The PowerVR technology may also include MPEG2 decode (DVD anyone ?), and Imagination technology has stated that they want their technology to compete with PC's and games consoles.

The ARM10 chip has the very fast VFP10 Vector co-processor (which at just 400MHz manages to equal a 1GHz Pentium for FP). The inclusion of the PowerVR onto the ARM (and possibly alongside ARM Jazelle JAVA extensions) means we could see a quite radical improvement in both 3D graphics, games and JAVA execution on ARM chips by as early as 2002.

More significantly new machines will not need PCI to provide good graphics (as PowerVR is on chip and has a much faster and wider co-processor bus available to it than the rather slow and narrow PCI bus).

Two things will need to happen though (i) RISC OS will need to add 3D support (just imagine the DrawModule with 3D added !!!) and (ii). A common API will have to be documented so as to allow software vendors to take advantage of the new oppertunities.

We may be faced by RISC OS users upgrading their xScale to an ARM10+PowerVR now there's a thought.

[Edited by ams at 15:02, 27/1/2001]

I have a friend who has been working on OpenGL drivers for the Kyro (PowerVR2) boards. He reckons they're pretty nifty. They're not quite up there with the top end nVidia chipsets but then not much is.

As far as 3D support goes it'd be utter madness not to support at least a subset of OpenGL. An ARM10 with the FPU and PowerVR2 would be incredibly powerful with an API like OpenGL. Throw in Jazelle and things look pretty darn good.

We may be faced by RISC OS users upgrading their xScale to an ARM10+PowerVR now there's a thought.......

Exactly what I thought! Maybe the XScale is only going to be temporary speed increase while we wait for the ARM10 with 3D acceleration, MPEG2 decoding and some serious FP processing? The end of Intel in RISC OS land perhaps?

Btw, wasn't ARM owned by intel?

Hmm, I don't think so. ARM was owned by Acorn, VLSI, and Apple. When Acorn was sold to some bank whose name I fail to remember, Acorn shareholders where given the ARM shares owned by Acorn as a payment for Acorn Computers plc. Whatever happened to the rest of the shares I have no idea but I don't think Apple has any shares in ARM now.

Intel on the other hand owns a right to produce processors by the ARM design and specification, hence StrongARM and XScale. They aquired that right when they bougth Digital's production line.

Not very accurate, I know, but I think you know what I mean.

Gulli

But a PCI (or another fast bus) will still be required for other cards, video editing springs to mind!
Gulli

Aggreed a fast bus is required, but why PCI ?

In PC land moves are afoot to PCI-X and by the time RISC OS machines have PCI it may already be on the way out !

Besides it is only likely that drivers for (possibly) a SCSI and network card will be available initially and devices that use proprietary technology may not have information publically available to allow RISC OS developers to produce the requisite drivers.

Why bias the whole hardware design just to allow TWO types of card to be available ?

RISC OS hardware developers should develope a more upto date standard than PCI with the possibility of a "bridge" to PCI should some hardware developers deem it a necessity. Additionally there have been issues with PCI chipsets in the past so why add that problem where its not needed (or where the benefits are so limited) ?

PCI-X might be a possibility (in 2002) or (better still) a fast wide bus agreed between the RISC OS hardware manufacturers with a low cost interface to allow backward compatibility with Podule and PCI busses. If they share the cost the project should be quite feasible.

Think radical guys the new processors with integrated Java, 3D support and fast wide 64 bit databusses demand a fast I/O bus as well, and PCI just isn't it.

[Edited by ams at 17:50, 28/1/2001]

But a PCI (or another fast bus) will still be required for other cards, video editing springs to mind!
Gulli

Aggreed a fast bus is required, but why PCI ?

My guess is that the main reason is because it exists! A more stable design would of course be preferred but that would require someone to sit down and design it, that costs money and money doesn't exactly flow out of the developers ears ATM.

In PC land moves are afoot to PCI-X and by the time RISC OS machines have PCI it may already be on the way out !

Well, ISA has long since been supposed to be wiped out by PCI but it's still there and although on the way out it will be there for a couple of years more. PCI will not disappear overnight since PCI-X hasn't really taken off yet.

Besides it is only likely that drivers for (possibly) a SCSI and network card will be available initially and devices that use proprietary technology may not have information publically available to allow RISC OS developers to produce the requisite drivers.
Why bias the whole hardware design just to allow TWO types of card to be available ?

I doubt that it will be a question if SCSI and network cards will available for PCI, I think the biggest question will be how many and how soon. As for other cards, there are thousands of card developers out there, designing and making all sorts of PCI cards. I'm quite confident in that some of them will be more than willing to assist in writing drivers for their hardware.

Secondly and seemingly totally overlooked possibility, what about RISC OS hardware developers? There's little stopping them designing and making their own cards and up their potential sales by writing drivers for Windows and Linux to name only two options. Giving them the revenue to develop better cards with a possible sales of tens of thousands of units instead of a few hundred.

RISC OS hardware developers should develope a more upto date standard than PCI with the possibility of a "bridge" to PCI should some hardware developers deem it a necessity. Additionally there have been issues with PCI chipsets in the past so why add that problem where its not needed (or where the benefits are so limited) ?

The best option without a doubt but costs more than using a "standard" that already exists and cards would almost certainly be more expensive than PCI cards (if only for less volume sold)

Gulli

[Edited by Gulli at 12:46, 29/1/2001]

Is the Frame buffer technology of PowerVR the same thing as Linx Framebuffers? That'd give a dual-boot son-of-Omega a boost until the RiscOS drivers caught up ;-)

Umm the framebuffer technology is PowerVR specific. What you need are drivers for any given OS in order to utilise it. I'd be very surprise if linux drivers didn't exist for PowerVR cards in some form, even if only development at the moment.

btw, what does Omega do about IOMD? (I know it bypasses VIDC, hooray).

They've designed their own controller. I would assume that they've either patched RISC OS to work with it or it's backwards compatible with the IOMD.

btw, what does Omega do about IOMD? (I know it bypasses VIDC, hooray).

They've designed their own controller. I would assume that they've either patched RISC OS to work with it or it's backwards compatible with the IOMD.

That jogs my memory, thanks. I believe their controller 'emulates IOMD' to the system bus.

The PowerVR technology may also include MPEG2 decode (DVD anyone ?)

More significantly new machines will not need PCI to provide good graphics (as PowerVR is on chip and has a much faster and wider co-processor bus available to it than the rather slow and narrow PCI bus).

We may be faced by RISC OS users upgrading their xScale to an ARM10+PowerVR now there's a thought.

Gulli

sigh, just *another* reason to wait, wait and wait.... blimey, I want to see some real samples!

Why wait? The new machines will be compatible with the ARMV5 architecture so get and Omega or something and upgrade later

We're all waiting for the RS portable (btw, if Psion can put an SA in their series7/netbook, why can't RS in their laptop.. or is that maybe the reason it's still not there.. they're putting an SA in it??),

No, RS aren't using the SA because it is difficult to get hold of the VIDC and IOMD chips RISC OS requires. The ARM7500FE has them built in. Psion are ok because EPOC isn't tied to VIDC or IOMD.

I suspect the laptop has hit a minor hitch - maybe the software isn't quite ready, or they're waiting for something else. Given the recent announcement in Acorn User about the faster RAM finally being available for the RS7500s I wonder if the laptop is being altered to take advantage too. This would make it pretty nippy and along with the ability to slow the refresh of the screen down (thus cutting the bus bandwidth consumed by the VIDC) it could well surprise a few people.

sigh, just *another* reason to wait, wait and wait.... blimey, I want to see some real samples!

Why wait? The new machines will be compatible with the ARMV5 architecture so get and Omega or something and upgrade later

Is the Frame buffer technology of PowerVR the same thing as Linx Framebuffers? That'd give a dual-boot son-of-Omega a boost until the RiscOS drivers caught up ;-)

We're all waiting for the RS portable (btw, if Psion can put an SA in their series7/netbook, why can't RS in their laptop.. or is that maybe the reason it's still not there.. they're putting an SA in it??),

No, RS aren't using the SA because it is difficult to get hold of the VIDC and IOMD chips RISC OS requires. The ARM7500FE has them built in. Psion are ok because EPOC isn't tied to VIDC or IOMD.

btw, what does Omega do about IOMD? (I know it bypasses VIDC, hooray).