Conroes are out .... so how about Kentsfield and K8L?

[Lazaredz]

What are the probabl timings (for release) and gains by the next gen chips?

I believe that the Kentsfield and K8L and the next "Big Steps" for both Intel and AMD respectively, right? So when are they supposed to come out and what are they to offer???

Honestly, just trying to figure out if a Conroe system would really be worth it if the K&K are both coming out when the next gen engines are due (Crysis and U3/UT2007).

[Banko]

Kentsfield is just Conroe with two additional cores, so in single-threaded apps it would perform exactly the same.

Thing is I think Intel is purposefully holding back Kentsfield because AMD doesn't seem to have anything to counter Conroe with, hell some people at xtremesystems.org already have engineering samples.

[Lou Ceifer]

Kentsfield is just Conroe with two additional cores, so in single-threaded apps it would perform exactly the same.

Thing is I think Intel is purposefully holding back Kentsfield because AMD doesn't seem to have anything to counter Conroe with, hell some people at xtremesystems.org already have engineering samples.

No, Kentsfield is 2 Woodcrest cores slapped together like the Pentium D, expect more of the same congested bus I/O and even poorer scaling beyond 2P as 4 cores thrash the front-side bus fighting over bandwidth. Thankfully the Opteron K8L will not suffer this fate as it will be a true 4 core design and will also have an L3 shared cache and most importantly, the cache-coherent Hypertransport bus.

AMD doesn't have anything to counter Conroe with??? Haha! You think they are just going to rest on their laurels as Conroe finally gets out the door? What about 4x4, or AM2 Rev. G, the Turion X2 rev. (codename Bulldozer) or later on, K8L??? The real story is we don't know yet who is going to "own" who in about a year as it's all conjecture at this point in time. I'm sure however, it's safe to say AMD is not going to let Intel's response go unanswered or uncontested for too long, as thier market share on the desktop and notebooks is at stake.

The recent AMD Developer's Conference shed alot of light on the K8L and it's shaping up to be more than a worthy adversary for Conroe. Even AM2 rev. G could have improved IPC (implementing some K8L features?) when released at 65nm in December and close the gap with Conroe. Go Google the K8L for yourself though, or go hit up AMDzone.com.

[retsam]

edit..opps wrong thread...

[Tikiman44]

No, Kentsfield is 2 Woodcrest cores slapped together like the Pentium D, expect more of the same congested bus I/O and even poorer scaling beyond 2P as 4 cores thrash the front-side bus fighting over bandwidth. Thankfully the Opteron K8L will not suffer this fate as it will be a true 4 core design and will also have an L3 shared cache and most importantly, the cache-coherent Hypertransport bus.

AMD doesn't have anything to counter Conroe with??? Haha! You think they are just going to rest on their laurels as Conroe finally gets out the door? What about 4x4, or AM2 Rev. G, the Turion X2 rev. (codename Bulldozer) or later on, K8L??? The real story is we don't know yet who is going to "own" who in about a year as it's all conjecture at this point in time. I'm sure however, it's safe to say AMD is not going to let Intel's response go unanswered or uncontested for too long, as thier market share on the desktop and notebooks is at stake.

The recent AMD Developer's Conference shed alot of light on the K8L and it's shaping up to be more than a worthy adversary for Conroe. Even AM2 rev. G could have improved IPC (implementing some K8L features?) when released at 65nm in December and close the gap with Conroe. Go Google the K8L for yourself though, or go hit up AMDzone.com.

Someone is a fanboy...LOL!

[Banko]

No, Kentsfield is 2 Woodcrest cores slapped together like the Pentium D, expect more of the same congested bus I/O and even poorer scaling beyond 2P as 4 cores thrash the front-side bus fighting over bandwidth. Thankfully the Opteron K8L will not suffer this fate as it will be a true 4 core design and will also have an L3 shared cache and most importantly, the cache-coherent Hypertransport bus.

AMD doesn't have anything to counter Conroe with??? Haha! You think they are just going to rest on their laurels as Conroe finally gets out the door? What about 4x4, or AM2 Rev. G, the Turion X2 rev. (codename Bulldozer) or later on, K8L??? The real story is we don't know yet who is going to "own" who in about a year as it's all conjecture at this point in time. I'm sure however, it's safe to say AMD is not going to let Intel's response go unanswered or uncontested for too long, as thier market share on the desktop and notebooks is at stake.

The recent AMD Developer's Conference shed alot of light on the K8L and it's shaping up to be more than a worthy adversary for Conroe. Even AM2 rev. G could have improved IPC (implementing some K8L features?) when released at 65nm in December and close the gap with Conroe. Go Google the K8L for yourself though, or go hit up AMDzone.com.
Woodcrest is Conroe pretty much, and people already got a hold of the ES of Kentsfield and actually ran benchmarks at xtremesytems.org.

Edit:

Sucks that the Woodcrest benchmarks showed it completely killing any Opteron.

[a12ctic]

Someone is a fanboy...LOL!
coming from tiki...anyway conroe isnt even that big of a deal, its sorta a in between chip, sure its the best for a couple months, but itll be gone in a year, just the the p4s, they were the best for a few months and then amd blew them away

[Zelda_fan]

coming from tiki...anyway conroe isnt even that big of a deal, its sorta a in between chip, sure its the best for a couple months, but itll be gone in a year, just the the p4s, they were the best for a few months and then amd blew them away

LOL Conroe is a HUGE deal. It bent AMD over the barrel and gib-erized it.

[a12ctic]

15% performance on a 4+ year old chip=/=a big deal

[Lazaredz]

Hmm .... well I personally don't see any reason to hold out for a Kent and with the K8 on hold till 2008, maybe a Conroe wouldn't be a bad upgrade for someone still with a Northwood. Well just have to see how much those AMD prices drop.

[Zelda_fan]

15% performance on a 4+ year old chip=/=a big deal

Really? The FX-62 is 4 years old? ....

If you compare the Conroe to what AMD had out four years ago, the performance delta would be absurd. And it's a good bit more than 15% btw.

And the speed isnt' even the biggest perk, it's the price. There is a $350 chip that outperforms AMDs top dawg.

[jAkUp]

15% performance on a 4+ year old chip=/=a big deal

As I brought up to you earlier, the 939 CPU was about 15% faster than a 4 year old chip.

[Sazar]

No, Kentsfield is 2 Woodcrest cores slapped together like the Pentium D, expect more of the same congested bus I/O and even poorer scaling beyond 2P as 4 cores thrash the front-side bus fighting over bandwidth. Thankfully the Opteron K8L will not suffer this fate as it will be a true 4 core design and will also have an L3 shared cache and most importantly, the cache-coherent Hypertransport bus.

Kentsfield early lab test actually show excellent scaling and excellent use of the memory controller. Even if it is not on-die, it has excellent throughput. Consider the release model with final hard-ware to be much better overall.

The k8l has some nice potential, remains to be seen how evolutionary it really is.

Pentium d was indeed true dual-core. It just wasn't nearly as good as the k8 based solutions from amd.

AMD doesn't have anything to counter Conroe with??? Haha! You think they are just going to rest on their laurels as Conroe finally gets out the door? What about 4x4, or AM2 Rev. G, the Turion X2 rev. (codename Bulldozer) or later on, K8L??? The real story is we don't know yet who is going to "own" who in about a year as it's all conjecture at this point in time. I'm sure however, it's safe to say AMD is not going to let Intel's response go unanswered or uncontested for too long, as thier market share on the desktop and notebooks is at stake.

Right now, AMD does not have a solution to compete on the server space or the consumer space when you see a wide-spread release of the core 2 based procs.

I have said for about half a year or so that Intel has done the right thing and is continuing to move forward with low TDP products. They have a lead over AMD currently and Intel/Israel continues to innovate. Kudo's to the design team there.

The recent AMD Developer's Conference shed alot of light on the K8L and it's shaping up to be more than a worthy adversary for Conroe. Even AM2 rev. G could have improved IPC (implementing some K8L features?) when released at 65nm in December and close the gap with Conroe. Go Google the K8L for yourself though, or go hit up AMDzone.com.

AM2 is a DIFFERENT marchitecture from k8l. What kind of k8l features do you expect to be implemented on something like that?

K8l is interesting but right now its all about woodcrest and kentsfield BECAUSE they are not only here, right now, but they scale well and the power requirements are very acceptable for the performance.

Potentially improved IPC? Perhaps. Lower TDP? I think not, perhaps, till k8l debuts and we see if there are new techs employed by AMD. Their design teams do not appear to have been working anywhere close to as dilligently as Intel's have to lower power draw.

Both will have good performance/watts, but Intel currently has a lead in r&d that is substantial enough to last for a while, perhaps beyond k8l.

I understand you are a little fanboy, but give Intel credit where it's due. They have an excellent product on the market. And this is all credit to AMD.

[jAkUp]

What about 4x4, or AM2 Rev. G, the Turion X2 rev. (codename Bulldozer) or later on, K8L??? .

4x4? Are you joking? Why on earth would anyone need 2 120w CPU's in their system? $2000 for slower performance in single threaded apps compared to Conroe? I'm sorry, 4x4 is not an answer to Conroe.

[DCult]

While I'm jumping to the Core2 bandwagon too, anyone who thinks that this performance difference between core2 and athlonAM2 is something to cheer about, think again. AMD has a very small market share and it will shrink even further. Monopoly is never a good think. It slows down development, and allows the dominant force to introduce technology at will and at any price point that brings them maximum money. In the worst case scenario, Intel will keep their prices unchanged until AMD reaches or surpasses them in performance. That could be in a year and a half, as someone in IDF said. :(

[Gnu_Raiz]

All I can say is show us the money! Have any of you tried to get a WoodCrest CPU. If you think preorder for Conroe is bad, try getting a WoodCrest!

Then you have to consider the motherboards, Come on Nvidia show us the goods. Heck getting a workstation board ie more than one cpu slot, for WoodCrest is harder then getting the cpu.

What I really want to know is where is the multithreaded apps? What good is all this super hardware without the software?

How many of you still play Oblivion, as that seems to be the only game that stresses the dual gpus, dual core systems. As far as 4x4 is concerned, I like the idea, might not be as fast as a similar WoodCrest system, but show me a motherboard that allows dual Conroes for that targeted price range?

What it comes down to is the software, do you really think most games will stress a 4x4 system? After all you still have to consider the IO, such as harddrives and load times.

Then you have to consider price/performance all I can say is it's a good time to be a consummer, as I win with both systems.

I just have to wonder why any of you guys aren't spouting more about WoodCrest. Surley two WoodCrest's would be better then a single Conroe. I am sure that at some point sli, and Xfire will be available.

[Lou Ceifer]

Someone is a fanboy...LOL!

Pot calling the kettle...

You ain't got any room to talk there Tiki boy, you've been pimping Conroe harder than the hookers downtown. I say to each his own and try dropping the blue colored glasses sometime and see what else is out there. While I am loyal customer of AMD, Conroe is good,(far improvement over the P4) but not great IMHO and this is merely a battle in a long campaign between AMD and Intel.

In reality, nobody "OWNZ" anyone, Intel ain't going anywhere, AMD ain't going anywhere and these product cycles are what spurs competition and also the industry. Enjoy your Conroe and spare us the "OMGZ CONROE OWNZ" fanboyism yourself, thanks. I'm trying to stick to an actual discussion and conjecture here if that's too much to ask.


Kentsfield early lab test actually show excellent scaling and excellent use of the memory controller. Even if it is not on-die, it has excellent throughput. Consider the release model with final hard-ware to be much better overall.

The k8l has some nice potential, remains to be seen how evolutionary it really is.

Pentium d was indeed true dual-core. It just wasn't nearly as good as the k8 based solutions from amd.

While Kentsfield will raise the bar from Woodcrest a bit it still suffers from the same aging FSB problem with Woodcrest that Intel is so stubbornly clinging to.(this is why it NEEDS massive amounts of cache to hide it's bandwidth problems) I will say this, until Intel moves to CSI their bus will not handle the traffic HT is easily capable of with quadcore processors. This is why they will not outperform Opteron on 4P or higher indefinately.

Right now, AMD does not have a solution to compete on the server space or the consumer space when you see a wide-spread release of the core 2 based procs.

I have said for about half a year or so that Intel has done the right thing and is continuing to move forward with low TDP products. They have a lead over AMD currently and Intel/Israel continues to innovate. Kudo's to the design team there.

AMD doesn't have a solution!?!? What? You obviously haven't gotten the memo that 2P is quickly becoming a thing of the past in mainstream servers in my research, 4P is the way it's becoming thanks to Opteron's price and scalability thanks to Hypertransport. For the price you pay for a 2P Intel server you can get a 4P Opteron setup with better performance and scalability, that's a fact and a reason Sun is doing so well selling Opterons right now.


Potentially improved IPC? Perhaps. Lower TDP? I think not, perhaps, till k8l debuts and we see if there are new techs employed by AMD. Their design teams do not appear to have been working anywhere close to as dilligently as Intel's have to lower power draw.

Both will have good performance/watts, but Intel currently has a lead in r&d that is substantial enough to last for a while, perhaps beyond k8l.

Sorry Saz, but I have yet to see AMD really falter on thier promises, unlike another company's name I will not mention. You also fail to understand AMD is also employing the likes of a design team Intel would only DREAM of, thier Isreal team is good, but not good as some of the pros in the industry who have doing it far longer.(and are now working for AMD!) Even one of Intel's former Itanium designers is on thier staff now, along with ex DEC, IBM, Alpha and Sparc engineers. You can't tell me they don't have some brilliant ideas to counter Intel, these guys are some of the best in server architecture design with many years under thier belts in thier respective areas.

Also a big point of my contention with throwing around the concept of TDP between AMD and Intel is AMD is MAXIMUM TDP and Intel's is AVERAGE TDP. Big difference here. I don't see Intel having that much of a lead in this area honestly as AMD's move to 65nm will help immensely in this area and the silicon will be here in December. I have read from this process change they will reduce power consumption by 40% from the reported estimates.

The only reason Intel is ahead in these areas right now Sazar is because you are comparing a 5 year old processors design and thermals(and performance) with a brand new processor which isn't much of a comparison period. Once AMD hit's 65nm hits the shelves I feel you may be in for a surprise, you will see just how advanced AMD's 3rd generation SOI will lower leakage and power draw. Unless AMD pulls a fast one and slips in some other core revisions to increase performance then AMD will have to pump the MHz to regain performance parity with Core2.

AM2 is a DIFFERENT marchitecture from k8l. What kind of k8l features do you expect to be implemented on something like that?

K8l is interesting but right now its all about woodcrest and kentsfield BECAUSE they are not only here, right now, but they scale well and the power requirements are very acceptable for the performance.

Thanks for the heads up Capt. Obvious. As I have said this is all conjecture, who is to say AMD isn't going to respond to Core2 with some core changes along with the 65nm process shrink to increase IPC? I would think AMD is capable of doing changes on short notice thanks to thier manufacturing advantage with thier APM software in FAB 30 and FAB 36.

For the uniformed here is a excellent thread on AMD's K8L plans on Extremesystems by poster "Oldblue":

http://www.xtremesystems.org/forums/showthread.php?t=107308


I understand you are a little fanboy, but give Intel credit where it's due. They have an excellent product on the market. And this is all credit to AMD.

I realize you are nothing but a fanboy yourself Sazar but I have already given Intel where's it credit is due, I just don't see Conroe, Woodcrest or Merom as the second coming of Jesus like some of you guys do, so whatever. I won't be upgrading my rig for quite some time anyway, all of this is merely amusing to watch honestly and I enjoy the banter. I'm all for competition so kudos to Intel for spuring on AMD to reduce prices and raising the performance bar FWIW.

4x4? Are you joking? Why on earth would anyone need 2 120w CPU's in their system? $2000 for slower performance in single threaded apps compared to Conroe? I'm sorry, 4x4 is not an answer to Conroe.

Who says they will be 120 watt CPUs when 65nm AM2's are here? Who says AMD won't give massive price cuts to all thier processors to answer Core 2's pricing? Slower performance in single threaded apps compared to Conroe, ha! By the time Vista hit's the market, the mainstream 64-bit OS will have ARRIVED.(and this is where AMD64 really shines!) The 4x4 board is going to be NUMA aware and Vista's implementation of it will be far superior than Win XP's. You are applying Intel's dated FSB architecture mentality to a vastly superior Direct Connect architecture and Hypertransport of AMD64 which will totally blow it out of the water for overall bandwidth and communication. Each socket on a 4x4 will have equal bandwidth and direct access to each other through the HT bus and linked together almost like an X2 processor. It's not much different than Nvidia's SLI from the infomation I am reading, add a processor to your existing one and gain a 80% increase in performance.

FWIW, 4x4 was never a true answer to Conroe, it's a platform of it's own, aka, a workstation and cheap 2P server platform since it does not need more expensive registered ram. This is the whole reason for 4x4 IMHO and it's a good one, being able to potentially best Core 2 with the platform is just a side affect.


To all you Core 2 fans, Intel's done good with it's new processor line.(on desktop and notebooks, in the server area remains to be seen) However, this is not to say AMD is down or out, by any means. They too have some great stuff cooking and I know they won't settle for being #2 in performance for long.(much less lose the marketshare they have taken from Intel over the past few years) Enjoy your processor and feel free to tone down the hostile attitude to anyone who doesn't bow before Conroe, it's just a friggin processor for God's sake, not everyone is going to agree with your viewpoint.(you know who you are)

[Tr1cK]

:owned:

[jAkUp]

Who says they will be 120 watt CPUs when 65nm AM2's are here? Who says AMD won't give massive price cuts to all thier processors to answer Core 2's pricing? Slower performance in single threaded apps compared to Conroe, ha! By the time Vista hit's the market, the mainstream 64-bit OS will have ARRIVED.(and this is where AMD64 really shines!) The 4x4 board is going to be NUMA aware and Vista's implementation of it will be far superior than Win XP's. You are applying Intel's dated FSB architecture mentality to a vastly superior Direct Connect architecture and Hypertransport of AMD64 which will totally blow it out of the water for overall bandwidth and communication. Each socket on a 4x4 will have equal bandwidth and direct access to each other through the HT bus and linked together almost like an X2 processor. It's not much different than Nvidia's SLI from the infomation I am reading, add a processor to your existing one and gain a 80% increase in performance.

From what I have read, 4x4 will work great for MULTITHREADED apps. Not single threaded apps. It will NOT be like SLI.

Also, AMD said they will not drop their FX CPU's! That is $2000 for 4x4. And if Vista implementation of mutliple cores will be better, I suppose this will be better for Kentsfield as well.

4x4 will be beneficial in a limited number of games, Crytek announced that Crysis will work with 4x4. That tells you right there that 4 cores will be limited to a small amount of games.

[Gnu_Raiz]

I would not count on Vista getting a really good implementation of multi-cored cpu's, it might be better then XP, but that is not saying much.

What really needs to be done is to have Quake Wars, and Crysis have a Linux client then we would have the best of both worlds. Like I said in my other post show us the software?

[a12ctic]

From what I have read, 4x4 will work great for MULTITHREADED apps. Not single threaded apps. It will NOT be like SLI.

Also, AMD said they will not drop their FX CPU's! That is $2000 for 4x4. And if Vista implementation of mutliple cores will be better, I suppose this will be better for Kentsfield as well.

4x4 will be beneficial in a limited number of games, Crytek announced that Crysis will work with 4x4. That tells you right there that 4 cores will be limited to a small amount of games.
huh? the fx62 was definatly on their price cut list...

[grimreefer]

hmm, should i buy a kentsfield(4cpu's), or should i pay extra for the dual clovertowns(that would be 8 cpus)? depends if asus makes a good overclocking server board...
having only 2 cores is sooo last year lol(and my rig is over a year old, but it still has a pentium d)

[Banko]

huh? the fx62 was definatly on their price cut list...
Yes but it was cut down to $799, when an e6600 which is still 2x cheaper beats it.

[licker]

Kentsfield will be arriving in our lab sometime this week for testing, All the testing we do here is server side but hopefully I can gather some numbers for you guys when they get here :D. Another great advantage of the kentsfield is hyperthreading.. so think of it this way (4 cores with hyperthreading = 8c) This chip will do wonders with apps that can actually utilize multiple cores but unfortunately not really any games out there that can do the same, so for the meantime this chip is really targeting the server market :(

[evilchris]

coming from tiki...anyway conroe isnt even that big of a deal, its sorta a in between chip, sure its the best for a couple months, but itll be gone in a year, just the the p4s, they were the best for a few months and then amd blew them away
yeah AmD Si BSTteEr CUZX InTel HAA sLOTSSA LOTsA MOnEy1!! HEHehEH IMA rcTIC I OnlY BUy The UdnerdOG!!