Intel Conroe

[Avalon]

Originally posted by: ddeder
And what happens when Intel delays Conroe by 6 - 12 months? This is a desperate Intel ploy that we have all seen before. Maybe AMD should respond with a vaporware launch of a 10GHZ 64 core Athlon that triples the performance of Conroe - to be debuted in 6 months.

Why would you reasonably expect Conroe to be delayed beyond Q3? It was working just fine at the IDF.

 

[ddeder]

"Why would you reasonably expect Conroe to be delayed beyond Q3? It was working just fine at the IDF."

Well gee, if it is working so well right now, why not just release it right now? Take the speed crown today.

 

[HmmmDonut]

Originally posted by: ddeder
"Why would you reasonably expect Conroe to be delayed beyond Q3? It was working just fine at the IDF."

Well gee, if it is working so well right now, why not just release it right now? Take the speed crown today.

I would assume that Intel is waiting till they have decent yields before they launch Conroe. They have no rush to release it anyway. Those benchmark tell me that K8 is as good as dead.

 

[dmens]

Well gee, if it is working so well right now, why not just release it right now? Take the speed crown today.

Umm... validation?

 

[Cooler]

Originally posted by: dmens

Well gee, if it is working so well right now, why not just release it right now? Take the speed crown today.

Umm... validation?

Beacuse they want to milk there P-D some more as im sure they are much cheaper to produce.

 

[Duckzilla]

Originally posted by: jiffylube1024
Damn right Intel is doing a good job -- they're comparing a tech that's 6 months away, on a smaller manufacturing process (65nm), and clocked way higher than they ever got the Pentium M (Conroe's daddy). It better compare well to current gen AMD!

If nothing changes in 6 months, Intel is going to kick butt. But I doubt AMD just releases AM2 and calls it a day, and for the next 6 months, AMD is untouchable...

Thats not true. People that might have bought an AMD rig are now saving their cash for Conroe. AMD has already been touched and Conroe is months away.

 

[robertk2012]

Originally posted by: Cooler

Originally posted by: dmens

Well gee, if it is working so well right now, why not just release it right now? Take the speed crown today.

Umm... validation?

Beacuse they want to milk there P-D some more as im sure they are much cheaper to produce.

Bingo. If they say six months people will still buy new processors for the next three. Then they release conroe early. If they said it would be out in 3 why would anyone by a new system that is going to get stomped in 3 months. Apple wanted those chips early (as in now) but intel wouldnt give in. Also intel is trying to get the new 4 core processor pushed up to the end of this year. AMD is in trouble unless they have something big up their sleave.

 

[dexvx]

Originally posted by: robertk2012

Originally posted by: Cooler

Originally posted by: dmens

Well gee, if it is working so well right now, why not just release it right now? Take the speed crown today.

Umm... validation?

Beacuse they want to milk there P-D some more as im sure they are much cheaper to produce.

Bingo. If they say six months people will still buy new processors for the next three. Then they release conroe early. If they said it would be out in 3 why would anyone by a new system that is going to get stomped in 3 months. Apple wanted those chips early (as in now) but intel wouldnt give in. Also intel is trying to get the new 4 core processor pushed up to the end of this year. AMD is in trouble unless they have something big up their sleave.

Wrong. Given die size, the Conroe is just as cheap to produce, if not cheaper than the Pentium-D Pressler.

The real reason is most likely validation. It is one step of the process you cannot get away with, and usually takes months.

 

[Quinton McLeod]

Originally posted by: robertk2012

Originally posted by: Cooler

Originally posted by: dmens

Well gee, if it is working so well right now, why not just release it right now? Take the speed crown today.

Umm... validation?

Beacuse they want to milk there P-D some more as im sure they are much cheaper to produce.

Bingo. If they say six months people will still buy new processors for the next three. Then they release conroe early. If they said it would be out in 3 why would anyone by a new system that is going to get stomped in 3 months. Apple wanted those chips early (as in now) but intel wouldnt give in. Also intel is trying to get the new 4 core processor pushed up to the end of this year. AMD is in trouble unless they have something big up their sleave.

Have you even seen the Quad Core pictures? It's a joke. They literally have 2 dual core dies placed in a single bed. Intel has got to be joking. It physically looks like a bottleneck. In order for the 4 cores to talk to each other, they are still going to have to go through the FSB which has yet to gain enough bandwidth to consider itself not bottlenecked.

Any person could tell that Intel was desperate and just wanted to be ahead of AMD. We don't need Quad Core right now. We're barely utilizing Dual Core!

 

[dexvx]

Originally posted by: Quinton McLeod
Have you even seen the Quad Core pictures? It's a joke. They literally have 2 dual core dies placed in a single bed. Intel has got to be joking. It physically looks like a bottleneck. In order for the 4 cores to talk to each other, they are still going to have to go through the FSB which has yet to gain enough bandwidth to consider itself not bottlenecked.

What the heck do you think an X2 is? All it is is 2 cores linked via a higher latency crossbar. The Yonah actually has shared cache. And supposedly the Quad Core Kentsfield are 2 shared Cached dual cores. And how do you get past the FSB bottleneck? Just add more lanes.

Originally posted by: Quinton McLeod
Any person could tell that Intel was desperate and just wanted to be ahead of AMD. We don't need Quad Core right now. We're barely utilizing Dual Core!

Yea, plenty of servers out there just don't need Quad Core. I mean geez, what kind of moron would design multi-processor servers??

 

[dmens]

It physically looks like a bottleneck.

I don't even know what that is supposed to mean.

 

[dexvx]

Originally posted by: dmens

It physically looks like a bottleneck.

I don't even know what that is supposed to mean.

He means that because somehow because they dont share the same die, it is not a true multi-core CPU.

 

[robertk2012]

Originally posted by: dmens

It physically looks like a bottleneck.

I don't even know what that is supposed to mean.

He was saying it looked like a coke bottle.

 

[dmens]

That canard again? Yeesh. Lots of "analysts", professional or amateur, FSB bus will kill any intel product for the high end. I figure it is a weakness, but not killer on uni, dual or even quad core. Maybe octo. Will this myth continue to survive?

 

[Quinton McLeod]

Originally posted by: dexvx

Originally posted by: Quinton McLeod
Have you even seen the Quad Core pictures? It's a joke. They literally have 2 dual core dies placed in a single bed. Intel has got to be joking. It physically looks like a bottleneck. In order for the 4 cores to talk to each other, they are still going to have to go through the FSB which has yet to gain enough bandwidth to consider itself not bottlenecked.

What the heck do you think an X2 is? All it is is 2 cores linked via a higher latency crossbar. The Yonah actually has shared cache. And supposedly the Quad Core Kentsfield are 2 shared Cached dual cores. And how do you get past the FSB bottleneck? Just add more lanes.

Originally posted by: Quinton McLeod
Any person could tell that Intel was desperate and just wanted to be ahead of AMD. We don't need Quad Core right now. We're barely utilizing Dual Core!

Yea, plenty of servers out there just don't need Quad Core. I mean geez, what kind of moron would design multi-processor servers??

Wow, aren't we behind on technology.
First of all, the way Intel boards are made, the lanes are Half Duplex. Meaning it can only send data in one direction at a time. Intel is not going to add another lane. You could suggest they do it all you want, but the fact to the matter is, Intel is too stubborn.

Shared Cache? Maybe... But the 4 cores aren't sharing the cache together. Only 2 cores are able to access a single cache at a time. It pretty much means that Core 3 and 4 share a cache and core 1 and 2 share a cache, but core 4 and 2 do not share a cache. Neither does Core 3 and 1. So, they will have to use the FSB to talk to each other.

Not only that, but what you don't realize is that if the cores use the cache to communicate, there's more data clogging the cache and thus causing another bottle neck. Correct me anytime ya want, buddy.

AMD's high latency crossbar switch?! The crossbar runs at CPU speed, you n00b.

 

[dexvx]

Originally posted by: Quinton McLeod
Wow, aren't we behind on technology.
First of all, the way Intel boards are made, the lanes are Half Duplex. Meaning it can only send data in one direction at a time. Intel is not going to add another lane. You could suggest they do it all you want, but the fact to the matter is, Intel is too subborn.

Ever hear about dual independant FSB's?

Originally posted by: Quinton McLeod
Shared Cache? Maybe... But the 4 cores aren't sharing the cache together. Only 2 cores are able to access a single cache at a time. It pretty much means that Core 3 and 4 share a cache and core 1 and 2 share a cache, but core 4 and 2 do not share a cache. Neither does Core 3 and 1. So, they will have to use the FSB to talk to each other.

Yea learn to read... I said 2 shared cache dual cores.

Originally posted by: Quinton McLeod
AMD's high latency crossbar switch?! The crossbar runs at CPU speed, you n00b.

The crossbar still has a latency on the order of 100s of ns. The shared cache has a latency on the order of 10s of ns.

 

[HurleyBird]

This is getting heated.

 

[TrevorRC]

Originally posted by: akers
Intel officially announced the "Conroe" today. They promise a 40% increase in performance over the current P-4. In other words, if you are willing to wait about six more months you can finally buy an Intel chip that is as fast as last years crop of Athlon 64s. And it will probably only cost twice as much as the comparable AMD chip.

What a DEAL!

Go Intel!!

Wow. Fanboyism to a new level.

I like AMD, I'm currently running an AMD processor and it is probably one of the best I've ever used [strike that, THE best I've ever used].

Conroe will be priced REASONABLY. There is a pricing chart showing 201 dollars for the low end model (2MB cache, 1.81Ghz, if I remember) and 510 for the high end. [2.67 w/ 4MB of cache for the high end.

Edit: Here:
Conroe E6700
2.67GHz / 1066MHz 4MB Q3'06 $530 (???)

Conroe E6600
2.40GHz / 1066MHz 4MB Q3'06 $316 (???)
Conroe E6400
2.13GHz / 1066MHz 2MB Q3'06 $241 (???)
Conroe E6300
1.86GHz / 1066MHz 2MB Q3'06 $209 (???)

209, and most likely overclockable. Looks good to me.

I don't see a speed increase from AMD (It would have to be ~3-3.2 to challenge the 2.67 Conroe) helping, nor do I see the switch to DDR2 doing much. Shame for AMD.

 

[Quinton McLeod]

Originally posted by: dexvx


Ever hear about dual independant FSB's?

Yes, I've heard of Dual FSB motherboards... But these motherboards were designed
to run two processors.

Wishful thinking on your part if you think that Intel will incorporate this in a quad-core system and not in their current Dual Core systems. (Which they are not doing... All they're doing is pumping their FSB)

Originally posted by: Quinton McLeod
Shared Cache? Maybe... But the 4 cores aren't sharing the cache together. Only 2 cores are able to access a single cache at a time. It pretty much means that Core 3 and 4 share a cache and core 1 and 2 share a cache, but core 4 and 2 do not share a cache. Neither does Core 3 and 1. So, they will have to use the FSB to talk to each other.

Yea learn to read... I said 2 shared cache dual cores.

Umm... You didn't understand what I said. Let me make it simple, ok?


1 <--> [C1] <--> 2


3 <--> [C2] <--> 4


Currently, this is what Intel's Quad Core systems look like. Core 1 and 2 share cache 1. Core 3 and 4 share cache 2. Notice how core 1 and 3 don't share a cache. Also notice how core 2 and 4 don't share a cache.

The crossbar still has a latency on the order of 100s of ns. The shared cache has a latency on the order of 10s of ns.

Link please?

 

[dexvx]

Originally posted by: Quinton McLeod
Yes, I've heard of Dual FSB motherboards... But these motherboards were designed
to run two processors.

Wishful thinking on your part if you think that Intel will incorporate this in a quad-core system and not in their current Dual Core systems. (Which they are not doing... All they're doing is pumping their FSB)

Theres no reason to believe this wont work for 4 socket systems. Moreover, there is technology available today that alleviates the FSB bottleneck, see IBM Hurricane for Xeons.

Originally posted by: Quinton McLeod
Link please?

Why do I need a link? Do the simple math. If the crossbar runs at 2.8Ghz, its latency is nowhere near the shared latency at 14ns (for Yonah).

 

[Tangerines]

Quinton:

- Two separate pieces of silicon for once CPU doesn't mean much when it comes to CPU performance, especially Intel's, since it lacks an on-die memory controller.

- I didn't realize info had been released for Intel's quad-core processor yet. Unless you happen to work at Intel (which seems highly doubtful given your AMD bias) could you please stop passing off speculation as fact?

- Would it kill you to admit that Intel has one-upped AMD for once?

 

[Quinton McLeod]

Originally posted by: Tangerines
Quinton:

- Two separate pieces of silicon for once CPU doesn't mean much when it comes to CPU performance, especially Intel's, since it lacks an on-die memory controller.

What are you talking about?

- I didn't realize info had been released for Intel's quad-core processor yet. Unless you happen to work at Intel (which seems highly doubtful given your AMD bias) could you please stop passing off speculation as fact?

Do you even know what the quad core processor looks like?

- Would it kill you to admit that Intel has one-upped AMD for once?

It hasn't. There are far too many questions about the benchmark Intel gave that are unanswered. Check out this place for more info:

http://voodoopc.blogspot.com/

 

[kknd1967]

This thread by voodoopc CTO is quite informative.
Inte's FX60 rig is not as good, intentionally or not, by a slight margin. Still Conroe beats a "properly" configured FX60 rig from Wil Harris. If AM2 platform does not provide significant gain, Conroe is likely to be the winner.

Conroe vs FX60 vs "intel's" FX60

Plus for those doubting the benchmarks via cross-platform comparison, please understand FX60 can be quite slow sometime

It hasn't. There are far too many questions about the benchmark Intel gave that are unanswered. Check out this place for more info:

http://voodoopc.blogspot.com/

 

[nycdude]

Originally posted by: Quinton McLeod

Originally posted by: Tangerines
Quinton:

- Two separate pieces of silicon for once CPU doesn't mean much when it comes to CPU performance, especially Intel's, since it lacks an on-die memory controller.

What are you talking about?

- I didn't realize info had been released for Intel's quad-core processor yet. Unless you happen to work at Intel (which seems highly doubtful given your AMD bias) could you please stop passing off speculation as fact?

Do you even know what the quad core processor looks like?

- Would it kill you to admit that Intel has one-upped AMD for once?

It hasn't. There are far too many questions about the benchmark Intel gave that are unanswered. Check out this place for more info:

http://voodoopc.blogspot.com/

Interesting....

 

[Maximilian]

Originally posted by: nycdude

Originally posted by: Quinton McLeod

Originally posted by: Tangerines
Quinton:

- Two separate pieces of silicon for once CPU doesn't mean much when it comes to CPU performance, especially Intel's, since it lacks an on-die memory controller.

What are you talking about?

- I didn't realize info had been released for Intel's quad-core processor yet. Unless you happen to work at Intel (which seems highly doubtful given your AMD bias) could you please stop passing off speculation as fact?

Do you even know what the quad core processor looks like?

- Would it kill you to admit that Intel has one-upped AMD for once?

It hasn't. There are far too many questions about the benchmark Intel gave that are unanswered. Check out this place for more info:

http://voodoopc.blogspot.com/

Interesting....

Very Interesting