AMD to enter PDA chip market [Archive]

Ed Hansberry

04-10-2002, 12:05 AM

<a href="http://www.theregister.co.uk/content/3/24758.html">http://www.theregister.co.uk/content/3/24758.html</a> Ok, we've received eleventy-two emails on this at last count, so we figured we'd better post on this. "AMD today launched a new chip, the first fruits of its recent takeover of Alchemy Semiconductor, and its first foray into the low-power PDA and information appliance CPU market. This is the Alchemy AU 1100: it runs on a MIPS32 instruction set." MIPS? So, what PDA uses MIPS? If AMD wants to gain significant marketshare inthe PDA market like they have on the desktop, they would have to do it the same way, emulating the king of PDA chips, which is the ARM core that Pocket PC's and Palm's OS5 will be using. MIPS is being almost completely abandoned by PDA makers, save the Casio BE-300, and Casio has an ownership stake in the company they buy their chips from, so they won't be using AMD either. I suspect their market will be more vertical, or proprietary like the BE-300. And the marketshare of the BE-300 is so small, I am not sure that is a business model I'd chase after. If AMD wants to go after consumers and corporate IT departments, they are going to have to support Palm or Pocket PC, and that means ARM.

Sslixtis

04-10-2002, 12:28 AM

I always knew you were a Genius! That is exactly what I have been saying since they announced their new chips. Hey maybe Hitachi will come out with anew SuperCool SH processor! AMD may how to use this new chip in the Smart Phone wars that are looming on the horizon, but I predict those will be mostly ARM based as well. 8O

Will T Smith

04-10-2002, 03:52 AM

AMD MUST use the x86 instruction set to compete in the PC computer space.

Likewise, in order to competitive in the Handheld market, they must implement an ARM instruction set. Two years ago, they may have made a go at MIPS. However, the market has spoken and consolidated on ARM.

I really want to see AMD in the handheld space. A multi-provider market for ARM CPUs will improve the handheld space substantially. I believe that pushing MIPS at this point will only stratify the market place an burn through valuable AMD cash.

AMD please don't go against the grain. Switch to ARM cores.

Dave Conger

04-10-2002, 03:54 AM

Isn't it this chip is just a "left over" from Alchemy Semiconductor. Granted it is kind of dumb to release a product that you have to know would probably not be a large sell, I just wonder if AMD is just trying to hold up a little flag that says "we are doing mobile stuff." I would think if they made an ARM based product that Palm would probably throw it in their device. Supposedlytheir next gen is TI based, but I would think they might look for a more "main stream" processor name to try to hook some people on that. Microsoft is forever Intel, so I don't think that AMD really has that much room to grown in the Pocket PC market.

Kirk Stephens

04-10-2002, 04:13 AM

eleventy-two? What number is that :P

Aceze

04-10-2002, 04:37 AM

Just to play a little devils advocate here - but why not MIPS? CE .NET is not ARM specific - in fact the ONLY thing that is ARM specific is PPC2002. How many PPC2002 devices are really out there - a pittance right now. It looks like they may targe .NET with this one (they even say that on their website), so I see no reason that they couldnt make the instructions backward compatible to enable all previous MIPS compiled software run compatible.

And as for Palm - backwards compatiblity will only be apps run on an emulator (so it really makes NO difference whether it's ARM or MIPS).

I, for one, am not going to believe wholeheartedly that there can only be one player (ARM chips in this case) in the PDA market (in fact I dont want one).

And I think that MIPS was "abandoned" by Microsoft - not PDA makers.

Aceze

Ed Hansberry

04-10-2002, 04:48 AM

Just to play a little devils advocate here - but why not MIPS? CE .NET is not ARM specific - in fact the ONLY thing that is ARM specific is PPC2002. How many PPC2002 devices are really out there - a pittance right now.
Over 2M is not a pittance. ARM isn't Intel either. ARM is just a standard. If HP switched from SHx to ARM and Casio switched from MIPS to ARM for Pocket PC's, what makes you think AMD, who inherited this chip, could change the tide? Competition is good, but too many standards is not. I think it is fantastic HP, Casio, Compaq, etc. can buy ARM chips from a variety of vendors without making things more difficult on the Pocket PC team, the PPC dev tools, the hardware makers with drivers, etc.

Having 2M devices or 10M devices is really too small to justify multiple platforms just for the sake of doing it, IMHO.

TQBrady

04-10-2002, 05:08 AM

MIPS brings a lot to the table. a 400 MHz processor that consumes 250 mW is nothing to sneeze at. And like Aceze said, there's no reason to believe CE.Net wouldn't work on these chips. The other capabilites STANDARD on this chip are better than current ARM standards as well. And who says AMD is even looking to do business with Microsoft concerning these chips(altought that is most likely the case). I'm far from business savvy, but I wonder if it might be smart for AMD to team up with another OS player to offset the Intel/Microsoft alliance. Perhaps a Linux distro. Perhaps a UNIX distro. Perhaps Apple. There are many, many, rumors about on that last one, culminating in the "news" that June will see OSX running on AMD desktops. Wouldn't THAT be something! If that's the case, Apple would be a formiddable handheld opponent. They're style, intuitiveness, and understanding of sound hardware/software design would lead to a really solid handheld. Though they got a lot of bad press for early handwriting recognition(give them a break . . . they were the FIRST!!!), the Newtons were amazingly capable of recognizing even bad handwriting in late revisions. Better, many say, than the best current offerings for Pocket PC and Palm handhelds.
And let's not forget Symbian. That's a bulletproof platform that AMD could employ.

(like the devil needed another advocate 3:)

charlie

04-10-2002, 05:10 AM

Just to play a little devils advocate here - but why not MIPS? CE .NET is not ARM specific - in fact the ONLY thing that is ARM specific is PPC2002. How many PPC2002 devices are really out there - a pittance right now.
Over 2M is not a pittance. ARM isn't Intel either. ARM is just a standard. If HP switched from SHx to ARM and Casio switched from MIPS to ARM for

I'm sure you know Ed, but for everyone else, ARM is a company, not just a standard (the ARM ISA standard). They designed the ARM instruction set and they've designed a lot of synthesizable CPU cores for it. If AMD wanted to design and sell their own ARM chips, they would have to license the ISA from ARM. If AMD wanted to really get serious, they might want to just license a core from ARM, add on some custom features to set it apart from the StrongARM and start making them, I don't imagine it would take a lot more than 6 months to have chips out if they started with a complete core.

Aceze

04-10-2002, 05:15 AM

Over 2M is not a pittance. ARM isn't Intel either. ARM is just a standard. If HP switched from SHx to ARM and Casio switched from MIPS to ARM for Pocket PC's, what makes you think AMD, who inherited this chip, could change the tide?

Well, this is not quite right here - I dont think HP and Casio's switch was particularly voluntary here. Microsoft decreed, and HP and Casio HAD to move to ARM to preserve future brand mindspace with the PPC2002 devices. But .NET is a whole new ballgame - it is not going to be platform dependent just like CE3 wasnt. The whole reason CPU uniformity right now is important is for the next platform to have backward compatible software (dev tools will be upgraded anyway, so how much of a difference is it really going to be when you're moving to a MORE capable platform anyway - assuming no huge gaffes like the old MIPS compiler problems). I argue that the MIPS platform does have a backwards compatible market there already. There is no low level programming that needs to be redone to compile software to the different chip.

Oh and btw, ARM is not Intel, but the version that PPCs are using is. StrongARM is an Intel product - as are the Xscale CPUs.

Competition is good, but too many standards is not. I think it is fantastic HP, Casio, Compaq, etc. can buy ARM chips from a variety of vendors without making things more difficult on the Pocket PC team, the PPC dev tools, the hardware makers with drivers, etc.

I dont think 2 CPUs are "too many standards". And as I said before, StrongARM chips are Intel only - so they cannot buy them from any old vendor. Even so, what really, does this change? I dont believe dev tools are a big deal - they were a big deal in the past because of problems never fixed that caused big disparities in performance. On an even keel - why should CPU matter? Compile for two code branches, make your installer aware of it, and off you go. In one fell swoop, Microsoft also makes sure all their eggs arent all in one basket as well (e.g. what happens if vendors find Intel chips too expensive, etc, etc).

Having 2M devices or 10M devices is really too small to justify multiple platforms just for the sake of doing it, IMHO.

That's a bit of a copout. It's not for the sake of doing it. If that were the case, why does MS come out with all the various builds of CE that allow it to run on MIPS, SH, x86, Motorola PPC, etc? You cant argue there - the more platforms, the larger the distribution - obviously, in a market where consumers are directly involved, you dont want excessive confusion, but I do think that much of the problems stemmed from poor vision concerning the various original PPC2000 initial specs (too different in hardware speeds, dev tools were too slanted to the SA chips performance wise).

If you want to argue numbers like that, you could argue that we should never innovate and should be happy with all our 16mhz Dragonballs!!

Aceze

Aceze

04-10-2002, 05:21 AM

Oh yes, and as TQBrady mentioned, there is also a large possibility that AMD may move the way Sharp has - toward the linux camp or even toward the Symbian camp with a player like Sony/Ericsson or Nokia.

They may even be (shock! horror!) targetting cross over PDA/Cell combo devices with the major Cell players. now wouldnt THAT be something...

AMD CPU + ATI video GPU... that's sweet!

Aceze

charlie

04-10-2002, 05:25 AM

AMD CPU + ATI video GPU... that's sweet!

Aceze

bah ATI sucks! with a little bit of uranium, there would be no problem powering a pocket GEForce4 (2go?)

chxxrlie

Aceze

04-10-2002, 05:34 AM

bah ATI sucks! with a little bit of uranium, there would be no problem powering a pocket GEForce4 (2go?)

No thanks - I'd rather not pay 200% more for a GF chip and get performance that was marginally better. Add to the fact, Nvidia cant make a mobile chipset if their lives depended on it - have you read any reviews of the GF2go lately? Bad power usage, and major inconsistencies - thanks but my PPC crashes enough already. ATI may not be the fastest, but the price/performance ratio on their solutions are MUCH better than Nvidia - not to mention higher visual quality and features.

BTW, the AMD/Alchemy chips got a good writeup at infosync.no as well:
http://www.infosync.no/show.php?id=1655

Better clockspeed, more features right off the bat (integrated LCD controller, 10/100 Ethernet, 2 SD controllers, USB host/device support - all right on the chip), and battery performance as good as the XScale. Exciting times!

Aceze

Chris Edwards

04-10-2002, 06:12 AM

Aceze

04-10-2002, 06:39 AM

TechJosh

04-10-2002, 06:47 AM

Many people brought up the point that AMD ought to have released an ARM compatible processor or emulated ARM instructions so that they could compete with Intel's StrongARM and Xscale processors. They also cite AMD's strategy and success in the desktop world as an example. However, this really isnt as big of an issue as it seems. In the desktop market anybody could pop open their pc and replace their Pentium 266 with a k6-266. Pocket PC's are not designed to allow this. It will not be a grass-roots sort of processor rebellion.

AMD has, instead, focused on creating a powerful processor that is based on a completely different instruction set. Is this a problem for them? No. When a company decides to produce a PDA they evaluate their options and find a harmony between price, quality, and development time. With PDA's each device is so proprietary that if AMD's processors are fast, more integrated, and easy to develop with then companies will use them.

Contrary to popular belief, software support really isn't an issue. as has already been mentioned, Microsoft's Windows CE.NET will support MIPS processors, and their Common Language Runtime will allow the same programs to run on any WinCE.NET device. As for older software, it is really easy to recompile a program to work on another architecture as long as there is a compiler availiable for it. You can even use the same source code and just use compiler instructions for processor specific instructions.

Look at HP... They moved the Jornada series from SH3 to StrongARM... why not move to MIPS? (if it survives the merger)

After saying all that I'd like to say that I dont know how well this processor will do for AMD. I have no useful information to base a judgment on one way or the other. I do know that competition in itself is a good thing but I doubt that it will decrease the overal pricing of the PDA's that much. These babies cost less than $30 a pop in large enough lots.

Marc Zimmermann

04-10-2002, 08:08 AM

This is an embedded processor and this market is a lot bigger than only the Pocket PC part... they may not even be targeting this device class. Besides, if several OEMs are asking for support of this CPU, do you think that Microsoft is going to resist the pressure?

Ed Hansberry

04-10-2002, 12:23 PM

But .NET is a whole new ballgame - it is not going to be platform dependent just like CE3 wasnt.

Just curious. If Windows CE 2.x and Windows CE 3.x supported multiple processors (SHx, MIPS, ARM and x86) and CE.NET does too, how is CE.NET a whole new ballgame? Sounds like it is just an upgrade.

And I know the PPC's now use the StrongARM chip from Intel, but that is not a requirement. In fact, doesn't the HP 928 use someone elses ARM chip?

Janak Parekh

04-10-2002, 04:10 PM

Just curious. If Windows CE 2.x and Windows CE 3.x supported multiple processors (SHx, MIPS, ARM and x86) and CE.NET does too, how is CE.NET a whole new ballgame? Sounds like it is just an upgrade.

I believe (suspect) that CE.NET intends to improve the seamless exchange between architectures (and platforms, for that matter). That's core to MS's .NET philosophy... whether it will be realized remains to be seen.

BTW, MIPS is not a terrible architecture from a programmer's standpoint. ARM has some nice features but it is still fundamentally pretty similar to MIPS in design philosophy.

It's worth mentioning that, btw, Palm will NOT be using StrongARM, but rather a different ARM core. Moreover, there's no reason to believe StrongARM will be the only choice 5-10 years from now, and if anything, I'd imagine better ARM and MIPS cores/instruction sets will be available to take care of the new advances in mobile processor technology.

The market is much bigger than PPC - AMD is clearly not aiming for it. Whether they achieve any form of dominance will not be answered overnight.

--bdj

Ed Hansberry

04-10-2002, 04:47 PM

Think of this this way. Windows CE is an embedded operating system that works on at least four processors. SHx, MIPS, ARM and x86. Not sure if PowerPC is supported or not, but doesn't matter.

Now, you are a device manufacturer. You are going to make portable wireless devices that will be rented out at NASCAR events. You will be able to keep up with everything going on on the track just like users at home can with all of the stuff scrolling on the screen. You'll know how far Joe is behind Carl, who is out of the race and why, etc. That is it. Nothing more, nothing less.

Now, you decide on Windows CE as your OS. You want a reflective screen so it can be seen outdoors, you want 802.11b, it needs to be rugged and at least 6 hrs of battery life.

Now, time for a processor. You want speed, battery efficiency and as low of a price as possible. Lets say you settle on an SH4 processor.

Fine and dandy. Now, why in the world would you even think about making a 2nd version of the product with another processor, say a MIPS? That is a different Mobo, die, possibily RAM, different hardware drivers, etc. Of course you wouldn't. Not unless the MIPS offered something very compelling, and then you would switch, not maintain two versions of the NASCAR Tracker.

So I have one question for you. Why do you think the Pocket PC team, just another customer of the CE group, just like NTD Inc. (NASCAR Tracker Devices - Nasdaq symbol CRSH), why do some of you think they would want to divert the money and resources to keep multiple versions of their software (the PPC UI is not an OS, it is software, just like Office is - it has its own shell and some API's of its own, but it is just software) running?

I know there is a love of AMD devices, but it isn't rational thought to just say "this new MIPS is cool. MS should do an AMD version of the PPC." Why? It makes no sense. It is just extra work. It has no speed difference over the X-Scale (ARM core) which has already been tested at 600MHz.

And if AMD developed this chip to be the best mobile processor absolutely blowing away anything the ARM group has on the drawing boards, I'd look for the PPC team to switch and drop ARM. Not maintain two versions. That just makes no sense.

Will T Smith

04-10-2002, 05:46 PM

Toms Hardware [http://www.tomshardware.com] has a review of the latest from NVidia -- GeForce 4 Go. It beats out ATI in most of the benchmarks, and NVidia has a good rep. of providing good drivers that are updated often (bringing performance increases).

Anyways, back to Pocket PC's...

That Nvidia rep is non-existent for the GF2Go chipset - which STILL suffers massive problems more than a 1.5 years after release, while the Radeon Mobility and Radeon Mobility 7500 came out practically flawless (hey go read about it at Toms Hardware and many other places). And as for their "rep" - I think it's pretty overblown. Looking at all the troubles the GF3 had to go through (6 months+ of lousy drivers), I cant ever find where the rose coloured glasses are coming from - their drivers are every much as buggy as ATI (who have made great strides in their desktop market).

BTW, ATI already has a succesor to the mobility 7500, the mobility 8700 which is in sampling. ATI is far ahead of nvidia at mobility solutions. Add to the fact that Nvidia's pricing structure sucks.

Aceze

Aceze,

I honestly think that your out to lunch.\

Starting with this post. GeoForce2 Go chipsets are no longer an issue. However, please remember that this chipset was their FIRST into the laptop market. Release one products always suffer from problems that experience fixes. GeoForce 4 has been out for month. Judge them based on this.

Nvidia has an EXTRAORDINARY ability to enter existing markets and rapidly gain widespread acceptance. At one point, everyone thought that 3dFx could not be beaten. Nvidia literally ATE 3dFx's lunch, breakfast, and dinnner ... then 3dFx itself. This is no small task.

Nvidia has the best driver support of ANY vendor out their. Their Detonator driver set is a model for utility, ease and stability. ATI is universally noted for having poor drivers AND poor driver support, infrequent updates, etc...

Regarding MIPS and ARM.

ARM is not a CPU fab. It is an instruction set. For compatibility and widespread utility, it is essential that a given set of devices all use a common instruction set. Otherwise binaries for each target must be individually compiled and tested for multiple CPU instruction sets.

AS a software developer, I can tell you that it's hard enough to get an application compiled and tested for ONE CPU. More than one doubles your efforts.

One processor enables small developers to create applications. It also allows medium and large developers to contain cost and speed to market time. As it was, the multiple CPU's were causing problems in the marketplace. CE apps were fewer partially due to the multiple CPU issue. Many apps would be compiled for all 3 but be tested on only one. Many simply were simply released on only on processor.

Compared with Palm and their standardized Dragonball 33Mhz processor they were at a disadvantage. I dare say that the StrongARM processor (and subsequent standardization upon) vaulted the PocketPC to the point where analysts are now predicting Palms doom.

This new MIPS processor may very well be cool. However, AMD will never profit by a MIPS based processor at this stage of the game. AMD has more CPU smarts than Intel. Mobile devices are the future of the PC marketplace and all the players KNOW it. A MIPS processor might be a nice "me too", however it does little to get AMD into the game.

Regarding MIPS based SmartPhones. They will all suffer from a lack of compatibility with PktPC 2k2 software which is 100% ARM. Try putting that on a brochure an selling it.

Will T Smith

04-10-2002, 05:56 PM

...

Contrary to popular belief, software support really isn't an issue. as has already been mentioned, Microsoft's Windows CE.NET will support MIPS processors, and their Common Language Runtime will allow the same programs to run on any WinCE.NET device. As for older software, it is really easy to recompile a program to work on another architecture as long as there is a compiler availiable for it. You can even use the same source code and just use compiler instructions for processor specific instructions.

Run time's do not fix everything. Look at Sun and Java. Cross platform compatibility is tough to implement and ALWAYS suffers some performance hit regardless of how small. In the case of a brand new run-time my guess is it will be fairly significant.

As for recompiling older softare, hehehe. Thats easy to say. However when you try this you end up with bugs at best, AND compiler errors at worst. Microsoft's compiler's have never been MAGIC. They have historically had problems running on just ONE platform. Try it before you pimp it.

Janak Parekh

04-10-2002, 06:50 PM

And if AMD developed this chip to be the best mobile processor absolutely blowing away anything the ARM group has on the drawing boards, I'd look for the PPC team to switch and drop ARM. Not maintain two versions. That just makes no sense.

I do agree with your post, but one point: it depends on CE.NET's lifetime. If it's going to last several generations of processors, they'll have to maintain several... ARM as an ISA (and its many implementations) is pretty good, but there's no guarantee a better competitor's not down the road.

--bdj

Janak Parekh

04-10-2002, 06:53 PM

Run time's do not fix everything. Look at Sun and Java. Cross platform compatibility is tough to implement and ALWAYS suffers some performance hit regardless of how small. In the case of a brand new run-time my guess is it will be fairly significant.

Absolutely true, but CLR is not Java. MS has made some concessions to speed up performance. Only time will tell how well it works. I do agree that the first implementations will need smoothing out, but MS is betting a lot on this strategy long-term.

BTW, Java has gotten much better. It's not really slow anymore. Memory overhead, OTOH, is another problem... and is why we have J2ME (which, btw, is another interesting player in the smartphone market).

--bdj

Aceze

04-10-2002, 08:07 PM

Damn, I've lost track of how many times this website has thrown my reply away and asked me to re-log back in (after logging in 5-10 minutes earlier). It's getting to the point where I'd rather not even reply anymore. sheesh.

Nvidia stuff deleted because I dont want to REWRITE everything I already just did...

GF2Go is still very problematic - have you tried one of these things? Nvidia has dropped the ball, and there's nothing in sight that indicates that they give a rats ass about it anymore. GF4Go - still not on the market, costs $100 more (in 10000 lots) a piece, and is pretty much equal (marginally better at some things, poorer at others) to the M7500/7800 (which are out). I dont want to get into a pissing contest about desktop chips (and I regret even starting this) because it's irrelevant here.

As well, ATI already has a chip coming out targetted at PDAs. Nvidia is still on the drawing board (as far as I know).

As for your stuff about how Nvidia's "extraordinary" abilities, take off the rose coloured glasses. 3dfx largely destroyed themselves (did not make product cycles, overextended and alienated allies by purchasing STB, overrelied on their shipping products with limited feature spec). Nvidia's drivers are no better or worse than ATIs at this point in time - your view of the situation is dated at best.

As for the SA situation on the CE platform - let me restate my point:
I think that what happened with the initial launch that soured everyone on multiple CPUs was largely due to bad tools (flawed compilers, non-existent tools at the beginning, etc) coupled with specifications that allowed wildly different (performance wise, eg. SHx vs MIPS/ARM) CPUs to be used on the platform - this made it difficult to justify spending time optimizing software to behave the same on the different platforms. Of course, testing will lengthen development when you have 2 (or more) CPU branches, but I think that for the future of this platform, you cannot lock yourself in to a single CPU mindset (look at Palm for God's sake!).

With GOOD dev tools (the kicker) and equivalent in perfomance CPUs, I see no reason why 2 CPU devices could not do well.

This new MIPS processor may very well be cool. However, AMD will never profit by a MIPS based processor at this stage of the game. AMD has more CPU smarts than Intel. Mobile devices are the future of the PC marketplace and all the players KNOW it. A MIPS processor might be a nice "me too", however it does little to get AMD into the game.

You make the mistake of thinking AMD is targetting PPC only here. What makes you think they're not interested in some other rival platform (Linux, Symbian, etc)? Not even that - the custom OS market is still alive and well (look at the japanese PDA market). That stage of the game right now is still pretty early - looking at the various capabilities of PDAs and Palm-tops (or rather, lack thereof).

Aceze

Aceze

04-10-2002, 08:11 PM

Ed Hansberry

04-10-2002, 08:41 PM

BTW Ed, when I said that CE.NET is a "whole new ballgame" - that's because I'm referencing it to the CE3.5 OS (PPC2002) that was only released for SA chips - thereby forcing everyone to standardize.

CE.NET is a "whole new ballgame" because it allows vendors to once again (like CE2 and CE3) to choose the best CPU that fits the bill (as well as other components).

No, this is simply wrong. Pocket PC 2002 devices use Windows CE 3.0. There is no such thing as CE 3.5 and I wish PocketNow would delete that asinine article that references it! He as much admitted he made the "3.5" number up.

Pocket PC, a software suite with a shell, sits on CE. The PPC software only works with ARM starting with the 2002 version.

CE.NET does nothing to change that. Zero, zilch, nada.

Let me explain it one more time.

Pocket PC 2000 was built on CE 3.0. CE 3.0 supports multiple processors. The PPC team built 3 versions of the PPC UI and apps- ARM, SHx and MIPS.

Pocket PC 2002 was built on CE 3.0. CE 3.0 supports multiple processors. The PPC team built 1 (ONE) version of the PPC UI and apps - ARM.

Pocket PC 200x will be built on CE 4.0. CE 4.0 supports multiple processors. The PPC team will very very likely build one version of the PPC UI and apps - ARM.

Same ball game. Just a new inning.

Ed Hansberry

04-10-2002, 08:46 PM

Damn, I've lost track of how many times this website has thrown my reply away and asked me to re-log back in (after logging in 5-10 minutes earlier). It's getting to the point where I'd rather not even reply anymore. sheesh.

I've never seen this, and I access this site from about 5 different PC's. What are your IE cookie/privacy settings?

Aceze

04-10-2002, 11:09 PM

No, this is simply wrong. Pocket PC 2002 devices use Windows CE 3.0. There is no such thing as CE 3.5 and I wish PocketNow would delete that asinine article that references it! He as much admitted he made the "3.5" number up.

Oh for pete's sake, lighten up. I've never read the PocketNow article, and I call it CE3.5 only as a jibe (as in, should have been a full version higher, but ended up only being a half). It's got nothing to do with any lack of perception that the PPC GUI sits on top of the same base OS (CE 3.0). It's just more convenient to call PPC the "OS", instead of constantly referring to it as the "GUI that sits on top of CE3" - sheesh.

Pocket PC, a software suite with a shell, sits on CE. The PPC software only works with ARM starting with the 2002 version.
...
Pocket PC 200x will be built on CE 4.0. CE 4.0 supports multiple processors. The PPC team will very very likely build one version of the PPC UI and apps - ARM.

Same ball game. Just a new inning.

You're taking an off the cuff remark ("whole new ball game") a tad bit seriously there, methinks.

I'd be dissapointed if the PPC team didnt reconsider the MIPS platform (well, that wouldnt make it the first time I was dissapointed in them!) - but it doesnt stop anyone from using the base CE.NET OS and rolling their own GUI for it (like the BE300). Wasnt the Hitachi Xscale unit running the base CE.NET some time ago?
Either way, I was merely playing devil's advocate to your statement that any new processor support would be a dumb move. Regardless of what I think, the PPC team is going to do whatever the PPC team want to do - but I thought it would be interesting to discuss merits that a second CPU branch (without all the inequalities and screwups that marred the process the first time around) might bring to the table - for both MS, and definitely the consumer.

Aceze

Aceze

04-10-2002, 11:22 PM

Damn, I've lost track of how many times this website has thrown my reply away and asked me to re-log back in (after logging in 5-10 minutes earlier). It's getting to the point where I'd rather not even reply anymore. sheesh.

I've never seen this, and I access this site from about 5 different PC's. What are your IE cookie/privacy settings?

Haha, it's ironic - I had just written up another message to reply to yours about this problem, and voila it did it again and swallowed up another of my messages.

For the record (again) Win2k, IE5.5SP2, "Medium" security.

Vbulletin boards (the only other board I go to very regularly) has never given me the problems this one has, and I've been one there for a much longer time. It's quite frustrating, as I frequently cut and paste my replies/posts to notepad before risking posting anything here. Once or twice, I forget, and invariably, I will get hit, and the damn session timeouts or expires (or ghod knows what) and it will dump my article again.

Aceze

Ed Hansberry

04-10-2002, 11:31 PM

For the record (again) Win2k, IE5.5SP2, "Medium" security.

Vbulletin boards (the only other board I go to very regularly) has never given me the problems this one has, and I've been one there for a much longer time. It's quite frustrating, as I frequently cut and paste my replies/posts to notepad before risking posting anything here. Once or twice, I forget, and invariably, I will get hit, and the damn session timeouts or expires (or ghod knows what) and it will dump my article again.

Hmmm... XP Pro here on Medium-High security with IE6. No problems. (if you can, you should really consider IE6. Same speed but I have found it more stable than IE5.5.)

Consider closing all browser sessions and whacking your pocketpcthoughts cookie. It could be corrupt.

Jason Dunn

04-11-2002, 02:38 AM

I, for one, am not going to believe wholeheartedly that there can only be one player (ARM chips in this case) in the PDA market

I'm 90% sure you're running a CPU that uses an X86 instruction set...sometimes standards are good. :-)

And I think that MIPS was "abandoned" by Microsoft - not PDA makers.

True, but not so much abandoned as MS simply picked the fastest horse and bet everything on it - it was taking them too much time to test code on all three CPU types. Streamlined testing = more time for development = better Pocket PCs for all of us. I'm all for it!

Jason Dunn

04-11-2002, 02:41 AM

Compile for two code branches, make your installer aware of it, and off you go.

It might seem simple to you, but words can't describe how confusing and chaotic this was for consumers. Few knew what kind of CPU they had, and then you had developers only releasing for one CPU type (because they didn't own all the hardware needed for testing).

One CPU type is the smartest move Microsoft could have made, and I'm very glad they did it.

Jason Dunn

04-11-2002, 02:44 AM

The market is much bigger than PPC - AMD is clearly not aiming for it. Whether they achieve any form of dominance will not be answered overnight.

I don't think anyone would disagree with you, but the reason we posted on this is that from a Pocket PC standpoint, this is 99% meaningless. We posted mostly to stop people from emailing us about it over and over. :wink:

Jason Dunn

04-11-2002, 02:47 AM

Jason Dunn

04-11-2002, 02:52 AM

CE.NET is a "whole new ballgame" because it allows vendors to once again (like CE2 and CE3) to choose the best CPU that fits the bill (as well as other components).

Once again? :wink: They can do that now with CE 3.0 - there are tablet devices, PDA devices, etc. that all use CE 3.0 and use MIPS, SH3, etc. CE.Net will be no different in that regard. But when it comes to the Pocket PC, it's ARM only, period. CE.Net will not change that in any way.

Aceze

04-12-2002, 04:40 AM

Hmmm... XP Pro here on Medium-High security with IE6. No problems. (if you can, you should really consider IE6. Same speed but I have found it more stable than IE5.5.)

Consider closing all browser sessions and whacking your pocketpcthoughts cookie. It could be corrupt.

Dont care much for XP, as I'm quite happy with 2K from both a speed and security POV, so that's not really an option (and I sure as heck aint upgrading an OS just for one website to run a little better! :) ), but I'm considering IE6. I'm a little iffy on IE in general, I think I might try paid Opera sometime...

As for the cookie thing - I do that periodically anyhow, and it's never fixed whatever the heck is timing out my session so quickly. Just a thought, I've seen various people bring this up (session timing out and killing posts) a bit more often that before - so there may be a problem here somehow that's affecting a few of us. *shrug* Where there's smoke...

Aceze

Aceze

04-12-2002, 05:05 AM

I, for one, am not going to believe wholeheartedly that there can only be one player (ARM chips in this case) in the PDA market

I'm 90% sure you're running a CPU that uses an X86 instruction set...sometimes standards are good. :-)

A-ha! While that's true for the short term - look at all the trouble Intel and AMD have to go through to deal with the fact that future solutions MUST have solutions that emulate or handle backward compatibility. The inertia that all this software holds is to great - to the point where future solutions must be crippled to deal with the emulation of this software (e.g. Itanium, and it's really bad handling of 32bit code).

With PDAs being non-upgradeable, I do think that because of the built in "throw away and upgrade" mindset (come on, how many of us will be using these toys for that much longer?!), I dont believe that we should stick to this mantra of "One Instruction Set" only - as I made an example of PALM just a short while ago. Palm stuck with the one instruction set policy with upgrades only in speed (albeit not very much upgrades!), and we view them as being a technologically "stagnant" company.

The Alchemy already offers tangible/significant benefits for the PDA market:
1) lower power requirements at fullspeed than the Xscale
2) way bigger integrated featureset - which leads to even more power savings as we wont have to waste power on complementary chipsets to deal with features that are not built in to the Xscale (such as Ethernet, 2 SD controllers, etc).
3) cheaper

... and we're throwing all these potential benefits for backwards compatibility for apps; a large proportion of which will be almost certainly rendered useless (in current form) due to the large jump in processor speed! The apps _need_ to be recompiled, why not recompile for the better chip? Why not have a choice?

And I think that MIPS was "abandoned" by Microsoft - not PDA makers.

True, but not so much abandoned as MS simply picked the fastest horse and bet everything on it - it was taking them too much time to test code on all three CPU types. Streamlined testing = more time for development = better Pocket PCs for all of us. I'm all for it!

See, that's what I was talking about. MS made a mistake (heck everyone does) when it specified so many different performing chips - it made it a headache for developers to build apps that would perform the same on all three chips - so much so that developers stopped trying (what burns me more, is that Microsoft, for whatever reason, did not bother to optimize the platforms to address some of these problems - eg MIPS compiler). However, we have a situation where two competing chipsets could be integrated that could start at more or less a "even" level - the experiment would not necessarily fail this time. I am afraid that the PPC will "stagnate" just like Palms did, by forcing a forward path down ARM avenue - without giving a thought to advancements being made in competing chipsets.
Of course, this assumes that the Alchemy chip will perform as well as the Xscale - but I hope you can see where I'm coming from.

Aceze

Aceze

04-12-2002, 05:19 AM

ATI is universally noted for having poor drivers AND poor driver support, infrequent updates, etc...

HERE HERE! My Radeon has horribly unstable drivers - works fine for 2D stuff, but the drivers frequently TANK XP when I play videos or DirectX games. NVidia's model for driver updates (all drivers work on all cards) is nothing short of REVOLUTIONARY in this industry of "screw the consumer when it comes to driver updates."

I will never buy another ATI graphics card...but dang, I wish the NVidia stuff wasn't so expensive!

Sorry to hear that about your Radeon, but are you sure that's not a factor of your system? I've bought lets see, 5 older Radeons (two original retail Radeons, and 3 Radeon 7200's - all the same chip), and every single one of them runs perfectly (on XP systems, 2k, and 98/98se). From direct recommendations, I've probably help sold about 20 Radeon 8500's with nary a complaint.

In fact, I almost exclusively buy ATI (and I'm a corporate buyer, and IT specialist at work), and they've been great. My Radeons, and Radeon 8500 has been ROCK solid for everything I've thrown at it (including 3d games, and higher end 3d cad design/renderings). Especially with the price point of these cards, I have nothing but good things to say about the Radeon line (Radeon and up). The older cards were not too good for much more than 2D - I will say - mainly because they were never really designed and marketed as such - they make great workstation cards though (that's with hundreds of older ATI cards I've rolled out).

The one experiment I tried with Nvidia chipsets was a horrendous nightmare - OEM video and QA quality was a crapshoot, price was murder, support was a joke (anyone for Nvidia dealer of the day?) as Nvidia does NO support whatsoever other than "if you havent tried our XYZ driver, wait for the next one". One thing you _know_ you're going to get is build quality and QA from ATI - something you cannot say for the "flavour of the week" Nvidia card. However, I dont want to really comment much on this, as this was more in a work environment, not in an enthusiasts.

As for your own Radeon problem, I know these cards run rock stable - maybe you should look up what may be the problem on your system? Try www.rage3d.com - it has possibly the best forum for ATI users on the net (just ignore some of the advocacy jackasses - why do these people gotta spoil it for everyone??). Much of the problem could be caused by existing Nvidia/3dfx drivers before the Radeon was installed, etc etc (such simple things).

Aceze

ps. BTW, all official Radeon drivers are unified (they work for all radeons and most older ATI cards - not the ancient ones though!).

pps. Damn, this site is weird - I just wrote the long reply and it went through fine. I try to edit and add a one line edit, it logs me out!! ?!?!?