Intel CPU Roadmap Update ... it's looking bad....

[Mephistopheles]

Here's the <A HREF="http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2256&p=1" target="_new">link</A> to the latest roadmap from Intel.

Nothing bright. Nothing shiny. The only moderately exciting thing I can see is the added dothan muscle. I'm a little suspicious about the 6xx having a lot to offer in terms of performance... That would only be the case if Intel managed to really, really add a lot of tejas' features in there and deflate the scotty design... and that's doubtful.

Dothan is still looking quite good, however, like I said. A 2.26Ghz and beyond version of the processor has been added to the roadmaps; the clock stagnation and upping FSB suggests that a 533Mhz FSB probably brings extra performance to an already great platform. Mind you, the 2.1Ghz has already been released and still consumes only 21W. And it's fast. It's something to keep an eye on: dothan will have support for DDR2-533 (currently only supports DDR-333) and extra FSB bandwidth; if this brings a lot of performance, we can more safely speculate that bringing dothan or yonah to LGA775 with desktop-class features would unleash great potential.

About Itanium: Madison 9M isn't even at 1.7Ghz, it's at 1.6Ghz... And even then, its usefulness is doubtful at best. The performance of the current 1.5Ghz is nothing to be shy about, but it has no market penetration. And it's not a 50% increase in cache and 7% increase in clock can change.

Maybe the 533Mhz FSB (8.5GB/s) for Itanium DP systems will make a difference, though. DP Itanium systems are better because the DP processors themselves don't cost like $4000 each, and maybe if the values for such systems fell, they could just compete adequately in floating point-intensive scientific computation and such... Still a far bet. In any case, there's always montecito... which should be one hell of a processor... too bad there are little apps that support this architecture.

 

[scottchen]

I'd love to see Intel finally bringing Dothans' to the desktop.

 

[Mephistopheles]

Woooowww.....

Did you see that???

That other thread? ***drool***

Dothan can reach clock speeds in excess of 3Ghz, depending on how you cool it! I wonder how it can clock with a desktop-grade cooler like XP90 and XP120 from thermalright or zalman cooler?



<P ID="edit"><FONT SIZE=-1><EM>Edited by Mephistopheles on 11/01/04 00:14 AM.</EM></FONT></P>

 

[scottchen]

I've always loved the Dothan chips, why the f*ck did intel bother with the sh*tty prescotts' seriously, Make a chipset that supports AGP/PCI Express for Dothan chips, and use DDR400 or DDR2 rams. If they did that i never would've purchased Athlon64, and Intel probably would still have the edge on the high-end CPU market.

 

[Mephistopheles]

Can you imagine those things running at 2.5+Ghz speeds and equipped with a fully-fledged 800Mhz FSB or 1066Mhz? Wow...

 

[scottchen]

I don't see anything on that roadmap that mentions the word Dothan.

 

[Mephistopheles]

You'd have to go to the next page, "mobile processors".

Agreed though, if your point was that noone mentioned dothan on the desktop side of things. Our point is: that's just sad.

 

[scottchen]

D@mn you b@stards for raising false hopes!!

Guess i'll be picking up my Winchester 3200+ soon.

 

[HardWareBoss]

Nice of them isn't it.

 

[priyajeet]

"I wouldn't trade places with AMD--ever. They are a good competitor; they make us better; we make them better. Together, we make better profits for the world. That's competition, that's good. But the only way we are able to stay in that fray is because we invest a lot in R&D and so does AMD." - Craig Barrett

:tongue: <A HREF="http://www.geocities.com/priyajeet/fing.jpg" target="_new"><i><font color=red>Very funny, Scotty.</font color=red><font color=blue> Now beam down my clothes.</font color=blue></i></A> :tongue:

 

[trooper11]

That roadmap puzzels me. Why is Intel so 'conservative' with dothan increases in clock. Since when do they only increase clock by 160mhz in an entire year. if dothans can indeed clock so high, why cant intel hit 2.5ghz next year? they cant be that foolish and wanting to 'reign' in dothan for some reason. i guess thats possible, but maybe there is some problem we arent aware of...

 

[Johanthegnarler]

We've all been watching dothan grow.. knowing that it will be a godly cpu, but it still feels like stolen amd technology to me. Intel has always been a power hungry monster with huge fsb and high clock rates.I just can't see passed that.

But i will accept that little guy in my arms if they release agp/pci x ddr1 technology. As long as it's priced good.

<A HREF="http://arc.aquamark3.com/arc/arc_view.php?run=277124623" target="_new">http://arc.aquamark3.com/arc/arc_view.php?run=277124623</A>
46,510 , movin on up. 48k new goal. Maybe not.. :/

 

[wolverinero79]

Dothan was designed to work as a long battery, high performance CPU. Raising the FSB and clock rates might defeat that purpose.

I'm just your average habitual smiler =D

 

[jammydodger]

Yeh but if it is used as a desktop CPU who cares about battery life?

 

[dobster99]

Intel has a division devoted to mobile cpus and one devoted to desktop cpus. You think that the head of the desktop division is just going to let the dothans appear in the desktop world and invalidate all the R&D and money he just wasted on the Prescott and Tejas. Corporate politics are in play. How do you explain to the big man upstairs that you just wasted billions of dollars for 400 mhz and SSE3

 

[Mephistopheles]

it still feels like stolen amd technology to me.

<i>What?</i>

AMD doesn't have that kind of tech, and even if it did, you can't just steal tech. Remember, centrino is a huge success.

Also, what you are saying there essentially means that if it's good, it's from AMD - that's not reason, that's bias.

 

[jammydodger]

But prescott isnt going to be going over 4Ghz, and I was under the impression tejas has been scrapped?
They are going to need something to replace these.

 

[trooper11]

on that point, I would make the case that intle needs to swallow some pride and consider some of the advancements amd has made. think if you paired a dothan with an onbaord memory controller.... that would be a quite a chip.

 

[Xeon]

They won't tima was enough.

Xeon

<font color=red>Post created with being a dickhead in mind.</font color=red>
<font color=white>For all emotional and slanderous statements contact THG for all law suits.</font color=white>

 

[endyen]

It was a bad design, time on the other hand, will be enough.

 

[Xeon]

Yet the Pentium M with no on die memory controller holds it own. Good engineering will always win over novel design.

Xeon

<font color=red>Post created with being a dickhead in mind.</font color=red>
<font color=white>For all emotional and slanderous statements contact THG for all law suits.</font color=white>

 

[Mephistopheles]

Wow.... dothan with an ODMC?

If they managed that within a year, then AMD would have some serious trouble in their hands.

 

[slvr_phoenix]

Wow.... dothan with an ODMC?

If they managed that within a year, then AMD would have some serious trouble in their hands.

Maybe in a theory. In reality it's a different story.

AMD's prefetching isn't so stunning. So their on die memory controller gives them a huge boost by minimizing the lag whenever a prefetch fails.

Intel's prefetching however is pretty solid. So even if Intel used an on die memory controller, it wouldn't gain them nearly as much as the performance boost that you saw it give AMD. Oh, sure, there'd probably be a measurable improvement, but measurable and useful (especially at the expense to implement it) are two different things. It likely isn't worth the cost to Intel for the small advantage that it would give them.

Besides, one has to wonder if AMD didn't patent the ondie controller anyway...

<pre><b><font color=red>"Build a man a fire and he's warm for the rest of the evening.
Set a man on fire and he's warm for the rest of his life." - Steve Taylor</font color=red></b></pre><p>

 

[trooper11]

you say that even as that 'novel design' gained amd the market in performance. give amd a bit of respect, they have some good engineering teams there too that do know what they are doing.

 

[dobster99]

Just because the Tejas project was scrapped doesn't mean millions weren't invested in its research and development. Weren't we suppose to see it in 2006 or late 2005.

why are the prescott with their outrageous thermal and power requirement being used for dualcore. Makes no sense does it. Unless the desktop division doesn't want to admit complete failure and using the dothan architecture would be an admittance to failure.

The P4 Prescott are 64 ready, so there may be a need for it when XP 64 is launched. However, that no excuse not put out a 32-bit Desktop Dothan optimized with a higher FSB and faster memory. This would allow you to have at least two solution. One that completes with AMD-64 and one that might out perform a A64 in a 32bit enviroment. The 64P4 and Desktop Dothan could be used as a temporary solution until a cpu that combines the best of prescott/tejas and dothan (64, SSE3, less heat/power needs and shorter pipelines with high ipc) is released.

Having both the Prescott and the Dothan available for desktop is alot more appealing to us the customer than just having prescott alone. Intel can't be that stupid enough not to know this. What we are probably seeing is an internal struggle between the two division heads. Who knows the Dothan architecture better. Who would be better suited to get the most out Dothan. Would you chose the people who develop the Dothan to optimize for its desktop form. Or would you let the people, who just waste millions in terms of R&D and lost market share, have control of desktop migration of the Dothan. The answer is obvious. The only way for Desktop D. Head to keep control is to fight against migrating Dothan into the desktop enviroment.<P ID="edit"><FONT SIZE=-1><EM>Edited by dobster99 on 11/05/04 12:42 PM.</EM></FONT></P>