002 9:59:50 AM #
oh, this is interesting. so, basically, the PPC devices will be (for the time being) stuck at 200 MHz while palms will be going at 400 MHz (likely with less overhead)

This is rather amusing, considering that the PPC users have always been proud of their faster processors

RE: interesting
mtg101 @ 6/18/2002 10:07:57 AM #
I think it's more of a case of using huge amounts of extra power to run at 400MHz - rather than not being able to - but I'm not 100% on that.

---
russ@russb.fsnet.co.uk
RE: interesting
I.M. Anonymous @ 6/18/2002 10:11:07 AM #
you can buy a 400mHz PPC Xscale already.
RE: interesting
I.M. Anonymous @ 6/18/2002 10:12:33 AM #
People who already have the Toshiba e740 Pocket PC with the PXA250 say it isn't any faster than the e310 with a 206 MHz StrongARM.

You can read about it on PDABuzz if you think this site is biased against PPC.

RE: interesting
I.M. Anonymous @ 6/18/2002 11:44:59 AM #
The principal advantage of the Xscale architecture at this time is lower power consumption, not speed. The StrongARM based PPCs have battery life barely better than some laptops, as any PPC owner will admit. Xscale is more power-efficient.
RE: interesting
I.M. Anonymous @ 6/18/2002 12:55:25 PM #
>>The principal advantage of the Xscale architecture at this time is lower power consumption, not speed.

No - the principal purpose of x-scale is both. You can run at 400mhz with the same power as 206mhz StrongARm, or at 200 mhz using half the power of StrongArm. The problem for PPC users is that - with no optimization for x-scale in the os - you have to run at 400 just to get the speed of a 206mhz StrongARM. But yes, if you are willing to take the cut in performance, you can increase yr battery life.

RE: interesting
I.M. Anonymous @ 6/18/2002 1:43:38 PM #
Is this true? Or have Toshiba chosen to extend battery life at the expense of speed. Considering that the current crop of 206MHz PPC2002s are pretty damn fast.
From a marketing point of view this makes good sense.
The new model has better battery life, is as fast as the best current models and also has a headline of a new 'faster' processor. Regardless of the actual performance many consumers will be persuaded this alone.

Reality check

I.M. Anonymous @ 6/18/2002 10:04:43 AM #
Ok, you can love Palm, but Ed is a little bit too optimistic. We are talking about HAL on device that must EMULATE a 68K processor, and you are desuming that it would be faster than a PPC? C'Mon, you are loosing credibility.
RE: Reality check
I.M. Anonymous @ 6/18/2002 10:12:25 AM #
don't say that outloud.

(on top of that this device is unannounced and rumored to be released in 2003. There are a lot of unannounced PPC machine too. For eg. NTT Docomo, NEC/Compal. All in the same time frame as Asustech)

But all being said, this rumored Xscale powered by yet to be seen in real world OS5.0, would certaily take over the PDA universe.

RE: Reality check
mtg101 @ 6/18/2002 10:15:07 AM #
The OS itself runs native ARM code. So whenever you make a call to the OS - which is most of what most apps do - you're running at full ARM speed.

It's only when you do intensive work within your code (the sort of thing that means you need a progress bar on PalmOS 4.1) where you'll have problems with the emulation speed. However - in PalmOS 5 you can use ARM code specifically for these intensive operations.


---
russ@russb.fsnet.co.uk

RE: Reality check
I.M. Anonymous @ 6/18/2002 1:00:22 PM #
whoa,whoa,whoa! we are talking about HAL here - not PACE. HAL is not an 'emulation layer' - it completes the OS with a layer - IT DOES NOT 'translate' from one set of processor codes to the next (like PACE). In OS 5, most software will require PACE - which will mean slower speeds as it is emulating motorola 68k. But anything written as native ARM (w/ os 6 this is the main approach) wuill not use 'emulation'.
RE: Reality check
I.M. Anonymous @ 6/18/2002 11:45:28 PM #
there is no emulation in PalmOS 5!
"PACE does not emulate the 68k chip or other hardware, nor does it run the old OS. Instead, it interprets the 68k instructions itself, and handles 68k trap instructions (used by applications to call OS APIs) by making calls into the native Palm OS 5 system. It is therefore quite efficient in terms of memory and processing power, and most developers will find that their applications and debuggers work just as before, other than the speed difference."
"Because Palm OS 5 is compiled to run natively on ARM processors, every call to the operating system will run at the processor's full speed while still providing excellent compatibility for existing applications which were coded for the Dragonball processors. This means that high performance is expected in nearly all circumstances, since most applications spend the vast majority of their time inside OS calls."
RE: Reality check
I.M. Anonymous @ 6/19/2002 4:38:14 PM #
PACE does not emulate the 68k chip or other hardware, nor does it run the old OS. Instead, it interprets the 68k instructions itself, and handles 68k trap instructions (used by applications to call OS APIs) by making calls into the native Palm OS 5 system. It is therefore quite efficient in terms of memory and processing power, and most developers will find that their applications and debuggers work just as before, other than the speed difference.

That's marketing speak. The facts are pretty simple:


  • Palm applications all use 68k machine code. These
    instructions are interpreted in software by PalmOS5.
    That's a slow process.
  • The operating system itself has been compiled for,
    and runs in, native ARM code.

What it comes down to is that the OS and the GUI will
probably feel quite fast, but any application that does
a lot of work will be much slower than a native ARM
application. I wouldn't be surprised if CPU-intensive
PalmOS4 applications didn't run slower on the PalmOS5
machines than on a 66MHz Palm.

Palm's use of phrases like "does not emulate" etc. seem deliberately misleading and confusing. PalmOS5 emulates a PalmOS4 environment, and there will be no native ARM applications for PalmOS5. It's basically an incomplete kludge, not a new OS.

RE: Reality check
I.M. Anonymous @ 6/19/2002 4:48:46 PM #
Don't ignore that all the APIs are native ARM code. A good chunk of what apps do is call APIs.

Applications that do a lot of work can use native ARM code, which OS 5 does allow.

Somewhat misleading?

Scott R @ 6/18/2002 10:53:54 AM #
Ed, perhaps I've misinterpreted the information coming out from the Toshiba PPC reviews, but I think you've misstated the problems (correct me if I'm wrong). Here's how I understand it:

The PPC OS is not optimized for XScale. As such, the newest XScale PPC (by Toshiba) doesn't take full advantage of the speed (and power?) advantages of the CPU. It is my understanding that there _is_ a speed increase (at most tasks) but that it is more of an incremental increase, rather than being twice as fast, for instance.

That aside, I think it's also worth bringing up again that the Palm OS is, by and large, more efficient than the PPC OS given it's measly CPU power currently. I'm hoping/expecting that a 400MHz XScale Palm will really fly.

Scott

RE: Somewhat misleading?
Ed @ 6/18/2002 10:59:07 AM #
I'm basing this on what I've read about the issue. According to the silicon.com article, "Microsoft's current PDA operating system, Pocket PC 2002, is based on a Windows CE 3.0 kernel, which is optimised for a class of processors which includes StrongARM, but not Xscale. So the rumours that the first Xscale products don't run any faster than StrongARM are almost certainly true."

Here's a quote from a review of the Toshiba e740 on PocketNow.com:
"Now I know this is upsetting news, but when it comes to doing common tasks and using the Pocket PC OS, you will not notice any performance improvement between the Toshiba e740, and say, an iPAQ. That's simply because the Pocket PC 2002 OS hasn't been optimized for the XScale CPU speeds at this point in time. "

---
News Editor

RE: Somewhat misleading?
mtg101 @ 6/18/2002 11:06:05 AM #
I think the point is mroe that the increase in speed comes at the expense of extra power usage.

That comes on top of the fact that the PalmOS is much more efficient in terms of power than the PPC OS even without specific Xscale support.

---
russ@russb.fsnet.co.uk

RE: Somewhat misleading?
Scott R @ 6/18/2002 12:12:29 PM #
Well, I'd be wary of trusting a "magazine"-type source like Salon. As far as PocketNow's review, if you look at the detailed speed chart it's a mixed bag. The Toshiba e740 (XScale) nearly doubled some speed numbers compared to the competition, for a couple of tests. In others, it fell way behind (mostly graphic-related). I don't know if these devices uses custom graphics chips or not, so it's hard to say whether the speed problems there were attributable to the XScale or the graphics chip. Battery life was definitely atrocious, but you have to take into account the fact that the reviewed model had built-in Wi-Fi. I'd like to see some more scientific reviews done before I could say whether or not the XScale is bringing a significant improvement, small improvement, no improvement, or worse performance to some/most tasks and battery life.

Again, I'm not disagreeing with you that the PPC OS is not optimized for XScale. Even the MS MVPs have admitted this. I'm just saying that there's a big difference between "not being optimized for" and "not offering any significant performance or battery life improvements."

Scott

RE: Somewhat misleading?
I.M. Anonymous @ 6/18/2002 1:52:38 PM #
Also if the manufacturers are using these chips how long will it be before MS release some sort of update or patch to the PPC2002 OS?. If there were no performance OR battery life improvments whey would Toshiba and the others make the switch. non-ARM XScale processors must be pretty cheap now.
I also think that, as with Windows, graphics performance counts for a lot in the users perceived experience. ie, the machine will seem much more responsive with a fast graphics chip.
RE: Somewhat misleading?
I.M. Anonymous @ 6/18/2002 4:35:30 PM #
"Palm OS 5 Can Take Full Advantage of Intel's XScale" is a pretty blanket statement to make considering these XScale Palms will not exist for another 6-9 months. Who knows if that is true or not until they are out?

RE: Somewhat misleading?
I.M. Anonymous @ 6/18/2002 9:13:37 PM #
>>"Palm OS 5 Can Take Full Advantage of Intel's XScale" is a pretty blanket statement to make considering these XScale Palms will not exist for another 6-9 months.

But it is true. PocketPC 2002 (and the next version) won't take advantage of x-scale because the OS doesn't. Palm has the concept of the Hardware Abstraction Layer which leaves it to manufacturers (in this case Intel) to write the HAL layer. Intel has said they will have this ready. SO: PPC does not have any ability to optimize for xscale while Palm OS does - provided Intel does it's job. I hardly think Intel would mess this up in such a way as to make their implementation of HAL cause Palm OS to be less optimized for xscale then PPC - which provides no mechanisim what-so-ever to optimize for these processors.

RE: Somewhat misleading?
Palm_Otaku @ 6/20/2002 8:51:20 PM #
Well, I'd be wary of trusting a "magazine"-type source like Salon

Scott, the source was silicon.com, an engineering-based site: check out the article as originally posted.

Ed, this part is not that accurate:

I.M. Anonymous @ 6/18/2002 2:10:17 PM #
Ed, this part:

"Pocket PC 2002 isn't and therefore will be limited to 200 MHz."

Pocket PCs can run at 400MHz. Just that it might not be optimized at the moment. Look at this photo from OurPocketViews.net: http://www.ourpocketviews.net/articles/captures/genioxscale_uc.jpg

This is from the Japanese e550G. The control applet is standard on PocketPCs that can run at 400MHz.

Or perhaps I misunderstood you?

RE: Ed, this part is not that accurate:
Ed @ 6/18/2002 2:36:52 PM #
While there may be PPC devices running on 400 MHz processors, they don't actually run at 400 MHz. As you said, PPC 2002 isn't optimized for the new processors and therefore can not run at the processor's maximum speed.

According to reports from people who have tried out these supposedly 400 MHz handhelds, they are no faster than 206 MHz StrongARM ones.

It is possible for Microsoft to release a new version of Pocket PC that is optimized for the XScale chips which will allow them to run at 400 MHz but, according to the silicon.com article, this isn't going to happen until 2004.

---
News Editor

RE: Ed, this part is not that accurate:
I.M. Anonymous @ 6/18/2002 3:51:03 PM #
Ahh.. I see that the emphasis here is on the OS and applications, not the actual hardware itself.

Thanks for clearing that up Ed.

RE: Ed, this part is not that accurate:
I.M. Anonymous @ 6/18/2002 4:15:46 PM #
Toshiba e740 benchmark. I am sure PIC has their own official "source" but generally we use VObenchmarking in PPC, instead of gesting all over the place with handwaving argument.

http://www.sprinklerhead.com/toshe740b.htm

RE: Ed, this part is not that accurate:
I.M. Anonymous @ 6/18/2002 4:32:17 PM #
It benchmarks higher but here's what PocketNow said:
"When it comes to doing common tasks and using the Pocket PC OS, you will not notice any performance improvement between the Toshiba e740, and say, an iPAQ. That's simply because the Pocket PC 2002 OS hasn't been optimized for the XScale CPU speeds at this point in time. "

www.pocketnow.com/reviews/toshibae740/toshibae740b.htm

What use is a high benchmark score if it is no faster than a 200 MHz iPaq in the real world?

Good luck trying to convince us PocketNow is a pro-Palm site. Their description of themselves is, "pocketnow.com's vision is to educate, enlighten, and enlist new users to the world of Windows CE devices. "

RE: Ed, this part is not that accurate:
I.M. Anonymous @ 6/18/2002 4:39:09 PM #
> While there may be PPC devices running on 400 MHz processors, they don't actually run at 400 MHz. As you said, PPC 2002 isn't optimized for the new processors and therefore can not run at the processor's maximum speed.

PocketPC DOES run at 400MHz. It benchmarks reasonably higher than the 206 SA's. Not being optimised is another issue, but it is 400MHz regardless.

RE: Ed, this part is not that accurate:
I.M. Anonymous @ 6/18/2002 5:04:07 PM #
You say it runs faster but the only way you can tell is to run a benchmark program? Do you have any idea how ridiculous that sounds?

Thanks for the laugh.

RE: Ed, this part is not that accurate:
I.M. Anonymous @ 6/19/2002 3:24:22 AM #
Look at the video at PocketNow and SEE for yourself that the Toshiba is FASTER doing certain things while actually SLOWER doing others. What does this mean? The process DOES speed things up CONSIDERABLY on certain things, and may slow things down doing other things depending on how the programs are coded for the processor.

Silicon.COM is WRONG

I.M. Anonymous @ 6/18/2002 2:40:09 PM #
As is pointed out above, PPC won't take full advantage of XScale. However, the silicon.com article is incorrect when it states:

"Having only recently ported to StrongARM, Microsoft's main rival Palm is not likely to release a genuinely Xscale-optimised operating system any time soon either."

I have written Ben at silicon, but he seems to want to stand by the eroneous point. Palm OS 5 is not 'optimized for StrongARM' - because of the HAL layer, it is equally optimized for all of the current class of ARM processors - including x-scale, because Palm did the smart thing and didn't tie optimization of a particular ARM processor to the OS (unlike MS). So, when the cpu manufacturer writes the HAL layer (all OS 5 compliant cpus have to have this) - then it is optimized. The silicon.com artical makes it sound like Palm is 'just now getting to StrongARM' and will have to release a newer optimized OS to work with xscale. Not only is this completely false - i haven't even heard ANYTHING from ANY manufacturer claiming that they intend to produce an OS 5 device that uses Intel's StrongARM (though, with appropriate HAL layer it could). I suggest that everybody write to Ben and encourage hime to correct this mistake: bking@silicon.com

Silicon.com is not the only place perpetuating this myth (based - innocently enough - on a lack of understanding), i've seen articles elswhere that say the same thing (including one on Wired).

RE: Silicon.COM is WRONG
I.M. Anonymous @ 6/18/2002 11:53:14 PM #
the rex and phillips nino were kick #@! weren't they, oh i wish they were still around.

Palm OS 5

I.M. Anonymous @ 6/18/2002 11:51:30 PM #
It's amazing the number of people who post
replies here who haven't a clue about processors
and compilers. It's very simple, you have a
compiler for say a target CPU, let's call
it an ARM compiler. The target CPU is an
ARM9 family, as in V4T instructions. That ARM
compiler is optimized for those cores and those
instructions. Now, let's say you have a
new CPU that has new instructions and is not
based on the ARM9 family but has many changes
to its design. Using the same compiler without
optimizations on the new processor == slower code.
RE: Palm OS 5
I.M. Anonymous @ 6/19/2002 12:08:22 AM #
OK - but how does the Palms 'Hardware Abstraction Layer' fit into this?

marketing...

I.M. Anonymous @ 6/19/2002 12:54:34 AM #
This reads like an Intel marketing infomercial....
I assume they are spinning damage control after the
rumors of poor performance with WinCE compared to
existing StrongARM.

Be ensured of Quality Products

I.M. Anonymous @ 6/19/2002 1:22:12 AM #
Be assured that Microsoft is not a company that rushes out poor quality Software and hardware to seemingly show that it can optimise the X-Scale procesor. Both Intel and Microsoft have good and long working relationship and hence the next generation of PocketPC will definitely surpass the rest of the underachieving competitors.
RE: Be ensured of Quality Products
I.M. Anonymous @ 6/19/2002 2:57:39 AM #
huh? what the @!$#? You watch too much tv - 'MS is this big omnipotent deity of a flawless organization that never makes mistakes' - please. the first couple of versions of CE were a total flop (BTW - Palm OS had success 1st try), then PPC is designed (by morons apparently) with the notion that apps should never be closed, then they backtrack and add 'Smart Minimize' so that users don't know whether an app is closed or not. Clever bunch. And for the record - MS not optimizing for xscale is not likey to be the result of an Intel/MS 'love-in': Intel has been ticked at MS in the past for failing to properly utilize chipset features in Windows. Be assured - assured that somebody always has something useless to add to the discussion.
RE: Be ensured of Quality Products
I.M. Anonymous @ 6/19/2002 2:29:33 PM #
When was the last time you 'closed' an application on a Palm? Never that's when, because it's unnecessary. Same for PocketPC.
The only reason for the pseudo-close button on PPC is that Windows users are used to it and therefore feel better when a program has been 'closed'.
This is a matter of user experience rather than function.
RE: Be ensured of Quality Products
I.M. Anonymous @ 6/19/2002 2:36:12 PM #
>>>When was the last time you 'closed' an application on a Palm? Never that's when, because it's unnecessary. Same for PocketPC.
The only reason for the pseudo-close button on PPC is that Windows users are used to it and therefore feel better when a program has been 'closed'.
This is a matter of user experience rather than function.


Wrong. When was the last time i 'closed' an application in Palm OS? - the last time i tapped the launcher button or any of the other hardware/silkscreen buttons, that when! As soon as you leave yr app - it's closed. On PPC IT IS NOT - unless the app was written to do so - which MS actually discouraged before doing an about face and changing their minds (because many developers were ignoring them anyway). Now when you hit the close button, the app either closes OR minimizes - the user doesn't know or get to decide which. Ever noticed yr PPC suddenly getting too slow? Tap start>settings>system>memory>running programs - what do you see? a list of apps currently running THAT DIDN"T AUTOMATICALLY CLOSE WHEN THE USER EXITED THE APP. LAME. If tapping 5 times to get to this screen seems a little bit much - get used to it - because for a lot of apps, this is the only way to close it down (BTW: you'll need to then select the app and tap 'stop running program' - thats another couple of taps).

RE: Be ensured of Quality Products
I.M. Anonymous @ 6/19/2002 5:06:59 PM #
>"Be assured that Microsoft is not a company that rushes out poor quality Software and hardware..."<

Give me a break! Nearly every piece of software MS has put to market has had serious bugs, been notoriously ineffecient, and lacking in security. And let's not forget prone to crashing (PPC does this too, I've seen it happen). Microsoft has always rushed out poor quality software. They are the company who really brought the "release and patch" scenario to the mainstream.

RE: Be ensured of Quality Products
I.M. Anonymous @ 6/20/2002 12:27:58 AM #
Quality Microsoft products? LOL biggest joke I've heard all year. Maybe the viruses that run under Windows are of high quality but not much else.
RE: Be ensured of Quality Products
I.M. Anonymous @ 6/20/2002 7:35:48 AM #
It is because those Techno geeks who thinks they have a bit of knowledge and always spend so much time to hack Microsoft that makes Microsoft products look lousy. If Microsoft products are no good ... why is the world using them? Why people choose to use Windows over Linux, IE over Netscape ... Doesn't all these prove that Microsoft is a cut above the rest? See all the competitors who produced inferior products in the end had to seek Microsoft for a bailout and rebranding like Solomon, Frontpage ... etc etc. Microsoft is the best as they donate many old computers to the schools with genuine old Microsoft softwares. So please open your eyes to know who is the real contributor to the US economy with good and efficient products. Even some of the states also support Microsoft by not contesting in the Antitrust Case.
RE: Be ensured of Quality Products
Palm_Otaku @ 6/20/2002 8:38:27 PM #
Heh heh heh - don't you people recognize satire when you see it!??!!

highly misleading

I.M. Anonymous @ 6/19/2002 4:30:17 PM #
Right now, it looks like PalmOS5 will run no applications natively--everything will be interpreted 68k code (if you must, you can write a few loops in ARM assembly language).

To claim that PalmOS5 will take full advantage of XScale is ridiculously misleading. Yes, you can say that all native ARM applications on PalmOS5 will take advantage of XScale--all zero of them (or one, if you count the emulator).

If speed an battery life are your thing, you are almost certainly going to get better performance running ARM code that hasn't been optimized for XScale on the XScale processor than running PalmOS applications using a 68k emulator on the Palm handhelds.

I like my Palm, but the company really needs to start deliverinng new technology, not lots of hot air.

RE: highly misleading
I.M. Anonymous @ 6/19/2002 4:41:03 PM #
OK, for the benefit of yet another PPC supporter who didn't notice that 5 or 6 people have tried to make this same arguement in previous comments and been shot down, let's go over this again.

I think mtg101 said it best:
The OS itself runs native ARM code. So whenever you make a call to the OS - which is most of what most apps do - you're running at full ARM speed.
It's only when you do intensive work within your code (the sort of thing that means you need a progress bar on PalmOS 4.1) where you'll have problems with the emulation speed. However - in PalmOS 5 you can use ARM code specifically for these intensive operations.

Intel has created a HAL that works between the OS and the XScale processor which allows the OS and APIs to take full advantage of the XScale's speed and battery saving capabilities.

RE: highly misleading
I.M. Anonymous @ 6/19/2002 4:47:40 PM #
You are just saying the same thing I am saying: all applications on PalmOS5 will run interpreted (with some very limited ability to use ARM assembly in a few places). Only the operating system will run in native code.

You seem to think that this is almost as good as an all native environment. But that's ridiculous. If you do execution profiling on real applications, you'll see that they do not spend a lot of time in the operating system. By being based on interpreted 68k code, PalmOS5 applications will run at the speed of interpreted 68k code, which may well end up being slower than fast Palm handhelds are today.

Trying to turn this into an "us vs. them" kind of argument is silly. I don't have a PPC and I don't like PPC. I want Palm to release a completely native ARM environment ASAP. As far as I'm concerned, PalmOS5 is pretty useless.

RE: highly misleading
I.M. Anonymous @ 6/19/2002 4:52:55 PM #
>>The OS itself runs native ARM code. So whenever you make a call to the OS - which is most of what most apps do - you're running at full ARM speed.
It's only when you do intensive work within your code (the sort of thing that means you need a progress bar on PalmOS 4.1) where you'll have problems with the emulation speed. However - in PalmOS 5 you can use ARM code specifically for these intensive operations.

Yes - I was about to disagree with this, but then I re-read Palms PalmSource2002 presentation pdf ( http://210.153.100.248/slides-2002/100.pdf ). You are completely right. The more that code relies on the OS API, the faster the performance gains. The more an app relies on internal application level algorythims, the slower it will be - until the next OS when all apps will be native ARM. For a Palm OS that completely supports native ARM apps, we will have to wait for the next OS - supposedly coming soon as Palm plans a 1-2 punch with an agressive release schedule (we'll have to wait and see about this point of course) - which is still better than the PPC situation because the next PPC OS (the one based on ce.net) will not be optimized for ARM processors other then StrongARM - and thats according to MS itself. Even the PPC OS set to be released in 2004 is not going to be - according to MS - 'completely' optimized for xscale.

RE: highly misleading
I.M. Anonymous @ 6/19/2002 4:53:11 PM #
Oh, one more thing: dropping into ARM assembly language for PalmOS5 is a non-starter for almost all developers: it's way too much effort for something that's a temporary band-aid, that's not going to port anywhere else, and that's going to change again in a few months anyway. Why waste any time on that?
RE: highly misleading
I.M. Anonymous @ 6/20/2002 7:44:24 AM #
Sorry but Ed et al still have it wrong. The HAL
is in "C" code for the most part. PalmOS5 devel is
using ARM ltd. tools. At this time ARM tools do
not optimize XScale very well at all. You need a
new compiler. That is why PocketPC StrongARM will
outperform PocketPC XScale. That is why PalmOS5 XScale
is no better than any other flavor of ARM.

Again, it's the compiler NOT the "C" code in the
HAL. There is very little assembly in the HAL.

thinking beyond xscale

I.M. Anonymous @ 6/20/2002 12:31:23 AM #
not only will PalmOS be optimized for xscale but also for TI, motorola, mediaq and anyone else who comes with an interesting ARM based design. For example, a future MediaQ architecture "Katana" ties ARM core, 2D graphics acceleration, MPEG-4 engine, Java acceleration, and embedded SRAM with a special memory-interface unit. That's a killer product that PalmOS will be ready for while PPC is still worried about xscale.

another limitation of PalmOS 5

I.M. Anonymous @ 6/20/2002 4:29:03 AM #
Issues of speed aside, PalmOS 5 has the same limitations on memory and pointers that PalmOS 4 has, because it has the same operating system interface and because applications use the same old 68000 instructions.

Palm really needs to speed up development and come out with a version of PalmOS that allows full, native ARM applications and that allows protected mode 32 bit applications to be written.

I don't see why they even bother releasing PalmOS 5. I think doing so may do more harm than good. And saying that it "takes full advantage of XScale" is a mockery.

Top

Account

Register Register | Login Log in
user:
pass: