AMD has changed release of AM2

[Starscream]

Seems AMD isnt gonna release AM2 at 6/6/06

but at the end of may. rumor has it that these are the dates:

* May 16, 2006: Global announcement of Energy Efficient Processor roadmap and pricing
* May 23, 2006: Global announcement of Socket AM2 and new desktop product availability and pricing
* May 21, 2006: Global announcement of AMD LIVE! desktop system availability

reason: CPUs and motherboards are already finished.

link to the article at dailytech:
http://www.dailytech.com/article.aspx?newsid=1854

[physics_geek]

6/6/06!!! AMD is being superstitious here. :P
Seriously, Conroe might be another reason.

[nn_step]

sounds good to me

[BSill]

AM2 procs right now only provide a 1 to 7% performance increase while in game. And that is with higher end ram. Ex: OCZ's new DDR2 xtc. I still dont see the point of switching over from 939 until 2007. It seems they are rushing the release.

[Starscream]

AM2 procs right now only provide a 1 to 7% performance increase while in game. And that is with higher end ram. Ex: OCZ's new DDR2 xtc. I still dont see the point of switching over from 939 until 2007. It seems they are rushing the release.


i dont think their rushing it.
boards and CPUs have been floating around for a while now i think the products r simply already finished

[BSill]

Maybe I confused you. My point meaning 1 to 7% performance increase seems pretty pointless to release the product.

[DaWaN]

Maybe I confused you. My point meaning 1 to 7% performance increase seems pretty pointless to release the product.

No it isn't, Intel is atm almost for 100% on DDR2, AMD can't stay with DDR1..

[Cobalt]

I think they just want to setup a user base so when the real improvements come those people will be more willing to upgrade and any problems with the socket can be ironed out. They also make more money this way because people will essientially be paying for the same thing twice, now and whenever K8L comes out. Of course a lot of people will probably hold out till then anyway but it still gets the product into the market.

[OmegaMerc]

Maybe I confused you. My point meaning 1 to 7% performance increase seems pretty pointless to release the product.

If this is true, then you wouldn't find half the video cards in the market available, since their performance gain is so low from one to another that it voids any reasoning.

[BSill]

So you are saying the performance difference from the x800 series to the x1900 series is minimal?

The release of the x1800 to the x1900 is a good example of 939 to AM2.
The release of the 7800 to the 7900 series is a good example of 939 to AM2.

I'm not saying its wrong, I'm just saying it would be a better idea to save about 800 dollars (AM2 CPU + AM2 Compatable Mobo) and put that 800 dollars towards a crossfire or sli setup that would give you a MUCH HIGHER in game performance increase.

[Thorry]

Why the hell would a new release automatically mean better performance???

In the CPU world this has NEVER been true.

To tell you guys the truth I'm getting damned sick of the entire thing. It's like everybody has forgotten the last 15 years...

Also AMD can't lag behind Intel, and I'm sick of seeing AM2 compared to Conroe. AM2 is next month, Conroe is unknown (but slated Q3 or Q4)... my Pentium 3 also was slower as my X2 now... :S

[BSill]

Why the hell would a new release automatically mean better performance???


Thank you for proving my point. I dont understand either why people think newer is better. Yes there are SOME examples disregarding processors.

[[XC] leviathan18]

when intel changed to ddr2 it was backstep in perfomance amd has little gain conroe is a new Marchitecture AM2 is just a new socket with a new IMC no new Marchitecture or new process i..... compare AM2 with current gen chips 939 or intel p4 not conroe... for conroe is coming another thing

[Thorry]

Very true, AMD has just been waiting for the performance of DDR2 to catch up with the performace they were getting out of DDR. Now that pricing and performance is basically equal to DDR (even a 3 to 5% increase) AMD is switching.

The AM2 CPUs will be just the same as the 939 CPUs except for the memory controller. (And some improvements on some but not all models)

Now people are flipping saying: Well the AM2 3800+ is no faster as the 939 3800+ or a AM2 CPU clocked at 2.5 ghz is only 3 to 5% faster as a 939 CPU clocked at 2.5 ghz...

AMD is going to introduce some higher clocked CPUs, also when buying a new PC now you can hang on to your memory and maybe even your mobo for some time. Also if you are now running a preshott with DDR2 memory the switch to AM2 is easier.

Just hold on to your hats when the 65nm production reaches the market, can anybody say 4 ghz X2 with 16mb L3 cache?

[dave_graham]

Maybe I confused you. My point meaning 1 to 7% performance increase seems pretty pointless to release the product.

lol

you're kidding, right?

ANY platform change is accompanied by the requisite failures/strengths of that new platform. For AM2, higher bandwidth applications will benefit more from DDR2 than the S939 platform. incidental latency aside, the average user (read: 99.3% of customers out there) will notice absolutely nothing different than the .7% of customers who think that benchmarking is indicative of complete system performance. :rolleyes:

sorry, bud, but Intel and AMD make money off of "real life" applications, not power users that try to scape another second of of SuperPI. 1-7% performance increase is what I expect from any evolutionary change to the AMD platform design.

in any case, spending $800 on a new AM2 platform isn't exactly the cost it'll be... pricing parity is there with S939 on the processor parts and the mainboards will be a little pricey initially.

dave

[|3ourne]

ok , so at stock , AM2 is roughly equivalent to Skt 939. But you all fail to realise that the true power of DDR2 shows at high speeds. At those speeds ( 800 mhz+) the bandwidth is more than you see from DDR most of the time. Combine that with the super efficient mem controller of AMD , and overclock that combo ( what we all do on XS neways ) , hopefully we will see some good ocs higher than current skt 939. A couple of tests dont prove anything. Wait till real world users try and overclock them.

Im not trying to take AMD's side but speculation is running rampant lately here.

[NickS]

Yeah but is that going to increase the performance enough to justify a rather expensive upgrade?

The performance difference between Socket A and socket 754/939 was CRAZY. IMO it's stupid to re-release the same architecture on 3 sockets. s754, s939, now AM2 and more all based on K8? AMD's pulling an intel ;) I think I'm going to wait for the next mainstream socket, as socket AM2 seems to be like the "new" s754 :(.. or maybe I'll pick up a Conroe when they're released.

Nick

[johnanderson]

hmm. i dont know how much amd can improve their cpu from now until june. in any case if they dont come out with a good conroe competitor then i guess theyll just have to lower prices, which = win for me :D

[Thorry]

Yeah but is that going to increase the performance enough to justify a rather expensive upgrade?

The performance difference between Socket A and socket 754/939 was CRAZY. IMO it's stupid to re-release the same architecture on 3 sockets. s754, s939, now AM2 and more all based on K8? AMD's pulling an intel ;) I think I'm going to wait for the next mainstream socket, as socket AM2 seems to be like the "new" s754 :(.. or maybe I'll pick up a Conroe when they're released.

Nick

Have you been with your head in the s hit all this time?

For your information:

S754 has been around a long while and will be around for some while to come, what you mean to say is AM2 is the new 939.

Also what do you want? AMD to stick with DDR? That just not realistic.

Conroe.... yeah when it comes, that'll be a while.

I agree that the upgrade costs are high, but if you have a nice high performance 939 system with a PCIe you really don't need to upgrade.
This platform is now mainly aimed at people that now have a 775 system and don't want to wait for Conroe, they only need to upgrade the CPU and the mobo (which is pretty normal when getting a new CPU anyway).

If you now have a 754 or 939 with AGP you'll be in for a world of hurting, you need a new CPU, new mobo, new GPU and new memory... but it'll be worth it.

A new CPU platform has in the past never ment a performance upgrade, but you will have the maximum upgrade time if you get a AM2 system right away.

[BSill]

in any case, spending $800 on a new AM2 platform isn't exactly the cost it'll be... pricing parity is there with S939 on the processor parts and the mainboards will be a little pricey initially.

dave


AM2 3800+ = 500 to 550 bucks (what do u think the fx AM2 will cost man?)
Memory = a decent 2gig set will cost 180 bucks on average
AM2 motherboard = 200 bucks average

thats over 800 right there

my point being that money can be spend differently to obtain a larger performance jump

[ozzimark]

AM2 3800+ = 500 to 550 bucks (what do u think the fx AM2 will cost man?)
:confused:

same PR ~= same price.

try around $300 for a 3800+ X2 :stick:

[nn_step]

If quad cores isn't a good enough reason to switch to AM2 I'm not sure what Is.

[BSill]

Oh my bad.

AM2 Proc = 300
Memory = 180 average
Board = 190 average

670 bucks average total
could use it on a second gpu, or a vapo

nn_step - serious question, doesnt intel have quad cores coming out as well?

[Piotrsama]

nn_step - serious question, doesnt intel have quad cores coming out as well?
Not this year, at least not on retail.

[nn_step]

Oh my bad.

AM2 Proc = 300
Memory = 180 average
Board = 190 average

670 bucks average total
could use it on a second gpu, or a vapo

nn_step - serious question, doesnt intel have quad cores coming out as well?
Yes but like Pentium D, it is just going to be two dies fighting over a SINGLE FSB. Which is going to kill the performance.

[dave_graham]

Not this year, at least not on retail.

late Q1, 2007. the tech demo is done already in Taipei. :)

dave

[metro.cl]

its official the embargo ends on the 23th of may.

[zabomb4163]

Yes but like Pentium D, it is just going to be two dies fighting over a SINGLE FSB. Which is going to kill the performance.
and around that time intel will have 45nm quad core.

http://www.intel.com/technology/silicon/new_45nm_silicon.htm

"20 percent improvement in transistor switching speed or more than a five-fold reduction in transistor current leakage."

[nn_step]

and around that time intel will have 45nm quad core.

http://www.intel.com/technology/silicon/new_45nm_silicon.htm

"20 percent improvement in transistor switching speed or more than a five-fold reduction in transistor current leakage."
:stick: Dude we are talking about the CPU to be limited to 133Mhz of Bandwidth per Core. The last Intel processer that had that little bandwidth was the 80486.
We are talking about a Strangle hold that Will KILL the performance. it is kind of like having a 10,000 horse power engine but only feeding it 1cc of Gas per Second.

[Poodle]

:stick: Dude we are talking about the CPU to be limited to 133Mhz of Bandwidth per Core. The last Intel processer that had that little bandwidth was the 80486.
We are talking about a Strangle hold that Will KILL the performance. it is kind of like having a 10,000 horse power engine but only feeding it 1cc of Gas per Second.


I hear ya. All the talk about Conroe is impressive but the Amd tehnology seems better for multi core. Intel might take back some of the desktop market though, with the conroe. On the server market Amd will probably continue to grow or at least not drop sales becuase of the good reputation they've gained.

I'm personally changing sockets due to the new more efficient chipsets. The NF4 chipset is dated with it's high energy consumption, which is a pain in the...

[incurable]

:stick: Dude we are talking about the CPU to be limited to 133Mhz of Bandwidth per Core. The last Intel processer that had that little bandwidth was the 80486.
'133MHz of Bandwidth'? What are you, and id ... oh wait, nn_step, it's YOU! Nevermind. :slapass:

[nn_step]

Dude Clovertown, will require Two processors (thus 8 cores) to share a 1066 mhz FSB.. thus 133mhz per core. Servers are the bread and butter of AMD and Intel really wants a bigger slice of that Pie.

[incurable]

Dude Clovertown, will require Two processors (thus 8 cores) to share a 1066 mhz FSB.. thus 133mhz per core.
Clovertown (which btw is a processor and doesn't require a sibling to work) will be launched on the Bensley platform which features dual independet busses (DIB), connecting each socket via its own private FSB to the northbridge. Depending on the FSB speeds it'll debut at (1066/1333), there'll be at least 2.13 GB/s or 2.67 GB/s available to each of the four cores in each socket.


(The sad thing is that this has been explained to you before.)

[Krug]

I hope that AM2 dosent end up as another s754...I got suckered into that back when AMD64 first came out...

[ozzimark]

I hope that AM2 dosent end up as another s754...I got suckered into that back when AMD64 first came out...
what's wrong with s754? i have a feeling my original s754 machine that i got will outlast my s939 one by a long shot... plus, AMD is putting EOL on s939 before s754.. :banana:

[T2k]

If this is true, then you wouldn't find half the video cards in the market available, since their performance gain is so low from one to another that it voids any reasoning.

Care to elaborate? I always tested/bought the latest top cards and I have never seen any new or prerelease card from ATI or NV with less than 10% faster performance than its predecessor. 10-15% - that's the usual minimum, even without architecture change.

[T2k]

and around that time intel will have 45nm quad core.

http://www.intel.com/technology/silicon/new_45nm_silicon.htm

"20 percent improvement in transistor switching speed or more than a five-fold reduction in transistor current leakage."

You want me to quote from papers Intel/HP weres sending me years ago before we bought our Itanics? ;)

[nn_step]

You want me to quote from papers Intel/HP was sending me years ago before we bought our Itanics? ;)
yes, I'll be really funny.. oh and do you have those Itanics Running R@H yet?;)

[Pegaso]

Why the hell would a new release automatically mean better performance???

I have to agree, when amd pased from 754 to 939 the perfomance boost wasnt really huge :nono:

[T2k]

So you are saying the performance difference from the x800 series to the x1900 series is minimal?

The release of the x1800 to the x1900 is a good example of 939 to AM2.


Umm excuse me? 4x extra PS power is in the range of 1-7%? :) Show me any area where AM2 will beat 939 by this margin...


The release of the 7800 to the 7900 series is a good example of 939 to AM2.


Noway either - IIRC 7900GTX is faster around 25-30% margin in 3dmark05 (say 1920@4xAA)


I'm not saying its wrong, I'm just saying it would be a better idea to save about 800 dollars (AM2 CPU + AM2 Compatable Mobo) and put that 800 dollars towards a crossfire or sli setup that would give you a MUCH HIGHER in game performance increase.

I think your commanding sentence is this one:

I still dont see the point of switching over from 939 until 2007.

And this is absolutely true, I think, unless AMD will pull some new trick around Fall. AM2 is nothing but a transition to DDR2 for future memory bandwidth. Unfortunately this time AMD couldn't make another magical memory controller like it did with DDR - this time mem controller sux big time. :(

[T2k]

yes, I'll be really funny.. oh and do you have those Itanics Running R@H yet?;)

Hehe, I have them saved somewhere... TBH I believed at the time for a short period - then they arrived and I was really pissed very soon.
OTOH we didn't really have a choice back then: at the time there wasn't any other platform you could get a machine w/ 16GB RAM and run your Windows-based app on it - I was only pissed by the fact that its 32bit performace was so ridiculous that it was totally useless for anything else than our single speacial IA64 apps (which was also buggy, of course.)
It was a good lesson - I never believed a word of Intel ever since.:cool:

[Ailleur]

:stick: Dude we are talking about the CPU to be limited to 133Mhz of Bandwidth per Core. The last Intel processer that had that little bandwidth was the 80486.


Funny how the utter stupidity of that comment went unnoticed. It goes to show how.. well... to put it nicely.. how retardedly stupid you are

[P8baller07]

Sounds like another s754 to 939 to me

Won't be a big difference at first, but AMD's newer beefy processors will
mostly AM2 processors.

[BSill]

I think your commanding sentence is this one:

And this is absolutely true, I think, unless AMD will pull some new trick around Fall. AM2 is nothing but a transition to DDR2 for future memory bandwidth. Unfortunately this time AMD couldn't make another magical memory controller like it did with DDR - this time mem controller sux big time. :(

I don't think they are going to pull anything out of their hat. Most likely will release AM2 as 80nm like we were informed sense they even boosted up the release date to 6/6, bascially same architecture with ddr2 memory controller. Getting ready for DDR2 before it takes way in the future. I think they'll get it out to us as they stated, see how it clocks and then focus their attention on the 65nm version (i believe its 65 correct?) for next year. By their 2007 releases we'll have sb600, r600 and g80 hopefully available to work with. Wouldnt bother upgrading till q1 of 07 which will most likely be pushed back to late Q1 maybe early Q2 just guessing?

Edit: Typo

[ozzimark]

Funny how the utter stupidity of that comment went unnoticed. It goes to show how.. well... to put it nicely.. how retardedly stupid you are
it makes perfect sense to me?

[darkhelmet]

i just know i'm pumped to upgrade my preshott 540 and P5AD2 Premium....although a nice mobo it is

[mesyn191]

Thank you for proving my point. I dont understand either why people think newer is better. Yes there are SOME examples disregarding processors.

this is a release for economic concerns, not so much performance IIRC, at least at launch anyways.

DDR RAM is starting to become expensive vs. DDR2 (good for thier OEM partners, saves them money building AMD machines..), performance on DDR will remain stagnant (at stock...) while DDR2 will still scale some too you know.

Also allows them to unite the dekstop sockets, no more 754 and 939, just socket 940 for value and performance markets.

Bear in mind of course if you've already got a A64 or higher clocked P4 upgrading to a AM2 system would be a waste of money, but lots of people have older machines and for those looking to upgrade before Conroe comes AM2 is a nice upgrade.

[mesyn191]

Unfortunately this time AMD couldn't make another magical memory controller like it did with DDR - this time mem controller sux big time.

Eh? Nothing wrong with AMD's DDR2 mem. controller at all. You've seen the Anadtech benches? Lower latency, more bandwidth, and support for DDR2 800. WTF is there to complain about?

[biohead]

Funny how the utter stupidity of that comment went unnoticed. It goes to show how.. well... to put it nicely.. how retardedly stupid you are
then what's your theory, einstein?

oh, and he is probably far more educated than you are.

[incurable]

How about (1) you don't measure bandwidth in Hz and (2) -diregarding (1)- his math is plain wrong (which he was told only days before this comment). Stupid enough?

[metro.cl]

clockspeeds can go higher that s939 :)

[FireDragon]

I would assume that there will be more initial benefits to AM2 then just higher bandwidth why else would they be PUSHING the release date sooner then later? You would think that for now they would do everything that they could even if it was only a few percent on the procs to compete with conroe...those r my thought...i am not expecting a miracle chip here nor one NEAR as fast as conroe but they arnt going to take it lying down...my guess is that by the time am2 is out and conroe is out there wont be more then a 10% diff in them unless you are spi freak (lol we know there is ENOUGH of them around here)

But if you ask me on the release it will be on average 10% faster then 939...they wouldn’t make a pointless move to integrate ddr2 right now with intels market share and conroe on the horizon it would be stupid to waste their resources to do that...it WILL be faster then 939 just not what we hopped even more so after seeing the the thrashing that conroe is going to give EVERY other cpu out there...We just had our hopes set high because we all saw what conroe can do...


Dragon

[nn_step]

All we have seen is Conroe is good at SuperPI, Conroe is just a tweaked Yohan.. So don't expect it to be great at gaming...

[largon]

Yea, Conroe gaming performance (http://www.anandtech.com/tradeshows/showdoc.aspx?i=2716&p=3) is truly woeful.

:slap:

[NickS]

That article is comparing the current gen. Athlon 64's afaik... so that is a completely un-fair generalization about Conroe vs AM2. Athlon 64 v3 (AM2 haha) could completely and utterly rape Conroe in gaming benches. We haven't seen enough non-biased benches/comparisons yet to correctly judge the new crownholder in the gaming arena.

Don't go on about SuperPi either. Intel has always gotten their :banana::banana::banana::banana:s and giggles by counting numbers.

Nick

[mesyn191]

I doubt A64 will rape Conroe in game, but I'd expect it to do OK.

My guess is AMD will end up having to compete on price rather than performance for a while, or for a long time if K8L aint all that its cracked up to be...

[JumpingJack]

I hear ya. All the talk about Conroe is impressive but the Amd tehnology seems better for multi core. Intel might take back some of the desktop market though, with the conroe. On the server market Amd will probably continue to grow or at least not drop sales becuase of the good reputation they've gained.

I'm personally changing sockets due to the new more efficient chipsets. The NF4 chipset is dated with it's high energy consumption, which is a pain in the...


Have you really studied up on the Conroe specifics (better generalized the Core uArch.) in terms of multi-core implementation. AMD's multicore approach requires internal cross-CPU connections for inter-cpu communictions (i.e. Cross bar -- nice, true), but the new architecture from intel shares and allocates at the cache level, in the end much more effective and frankly much more elegant. Now, data shared by two threads running can be accessed by either core within the same cache block. AMD still would have to swap data around between cache's, so in the end AMD will still have a certain degree of cache coherency to deal with.

[metro.cl]

All we have seen is Conroe is good at SuperPI, Conroe is just a tweaked Yohan.. So don't expect it to be great at gaming...


the more i hear from you the less i think you know....

[Ubermann]

I hope we get a total war btw AMD and Intel now so we can skip all minor 200mhz updates and get something totaly new.

[metro.cl]

I hope we get a total war btw AMD and Intel now so we can skip all minor 200mhz updates and get something totaly new.


am2 rev G is gonna be more than 200mhz but from what i know it will be out on 2007 that is goin to put amd on top again (after conroe is released).

[thunderstruck!]

the more i hear from you the less i think you know....
I have no idea what he's talking about either.

[largon]

metro.cl,
I gather you'r talking about K8 @ 65nm (= Brisbane)?

[metro.cl]

metro.cl,
I gather you'r talking about K8 @ 65nm (= Brisbane)?

there are 2 revisions of am2 the one that is coming out in the 23rd is rev f and is mostly a s939 with ddr2 mem controller.

on q1 2007 rev g is coming out that is a 65nm am2 with more cache

[Piotrsama]

Just a question.
Socket AM2 procs will bring the virtualization techs (Pacifica?), rigth?

[xenolith]

Just a question.
Socket AM2 procs will bring the virtualization techs (Pacifica?), rigth?

Yes.

[xenolith]

That article is comparing the current gen. Athlon 64's afaik... so that is a completely un-fair generalization about Conroe vs AM2. Athlon 64 v3 (AM2 haha) could completely and utterly rape Conroe in gaming benches. We haven't seen enough non-biased benches/comparisons yet to correctly judge the new crownholder in the gaming arena.

Don't go on about SuperPi either. Intel has always gotten their :banana::banana::banana::banana:s and giggles by counting numbers.

Nick

Let get this straight, so you're speculating a bump in bandwidth from DDR2 and mem controller tweaks could possibly help an equally clocked AM2 rape Conroe in game performance?