What would you do if you were in charge of Xbox 360 design?

[Karma Police]

Two things:

1. eDRAM to fit 1080p!
2. PPU!

you should add

3. new ramdac to display the 1080p

Thats a given.

 

[scooby_dooby]

What's with all the people already concluding 360 is less powerful than PS3?

That's just completely unfounded speculation.

Neither box has been benchmarked, and the world is still out on the GPU's.

The original PS2 did many times the T-FLOPs of PC's at the time, but it was still creamed by a 2 year old Celeron with more bandwitdh and a superior GPU.

1. We don't know which GPU will be superior
2. We don't know who will have the most bandwidth. PS3 has a slightly higher transfer rate, but the 360's EDRAM will remove a lot of data from the main bus.

So until benchmarks or games start coming out I don't see how anyone can come to any conslusiongs here..

btw - I voted good as is. Can't say I can criticize anything they've done really, I think this system will be extremely powerful, and it's just as good a deal hardware wise, as the original XBOX in 2001, if not more so!

I can't wait til Nov.

 

[Titanio]

What's with all the people already concluding 360 is less powerful than PS3?

For my part I'm not coming to any conclusion. I prefixed my statements by saying we still didn't have all info. I'm simply saying that from what we do know, PS3 looks like it will be more powerful. I don't think that would surprise anyone.

With every set of systems, it's rare for any one not to have advantages over the others in some respects, but that doesn't stop one being more powerful in more respects, or overall. PS2 had some advantages over Xbox, but on balance that certainly didn't make it more powerful.

The original PS2 did many times the T-FLOPs of PC's at the time, but it was still creamed by a 2 year old Celeron with more bandwitdh and a superior GPU.

You're comparing a CPU to a CPU+GPU combined? Also, PS2 wasn't doing Tflops

2. We don't know who will have the most bandwidth. PS3 has a slightly higher transfer rate, but the 360's EDRAM will remove a lot of data from the main bus. .

True on the latter point (eDram), but on the former, PS3 has over twice the bandwidth to main memory, not "slightly higher".

Anyway, this thread isn't about power debates - again, there's no conclusion to reach yet, of course. I'm just talking in terms of probabilities.

 

[Acert93]

The only reason to push the envelope on design would be to make it competitive against Sony. If they thought performance didn't matter and it was only important to get out of the gate first, they would have designed a less-powerful box, since it would save costs. So it seems like they were more interested in getting in the ballpark, not necessarily in hitting more home runs or hitting longer HRs.

See, I do not get this. What could have MS *realistically* done better?

GPU: R500 w/ eDRAM is 332M transistors. That is 10% larger than the PS3 GPU and over 100M transistors larger than the largest GPU on the consumer market.

Memory: 512MB of GDDR3 memory. It was not even 3 months ago when everyone had pretty much concluded we would only see 256MB of memory.

CPU: 165M transistors, 3.2GHz, 1MB L2 cache, 3 cores, beefed up VMX units, and 6 total HW threads. 115GFLOPs of performance. Custom L2 streaming to the GPU for procedural synthesis. Outside of CELL, name one CPU that can compare, for game development needs, as well or better overall? Maybe they could have added a 4th core, but you are looking at a ton of problems doing that. With 4 cores you will need more cache to keep all 4 processors fed, you have more power consumption, more heat, more die realestate thus higher cost, lower yields, and possibly due to all of this lower frequencies. I believe the original goal was "3.5GHz+" so the machine is already running almost 1GHz below the goal spec. Lets say a 4 core could only run at 2.8GHz. In my opinion fewer faster cores is better than more slower cores (all things being even).

HDD: It was long rumored MS would not include a HDD due to the cost of the HDD unit. It looks like Sony is not including a HDD standard. If MS was cutting corners they would have just bundled the HDD with Live Gold. That way every Live customer who plays online (see below) would have the necessary HDD, and their first years payment defers some of the HDD cost. And MS would not get stuck with the bill for those who may not be online (which I must say wont be many at first considering most early adopters have broadband seeing that almost 40M Americans and 215M worldwide will have broadband by the time the Xbox 360 launches).

Live: Free Live on weekends + a ton of Live features free all the time. If the bottom line was really important they could have sold all Live features, even at a small rate. And they could have deferred the HDD cost to the Gold package.

Cordless Controllers: The reports are that the new Xbox controller is on par or better than the Wavebird. The Wavebird is the best controller I have ever used, even better than my Pelican Chamileon PS2 controller. Having a wireless controller standard is pretty great. And offering USB 2.0 slots for wires controllers that will work on the PC is really nice too.

MS is proclaiming 3 pillars: Hardware, Software, and Services. Making a console is never about making the best hardware (see: M2, 3DO). It really is a balance of HW that maximizes software potential.

Are there areas MS maybe cut corners? Yes.

WiFi: Definately a cut corner. Does it matter? I do not think so. I know dozens of people with consoles and yet I am the only one with a router, let alone a wireless router :? And I do not plan to use wireless (definately hard wired for me... less latency). SOME people will be ticked it was not standard. But from a perspective of how many consumers will use it vs. cost, having a $20-$30 addon seems like a fair compromise. I know I do not want to pay for wireless when an ethernet cable has the same functionality.

HD optical drive: New formats, no established market leader, movie media not being sold in stores yet, etc... Good move for Sony to include it, I see no reason for MS to in 2005. In 2006/2007 including a HD optical drive, larger HDD, and digital outs as a new SKU to compete with the Sony "PS3: Entertainment Device" seems like a logical move. A cheaper game-focused device and a more expensive game+movie device. A 12x DVD + HDD caching will be faster than a 1x BR + No HDD for loadtimes anyhow (because if MS included a BR drive I doubt it would have been faster than 1x). Since the 12x DVD drive is a bit faster, and I hate load times, I will have to say at best this is a toss up. Some will hate it, others will like it.

Analog out: Definately a cut corner here. Was this a miscalculation on the market growth? I think that is part of it. Is it a calculated move for a 2nd game+HD movie SKU? Yes. Basically those who buy an Xbox 360 in 2005 are buying it for games. If you are a big movie buff and want HD movies my guess is they will have a SKU with digital out come fall 2006. I do not see this as a cost cutting move; more of a market positioning move.

No 1080p: This could relate to the new SKU, but honestly, I think 1080p is new enough that MS did not plan on it being important and definately did not expect Sony to trumpet it. So far 720p/1080i define HD TV. While 1080p will no doubt find its uses it will be more important to see the support it gets throughout the industry. As for games, even Sony did not say it was the minimum standard, so we will have to wait to see how many PS3 games support 1080p at solid framerates.

Media Ports: Would not mention this less Sony has a ton. If Sony had no included every port known to man this would not even be an issue of "cutting corners". As it is I expect MS to offer a basic multi format card reader. Again an extra most wont use, but nice to have. Really has little to do with the power of the system.

No point discussing dual HDMI outs, 3 Gbit ethernet ports, etc... nice features (dual HDMI could be nice for 4 player splitscreen on 2 displays), but they are not "performance" features and MS not including them does not make me think they were cutting corners. I actually thought many many years ago that having dual displays would be cool on a game console. I hope it gets supported more than the direct link was. But realistically, without knowing how these will be used (and since they have not been used before and I do not remember anyone saying, "I want 3Gbit ethernet ports!" I think they are more of a side issue. More does not always mean better/more powerful).

So after looking at the entire system I can think of only 2 areas where I would make real changes:

1. CPU. Out of Order Execution would have been nice. More cache. Both of those would have increased the die size a ton and also produced more heat and thus lowered frequencies. It would have been nice if MS could have been closer in the FP performance, but alas, you cannot have everything. I do not see a better alternative on the market for MS, but when your competitors key note is CPU performance it is nice to be as close as possible in all metrics. MS does well in many areas (more PPC cores, more L2 cache) but falls short of the hot topic measurement (this generations polygons, MIPs, etc.): Floating Point. 165M transistors is a pretty big chip--especially for a $300 console. As gamers we can always whine "It is not enough". But what would have? Where did MS cut corners on the CPU? Maybe someone can link me to a CPU that would be affordable, but better performing for games? Until I find that chip I cannot see how MS cut corners here.

2. Memory bandwidth. 23GB/s is not too shabby for a closed box system. We all know that memory bandwidth has lagged behind processing performance, and it looks to continue. MS scored big time by isolating the backbuffer on the eDRAM (and getting high quality IQ at the same time, like 4x AA). But it has always been my belief that the new consoles will be memory limited: both size and speed. 256bit memory would have been nice, but the cost and difficulty appear to have been too difficult for both MS and Sony.

So memory bandwidth is a corner that was cut, to a degree, by both--yet both found work arounds (MS: eDRAM, Sony: XDR can be accessed by the RSX).


Looking at what is on the market in 2005 and what is coming in 2006, I do not know how MS could have got more performance realistically out of the Xbox 360. I think a lot of people are overblowing the "they are saving money" angle. Allard has already outlined their cost saving agenda: process shrinks to 65nm and then 45nm, solidifying the ICs and reducing the number of DRAM modules.

The accusations that MS was only trying to be in the ball park are really misplaced IMO. They did not know every detail of the PS3 three years ago when they started to form hard targets. Simiarly, MS cut no corners on the GPU. It is larger than the largest $500 consumer GPU on the market and larger than the PS3 GPU. I do not hear people saying Sony has cut corners. MS went for a full 512MB of memory (the leak said 256MB+) and they included a HDD standard.

MS could have gone with 256MB of memory, no HDD standard, a dual core CPU, a smaller GPU (332M transistors is huge), and cut a lot of corners. But they did not.

The Xbox 360 is clearly a design that IS EXPENSIVE. But it is designed intelligently. It is designed to reduce in cost in process shrinks. The GPU is in 2 parts right now: 2 smaller parts should improve yields. The CPU is almost 100M transistors smaller than the DD2 CELL (250M). eDRAM supposedly takes up less die space than logic... so it appears MS has a significant edge on Sony on when they will be able to get the GPU and CPU on one chip.

Same with the memory. 512MB of homogenious memory vs. Sony's 512MB of hetergenious memory. Sony has more expensive memory (MS exchanged some logic for bandwidth).

Anyhow, I will end my rant on this. I just do not see how MS has cut a ton of corners to be "cheap" or that they never aimed to compete in performance. A 332M transistor GPU is more than competing. 512MB of memory and a HDD standard is competing and expensive. I think a lot of people have misunderstood MS's goal to be more affordable this generation. That did not come at the cost of HW performance. MS made better IP based deals this gen and has a very reasonable scaling plan for their chips. Their cost orientation has less to do with the initial cost of the system but more with the long term profitability.

 

[Shifty Geezer]

I agree with Acert here completely. What more could people ask for? XB360 is a monster powerhouse. It's only KK's crazy visions that have resulted in PS3's CPU having more input and so resulting in a math-friendly machine that looks awesome on paper. No 'normal' company would have spent so much on a console!

A more realistic console would be maybe an Athlon 64, top end single GPU, and a bit of customization to make it less PC like, give it a bit of console edge (eg. limit top resolutions and provide more effects).

KK's crazy hype for his dreams for PS3 drove MS into a crazy-power machine and power-for-price, I don't think gamers have ever had it so good.

 

[Acert93]

I agree with Acert here completely. What more could people ask for? XB360 is a monster powerhouse. It's only KK's crazy visions that have resulted in PS3's CPU having more input and so resulting in a math-friendly machine that looks awesome on paper. No 'normal' company would have spent so much on a console!

A more realistic console would be maybe an Athlon 64, top end single GPU, and a bit of customization to make it less PC like, give it a bit of console edge (eg. limit top resolutions and provide more effects).

KK's crazy hype for his dreams for PS3 drove MS into a crazy-power machine and power-for-price, I don't think gamers have ever had it so good.

Thanks for summing it up well Shifty

Gamers have it really good this gen. I would have never imagined that consoles would, on paper, have this type of performance all around. Top end GPU (performance and speed), a ton of memory, very game-oriented CPUs.

Like Shifty notes, if we take the CELL out of the PS3 there is no way to accuse MS of cutting corners. Ken Kutaragi's CELL Dream really is the difference, at least on paper. And CELL may very well be the difference in the long run (I see it as an awesome power move for PS4). But for this gen, MS has really made an impressive machine. Sony and MS gamers are getting the best deals ever on this gens consoles. More functionality, more power, more services. And hopefully better software!

 

[pc999]

Nice posts Acert69/Shifty Geezer

BTW R500 looks like a beast at GPGPU, and I can live with "only" 250Mpolys/second (~3,5-4M per frame@60FPS) and no trivial shaders.

 

[Dave Baumann]

250MPoly's is the tesselator output, the vertex shader throughput should be many times that (although lord knows what the setup engine can do).

 

[Titanio]

I don't think anyone doubts that X360 is a wonderful machine for the launch window it's being aimed at.

I still think they should have aimed for a 2006 release, however. This is a different point.

Xbox is quite comfortably the most powerful system on the block this gen, and could easily see itself through another holiday and 2006. There's a lot more technical potential there, and it's just being waved goodbye.

From a European perspective, Xbox will have only seen three holidays. It's felt like a very rushed ride to me. I'm not sure why we need another one now, especially when you consider how much more powerful X360 could be if it were coming out in Nov 2006 vs Nov 2005. With another 12 months and a budget to match the original's, where would X360 be technically? That's my question. It'd also have another 6 months over PS3 to play with in terms of technology, and retain the brand's technical preeminence.

Who here couldn't live without a next-gen Xbox in 2005? Be honest. I know new tech looks nice, but I'm sure most would be happy, indeed happier, if MS had announced a 2006 machine, expecially given that the original remains still the youngest hardware.

I understand the business case and reasoning for launching in 2005, but from my own perspective and that, I'm sure, of many other customers - and Xbox fans who place a premium on hardware quality/performance and superiority - a (much) better machine in 2006 would be a lot more compelling. This 2005 launch really isn't being driven by the product, which IMO is unfortunate.

 

[pc999]

250MPoly's is the tesselator output, the vertex shader throughput should be many times that (although lord knows what the setup engine can do).

By the tesselator output you mean processural work(?) from the CPU?(or arttechnica article confused me?)

If so they can fill (very well) the screen only with the CPU .

Anyway (which has what I meant), half of the power from R500 in GPGPU work should be a big jump in some games with intense maths, Right?

Thank you very much.

 

[Dave Baumann]

I've not read the CPU article, but Xenos has a fixed function tesselation unit that can take in simple primitives and subdivide them - it spits out 1 vert per clock, the CPU can also pass geometry directly to Xenos for it to use as input (general or to the tesselator). However, if we consider something such as X800 XT has a geometry throughput peak of 750M triangles per second, when we consider that (for geometry work alone) there are 8 times the ALU's available on Xenos for vertex processing that would give us about 6000M Triangles/s throughput through the shaders - it would of course be setup limited (now, think why Z only setup passes are good for this architecture!).

 

[pc999]

Thanks you very much for the clarification, I was thinking in the CPU.

 

[DuckThor Evil]

Who here couldn't live without a next-gen Xbox in 2005? Be honest. I know new tech looks nice, but I'm sure most would be happy, indeed happier, if MS had announced a 2006 machine, expecially given that the original remains still the youngest hardware.

I understand the business case and reasoning for launching in 2005, but from my own perspective and that, I'm sure, of many other customers - and Xbox fans who place a premium on hardware quality/performance and superiority - a (much) better machine in 2006 would be a lot more compelling. This 2005 launch really isn't being driven by the product, which IMO is unfortunate.

I want it now!, my hdtv want's something to play around with. You can always get better tech "next year" and it's hard to say how much it would have improved in a year, maybe not that much. Right now it looks to me atleast that X360 is fairly competitive with PS3 even though PS3 comes in 2006 and it's hard to see how MS could have made much better machine than PS3 in 2006.

 

[wco81]

I don't think anyone doubts that X360 is a wonderful machine for the launch window it's being aimed at.

I still think they should have aimed for a 2006 release, however. This is a different point.

Xbox is quite comfortably the most powerful system on the block this gen, and could easily see itself through another holiday and 2006. There's a lot more technical potential there, and it's just being waved goodbye.

From a European perspective, Xbox will have only seen three holidays. It's felt like a very rushed ride to me. I'm not sure why we need another one now, especially when you consider how much more powerful X360 could be if it were coming out in Nov 2006 vs Nov 2005. With another 12 months and a budget to match the original's, where would X360 be technically? That's my question. It'd also have another 6 months over PS3 to play with in terms of technology, and retain the brand's technical preeminence.

Who here couldn't live without a next-gen Xbox in 2005? Be honest. I know new tech looks nice, but I'm sure most would be happy, indeed happier, if MS had announced a 2006 machine, expecially given that the original remains still the youngest hardware.

I understand the business case and reasoning for launching in 2005, but from my own perspective and that, I'm sure, of many other customers - and Xbox fans who place a premium on hardware quality/performance and superiority - a (much) better machine in 2006 would be a lot more compelling. This 2005 launch really isn't being driven by the product, which IMO is unfortunate.

That was the other part of the point I was trying to make. Making the 2005 Holidays seems to be a higher priority than making the most powerful design. If they waited to 2006, they would have even more capable design.

Not saying the X360 won't be capable, because they are giving you more than anyone ever did before for $300. But people are saying it's roughly 15 times as powerful as the Xbox. You wonder what it might have been if they waited to 2006.

 

[Acert93]

I've not read the CPU article, but Xenos has a fixed function tesselation unit that can take in simple primitives and subdivide them - it spits out 1 vert per clock, the CPU can also pass geometry directly to Xenos for it to use as input (general or to the tesselator). However, if we consider something such as X800 XT has a geometry throughput peak of 750M triangles per second, when we consider that (for geometry work alone) there are 8 times the ALU's available on Xenos for vertex processing that would give us about 6000M Triangles/s throughput through the shaders - it would of course be setup limited (now, think why Z only setup passes are good for this architecture!).

Good info Dave! I am drooling over here while I wait for your Xenos article!

Question: Where did MS/ATI get the 500M Poly/s number in the official stats then? If the tesselator does 250M, where did the 500M number come form? Is that the setup limitations? Vertex Shader ALU peak? (I thougtht the 500M number was, based on the leak, a realistic figure of what to expect w/ non trivial shaders). Any clarification would be great if you have that info

 

[Titanio]

I want it now!, my hdtv want's something to play around with. You can always get better tech "next year" and it's hard to say how much it would have improved in a year, maybe not that much. Right now it looks to me atleast that X360 is fairly competitive with PS3 even though PS3 comes in 2006 and it's hard to see how MS could have made much better machine than PS3 in 2006.

With another 6 months or more over PS3, assuming a Nov 06 launch in the US, X360 could easily have been more powerful. And sigificantly more powerful than it will be now.

Of course, you can always "wait another 6 months" - fall into the old PC-upgrader trap - but for a machine of Xbox's age, '06 makes far more sense for its successor's debut than '05, from a technical/product point of view.

If the tesselator does 250M, where did the 500M number come form? Is that the setup limitations?

I believe so, one per clock.

 

[therealskywolf]

I don't think anyone doubts that X360 is a wonderful machine for the launch window it's being aimed at.

I still think they should have aimed for a 2006 release, however. This is a different point.

Xbox is quite comfortably the most powerful system on the block this gen, and could easily see itself through another holiday and 2006. There's a lot more technical potential there, and it's just being waved goodbye.

From a European perspective, Xbox will have only seen three holidays. It's felt like a very rushed ride to me. I'm not sure why we need another one now, especially when you consider how much more powerful X360 could be if it were coming out in Nov 2006 vs Nov 2005. With another 12 months and a budget to match the original's, where would X360 be technically? That's my question. It'd also have another 6 months over PS3 to play with in terms of technology, and retain the brand's technical preeminence.

Who here couldn't live without a next-gen Xbox in 2005? Be honest. I know new tech looks nice, but I'm sure most would be happy, indeed happier, if MS had announced a 2006 machine, expecially given that the original remains still the youngest hardware.

I understand the business case and reasoning for launching in 2005, but from my own perspective and that, I'm sure, of many other customers - and Xbox fans who place a premium on hardware quality/performance and superiority - a (much) better machine in 2006 would be a lot more compelling. This 2005 launch really isn't being driven by the product, which IMO is unfortunate.

That was the other part of the point I was trying to make. Making the 2005 Holidays seems to be a higher priority than making the most powerful design. If they waited to 2006, they would have even more capable design.

Not saying the X360 won't be capable, because they are giving you more than anyone ever did before for $300. But people are saying it's roughly 15 times as powerful as the Xbox. You wonder what it might have been if they waited to 2006.

Why not wait for Fall 2007 while we are at it, that way the Xbox 360 would have exactly the same advantage over the Ps3 has Xbox had over the PS2.

And people talk as if suddenlly Xbox 360 is coming early and the only true next gen system is the PS3.....great, so 6 months tops later we are set?

And please....Cell is wonderfull, like J ALLArd, is wonderfull for the tech geek that loves to run trought specifications and learn how it works, but for Real life performance, like Games, i doubt that we will se a significant difference between the 2 systems.

 

[DuckThor Evil]

I want it now!, my hdtv want's something to play around with. You can always get better tech "next year" and it's hard to say how much it would have improved in a year, maybe not that much. Right now it looks to me atleast that X360 is fairly competitive with PS3 even though PS3 comes in 2006 and it's hard to see how MS could have made much better machine than PS3 in 2006.

With another 6 months or more over PS3, assuming a Nov 06 launch in the US, X360 could easily have been more powerful. And sigificantly more powerful than it will be now.

Are you sure, and by how much, I agree that you could make more powerful machine, but maybe the difference wouldn't be as big as you think.

 

[Titanio]

Why not wait for Fall 2007 while we are at it, that way the Xbox 360 would have exactly the same advantage over the Ps3 has Xbox had over the PS2.

I addressed this point somewhat above. 2007 would be as inappropriate as 2005. 2006 is a far more reasonable timeframe than either. You must recognise that there could still be at least another year in Xbox if MS wasn't killing it off now.

And please....Cell is wonderfull, like J ALLArd, is wonderfull for the tech geek that loves to run trought specifications and learn how it works, but for Real life performance, like Games, i doubt that we will se a significant difference between the 2 systems.

There'd be a significant difference between the X360 of Nov 05 and the X360 of Nov '06. And we've little idea of the difference between X360 and PS3 yet..in certain areas it could be quite significant, in others less so.

Are you sure, and by how much, I agree that you could make more powerful machine, but maybe the difference wouldn't be as big as you think.

Of course I'm sure, there'd be something quite wrong if a machine coming out 6 or 12 months later wasn't more powerful. Launching in Nov '06 also would possibly raise the prospect of working with a more advanced manufacturing process too...which would reap large performance benefits beyond the norm. Beyond performance there'd also be features, like a (or THE) next-gen DVD format.

 

[wco81]

Well whether 2005 is too soon or not will be judged by history. If it gains them the big market share lead they hope for, then it's obviously not too soon and getting the best possible tech specs. for the next 5 years is moot.

My HDTV is ready for a new console too.

But my point was that MS had this hard goal for this Xmas. Gates talks about it in an interview, that they can't move Xmas so they have to cut off any development work to meet this hard target.

So besides the usual cost constraints in the design of a console, you have this hard date constraint.

One other design goal which apparently wasn't as high a priority was backwards compatibility. They waited until the last minute to say there would be some BC.