What kind of specs would the 360 has if it was released in 2006

[3dilettante]

AMD = ATI, but anyhow ATI (and I think even nvidia) are already producing graphics chips on an 80nm process.

Chips on TSMC's 80nm process were brought to market in September 2006 with Nvidia's smaller core mobile design.

Xenos is definitely not a small core.

It would have been a dangerous gamble to bet on 80nm years ago. I'd be that if MS had gone 80 nm, we wouldn't see the 360 until 2007.

edit:

ATI was even later than Nvidia, based on a well-established design no less. It would have been really down to the wire for a 2006 release of a new GPU architecture in any quantity.

 

[mboeller]

Higher cpu clock speed. Possibly OOE. Had they implemented OOE, I'd expect Xenon to be showing higher real world performance then Cell in most situations, but then it's likely IBM could have used the same design expertise to give at least the main core of Cell OOE as well. Xenon and Cell lack OOE because it wasn't a priority for IBM, not because Microsoft wouldn't have liked it within their $100 million design budget. IF OOE had made it in, I think you would have been looking at a processor under 3ghz due to the additional heat concerns and design complexity. It may have also had to drop down to dual core to save die space. IMO, it'd still be superior design since it'd be far easier (possible) to get better results out of, though only if they could do it without ballooning the design budget. I'd expect 50%-100% more performance per core even at the lower clock which would more than offset the loss of a difficult to use additional core.

I'm not sure about that. Look at the PWREfficient designs I mentioned in my "proposal". They are already dual-core with 2GHz and OOE but are small* and use only 7Watt per core
Therefore IMHO a 60-70Watt design could have been a quad-core with 3-3.2GHz and maybe one additional Altivec/VMX128 unit per core.

*= I'm not sure how big this cores really are but they are intended for the embedded market too so size matters.


Link: http://www.pasemi.com/

 

[Gubbi]

*= I'm not sure how big this cores really are but they are intended for the embedded market too so size matters.

They are 10 mm^2 in 65nm proces tech, L1 caches not included, so maybe 11mm^2 in total judging from the die floorplan. So they'd be around 20 mm^2 in 90nm. The core is listed to use a maximum of 7W @ 2GHz in 65nm. It can run at speeds up to 2.5GHz

They'd be clocked lower in 90nm proces tech, but I still think they would have spanked the current cores in the 360 being wider superscalar (3 instructions) and OOO.

They could've packed 4 such cores and 2MB of cache into the 173 mm^2 the current 3-core CPU takes up, have higher performance *and* lower power consumption.

Cheers

 

[LunchBox]

If the X360 launched on fall of 2006.

it would be the same spec but one SKU

the premium @ $299.99

 

[max-pain]

Better cooling, quiter DVD drive, bigger HDD, bigger memory card, HDMI output, and maybe a HD DVD version.

The basic spec would be the same...

 

[ADEX]

I posted a lengthy reply to this but it apparently got lost in the ether on the way to the server. Grrrrrr

I think likely improvements if the 360 had been designed to launch a year later...
Higher cpu clock speed. Possibly OOE. Had they implemented OOE, I'd expect Xenon to be showing higher real world performance then Cell in most situations, but then it's likely IBM could have used the same design expertise to give at least the main core of Cell OOE as well. Xenon and Cell lack OOE because it wasn't a priority for IBM, not because Microsoft wouldn't have liked it within their $100 million design budget.

A shorter version:
Not using OOO was MS's choice, if they really wanted it it would be there but yes clock would be lower and die size bigger.
Alternatively they could have gone for the existing 970MP which would have been smaller, cheaper and at 2.2GHz - 2.3GHz will have similar power numbers.

OOO only boosts the performance of *some* types of code, the sort of thing Xenon was designed for is high compute vector operations on predictable streams of data, OOO wont have much (if any) effect on that sort of code and the clock drop it'll introduce will hurt it.

MS wanted FLOPS and lots of them, by choosing to go for a higher clocked in-order design they get roughly twice what the 970MP would have given (most likely more as the Xenon vector register files are bigger).

Cell and Xenon are a lot more similar than they might appear - they are both designed to operate in exactly the same way. There's a good reason for that.

 

[blakjedi]

...Not using OOO was MS's choice...

I honestly thought it was IBM who convinced them that in-order was the way to go due to expense, time, heat and power...

 

[ADEX]

I honestly thought it was IBM who convinced them that in-order was the way to go due to expense, time, heat and power...

Probably was, but the end choice of what to go for would have been MS's.

 

[Fox5]

Probably was, but the end choice of what to go for would have been MS's.

If the alternative was delaying the system or spending another $500 million on development, I wouldn't consider that much of a choice. Microsoft did originally want OOE, whether they just plain couldn't get it within their time/money budget or whether any OOE design that IBM could have feasibly popped out would have just been plain inferior for games, we don't know.

 

[ADEX]

If the alternative was delaying the system or spending another $500 million on development, I wouldn't consider that much of a choice. Microsoft did originally want OOE, whether they just plain couldn't get it within their time/money budget or whether any OOE design that IBM could have feasibly popped out would have just been plain inferior for games, we don't know.

Adding OOO wouldn't have added much (if anything) to the development cost - it's an extra chunk in the middle of the processor which IBM have plenty of experience designing (40 years).

Every processor design makes a set of trade-offs, including or not OOO hardware is one of those trade offs. They traded it for an extra core and a higher clock speed, they didn't do that based on a guess, it was based on research.

If you can find it IEEE Micro ran an article on the 360 by the system's architects (both Microsoft people).

 

[ImaginaryIndustryInsider]

When one game has three zombies in one tiny room, and another has fifteen in an arena, you might suspect something is different. Sure, they may look the same, but something was obviously changed.

I don't think that example is all that accurate, because we're more likely to see something like 20 vs 40, or in a "Dead Rising-like" game, 150 vs 250...especially when you're talking about those kind of numbers, will the average consumer even care?

 

[Fox5]

Adding OOO wouldn't have added much (if anything) to the development cost - it's an extra chunk in the middle of the processor which IBM have plenty of experience designing (40 years).

Every processor design makes a set of trade-offs, including or not OOO hardware is one of those trade offs. They traded it for an extra core and a higher clock speed, they didn't do that based on a guess, it was based on research.

If you can find it IEEE Micro ran an article on the 360 by the system's architects (both Microsoft people).

I seem to remember that ebook talking about the 360 saying that MS wanted OOO, but it couldn't be put in within the timeframe MS wanted. Additionally, a longer dev timeframe would have meant more money put into the project, though I don't know if another $100 million would have been that big of a deal to MS, but significantly delayed dev time certainly was.

 

[assen]

Have the HDD standard, and keep that $50 mill-stone tied to their feet all the way to 2010? No way.

There's only thing I would add to the Core version besides the obvious "more of <insert your favorite number>", for which we don't know how feasible it would be (e.g. more ALUs in Xenos, or more L2 cache in Xenon etc.), and it's the only thing I like about Wii's hardware: 512 MB of flash memory.

 

[3dilettante]

I don't think that example is all that accurate, because we're more likely to see something like 20 vs 40, or in a "Dead Rising-like" game, 150 vs 250...especially when you're talking about those kind of numbers, will the average consumer even care?

Or 10 vs.20 more complicated zombies, or 5 vs. 10 really smart zombies.

 

[Gubbi]

really smart zombies.

erh.. ?

Cheers

 

[Fox5]

erh.. ?

Cheers

The possibilities... :drool

I'd like a pint of brains with me tea, old chap.

 

[Shifty Geezer]

erh.. ?

College educated zombies mumbling high-level maths and English Lit as they shamble along. Needs loads of CPU cycles to produce zombies like that, which is why to date they've been mostly school drop-outs with the IQ of a glass of water and diction that goes 'Urrrrrrhhhhhh.' When we get Smart Zombies, they'll chase after the player with cries of 'Behold my sinister shamblings, as in peripatetic pursuit my festering kinsfolk and I set upon you with grizzly rage.'

 

[Crossbar]

College educated zombies mumbling high-level maths and English Lit as they shamble along. Needs loads of CPU cycles to produce zombies like that, which is why to date they've been mostly school drop-outs with the IQ of a glass of water and diction that goes 'Urrrrrrhhhhhh.' When we get Smart Zombies, they'll chase after the player with cries of 'Behold my sinister shamblings, as in peripatetic pursuit my festering kinsfolk and I set upon you with grizzly rage.'

Or that kind of Zombies that write posts on computer forums. I doubt it would be a fun game though.

 

[NERO]

What kind of specss would the 360 have if it was released in 2006? Well, what kind of specs would an Atari Jaguar, NES, Coleco Vision would have if they were released in 2006? Maybe the NES would have an R600 that'd allow the Adventures of LoLo to truly shine. GNARLSSSS!!!!!!!

P.S. Those smart zombies might have answers to these questions.

 

[ImaginaryIndustryInsider]

Or 10 vs.20 more complicated zombies, or 5 vs. 10 really smart zombies.

Are we talking about High School drop-out Zombies vs just Community College/Technical School educated Zombies, or full-fledged Ivy League educated Zombies working on their doctorates in Hebrew Studies?