backtop


Print 72 comment(s) - last by uksupramk3.. on Jul 29 at 10:33 PM

"Tulsa," "Merom," "Kentsfield" and "Clovertown" moved up

Yesterday during Intel's Q2'06 earnings report, Intel CEO Paul Otellini revealed that Intel has moved up its launch schedule for quad-core processors to Q4 2006 instead of Q1 2007 as originally announced by Intel in roadmaps and public relations.

Otellini claimed "We notified customers that we're pulling in both a desktop and server of the first quad-core processors into the fourth quarter of this year from the first half of 2007."  These two processors, dubbed Kentsfield and Clovertown, respectively, are essentially twin-die packages of Core 2 Duo.

In late May, Intel announced that the company will ship a 3.2GHz Core 2 Extreme processor followed by quad-core Kentsfield in Q1'07Kentsfield has always been slated as an Extreme processor, meaning it will carry a larger price tag than the traditional Core 2 Duo series.  Intel's quad-core server processor, dubbed Clovertown, is virtually identical to Kentsfield but will use the Socket 771 package instead of Socket 775.

Furthermore, Otellini confirmed that quad-core isn't the only processor series moved up.  The CEO confirmed Merom has been moved up and is already shipping to revenue, as was reported by HKEPC (English) several days ago.  Intel's Tulsa processors for Xeon MP are also already shipping to revenue according to Otellini, but the availability of these processors has largely been overshadowed by yesterday's launch of Itanium 2 Montecito and the recent launch of Xeon DP Woodcrest

Typically there is a two to three week lag between revenue shipments and retail availability, so expect to see many of these new "shipping to revenue" processors before the end of the month.


Comments     Threshold


This article is over a month old, voting and posting comments is disabled

To combat 4x4 I take it.
By Bladen on 7/20/2006 12:19:01 PM , Rating: 2
I can't really see 4x4 or Kentsfield being super popular anyway - I guess it is just that old struggle for the title of being the creater of the most powerful CPU, or CPU configuration.

I don't know why they advertise 4 processing cores as the system that gamers should get. When a greater proportion of games become multi-threaded, then yeah knock yourself out, but otherwise it is basically a lie.

It would be good to see quad core in the mainstream though, one day that will come anyway.




RE: To combat 4x4 I take it.
By rrsurfer1 on 7/20/2006 12:40:28 PM , Rating: 2
I agree with you on the gaming issue. 4 cores probably won't be much of an improvement over 2. Many games now are somewhat multi-threaded, like the AI runs on one, and the GPU is kept fed on another. But for a game to be utilizing 4 cores at once it'd have to take that multi-threading to another level which would have to be carefully programmed.

That's why I'm thinking AMD's 4x4 design isn't going to translate into real-world performance for the majority of apps and especially not for games.


RE: To combat 4x4 I take it.
By PT2006 on 7/20/2006 12:42:22 PM , Rating: 2
I'd honestly be surprised if 4x4 even really becomes a retail solution. I think the only places you'll see it at is Alienware or Voodoo.


RE: To combat 4x4 I take it.
By TomZ on 7/20/2006 1:14:30 PM , Rating: 2
quote:
I'd honestly be surprised if 4x4 even really becomes a retail solution. I think the only places you'll see it at is Alienware or Voodoo.

I disagree - I think Intel puts all its desktop processors into the retail channel. I can't think of any recent exceptions, can you?

To be sure, this processor will have a price point higher than US$1000, but even there, there is still at least a small market. This has been proved by both AMD and Intel with their recent offerings for this segment.


RE: To combat 4x4 I take it.
By Strunf on 7/20/06, Rating: 0
RE: To combat 4x4 I take it.
By TomZ on 7/20/2006 1:39:31 PM , Rating: 2
Good point; I stand corrected.


RE: To combat 4x4 I take it.
By Griswold on 7/20/2006 2:19:29 PM , Rating: 2
I think they call it 4x4 (4 by 4) due to the SLI support - eventually becoming quad SLI. And if thats not the reason, 4x4 will make sense when you plug quad cores in the sockets at some point... or whatever.


RE: To combat 4x4 I take it.
By Strunf on 7/20/06, Rating: 0
RE: To combat 4x4 I take it.
By Griswold on 7/20/2006 3:26:30 PM , Rating: 2
This is supposed to be some enthusiast shizzle.. they dont care about "electricity bills". Wasnt that what the Intel fanboys used to say about their nuclear furnace?


RE: To combat 4x4 I take it.
By Strunf on 7/20/06, Rating: 0
RE: To combat 4x4 I take it.
By drmo on 7/20/2006 7:10:51 PM , Rating: 2
For those who think that it shouldn't be called "4x4", think about this: how any wheels does a 4x4 vehicle have? I think they are just playing off the image of the tough 4x4 vehicles. It implies that four times the power gets four times the work done (hence 4x4). What kind of marketing is 2x2? That wouldn't imply anything great to me, and could imply there are only 2 processors working.


RE: To combat 4x4 I take it.
By brownba on 7/20/2006 8:23:51 PM , Rating: 2
a 4x4 is a vehicle with 4 wheels and all 4 of those wheels are driven (all-wheel drive).
a 4x2 is a vehicle with 4 wheels and just 2 of those are powered (so either front or rear-wheel drive).



RE: To combat 4x4 I take it.
By drmo on 7/21/2006 5:09:25 PM , Rating: 2
"a 4x4 is a vehicle with 4 wheels and all 4 of those wheels are driven (all-wheel drive).
a 4x2 is a vehicle with 4 wheels and just 2 of those are powered (so either front or rear-wheel drive). "


Exactly, so a 2x2 would be a two processor system, where two processors work to the fullest (not a quad processor system). Of course, on single-threaded apps it would be 2x1. (You could say a P4 + HT would be 2 x 4 from that standpoint.) But really, 4x4 is just a marketing thing, IMO. Most people don't care how many sockets there are, just how much work gets done...


RE: To combat 4x4 I take it.
By Calin on 7/21/2006 9:22:24 AM , Rating: 2
There are plenty of people (plenty = thousands) that could put to good use a 4 core computer - paying $2000 for a processor that runs on a $200 common mainboard would be better than paying two times $1000 or $800 and buying a $500 workstation/server mainboard. The possiblity of upgrade would bring some to it.


RE: To combat 4x4 I take it.
By TK2K on 7/21/2006 5:19:20 PM , Rating: 2
not true at all
look at quad sli, that has really taken off even in the home builder section! it will take off, the question is, how much will it be?


RE: To combat 4x4 I take it.
By Strunf on 7/20/06, Rating: 0
RE: To combat 4x4 I take it.
By epsilonparadox on 7/20/2006 2:14:31 PM , Rating: 2
The Cell has one full functioning general core and the rest are SPEs that can only handle small specific task that must be written into the application code. the 4x4 and kentsfield has 4 full functioning general cores that can handle a multiple various-sized tasks. The Cell is a completely different beast from these two.


RE: To combat 4x4 I take it.
By Calin on 7/21/2006 9:26:33 AM , Rating: 2
The Cell has one common processor (one that is good at running mostly any code). There are another 7 SPE, which are limited in: Out of Order Execution (not available), main memory access (not available IIRC), some instructions. As long as you feed them code optimized to run on them, they can put to shame anything else. If you feed them code that is not optimized, they are a millstone around your neck.

The Core 2 Quad (or whatever) is composed of four identical common processors.


RE: To combat 4x4 I take it.
By Trisped on 7/21/2006 12:03:56 PM , Rating: 2
On the "Multi-Threaded Games" topic:

I don't think most games are truly threaded. Most were just hacked at until it was split into two threads.

Not all games have this limit though, as WarCraft3 creates a thread for every trigger, allowing it to scale to any number of processors.

From what I can tell the real way to thread a game is to put every major job into its own thread. Audio, AI, physics, mapping, preloading, GUI, networking, video, etc. Then, no matter how many cores you have, the game is ready. The problem is the cost of making sure the threads all talk to each other properly.


RE: To combat 4x4 I take it.
By TomZ on 7/20/2006 12:42:37 PM , Rating: 2
quote:
I can't really see 4x4 or Kentsfield being super popular anyway - I guess it is just that old struggle for the title of being the creater of the most powerful CPU, or CPU configuration.

I would want one - sign me up! I don't do any gaming (sadly), but I do tend to run a lot of apps at once and often times have several apps busy processing at the same time.


RE: To combat 4x4 I take it.
By clementlim on 7/20/2006 1:22:38 PM , Rating: 2
Is there anyway to analyse that quad cores actually works as 4 cores other than they quicken the application by 10-20%? Theoritically, they should push 400%...of course that is impossible, but just a theory.


RE: To combat 4x4 I take it.
By TomZ on 7/20/2006 1:38:41 PM , Rating: 2
quote:
Is there anyway to analyse that quad cores actually works as 4 cores other than they quicken the application by 10-20%? Theoritically, they should push 400%...of course that is impossible, but just a theory.

Sure: You can run 3 CPU-bound operations, then add a 4th, and the first 3 don't run any slower as a result. I've run similar tests on my dual-core and dual-processor machines before.

Most common applications aren't multi-threaded beyond some simple multi-threading in order to decouple GUI from "work" to keep the GUI responsive while the application is doing "work." For applications like this, the only benefit of quad-core is maybe that you could run more of these at the same time, but any individual application probably won't run any faster on quad-core.

If you had a particular application that was highly multithreaded and CPU-bound, then in theory you could get a 4x throughput increase with dual-core, but these applications are rare.


Intel's 7800GTX-512
By Fenixgoon on 7/20/2006 12:55:52 PM , Rating: 2
It doesn't matter that only a select few of these will actually exist. The important thing (for Intel) is that their quad core is debuting FAR ahead of K8L. Just like the 7800GTX-512... no one had one, but Nvidia nonetheless pwned ATI until they *finally* brought out the x1900




RE: Intel's 7800GTX-512
By raven3x7 on 7/20/2006 2:53:34 PM , Rating: 2
you forget that ketsfield is not a "true" quad-core chip and that we'll have to see if the fsb doesnt hold the system back or even if a new 1333 fsb mobo isn't required for this cpu. on the other hand you have k8l which is a native quad-core chip and HTX givs AMD processors larger memory bandwidth(granted k8 processors cant really utilize it, but can quad-core chips and the k8l in general?)but that wont be available before late q1 or early q2 2008. So this looks to me like Intel is pushing everything forward because they know that they have the advantage now but won't have it for long as AMD will catch up soon enough and Intel wants to get as much marketshare back as it can get


RE: Intel's 7800GTX-512
By Strunf on 7/20/06, Rating: 0
RE: Intel's 7800GTX-512
By gramboh on 7/20/2006 6:24:06 PM , Rating: 2
Why is 4x4 a dual socket/chip solution? Is it because AMD is too far away with K8L (assuming it is quad core chip)? No one wants to buy multi chip systems it's a pain in the ass and expensive both in terms of mainboard and CPU.


RE: Intel's 7800GTX-512
By GoatMonkey on 7/20/2006 7:57:52 PM , Rating: 2
Please share these benchmarks with us.


RE: Intel's 7800GTX-512
By SexyK on 7/20/2006 9:39:54 PM , Rating: 2
quote:
Please share these benchmarks with us.


http://www.xtremesystems.org/forums/showthread.php...


RE: Intel's 7800GTX-512
By TomZ on 7/21/2006 10:19:52 AM , Rating: 2
You found benchmarks somewhere in that mess?


RE: Intel's 7800GTX-512
By Strunf on 7/21/2006 12:09:58 PM , Rating: 1
Yes, they have posted some benchmarks but you have to look for them.


RE: Intel's 7800GTX-512
By GoatMonkey on 7/21/2006 2:36:50 PM , Rating: 2
Interesting. Thanks.

It looks fast, but I'm a little disappointed that they didn't put more cache on it.


RE: Intel's 7800GTX-512
By GoatMonkey on 7/20/2006 8:02:54 PM , Rating: 2
I would imagine that Intel will get around the bus speed problem by throwing more cache at it.

The new improvements that we've seen in Conroe will still be there in this Kentsfield chip. So, I'm betting that the Kentsfield will be equal or better than the 2 dual core chips that will be available on 4x4.

It sounds to me like it's going to be a very limited production run at first though. I'm no Intel fan, as I write this on one of my AMD machines, but you have to face the facts.


RE: Intel's 7800GTX-512
By TomZ on 7/20/2006 9:47:36 PM , Rating: 2
quote:
I would imagine that Intel will get around the bus speed problem by throwing more cache at it.

Kentsfield will be 2 x 4MB L2.


RE: Intel's 7800GTX-512
By GoatMonkey on 7/21/2006 8:31:01 AM , Rating: 2
That's not enough. Intel needs to go to 4 x 4MB of cache for this chip.


Another Pressler
By Goty on 7/21/2006 11:03:01 AM , Rating: 2
So nobody seems to realize that this is just another Pressler. Intel has just taken two dual-core chips and slapped them on a single package together. You know the whole performance benefit that the Core chips must have gained by not having to send data between cores through the chipset? Yeah, that's gone. In fact, the FSB is probably even more of a bottleneck now that you're trying to send two cores' worth of data between pieces of silicon instead of just one.




RE: Another Pressler
By TomZ on 7/21/2006 11:21:27 AM , Rating: 2
quote:
So nobody seems to realize that this is just another Pressler.

No, not really. Presler is two Netburst single-core processors packaged in an MCM. Kentsfield, on the other hand, is two Conroe dual-core processors packaged in an MCM. Not nearly the same.


RE: Another Pressler
By TomZ on 7/21/2006 11:23:18 AM , Rating: 2
quote:
In fact, the FSB is probably even more of a bottleneck now that you're trying to send two cores' worth of data between pieces of silicon instead of just one.

Why do you assume there is a lot of data flow from core-to-core? If two separate threads are assigned to different cores, which is typical, I don't see how that causes much core-to-core data flow.


RE: Another Pressler
By Motley on 7/21/2006 9:51:24 PM , Rating: 2
There normally isn't that much data flow. Hardly any at all actually.


RE: Another Pressler
By Goty on 7/23/2006 5:50:54 PM , Rating: 2
Unless both of you are hardware engineers that specialize in this field, I don't see how you're more qualififed than I am to make such statements.


RE: Another Pressler
By Motley on 7/21/2006 9:55:14 PM , Rating: 2
Yeah, the FSB and non-integrated memory controller are such a bottleneck for the conroe. /laugh


RE: Another Pressler
By ahto on 7/22/2006 7:44:01 PM , Rating: 2
actually Conroe performs surprisingly poor on 64bit environment.. sometimes 10% slower than in 32bit OS

http://www.overclockers.ru/hardnews/22814.shtml

If you dont understand Russian, try translateing it with Babelfish.. text remains still quite logical


RE: Another Pressler
By Goty on 7/23/2006 5:50:12 PM , Rating: 2
If you cared to read the post you were repsonding to, I didn't mention a FSB bottleneck on Conroe at all, and I NEVER mentioned an integrated memory controller.


Limitations
By boboisgreat on 7/20/2006 9:31:54 PM , Rating: 2
This 4 core solution will be limited in that the cores will have conflicts accessing the cache. The Pentium D 930 should offer comparable performance for much less $$$. The Conroe architecture has some limitations.




RE: Limitations
By TomZ on 7/20/2006 9:45:36 PM , Rating: 2
quote:
This 4 core solution will be limited in that the cores will have conflicts accessing the cache. The Pentium D 930 should offer comparable performance for much less $$$.

Well, Conroe outperforms the 930, and Kentsfield is basically two Conroes. The two Conroes have separate caches. So how will cache conflicts cause the performance of Kentsfield to be the same as the 930? It doesn't make sense to me.
quote:
The Conroe architecture has some limitations.

I'm sure it does, but please elaborate as to what you mean.


RE: Limitations
By hmurchison on 7/20/2006 11:30:13 PM , Rating: 2
I don't think you're right on this. Having 4 cores share a 8MB cache would be ideal but two Conroe cores on the same die with a fast FSB for cache coherency between the cores will still be faster than the slow Netburst D procs.

Conroe has little limitations that I can see but I'd be interested in hearing your opinion.


RE: Limitations
By AntiTomZandmasher2 on 7/21/2006 12:36:07 AM , Rating: 2
I'm a computer architecture student, and right now, the only weakness I see in Conroe is the bus possibly becoming a limitation. They're still on a shared bus solution and the AGTL+ bus is showing its age compared to Hypertrasport. That might become a problem when we get more cpus demanding more throughput.


RE: Limitations
By TomZ on 7/21/2006 10:21:23 AM , Rating: 2
Yes, that's true, and it has been stated in nearly every review of Intel architecture in the past 2 years. Nothing new there.


RE: Limitations
By AntiTomZandmasher2 on 7/21/2006 12:43:52 AM , Rating: 2
I'm a computer architecture student, and right now, the only weakness I see in Conroe is the bus possibly becoming a limitation. They're still on a shared bus solution and the AGTL+ bus is showing its age compared to Hypertrasport. That might become a problem when we get more cpus demanding more throughput.


By hmurchison on 7/21/2006 10:21:56 AM , Rating: 2
Intel's roadmaps never lied. Yonah was going to have 6-12 months of life before being usurped by Merom.

Yonah is a good step towards getting ISV to start making their apps multithreaded. Most of the chipset for the Pentium M was reused for Yonah so you don't have a huge platform change either.


Apple?
By dcollins on 7/20/2006 12:32:21 PM , Rating: 2
I wonder, with merom moving up, how soon apple will release new macbook pro's?




RE: Apple?
By Knish on 7/20/2006 12:33:26 PM , Rating: 2
Probably at WWDC Aug 7


RE: Apple?
By kelmon on 7/20/2006 1:33:12 PM , Rating: 2
Words fail to describe how much I hope an early announcement will take place but I have to confess that I suspect that later in the year is more likely. MacExpo may well be more likely but I'll be keeping my credit card at the ready in case WWDC does see the announcement happen.


RE: Apple?
By secretanchitman on 7/20/2006 12:55:23 PM , Rating: 2
exactly! i think that this will be good news for mac owners. faster way to get new computers with core 2 duo (desktop and mobile versions) in them.


By wingless on 7/20/2006 9:09:47 PM , Rating: 2
When Windows Vista comes out and eats single and dual core processors for it's 3 major meals and 2 snacks a day, we all will hope and pray we can get quad core. This argument about 4 cores seems just like the old 64bit argument way back in the day (only a year or two ago...). For us ENTHUSIASTS, we will find a use for all 4 cores along with every extension and the 64bit capabilities in the next 2 years simultaneously. Hell, finally I can play games at full FPS while running seriously intensive software in the background and not skip a beat. As far as concerns about power consumption, AMD and INTEL are moving to or already using 65nm by the end of this year, and 45nm will come next year before we know it. a 4 core 45nm CPU will probably consume less enegery than our old 90nm's.

Im amazed to see people whine about innovation and MORE POWER!! I personally cant wait until I can own a computer thats 100% pure overkill for anything I use it for. Ive never seen a sytem that I can live up to that expectation yet..




By wingless on 7/20/2006 9:10:14 PM , Rating: 2
wow...I love a good senseless rant..


By GoatMonkey on 7/21/2006 2:41:00 PM , Rating: 2
I can agree with that. I'm getting tired of people saying that it's pointless because games can't use all of the CPUs. But you know... they will. And it won't be that long from now.

If they build it, someone will write code for it.

I'll take two.


Take on Conroe, et al.
By maxusa on 7/21/2006 1:35:41 AM , Rating: 2
One thing that confuses me is why Intel released the original Yonah only to be killed in 7 months? Poor Apple with their decision to jump on the Core architecture... it was grossly premature move that will likely to aggravate a lot of new Mac Mini, iMac, and MBP owners.

Same situation is in the x86 camp. Dell, for one, has kept its Yonah offerings to just a few models. Sure this had lots of people wonder... no more. Merom/Conroe/Woodcrest is the real thing, which I'm sure will be here to stay for some time. This is good news for the corporate customers and the next upgrade cycle.




RE: Take on Conroe, et al.
By TomZ on 7/21/2006 10:23:57 AM , Rating: 2
quote:
One thing that confuses me is why Intel released the original Yonah only to be killed in 7 months?

Probably not a big deal in practice - at these production volumes, having to do a board rev (or new board design) is not really a big issue.

End-customers for the most part don't care if their processor is Yonah, Merom, etc. as long as it has good performance and good value.


RE: Take on Conroe, et al.
By ScottEllsworth on 7/21/2006 4:12:14 PM , Rating: 2
They released Yonah when they had to. Without such a chip, they have nada in the marketplace for six or seven months. With Yonah, they have _something_ that competes, if not as well as what was coming later.

From Apple's perspective, they got to move the Mac Mini and the powerbook lines over to Intel in a fairly short period. Their alternative was limping along with 1.6GHz single core G4s that were looking pretty anemic compared with the competition.

I bought a 17" MBP not long after they were announced, and am pretty happy with it. It runs fast, runs reasonably cool (compared with the G4 running the same tasks), and has no noise issues. Using the server VM, I am getting ten times the performance on my Java apps - ten times! I would have liked a Merom, but I also wanted a computer to do my work on between March and August, and thus I really could not wait.

So, speaking as one owner, it does not bug me that a successor with perhaps half again the speed is coming out in the next month or two. I got a factor of ten speed boost for six months, and that is what I needed.

(NB - Java apps benefit more than many others, as the server VM was not available for the PPC, and thus you get both the boost from -server and the speed boost from the new CPU.)

Scott


By ahto on 7/22/2006 7:40:02 PM , Rating: 2
Id like to know which home user windows would support 4 cores????

2000 and XP supports up to 2 processors.. I do know only one 32bit windows able support 4 processors at once.. and this is server OS.. Win2003 Standard

Afaik even different versions of Vista are planned do support only single or up to 2 processor systems...




By Garreye on 7/23/2006 10:44:18 AM , Rating: 3
Its 4 cores on 1 processor, so there shouldn't be a problem, you could even have 2 of these and be ok...I'm not sure what the limit for cores per processor supported by XP is, but it's at least 4 because the Intel's dual cores with hyper threading register 4 cores in XP.


Yield Issue?
By czarchazm on 7/20/2006 12:31:00 PM , Rating: 2
So it seems like Intel isn't having yield issues with its Core 2 Duo wafers.




RE: Yield Issue?
By akugami on 7/20/2006 12:37:08 PM , Rating: 2
I wouldn't bet on that just yet. It could be a very limited release that is priced sky high. Basically they can claim it's available, just that 90% of the people who would want one are turned off by the initial high price and then a couple of months later in Q1'07 they can release lower priced variants after working out the kinks. Marketing game. Nothing new. AMD has done the same thing.


Game developers need to take advantage of this
By kitchme on 7/20/2006 12:36:50 PM , Rating: 2
I'd like to see more game developers take advantage of these multi-core CPUs. As of now, you still have to disable 2nd core in some games to reduce game problems, or to even play. But, once hardware is efficiently used by game coders, that'll be a good day.




By segagenesis on 7/20/2006 1:11:38 PM , Rating: 2
This is currently the growing pains associated with expanding cpu power outward rather than upward, I'm not sure if we will see single core speeds get any faster in the immediate... likewise you cannot expect developers overnight to start doing thread friendly programming.

Dual core stuff is nice though for reasons TomZ stated above, for desktop use no longer does one application really bog the system down.


Xeons?
By inthell on 7/20/2006 1:17:12 PM , Rating: 2
are these only Xeon type server/workstation chips?




RE: Xeons?
By epsilonparadox on 7/20/2006 2:17:39 PM , Rating: 2
the clovertown is server. the kentsfield is desktop. Same chip but different pin count.


Why I see this as a good thing for gamers
By PAPutzback on 7/20/2006 3:09:06 PM , Rating: 2
I think with dual cores and quad cores coming out then we can dismiss the thought of a $300 dedicated physics card. Hopefully HAVOK or MS's next DirectX will be able to use the additional CPU's to handle physics instead of a dedicated video or graphics card. Perhaps split Effects Physics and Gameplay physics across two CPU's, the game on the third and all of the background services and apps on the 4th. Expect a tweakguid soon after for setting the affinity for all these processes




By kelmon on 7/21/2006 6:51:59 AM , Rating: 2
I have my doubts that the performance will match a dedicated chip for such processing but I'm much more in favour of using existing resources (especially if they are not being used) rather than needing to buy an additional card that only does a single task. At least with additional processor cores doing the physics processing you are making efficient use of the resources that you have bought rather than spending money on something that will see relatively little usage.

With operating systems only recently starting to take advantage of GPUs even your graphics card was a pretty inefficient use of money. I'm glad that this situation has changed so at least the expense is more justifiable.


Apple?
By a1016neo on 7/20/2006 12:36:29 PM , Rating: 2
WWDC 2006 (Aug 7th) or

Apple Expo (September)




yummy
By nfin1ty on 7/25/2006 1:52:33 PM , Rating: 2
mmm.. all i can envision for Intel's procs that go beyond 2cores is insane bus saturation. i bet the new dual cores are just pushing the bus to it's furthest.. sure they trump X2's now.. that's not the point.. intel will take a long time before they can devise a scalable bus transport that can go beyond 2-way without seriously saturating the bus to the point of making the system just a tidbit faster than a 2-core. I'm looking forward to AMD's quad-core system.. since they know how to do memory management. Although i must digress and admit that i highly doubt that any software will support quad core from either company considering that dual-core threaded apps are still fleeting in numbers.




The problem with this....
By uksupramk3 on 7/29/2006 10:33:49 PM , Rating: 2
...is that Intels insistence on using a Front Side Bus is effectively going to cripple them eventually. Okay, so at 1333Mhz, thats pretty fast and while the current Core 2 Duo's arent having a problem with it - in 4-way SMP, it's going to come back and bite them in the arse.

Not saying the Opterons are a better processor (although they are good), but they do have the best implementation when it comes to SMP, with NUMA and Hypertransport.




"If they're going to pirate somebody, we want it to be us rather than somebody else." -- Microsoft Business Group President Jeff Raikes

Related Articles
Intel's "Woodcrest" to Launch Monday
June 23, 2006, 1:42 PM
The Count of "Montecito"
July 16, 2006, 3:39 PM
Intel Bumps "Tulsa" Ahead to Q3'06
June 12, 2006, 5:40 PM













botimage
Copyright 2014 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki