Klaus, I would be interested to learn why Nokia chose the TI OMAP1710 processor. I assume it was because of its integral DSP support as well as support for multiple comms interfaces UTC has a new mobile phone/PDA manufactured by UTS marketed in the US as the 6700 by both Sprint and Verizon and running WM 5.0 PPC version. It has a much smaller screen size than the N770 but it is still very useable as a PDA. It seems quite fast and the graphics are excellent including video. In terms of comms interfaces it has three radios and it supports CDMA voice, EVDO broadband and 1xRTT data, 802.11b, bluetooth, infrared, and USB. I have been surprised by how responsive this device appears to be, especially since it has the more feature rich PPC version of Windows Mobile and not the stripped down mobile phone version. I do not know if it has an onboard DSP chip. Here are the specs: Specifications Operating System Windows Mobile 5.0 Processor Type Intel PXA 270 Processor Processor Speed 416 MHz Memory 64MB SDRAM, 128MB ROM Display Type 2.8" Color TFT Touch Screen Display Resolution 240 x 320 Dimensions 4.25" x 2.3" x 1" Weight 6.07 oz. Battery , Type: Removable 1350 mAh Lithium-lon Built-in Expansions 1.3 Megapixel cameraCamcorderBluetoothWi-FiMini SD Expansion Slot Add-on Expansions N/A Synchronization Included: Mini USB Port Best Regards, John Holmblad Televerage International GSEC Gold, GCWN Gold, GGSC-0100, NSA-IAM, NSA-IEM (H) 703 620 0672 (M) 703 407 2278 (F) 703 620 5388 primary email address: jholmblad at aol.com backup email address: jholmblad at verizon.net www page for texting: www.vtext.com/users/jholmblad text email address: jholmblad at vtext.com klaus at rotters.de wrote: > Am 20 Feb 2006 um 11:08 hat Clemens Eisserer geschrieben: > >> Yes I ment UI responsivness, not raw graphic performance. >> The whole UI simply feels a bit sluggish compared to Palms or >> Pocket-PCs, I know they have faster CPUs but i also think their >> software simply is more efficient. >> > > I don't think that there will be a huge speed improvement. Why? > > 1) The N770 has just a 16-bit data bus but it has to draw the whole GUI to a 800x480 16 bit > pixel memory buffer. This is much more than the 320x320 pixels on a palm! > 2) As far as I understand the OMAP doc, the LCD redraw had also be done using the 16-bit > memory data path. If its done at 60 Hz (I don't know), there is a constant memory dma at > 800x480x60 = 21,97 MWords/sec blocking the bus. In fact the OMAP has a shared memory > design! > 3) The bus is shared with the ARM cpu core and the DSP cpu core. Of course, there are > caches, but you can't cache everything. > 3) As often said, gtk+2 is very good, but not the fastest. I compiled some demos with fltk, and > I *feel* that they run faster. But it too late to change the main gui lib and fltk has other > disadvantages. > 4) The memory design: The N770 is more or less just a Linux device and uses a lot of > libraries. Most of the libs are designed for a linux desktop machine, where memory is > (because of swap) not a problem. This is maybe the biggest problem for Linux based PDAs! > One has to fine tune all those libs to perform better on small devices, which would be a big > task and maybe also force a development split, so the folks at Nokia must maintain those > libs. There is no manpower for that. > > It would be interessting, how a 128 MB RAM N770 would perform. I think 128 MB-DDR-RAM > is the limit for the build in OMAP processor. > > Don't get me wrong: I like the N770 (especially the bright display!), and I even don't think that > it is too slow. I compare it with my Sharp Zaurus and most times the N770 wins. The 800 > pixel display is a huge plus working with gnumeric. > > Please correct my if there are some facts wrong, I *really* would like to hear that! I got most > of this reading the OMAP5912 Design Overview, which should be similar to the OMAP1710 > used in the N770 device. > > > CU, -Klaus > -------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.maemo.org/pipermail/maemo-users/attachments/20060221/f298ffe8/attachment.htm