ARM releases 1GHz specs for mobile processors

[Ed Hansberry]

<a href="http://story.news.yahoo.com/news?tmpl=story&ncid=73&e=3&cid=73&u=/zd/20020429/tc_zd/5107019">http://story.news.yahoo.com/news?tmpl=story&ncid=73&e=3&cid=73&u=/zd/20020429/tc_zd/5107019</a><br /><br />ARM Holdings, who's designs are currently used by Intel as the foundation for the StrongARM and X-Scale processors, released details today on the ARM11 processor. "In its initial incarnation using a 0.13-micron manufacturing process, ARM11 will deliver 350MHz to 500MHz in the worst cases, with typical performance of 533MHz to 750MHz, ARM said. Using a more advanced 0.10-micron process the architecture will be able to run faster than 1GHz."<br /><br /><img src="http://www.pocketpcthoughts.com/images/armlogo.jpg" /><br /><br />We could see something as early as 2003. They plan on releasing their first ARM11 CPU in late 2002 which will allow licensees and OEM's a few months to integrate these speed demons into handheld devices. <a href="http://www.arm.com/news/ARM112904">Additional information can be found here.</a> Thanks to Jonathan Blackwell for the heads up.

[TomB]

Hmmm... with the state-of-the-art we have this is starting to sound like a jet engine on a go-kart. In order for this to make sense, things like voice recognition, battery, storage and display technology have to develop at a much faster pace. If not, people with 32MHz Palms will be laughing themselves silly watching us tapping out messages on PPCs 1000 times faster then our soft-keyboarding skills! :)

[Will T Smith]

Hmmm... with the state-of-the-art we have this is starting to sound like a jet engine on a go-kart. In order for this to make sense, things like voice recognition, battery, storage and display technology have to develop at a much faster pace. If not, people with 32MHz Palms will be laughing themselves silly watching us tapping out messages on PPCs 1000 times faster then our soft-keyboarding skills! :)


Well said. I'll add one comment.

I wish ARM would start integrating graphics acceleration to their cores. Handhelds really need graphics (2-d & 3-d) acceleration. The fact is that regardless of how fast a CPU is, without hardware acceleration for multi-media, you can never implement real-time applications.

[ChrisD]

Without Microsoft's support for additional processors that the OEMs can choose from we are limited to the StrongARM and XScale. I have yet to hear of another ARM processor that is supported for the Pocket PC 2002.

[Ed Hansberry]

Without Microsoft's support for additional processors that the OEMs can choose from we are limited to the StrongARM and XScale. I have yet to hear of another ARM processor that is supported for the Pocket PC 2002.

Most seem to be using Intel's ARM processors, but if someone wanted to, they could plop in a TI ARM chip. In fact, the new Jornada 928 does just that - http://asia.cnet.com/reviews/handhelds/0,39001703,39001200p,00.htm
Under the hood of the 928 is a Texas Instruments OMAP 710 processor.

[charlie]

Without Microsoft's support for additional processors that the OEMs can choose from we are limited to the StrongARM and XScale. I have yet to hear of another ARM processor that is supported for the Pocket PC 2002.


why should manufactures be limited to ARM processors supported by microsoft? shouldn't any processor that fully conforms to the ARM ISA work? It seems to me that anyone could license an arm core or make their own and fab processosrs with integrated mp3 decoding and 3d acceleration and wireless (with external components to handle the RF) and it should work fine. hell, even I made my own structural arm core from scratch. Its not that hard.

charlie

[ChrisD]

Sure OEMs could try to do that but they would have to create all their own drivers for the hardware. Right now that's a very expensive approach vs using standard drivers and tweaking them to suit the design.

So which would you choose? Spending tens of thousands of dollars on a reference design for a new chip and drivers or using an existing one? There's got to be some financially compelling reason to do so.

[charlie]

So which would you choose? Spending tens of thousands of dollars on a reference design for a new chip and drivers or using an existing one? There's got to be some financially compelling reason to do so.


moving several million units seems like a pretty compelling reason to me. the first ones to really integrate will have lower manufacturing costs, more features and possibly better power consumption. If you can sell better and cheaper devices, you're going to dominate. You won't have to rewrite all of the drivers and the ones you do have to rewrite aren't that bad if you conform to standards. Whats a few tens of thousands of dollars to write some drivers compared to selling a million units?