Tejas

[HardWareBoss]

6 Gig CPU 1.2 FSB. How fast can we go. How small 45 Nano? Prescott 90.

 

[imgod2u]

Teja is suppose to scale to 9ish GHz before Nehalem takes over.

"We are Microsoft, resistance is futile." - Bill Gates, 2015.

 

[juin]

90 NM at start supposte so scale down to 0.65.


Tja have rumor to include IA-64.Hard to believe

[-peep-] french

 

[imgod2u]

I wouldn't say so. Teja is suppose to be what, 150-200 million transistors? I'd say that that was the point in which the complexity of an ISA implementation would become trivial and it'd be pretty easy to include a front-end for multiple ISA's while maintaining a complex back-end. Plus Teja is suppose to be released at about the same time Intel moves to its BBUL packaging, which is almost perfect for a dual-core design. Perhaps we'll get an x86 core and an IA-64 core on one processor.

"We are Microsoft, resistance is futile." - Bill Gates, 2015.

 

[Mephistopheles]

I´d say that it´s highly likely that Tejas has some sort of 64-bit technology in it, and it´s not at all hard to believe. There are even rumors about prescott - which will compete directly with the hammer - having 64-bit extensions disabled on-chip (much like HT was disabled in 2.8Ghz and less CPUS) as a safeguard against the thread posed by Hammer. The whole thing sounds quite interesting, doesn´t it?...

I don´t think that Intel would just sit idly while AMD readies 64-bit tech to compete in desktops. I find it more likely that intel is waiting to see if hammer is or is not a disappointment, readying its counterpart for the worst case scenario. Either Nehalem, Tejas or Prescott should have something to do with 64-bit tech, I´d say. Intel should be aiming for flexibility - they´ve certainly got resources for it. Besides, Athlon-64 has been severely delayed... Its window for maximum impact lost...

And to think that, according to AMD, the whole thing about the 64-bit extensions could be achieved by no more than a 2% increase in die size...

 

[eden]

Though many will preach once again how 64-bit is uneeded for Intel any soon, just like it's pointless for desktop Athlon 64 systems, seems like even Valve has found one its most succesful product to actually benefit greatly from the simple 64-bit switch.
On ActionTrip's news, Valve announced that the Opteron 64-bit transition allowed Counterstrike servers to go up 30% clock per clock, which accounts as a great increase in stability and reliability as well as performance for the online gaming. Seems to me that Opteron may really hit the jackpot if game servers are THAT sensitive to x86-64 transitions:
<A HREF="http://www.actiontrip.com/" target="_new">http://www.actiontrip.com/</A>
Scroll down, or Ctrl-F and type in Counter and you'll get to it.

--
This post is brought to you by Eden, on a Via Eden, in the garden of Eden.

 

[Mephistopheles]

30%, clock per clock, is actually a great advantage indeed! (although the initial clock rates won´t be over 2Ghz, if I´m right, will they?) I´m curious to see the real benefits. Opteron will be here soon, much sooner than Athlon 64. The way I see it, Opteron is set to have a good launch, and many people will recognize its advantages. It won´t be stellar, though. On the other hand, I don´t see a clear picture for Athlon 64. If indeed prescott has the extensions, then it can be rather devastating: just picture a 3.6Ghz, 800Mhz FSB, doubled caches, Ht-enhanced, prescott with improved architecture and 64-bit extensions on a PAT-canterwood with dual DDR400... And even without the extensions, prescott is looking quite good on paper.

Then again, Athlon 64 has been "looking good on paper" for years now... It´s going to be an interesting year for CPUs, I guess!

 

[alpha_03]

64bit http://www.intel.com/pressroom/archive/releases/sp072700.htm

 

[eden]

Err, no, Prescott is definitly not gonna have any 64-bit extensions.
What it WILL have is LaGrande, the Palladium initiative.

--
This post is brought to you by Eden, on a Via Eden, in the garden of Eden.

 

[Mephistopheles]

Oh yes, the Palladium initiative. Hammer will have that too, right? I hear it´s causing quite a controversy because of security implications, right? Haven´t had the opportunity nor the time to read on the subject, though...

OK, I know Prescott has no 64-bit-extensions officially, but maybe they´re there and are disabled. I can´t remember anyone saying about disabled-HT on Willamette or Northwood chips, but it WAS there. I´m talking about the 64-bit Yamhill implementation, a secret intel skunkworks project that supposedly leaked and is a conterpart to x86-64 from AMD, adding the 64-bit extensions to intel processors. That´s what I mean. Of course, it could turn out to be a false rumor. Still, there are a number of links on the net about the yamhill rumor.

But then, there´s a lot on other conspiracies, like UFOs, as well...
What about LaGrande, what are actually the security implications? Anyone enlighten me on that subject?

I´ll check it out later anyway...

 

[imgod2u]

x86-64 is not just a shift to 64-bit. I find this a constant mistake. Was the increase in Counterstrike due to the added 64-bit capability? Or the extra registers available in the x86-64 ISA extension? I'd say the later. x86 is very limited as far as registers are concerned and having twice as many could definitely help.

"We are Microsoft, resistance is futile." - Bill Gates, 2015.

 

[eden]

Well the only thing I can add here is that the Yamhill project, as far as I am concerned, was stated by Intel as dropped.
Maybe it is still alive, but last I heard, they stopped it.

--
This post is brought to you by Eden, on a Via Eden, in the garden of Eden.

 

[Mephistopheles]

Oh, it was actually dropped? I thought they had no statement about it at all, and I´m pretty sure that whenever anyone asked them to comment, they said it didn´t exist.

It would be sad if they had dropped it, I must say. Lost R&D money. Which makes it all the more unusual. If they did drop it and it was a secret, why say anything about it in the first place? Maybe it quite simply didn´t exist or isn´t complete for prescott yet.

Check this out: <A HREF="http://www.geek.com/news/geeknews/2002june/bch20020627015159.htm " target="_new">http://www.geek.com/news/geeknews/2002june/bch20020627015159.htm </A>
If this is true, then AMD is in trouble... I couldn´t find any more recent info, though...

Plus, Intel has the IA64, which is true 64-bit - not just extensions. They could use that experience, right? Why not fuse the two technologies? Mark the sucessor to McKinley as the P5-"server" or something, and desktops using 64-bit...

"I have a dream...", I guess...

 

[juin]

a mackinley core and a p3 core on a 2MB of cache should reach close to 250 milliontransistor.A multiple front-end with a compatible back end can be easier but so much change as to be done.In P8 intel would have to offer a transistion cpu to the market.If I2 sale reach 25000 server like predict to IDC.Or just putting a dual p3 core on a deerfield 0.90.

Speaking of dual madison going to be dual core HP have been developing a dual core package for madison to be use in the future IA-64 super dome

[-peep-] french

 

[juin]

The best information i found about Yamill as ex-intel eng that write that he have work on a extention to IA-32 to 64 bit but he also write that the project was cancel in 2001

[-peep-] french