<Q>: Hello! Thank you very much
for spending your time answering our questions. We are glad that you managed
to give us a few minutes. To start with, could you please introduce yourself
to our readers and tell us in brief about your position in Imagination
Technologies?
<A>: I'm David Harold and I'm the PR manager for Imagination
Technologies and its divisions: Ensigma Technologies, VideoLogic Systems
and PowerVR Technologies.
<Q>: Could you tell us in brief
about Imagination Technologies?
<A>: Imagination Technologies Group plc is an international company
that develops, licenses and supplies market-leading 2D/3D graphics, digital
video, audio and speech technologies and products for consumer entertainment
and PC Markets. Since it was founded in 1985, the company has introduced,
via licensing arrangements or directly, a succession of innovative technologies,
silicon chips and add-in boards, which have played a major role in the
development of multimedia and computer-based entertainment systems. Indeed,
the company has deployed and planned technologies for games consoles,
PC's, arcade entertainment machines, digital set-top boxes, in car
information/entertainment systems, and mobile devices. Our corporate headquarters
is in the United Kingdom and we're publicly traded on the London
Stock Exchange. More information is available on the Imagination Technologies
web site at www.imgtec.com.
<Q>: So, it is not a secret that
most of questions will be about the new chip generation named KYRO developed
by Imagination Technologies. Also we know, STM is a manufacturer of KYRO.
So, the questions are the following: could you describe the part of work
made by Imagination Technologies? Did you only design architecture and
software (SDK, DDK) or did you also design the chip?
<A>: It's a collaborative process. We designed the KYRO
architecture and software. We worked together with ST on layout. ST does
all the manufacturing and sales of KYRO.
<Q>: What do you think about future
of tile rendering architectures? This architecture really reduces memory
bandwidth, but will it be effective for applications with a million or
more polygons per frame?
<A>: Tile based renderers will continue to be effective even
with the potentially huge number of polygons per scene. We have been developing
solutions to maintain the architectural benefits of our tile based renderer
and have applied for a number of patents in this area.
<Q>: Who do you think to be the
main competitors for KYRO based cards? What do you plan to offer to beat
them or to increase your presence at the market?
<A>: NVIDIA products are out main competitors. We plan to offer
much better feature sets, future-proofing and better price/performance.
<Q>: What is IMG going to do to
attract customers (low prices, high quality or warranties and support)?
<A>: Well, our customers are very few - NEC, STMicroelectronics
and the like. It's more accurate to call them partners. They then
sell processors based on our technology and they have a lot of advantages
doing so. They're huge companies, with massive resources and economies
of scale and they are dealing with technology in PowerVR that is inherently
very cost-effective, yet powerful too. This helps them to attract customers
for PowerVR.
<Q>: Could you name any graphics
card manufactures from first division interested in KYRO? Is it true that
NVIDIA doesn't allow vendors to make cards based on chips different
from their own?
<A>: I can't name any more manufacturers until they announce
products themselves I'm afraid. I can't speak for NVIDIA either.
Why not ask them what they think?
<Q>: A very popular question -
how will be final (retail) cards clocked (chip/memory)?
<A>: 115 core and memory.
<Q>: We've heard that current
KYRO design has some mistakes, so the clocking is low. Is it true? Do
you plan to redesign the chip?
<A>: That's not the case at all.
<Q>: Will we see a special KYRO
version to be integrated into mainboards?
<A>: I don't believe one is currently planned.
<Q>: Why doesn't Kyro have
HW T&L?
<A>: T&L is good technology and we are already using it in
out latest arcade solutions. However the applications just aren't
there on the PC right now. At the minute T&L isn't as useful
a benefit on a PC processor as Tile Based Rendering. Compare Dagoth Moor
at 32bit 1024x768 on a KYRO Board and a GeForce II MX and you can see
this.
<Q>: Do you plan to add HW T&L?
When can we expect announcements?
<A>: We are using T&L now in arcade solutions. We'll
bring it to other platforms when the applications are there to really
use it.
<Q>: Does KYRO use HW polygon sorting?
And what about next generation chipsets?
<A>: It depends what you mean by HW polygon sorting.
Two possibilities - depth sorting (aka hidden surface removal), which is done on KYRO. KYRO's difference is that is does HSR entirely on chip and in parallel on a per pixel basis. OR Translucent polygon sorting: This sorts translucent polygons from back to front order (again per pixel to do it right) before drawing to ensure the final image looks correct. No PC chips do this. It is done on Dreamcast though... <Q>: Could you describe Frame Buffer
Tile parameters?
<A>: As a scene based capture system we operate on a tile at
a time and only the components in that tile are operated on at any time.
These are the parameters and can be fetched from either the frame buffer
or AGP space.
<Q>: As far as we know, PowerVR
technology is easily scaled, e.g. we can place two KYRO chips in one PCB
and we see a linear increase of performance. Is it true? Do you plan multichip
cards for PCs? What is your opinion about multichip solutions?
<A>: PowerVR is indeed very scalable. We use multiprocessor boards
now in the arcade. However we prefer in the PC to increase performance
of single chip solutions rather than adding multiple processors, which
we think has proven a competitive disadvantage to some of our competitors
who were forced down the multi-chip route by bandwidth issues.
<Q>: What about DX8 support? Will
we see announcement that KYRO is pure Hal device (in DX8 terms)?
<A>: I understand DX8 support in KYRO is impressive. Certainly
independent testers like PowerVR Revolution have reported that that's
the case.
<Q>: Is PVRSGL dead?
<A>: PVRSGL has been moved from our SDK to the DDK, but like
the rest of the industry our focus is on OGL and DirectX.
<Q>: What do you think about the
current situation on 3D market? What do you think of its evelotution?
<A>: 3D is an enormously difficult problem to crack, from a technical
standpoint, and one that it's difficult to succeed in commercially.
The fast pace of change has taken its toll on many graphics companies
and more will fall in the near future. We're happy that we have an
amazing technology to exploit, with lots of patents applying to it, and
great partners in NEC and ST - huge companies who aren't susceptible
to the whiles of the graphics industry. We have some great plans for future
products and plan to make PowerVR a very significant presence in PC 3D
graphics - at least on par with our success in arcade and console.
<Q>: Which parameters do you consider
most important for modern graphics cards: performance, graphics quality,
innovations or something else?
<A>: These features all need to be present in balance. KYRO offers
a really good performance level at an affordable price. Its graphics quality
is excellent, with ITC, EMBM and 8-layer multitexturing. It's truly
innovative too in that it brings Tile Based Rendering to the PC mass market.
<Q>: What's your opinion on
integrated graphics (into mainboard chipsets, for example)? Is it the
only way for evolution?
<A>: Integration to the motherboard is useful for some solutions,
particularly low-cost, but there will be add-in accelerators for a long
time to come.
<Q>: Could we expect to see a new
leader or a new player in 3D graphics market in the nearest future?
<A>: I think in the PC arena ST will be very successful using
PowerVR Technology.
Thank you for answers!!!! Write a comment below. No registration needed!
|
Platform · Video · Multimedia · Mobile · Other || About us & Privacy policy · Twitter · Facebook Copyright © Byrds Research & Publishing, Ltd., 1997–2011. All rights reserved. |