How different it is with machines out in the market place. We have been interested to read the various comparisons between the Omega, RPC and the Iyonix. The most important set of results we have seen were posted by Steffen Huber.

Steffens results showed up an anomaly in the ARM processor (when used in the context of a desktop design) relating to performance with applications using large sets of data. We have spent time investigating exactly what was going on and the results of our findings are being currently incorporated into the ARMTwister system, of course these improvements are only possible because of our use of FPGAs in the Omega between memory and the processor.

What has also become apparent is that there is a performance drop when using high resolution modes with true colour. Our critics have said that there is not enough bandwidth available to support both video and cpu. In practice, they vastly overestimate the amount of bandwidth the StrongARM is capable of using, so in fact there should be some spare. There are several aspects of the operation of the StrongARM external bus that are not mentioned, much less explained, in the Intel documentation, but which can only be found out by careful observation with a logical analyser or digital storage 'scope. This performance drop relates to the way the present version of our SDRAM controller operates. We have made some improvements which will be in the next update release (15), which will be available shortly. This set of changes gives an improvement of perceived cpu bandwidth of up to 50%. There is scope for further work to give even more in future releases. This shows the benefit of our soft-hard design strategy.

By comparison, it is apparent when reading the datasheets of the newer XScale implementations that Intel have found some performance problems in some of the earlier integrated designs. However as we are not using any of the integrated XScale devices these problems do not affect us. These problems have been fixed, but the price to the manufacturer of systems using those chips is that the new replacement chips are physically incompatible, with different sizes and new pin-outs, forcing significant new board design with the inherant additional costs.

With regards to comparison testing anyone who has been involved might like to re-run the tests with the current issue 14 software as we have made several changes in this release which have resulted in considerable increases in the Omega's overall performance.

It has also been pleasing to read that, apart from the actual processor speed, the Omega out performed the competition which proves the point we already know from the early comparions between RPC and Pentiums that having a fast processor does not necessarily mean you have a fast system. Regarding comparision tests. you have to be very careful to understand what is being compared because it is very easy to produce totally misleading results. My feeling is that in the real world the 600MHz Iyonix is still overall a little faster than the 300MHz Omega, hopefully our next upgrade will overtake them and we still have the XScale to look forward to.

After consultations with our dealers and the majority of owners with whom we have had contact, we have prioritised the completion of the Omega project as follows :

1) Ethernet

2) 2nd processor (XScale)

3) USB

4) Manual

5) Graphics acceleration

6) SCSI

However, of course some of the above activities are being progressed in parallel.

Progress report

1) Ethernet: The Ethernet card and driver are complete. We have sent cards and software out to our beta testers and the full release version is close - watch this space.

2) 2nd processor: A lot of work has been completed but I would rather not comment further at this time.

3) USB: The low level driver (software), that’s the interface between the Southbridge chip and the Simtec USB stack is now finished but not completely tested. When we release the software to Simtec the completed system will basically depend on how busy they are with other work. But we will keep you up to date with progress.

4) User manual: This is now complete all three hundred pages and it is with the printers, when the books are finished we will send them off to the existing Omega owners.

5) Graphics acceleration is work in progress.

6) SCSI again this is work in progress.

Other issues:

Select: Last week Paul (Paul Middleton MD RISC OS Limited) kindly sent us a CD with the latest RISC OS 4.39, called Adjust (ie Select in ROM) and now after several days testing we can report that it works a treat both from plug-in ROM and in Flash ROM point of view. This new RISC OS really looks a winner, once you have used it you won't go back to RISC OS 4.0x - congratulations to Paul and Justin great work chaps.

Top of page

Home ·  Products ·  Consultancy ·  Contact us ·  NewsDesk ·  Education ·  Partners ·  Sales ·  Owners Club

Microdigital Website© MicroDigital Limited, 2002
Any comments please contact the Webmaster