Significance of the cancellation of Tejas
- Thread starter bbot
- Start date
I don't trust the 3 Ghz+ chip specs for a machine that's going to be sporting thee of them. Assuming IBM can even scale that high in time, it won't be cheap.
The xbox was already pushing the limits of what's acceptable for a console with regards to space, noise, and airflow, and that was on a single 733Mhz chip.
There would have to be some extreme engineering to cool off three monster chips running at five times the clock speed. Unless they pull some kind of heat-pipe assembly that attaches directly to a massive heatplate external to the chassis, the insides are going to cook. Let's not forget there's going to be a next gen video core thrown in the mix as well.
The xbox was already pushing the limits of what's acceptable for a console with regards to space, noise, and airflow, and that was on a single 733Mhz chip.
There would have to be some extreme engineering to cool off three monster chips running at five times the clock speed. Unless they pull some kind of heat-pipe assembly that attaches directly to a massive heatplate external to the chassis, the insides are going to cook. Let's not forget there's going to be a next gen video core thrown in the mix as well.
Wasn't it because "Ultra" sounds really funny (as in weird) when pronounced in Japanese? That's what I heard anyway. 
And 3dilettante, the CPU in the box is a MOBILE P3; it doesn't get particulary hot, as evidenced by the small device used to cool it. MS would have been much better off with a cooling device like what was used in the Cube for example; one heatsink for both chips and a fan dragging air across it.
Now there's like 3 fans in the Box, instead of just one, and the circuit board is just uber big for no reason. If you look at what they cram into those small form factor PC mobos, you see immediately exactly how badly designed the box is. Using it as some kind of metric for the limits of size, noise is only going to define the WORST-CASE situation; all the other hardware platforms do much better than MS managed in their first attempt.
And 3dilettante, the CPU in the box is a MOBILE P3; it doesn't get particulary hot, as evidenced by the small device used to cool it. MS would have been much better off with a cooling device like what was used in the Cube for example; one heatsink for both chips and a fan dragging air across it.
Now there's like 3 fans in the Box, instead of just one, and the circuit board is just uber big for no reason. If you look at what they cram into those small form factor PC mobos, you see immediately exactly how badly designed the box is. Using it as some kind of metric for the limits of size, noise is only going to define the WORST-CASE situation; all the other hardware platforms do much better than MS managed in their first attempt.
Guden Oden said:
The fact that the processor was a low-power variant of the P3 doesn't help the case for a triple-core chip running at five times the clock rate.
The reason why the x-box had a louder cooling solution was because MS knew that they couldn't guarantee the unit would get good airflow. If the Xbox2 has to exist within the same environment as the original, then they're going to need a wind tunnel integrated into it.
Something's amiss with the suggested set-up.
Current powerpc 970 processors run at 42 watts at 1.8GHz. Roughly doubling the clock speed and tripling the number of cores is not going to leave a cool-running chip.
I would argue Pentium-M (banias/dothan) is, in fact, a P3 descendant - just like P3 is a descendant of PII, which is itself based on PPro (which is the original "P6-Core"). Granted, it does have some differences - but it cannot deny its heritage. The differences are actually rather small IMHO. Just look at the block diagrams, and it's pretty obvious. Yes, it has a completely different bus interface (which, btw, is the most important reason its performance is higher than that of a PIII at the same frequency), and it has some improvements (power-consumption related tweaks, SSE2, other minor tweaks), but overall, this IS a P6 core (as opposed to the P4, which really is completely different).Saem said:
mczak
(btw the K8 of course is also not very different from a K7. Same story as above, lots of improvements (new i/o interface (HT, on-die memory controller), SSE2, x86_64), but you can clearly see that it resembles the K7)
Intel's Big Shift After Hitting Technical Wall
Staight out of their mouth:
http://www.nytimes.com/2004/05/17/business/17intel.html
Staight out of their mouth:
http://www.nytimes.com/2004/05/17/business/17intel.html
JF_Aidan_Pryde
Regular
How can scalaing be 'dead' and yet Cell reach 4GHz?
accidentalsuccess
Newcomer
I believe the scaling they are talking about is eg. you won't see the cell at 1x 8GHz chip (or the P5 at say 8 GHz), but you'll see 2x 4 GHz or more likely 4x4GHz chips.
I totoally agree w/ lithiography being basically dead b/c of current leaks, etc. I'm quite frankly very impressed that the industry has 90 nm as a realistic cutoff. . .i thought it would have been somewhere around 120 b/c of current leak and yield issues.
guess they didn't ask a chemist
I can't wait to see how we move forward from this. I really think that breaking away from the "more ticks, Scotty!" approach to processor design will help the industry immensely, as long as we don't go back to the proprietary hardware/software hell of the 70/80's.
I totoally agree w/ lithiography being basically dead b/c of current leaks, etc. I'm quite frankly very impressed that the industry has 90 nm as a realistic cutoff. . .i thought it would have been somewhere around 120 b/c of current leak and yield issues.
guess they didn't ask a chemist
I can't wait to see how we move forward from this. I really think that breaking away from the "more ticks, Scotty!" approach to processor design will help the industry immensely, as long as we don't go back to the proprietary hardware/software hell of the 70/80's.
Megadrive1988
Veteran
in August or September 1993, Project Reality was announced, and was slated to have a 100 Mhz CPU. release would be 1995. later there were rumors or reports that the CPU would be boosted to 150 Mhz. Nintendo 64 ultimately launched in Japan in June 1996 with the CPU running at about 93 Mhz. later and slower than expected.
accidentalsuccess said:
It looks like AMD saw the problems with "more ticks, Scotty!" approach and designing the original K7 core accordingly
Attually this exact point has be expanded apon by WaltC before here :
http://www.beyond3d.com/forum/viewtopic.php?p=285510&highlight=amd#285510
To clear up the confusion, there are actually many versions of xboxes. First version had a crappy dvd drive (Thompson) and 3 fans.
Newer versions have mostly good and silent dvd drives (Samsung or Philips) and only one fan. As far as noise is conserner, I'd put a modchip and then change hard drive to make the most difference.
Newer versions have mostly good and silent dvd drives (Samsung or Philips) and only one fan. As far as noise is conserner, I'd put a modchip and then change hard drive to make the most difference.
BlackAngus
Newcomer
I havent really seen anyone speculate on whether or not the XCPU in question is truly just a rejected power 5 Chip. I beleive Power 5 should run in the 3.5 ghz range and it has 4 cores per cpu. Maybe IBM are using this as a way to make the most of their power 5 output. (Perfect 4 core power 5's for computing, defect 3 core power 5's for Xbox2)
Only thing wrong with that is; I can't see power 5 being high enough of a volume to have enough chips to make for the xbox2 cpu. Then again maybe power 5 has a very poor yield.
Any thoughts?
Only thing wrong with that is; I can't see power 5 being high enough of a volume to have enough chips to make for the xbox2 cpu. Then again maybe power 5 has a very poor yield.
Any thoughts?
