http://www.nvitalia.com/articoli/Intervista_Richard_Huddy_ATI/index.php?pagina=6Nice
to meet you. First of all, may you introduce yourself to our readers?Hi!
And first of all a big thanks for allowing ATI to contribute to yoursite.
I understand this is the first time that ATI been invited in, but Ihope it
won't be the last. It's interesting and unusual to see a fan siteopen up a
dialog to a competitor in this way, and I think you deserveconsiderable
credit for being so open.OK, so back to the question... I'm Richard Huddy,
and I run the DeveloperRelations team here in Europe for ATI. That means I
get to work with all theEuropean developers and help them to understand
our technology and productroadmaps. I've been with ATI for over 3 years
now in this same role.Previously I worked at NVIDIA for four years in
pretty much the same role,so some of your readers may remember me from
then. I've been called a "3DGuru" at times, and although I don't really
much like labels like that Iguess that this is probably pretty fair
comment when you take into accountthat I've been doing this kind of work
with developers for over nine yearsand that I've been involved in 3D for
around fourteen years.This year it's occurring the twentieth anniversary
of ATI Technologies.Would you please remind us, in a few words, ATI's
history and the "mission"of this company?ATI has been in the PC graphics
business for a very long time. In thisindustry 20 years seems like
centuries!We were the first company to offer hardware accelerated motion
video (1994),hardware accelerated 3D (1996) dual graphics chip based
boards for consumers(Rage Fury Maxx in 1999), and we were the first to
produce DirectX 9graphics (2002).All through our history PC graphics have
been our primary focus, but it'sinteresting to look inside ATI and see the
diversity of the business too.Our company "Mission Statement" is simple and
clear: "ATI - delivering theultimate visual experience". Right now that
means that we focus on PCgraphics, consoles, mobile phones, handheld
gaming units, digital TVs, HighDef TVs, embedded displays, super high-end
workstation graphics and thingslike that. Wherever there are pixels to be
seen you'll find ATI pu****ng theboundaries and delivering both quality and
value.After the giant Intel, ATI holds the second position in the graphics
market,with about the twenty-seven percent of it. Which prospects does the
graphicscards segment offer today to ATI and in which direction is this
marketdeveloping?The graphics market in PCs is an interesting one because
the squeezes comefrom several very different directions. We're under
pressure from Intel(they are the number one producer of graphics for PCs)
to produce low-costgraphics as part of integrated chipsets for what we
call "cor****atedesktops", and at the same time we need to produce very
high end gaminggraphics on add-in graphics cards. Most of your readers
probably focus onthe high end graphics cards, but it's a surprise to
realise that thisamounts to only a few percentage points of the PC market.
The average PCprice is falling in cost (which produces downwards pricing
pressure on ustoo), and yet the technology involved moves forward very
rapidly, whichmeans we have to invest a great deal of cash into Research
and Development.All these pressures produce a market where competition is
spectacularlyintense. It's easy to gain the lead in one generation, and
lose it with thenext. I'm particularly looking forward to the impact that
out nextgeneration hardware will have when it arrives in the market. I
think a greatmany people will be duly impressed.The current graphics card
line Radeon X8x0 is born from the undeniablesuccess of the previous
series, based on successful graphics chip R300.Would you like to remind us
the labour of that project, if possible somebackstairs business, and how
this achievement actually brought to the up todate production?The R300
chip was a tremendous achievement for ATI. Up until then ATI hadfocussed
on parts of the PC market which involved large volumes of sales butdidn't
require total technolo
gy leader****p. In 2000 when ATI acquired alittle known Californian company
called "ArtX" ATI changed its plans anddecided to produce a world beater. A
couple of years later the R300 arrivedand astonished most people. The R300
was roughly twice as powerful asNVIDIA's NV30 (which arrived 6 months
later) and was much better accepted bythe market. That was the first time
ATI had aimed to capture the high end,and we achieved it in a way that
went far beyond my own personalexpectations. It turns out, of course, that
this was a good thing forconsumers and for NVIDIA too, since it meant that
there was renewed healthycompetition in the high end of PC graphics. That
means lower prices forgames players, and it meant that NVIDIA had to sit
up and take a close lookat what they wanted to achieve. Overall I'd say
the main gainer was theconsumer.Can you tell us in a few words the guide
lines and the strength points, andeven the possible improvable sides of
the current ATI's graphics cardsproduction?When I work with games
developers I explain where ATI is going and why, andI collect their
feedback and we use their input to help us design chipswhich genuinely
solve developers'problems. We use this feedback to focus onuseful features
and to help us avoid getting involved in what we call "TickBox Technology".
That's technology which is there just because it wouldallow us to boast
about features, even though they're not useful.My own direct experience
with games developers shows me that they reallylike the fact that our
hardware is easy to program and tends to work in avery predictable kind of
a way. One of the great strengths of our hardwareis that it always works as
a very fast generic DirectX 9 graphicsaccelerator. We don't need to produce
complex programming manuals to explainpainful details of how to make the
hardware run fast, and we don't need topromote obscure use of the
APIs.With every generation our hardware becomes faster and more powerful,
andwe'll add interesting new features when we introduce our next
generationhardware. Interesting I've seen David Kirk from NVIDIA saying
that in hisview some of our new features are years away, so I guess the
new hardwarecould make a bit of a stir. Our focus is always on usable
features which runat full speed, so I'm looking forward to the fact that
our new hardware willchange the games development landscape for the better
in ways that gamesdevelopers will welcome. Indeed the one word which games
developers commonlytend to use when we discuss the new hardware with them
is simply "Cool!"The single thing which will make games look better in the
next generation isfull on HDR. HDR stands for "High Dynamic Range" and what
it really means inour context is rendering 3D images using data which is
very like that foundin the natural world. "HDR" has become the buzzword in
the games industrybecause a few games developers have started to show demos
which use HDR andthe difference is simply stunning. But the trouble right
now is that all theHDR implementations in the market to date are limited
in one way or anotherwhich makes the decision to use HDR a somewhat
painful compromise. We planto eliminate the compromise and thereby do away
with the pain.In the late times have been coming out a lot of models based
on the ChipR4x0, as the Radeon X550, the Radeon X800 and the almost next
Radeon X800GT.Would you briefly explain to which scale of end users are
these productsaddressed to, and the targets that you are intending with
them?We tend to split the PC graphics market up into 3 primary
segments.The "value segment" focuses on cards which are priced at up to
about $99.This includes cards like the X550, 9250 and 9500. They deliver
performancewhich would humble a perfectly respectable graphics cards from
just two orthree years ago but they're far more affordable.The "mainstream
segment" focuses on cards which go from around $100 up to$250. This segment
is covered by ATI's X700 and X800XL cards. That's not agreat large amount
of money to spend on a graphics card, and buying one ofthese can give you
performance whic
h only a year ago would have cost morethan twice as much. This is what
most gamers buy. They will run all gamescomfortably at 1024x768 screen
resolution, and for most consumers thesecards represent the sweet spot of
price versus performance.Above that you find the "high-end segment". Cards
like the X800XT or thenewer X850XT are found here, and they cost anything
from $250 to $500.They're for gamers who are so serious about their gaming
experience thatthey won't compromise. This is the group of people who want
high performancegaming with all the display settings at their most
demanding like alwaysenabling anti-aliasing and expecting to run games at
resolutions up to oreven beyond 1600x1200... There are just a handful of
millions of gamersaround the world who will go to these extremes.And if a
gamer is willing to spend almost without practical limit, thentoday's top
choice from ATI would be to get a CrossFire system. Twin X850graphics with
a super-fast ATI motherboard that generates frame rates thatgive you a huge
edge in competitive gaming. To some people this representsthe lunatic
fringe of gaming - but I'm proud to include myself in thisgroup.Obviously
these segments are artificial. Some people fall into more than onegroup
because of the different ways they play different games, and one dayyou
could be in one group and next day in another. But these groupings helpus
to understand and target the main parts of the market. And in thediscrete
business (that's add-in graphics cards) each segment tends to bearound ten
times larger than the segment above it. That means that we cansell roughly
100 times as many X550 class cards as X850 class cards!The existing ATI's
production does not sup****t some features of thoserequested by the DirectX
9.0c. Skipping the matter that there are stilltoday not many applications
exploiting its advantages, would you explain ineasy words to the readers
the philosophy on the grounds of these choices?This question takes me back
to a point I made earlier about "Tick BoxTechnology". For us it has been
very im****tant to deliver features which areall genuinely useful. If new
features don't help games look better and playbetter then they're not a
good investment. And we think that 2005 and 2006are good times to
introduce some new features... In 2004 we wanted to focusprimarily on
performance. This year we're going to set new standards in bothperformance
and features. The fact that there are so few games in existencewhich
attempt to use the SM3 features of DirectX 9.0c indicates to me thatadding
those features to hardware last year was a relatively
poorinvestment.Clearly there are times when it's necessary to produce
hardware which opensup new possibilities, but the danger lies in producing
hardware which runsnew features so slowly that they're genuinely
unusable.Let's talk about the CrossFire technology. Would you tell us
about thefundamental idea and the targets, most of all in terms of
commercial offer,of this new technology?I'd say that there are three
primary strengths to CrossFire when compared toSLi.Firstly and most
obviously it works with every single game, out of the box,without any
hassle. Every single game runs faster with CrossFire... WithSLi, NVIDIA
have profiles for around 80 games, so if your game is not ontheir list
then you won't get any benefits from SLi.Secondly we are able to use
CrossFire to dramatically improve image qualityon all games too. If your
game is already running fast enough then you canuse CrossFire to make it
look better. And that's a great alternative since agame which is rendered
more accurately can give a gamer a clear advantage.And the third point is
that you can freely mix and match cards fromdifferent vendors when
building a CrossFire system. You can take any X850card, pair it with any
X850 CrossFire board and plug it into any CrossFiremotherboard and it will
just work. The problems persuading SLi to workreliably are well do***ented,
and we wanted to make sure that people who buyCrossFire systems get a
system which "just works". No one should have tomess with conf
igurations and settings to make a game run faster.Our fundamental target
with CrossFire was to produce a robust system whichaccelerates all games
and delivers great value. And I think we've done agreat job of that!In
your opinion, will the multi-board technology bring to a change in
theupgrade market, and in particular even in the reuse of the video
boards, andin which measure? Which concrete advantages will the CrossFire
technologyconvey for the "standard" ATI's end user?CrossFire is aimed at
the serious gamer. If you do see yourself as a seriousgamer, and you have
ATI graphics then you probably own an X800 or X850board. You have at most
two steps to the upgrade to get yourself CrossFireenabled. Firstly you
need to have one of the dual slot PCI-Express boards(the ideal is the
Radeon Xpress 200 CrossFire Edition) and then you need toget a CrossFire
edition graphics card. The process is as simple as that.There are a couple
of great advantages to "standard users". The first one isthat all games
suddenly run faster. CrossFire works on every single hardwareaccelerated
3D PC game without needing any intervention or fiddling from theuser.
That's in dramatic contrast to SLi which works on some games, and noton
others. Turning SLi on and off requires a reboot, CrossFire just lets
theuser switch settings and start their game. Standard users don't want to
haveto play around with obscure control panels and driver settings...The
other feature which makes CrossFire so attractive is that the
upgradeprocess is really easy. You can simply buy a CrossFire edition
graphics cardthat matches the existing graphics card in your machine. If
you have an X800then get a CrossFire edition X800. With SLi you need to
worry about matchedBIOSes, you have to worry about clocks speeds being
exactly the same, and tobe safer you need to buy a graphics card
manufactured by the same companythat made your first card. That turns out
to be a lot of unreasonabledemands made on unsuspecting PC users...One of
the key points of the next ATI's projects is surely the graphics chipR500
for the Xbox 360. Would talk in general of how ATI is planning toapproach
the relation****p between platform "Console" and the "PC" one? Whenthese
two worlds will meet and how will they affect to each other?The Xenos chip
(that was our internal codename for the Xbox 360 graphicschip) is now in
high volume production. So, we're now in a position where wecan start to
take chunks of that technology and apply it to the PC. Forpeople who
aren't familiar with the key Xenos capabilities I'll list justthree.
"Intelligent Memory", Unified Shaders" and "Generalised memory accessfrom
within shaders".One im****tant thing to understand is that not all console
technologies makesense on the PC. ATI sells over a million chips every
week, and most ofthese are in the value segment of the market. All our PC
chips need to besold at a profit. In contrast consoles typically sell at a
rate of around amillion per month and early on they can be sold at a loss
to build aninstalled base.That means that "Intelligent Memory" isn't
applicable to the PC. It's far tospecialised to use in a PC and we
couldn't readily produce a "value" boardbased on this kind of idea. So I
think that this is one technology that'sunlikely to affect PC graphics
architecture in the near future.The other two features are perfect for PC
designs. Unified shaders are justwhat Microsoft want for Windows Vista,
and they're a great solution to along term problem in PC graphics design.
In only a few years I expect thisto be the standard way of producing
graphics chips for PCs. And a littlefurther into the future the
"generalised access to memory from withinshaders" will become a hot topic.
This feature allows the GPU to do anythingthe CPU can do in a way that
programmers will find easy to use. It's a verystrong feature of the Xbox
360 and it will be one of the key features whichallow games to look much
more like movies. As far as I know it's presentlyunique to the Xbox
360.David Orton, at the conference to the investors of the third quarter
2005,a
ffirmed that ATI had some delay with R520 but that is trusting the
newgeneration R5x0 for the winter fall. Would you tell us where are we
standingnow with ATI's plans for the next months?I'm not allowed to
comment on unannounced products, so my answer here isgoing to be very
limited... But I can say that I think the remainder of thisyear is going
to be a very interesting one. I'm delighted that Dave spokeopenly about
this and I'm pleased to say that we have no changes to announceto Dave's
timetable at this time.That's a coded way of saying that I'm really
looking forward to the nearfuture. You're going to see some very
impressive hardware from ATI.Let's talk about Windows Vista. Many end
users, are questioning how theinformation technology world will change,
after the arrival of this newoperative system and its new graphics
standards. Can you tell the readershow ATI is paying attention to this
change and which strategies will it takefor the sup****t and for the
back-compatibility?One of the key statements about Vista was made recently
by a Microsoftspokesperson when they made it clear that integrated graphics
are not goingto be a good way of experiencing Vista. The reason for this is
simple... Theuser interface for Vista is going to be built upon DirectX
graphics and thatmeans that the operating system is going to demand a very
flexible andpowerful DirectX 9 graphics system just to work in an
acceptable way. Thisis obviously great news for those of us working in the
graphics hardwareindustry, but it's also good news for all PC users too.
With Vista youshould have a system which is always responsive, smooth and
slick.It's definitely not the case that these demands for graphics power
areexcessive in any way. What is needed is a graphics system which is in
manyways like current system sup****ted by CPUs. A modern CPU is
perfectlycapable of running the many threads and processes that modern
operatingsystems use to create a smooth user experience. But many graphics
cardsdon't work in the same way, and even those that do, are never driven
in thesame way - so you often end up with a choppy "sticky" user
experience. WithVista and an appropriate graphics card the system will be
capable ofresponding much more quickly. Vista will add eye candy for sure,
but thereal user benefits lie in an advanced graphics scheduler and a slick
andalways responsive user interface.If you had the chance to say something
directly to the whole end users ofgraphics boards, of ATI but not only,
most of all, what would you tell?I guess I have two messages.My first
message is to focus on features and performance rather than
brand.Simplifications like "ATI is better than NVIDIA" or "NVIDIA is
better thanATI" always miss the real point. When you buy a graphics card
for your PCmake sure that it has the ability to run your games fast and
look good.Upgrading your graphics card and installing more memory are
probably the twomost useful changes you can make to a PC which is
underpowered.Secondly I'd say to any of your NVIDIA users who really do
believe that"NVIDIA is good and ATI is bad" that it's well worth taking a
closer look.Recent evidence says that ATI's drivers are measurably more
robust and Ibelieve that when you experience CrossFire and the other
hardware that we'llbring forward in the near future then you'll appreciate
how hard we work toproduce great value in all the hardware that we bring to
market.Let's end up with cheers to NVITALIA community. Have a nice
work!Absolutely! My thanks to you for giving us the chance to contribute
to yourwebsite, thanks to all your readers who waded through this
lengthyinterview. And thanks also to NVIDIA for giving us competition.
Competitionin industry benefits everyone, and, because of this intense
competition,buyers of PC graphics have a tremendous op****tunity to get
great value.
|
