Re: Xbox360 More Powerful Than ATI X1800 At HDTV Resolutions


"Dog Bowl" <dogUNDERSCOREbowl@xxxxxxxxxxxxxxxx> wrote in message
news:MPG.1dc56e37feba6c79989981@xxxxxxxxxxxxxxxxxxxxxxxxxx
> In article <VpOdnZPrcZZbAMfeRVnyrg@xxxxxxxxx>, Donotemail@xxxxxxxxx
> says...
>>
>> "Dog Bowl" <dogUNDERSCOREbowl@xxxxxxxxxxxxxxxx> wrote in message
>> news:MPG.1dc42dae2508a16c989980@xxxxxxxxxxxxxxxxxxxxxxxxxx
>> > In article <9-idnU_pwcT2_8TeRVnytA@xxxxxxxxx>, Donotemail@xxxxxxxxx
>> > says...
>> >>
>> > <snip>
>> >
>> >> Are you OK for me to post your replies in a few months after the 360
>> >> is
>> >> launched? I'm doing some research into the effectiveness of marketing
>> >> as
>> >> part of of my Phd.
>> >>
>> >> Hopefully you will be correct in that all these technical specs will
>> >> translate into games that look better than if they were run on a
>> >> 2.5GHz
>> >> PC,
>> >> with a Radeon 9800pro.
>> >>
>> >> I've looked into the hype surrounding the original Xbox, when many
>> >> people
>> >> were sucked into the idea that a Celeron733 and an MX420 GPU could
>> >> somehow
>> >> be better than a 1GHz PC from that time. You guys are probably
>> >> different
>> >> though, and not sucked into the hype, or listening to what the
>> >> developers
>> >> are saying.
>> >>
>> >
>> > No they were not. What you seem to fail to observe is a) the GPU on the
>> > Xbox was _not_ and MX420 is that a hybrid of a GeForce 3/4. The MX
>> > chipsets do not contain programmable pixel shaders they have a fixed
>> > pipeline. The relative spec of CPUs was based on the fact that a 733
>> > Celeron dedicated machine was faster than a 1Ghz PC. This is by virtue
>> > of the mother board design and not having to run a separate operating
>> > system under the application. Having the game itself govern memory
>> > allocation, disk usage, security by use of kernel cores alows the code
>> > to be more specific and not have to be so generic. This results if a
>> > far
>> > more efficient executable image hence you can more (in terms of
>> > perceived user experience) than if you have a heavy OS underneath it.
>> >
>> > You need to compare like for like or else you will find that your
>> > research will be _seriously_ flawed. I hope you do take into account
>> > the
>> > usage of the systems, after all you can say a spoon is *** compared to
>> > a key becuase is doesn't fit into the lock to open a door.
>> >
>> > Pontificating on about doing a Phd won't make you look like an
>> > authority
>> > on this ng when there are plenty of people here who have been there
>> > seen
>> > it done it and more and when you have some issues with the basis of
>> > your
>> > research. Set yourself some good solid ground rules as I'm sure you are
>> > on to a winner with the basis for your Phd.
>> >
>> > The specs on the Power PC chips (Cell and 3 Core) far exceed the same
>> > clocked P4s. They are just starting to be used for defence applications
>> > due to their significant power increases. The big problem at the moment
>> > is development of cooling solutions. This I suspect will leave them for
>> > a good year or so as engineering workstations/servers (such as my
>> > Mercury Computers) rather than getting out into VME buss applications
>> > for a while.
>>
>> I'm only going off what the developers are saying.
>
> I'm coming from a developer point of view too. I have worked on systems
> from transistory based PDP-11s through to state of the art RISC based
> systems. The PC is much more involved compared to a closed system.
>
>> Anyone can talk all day about pipelines and buffers etc, but the real
>> world
>> performance is what matters, especially where coding is generic. (We have
>> yet to see that the specialist nature of the console hardware will in
>> fact
>> be taken advantage of, other than for an extreme minority of games)
>
> The specialist nature of consoles is the fact that they are not running
> generic code. Having to cater for a much looser security base for a
> start, multitudes of hardware variations adds a significant amount to
> code development. For a start you don't have to query device caps as
> they are at a known base configuration for a start. This also means that
> you don't have to check return status information. That cuts alot of
> code out straight away.
>
>> Part of my research is the 'fanbois' or 'us versus them' effect in which
>> normal people somehow become so obsessed with their particular hobby,
>> that
>> any detractors are labelled as "extremists for the other side", whether
>> or
>> not the detractor is part of the 'following' for one piece of hardware or
>> another. Such arguments have hitherto been rampant in the "Ford vs.
>> Chevy",
>> "Democrats vs. Republicans", "Black vs. White", "My Mum's better than
>> your
>> Mum" arguments. In Gulliver's Travels, people went to war over which end
>> of
>> the boiled egg to eat from, because of the innate Western problem of
>> having
>> to take sides (Polarisation), rather than to look at the bigger picture.
>
> Part of that is down to human nature. People will *** each over for one
> upmanship. The slime bag marketers just play on that. "Buy product X and
> you will be fucking over people with product Y becuase it makes you look
> better and will help you get the big titties blonde and a ferrari". You
> can use that quote if you like :-)
>
> On top of that, its a simple fact that the pecking order for console
> processing power is Xbox -> Game Cube -> PS2. Then you need to compare
> software, where PS2 rules on quantity, Game Cube has the quality, Xbox
> has it on FPS.
>
> Then on top of that you have the "Simpsons comic store guy" attitude of
> a lot of PC owners that all consoles are under peforming kids toys.
> Which seems to be where your stand point is judging from your intial
> comments about relative performance, pretty much pidgeoning you into the
> "PC fanbois" categroy.
>
> Part of the reason I bought an Xbox was simply that it was to escape the
> marketing "You must have this graphics card and this sound card else you
> are 'teh ghey fhag' and your games won't run". I got sick and tired to
> being bombarded with magazines which were little more than "Stuff" but
> for computers. Why pay 200-300 quid for a graphics card, 100 quid for a
> sound card to upgrade a 800 quid system to be not significantly better
> than an Xbox in terms of game play, and still have a, relativley, poor
> on line experience. voice coms in PC games are still a joke, even on
> Counter Strike:Source with people resorting to text chat anyway.
>
>> I too would like to keep an open mind here, and it is interesting that so
>> far no-one has pointed me in the direction of any physical proof that
>> would
>> disqualify my original assertion.
>
> There is a shed load of info. You have already been proven wrong on your
> assertions on the "MX420" GPU. I would do your own research, after all
> it is you that is studdying the Phd.
>
> ***, even Toms Hardware will give you a basis. You could also get the
> data sheets for the processors which tend to be more fact based rather
> than marketing ***.
>
> For basic ATI 9800 info, the Mac pro edition can do upto 380million
> polygons per second via 4 geometry processors. The new ATI chip has a 48
> way dynamically assignable shader units able to process 500 million pps.
> The dynamically assignable part means that the load balance can be
> shifted between pixel shading an vertex shading based on the load. This
> reduces bottle necks in the pipeling. The result of this being that the
> theoretical maximum processing capability is more of an achievable
> number. With fixed pipelines the reality is about half the capability.
>
> For the 2.5Ghz P4. They are the non hyper threading type for a start.
> This means 1 hardware thread (even HT only processes 1 thread but has
> two fetch unit). On a simple measure the 360 will be processing 3
> processes at 3.2ghz, but with 2 fetch units per core, this means 6
> hardware threads running comared to a HT cpu's 1). On top of that a CISC
> processor tends to take 4 to 10 cycles for an instruction where as RISC
> tend to take 1 cycle. Although RISC uses much simpler instructions the
> code tends to end up doing a lot of RISC based instructions. I know the
> P4 at its core is RISC based but it still processes alot of CISC
> instructions.
>
> Here is an abridged quote from someone who managed to clock a P4 to 4
> Ghz
>
> "One thing that I was amazed to see, is that even at 3.9Ghz the Pentium
> 4 can't beat the PowerPc G4 at Floating Point Operations!
> The P4 was able to do 4.9 Gigaflops... a Dual G4 with 1Ghz each
> processor can do 15 Gigaflops...The new PowerMac G4 DP 1.25Ghz can do 18
> Gigaflops".
>
> Well the 3 core Xbox 360 can do a terraflop. Which basically wipes the
> floor with the P4.
>
> Oh and a PIII 1 Ghz is only about 50% more power than a Celeron 733.
> Once you factor in the inefficiencies of the PC based mother board
> achitecture, running an OS, running security software, running libraries
> that provide the common interfaces for a multitude of chip sets the PIII
> becomes considerable loaded with this wasted overhead.
>
>>
>> According to you and others so far, Saddam *does* have WMDs,
>> metaphorically
>> speaking.
>
> I hope the basis for your research is considerably stronger than
> playground name calling method which you seem to be employing here.
>
> And based on your points on realitive performance using the same
> metaphor I could say your argument is "Saddam never did anything wrong
> and infact donates huge amounts to western charties every year and is a
> very fluffy nice bloke"

Again, three words: Real World Performance.

I ask for proof, and you give me more theory.


.