AMD: Barcelona "introduction" at end of Q2, in market Q3

An interview with Hector Ruiz and Henri Richard.

You've got to love a reporter who asks direct questions:

CRN: What will be the performance of Barcelona vs. Intel's quad-core?

RICHARD: It's in the speculation stage. [But] we know that we expect performance gains in the 40 percent to 70 percent range, and we know the scalability of our architecture is better than the competition. And it's no secret they have to introduce a new architecture at some point, so we know Barcelona will perform exceptionally well and will be outperforming the competition. The program is on track, .

RUIZ: The best measure of whether Barcelona is going to be good, great or fantastic is really our customers, and the feedback from them is so strong.

CRN: When will Barcelona ship?

RICHARD: It's slated for introduction at the end of the second quarter and will be in the market in the third quarter.

Those performance gains of 40 to 70 percent are probably rev. G quad-core vs. rev. F, which is what they've been saying all along.

40 to 70 percent? And still a 30% gap which doesn't make a whole lot of sense right now to be technical. Ok then, if this is remotely true we should see some (leaked) benchmarks within the next few weeks. Hopefully, they will be held here on this forum.

thats awesome

i wnana see some benchies as intel + crossfire = not so good of a future

and intel quad core + 680i = no fsb above 310fsb for aircooling

so mutli GPU + quad core is hard to choose

so yeah i hope K8L has good overclocking headroom. with conroe ur seeing chips OCing 50% at least

I want to see some K8L OC'd vs. OC'd Kentsfields/Conroe. Hopefully AMD gets its act back together and delivers some serious competition to the market.

It's k10 isn't it?

k8l is a low energy k8, iirc?

Quote:

---

Originally Posted by 3NZ0

It's k10 isn't it?

k8l is a low energy k8, iirc?

---

understatement of the year. Think major upgrade and massive improvements

i hope they deliver what they promise..in fact i hope they understated themselves and over deliver.

Then it'll put Intel on it's feet...dang nabbit, i won't be able to upgrade for another 2 years..i gotta wait till the dust settles and their's a defined winner.

AMD Quad Core @ 3.0Ghz + SLI'ed Geforce 8900 GTX's + 4 GB's of DDR2 RAM...mmm I can't wait!

awwww :banana::banana::banana::banana:.... I hate to bring this up, But What does Conroe do V K8 clock for clock, is it 30% faster than K8? If so, than were looking at 10% to 50% faster than Conroe?

Very interesting.

~Mike

Quote:

---

Originally Posted by 3NZ0

It's k10 isn't it?

k8l is a low energy k8, iirc?

---

It's not clear what this generation of cores should be called. The Inquirer has called it K8L, then Rev. H, then K10.

As far as I know, AMD has never given any official name to the set of processors based on the new cores. It sure would help if they did.

They have given names to the individual processors. The first of these will be Barcelona.

Quote:

---

CRN: What will be the performance of Barcelona vs. Intel's quad-core

---

Quote:

---

RICHARD: It's in the speculation stage. [But] we know that we expect performance gains in the 40 percent to 70 percent range, and we know the scalability of our architecture is better than the competition.

---

I'm sorry, but am I reading this wrong? From what it says from the OP it looks to me like the 40-70% will be over Kentsfield since isn't that "Intel's Quad-Core"?

Some one I know who works at AMD said we'll see ES with benchmarks before launch. He wouldn't say anything else, except that AMD will be back.

Quote:

---

Originally Posted by oldblue

It's not clear what this generation of cores should be called. The Inquirer has called it K8L, then Rev. H, then K10.

As far as I know, AMD has never given any official name to the set of processors based on the new cores. It sure would help if they did.

They have given names to the individual processors. The first of these will be Barcelona.

---

Actually, Henry Richard did refer to the new architecture as 'K8L' in his interview with digitimes.com....
http://www.digitimes.com/bits_chips/a20060314PR200.html

Quote:

---

Originally Posted by Chris Hall, DigiTimes.com, Taipei Tuesday 14 March 2006

Q: AMD has indicated it will have major announcements at Computex Taipei, in June. Would a new core be something you'll be announcing at Computex?

A: Well, if we’re going to be more open about 2007 plans, I think Computex would be a good time to start saying a little more about it since that’s where we are introducing our AM2 new infrastructure. That's not to say we're going to present K8L at Computex – don't get me wrong – but I think that that would be a good time to start to disclose more about the future because one of the strong attributes of our roadmap, both in 2006 and 2007, is socket compatibility. The nice thing we're going to do is to deliver to customers. Whatever improvements K8L will provide, they will be applicable to some of the sockets we will be introducing. Therefore, there's a certain logic, to my mind, in disclosing more at that time.

---

Quote:

---

Originally Posted by EternityZX9

I'm sorry, but am I reading this wrong? From what it says from the OP it looks to me like the 40-70% will be over Kentsfield since isn't that "Intel's Quad-Core"?

---

When they say "performance gains", I don't think they mean "performance advantage." I think they mean that this is the amount by which their performance will improve. This is consistent with the numbers they put up at the Virtual IT Experience. Skip ahead to 2:30 to 2:40. That's where the 40% and 70% probably come from.

Quote:

---

Originally Posted by VulgarHandle

Actually, Henry Richard did refer to the new architecture as 'K8L' in his interview with digitimes.com....
http://www.digitimes.com/bits_chips/a20060314PR200.html

---

That's true, but it's only after the Digitimes interviewer said, "There have been reports, notably on The Inquirer website, that AMD will in fact return fire with its own new core in 2007, the K8L." I believe Richard was just repeating the terminology that the reporter was familiar with.

More recently, in the Virtual IT experience there was a chat session in which an AMD representative said, "K8L is not a real product - this was something that was created as a guess by the Inquirer." And then there is this. AMD refers to the first quad core processor as "Barcelona," but I still don't know what to call the new family of processors.

Let me get this straight. Isn't the "Barcelona" version of the K8L/K10 a socket F+ Opteron? And the "Agena" (quad-core) and "Kuma" (dual-core) are desktop versions of K8L/K10, right? Isn't "Agena" a socket F+ too?

If so, will mobo makers be releasing socket F+ mobos in Q3 for the destop?

Quote:

---

Originally Posted by xenolith

Let me get this straight. Isn't the "Barcelona" version of the K8L/K10 a socket F+ Opteron?

---

Barcelona is an Opteron, rumored to be socked F+, and definitely compatible with socket F. Here's the official server roadmap.

Quote:

---

Originally Posted by xenolith

And the "Agena" (quad-core) and "Kuma" (dual-core) are desktop versions of K8L/K10, right? Isn't "Agena" a socket F+ too?

---

To my knowledge, AMD has revealed very little about the desktop K8L generation. Even the code names are just rumor at this point.

With that said, metro.cl recently dropped this. (Thanks metro!) So according to rumor we would have

Agena FX: Socket F+
Agena: AM2+.

I think it would be similar to how we currently have

FX-7x: Socket F
X2: AM2

Thanks oldblue. Agena FX w/Socket F+ and Agena w/AM2+ makes sense to me now.

Damn I wish AMD would leak some benchies or start sending out ES samples soon. I just hope this Quad Core chip is as good as AMD says it will be. I remember back in the day when everyone was waiing for the first A64 benchmarks:D and look how well K8 turned out.

bah! Now another thing giving me buyers remorse even before i buy... darn it!

Quote:

---

Originally Posted by LOE

That means 40% faster in some apps, faster in other apps, up to 70%

I don't think 40-70% faster than core 2 quad is possible, but 40-70% faster than K8 is somewhat doable.

Dunno bout you but I cant wait for some benches, I want to get a new workstation and want to know if its worth waiting

---

IT'S AMD :stick:

Quote:

---

Originally Posted by LOE

That means 40% faster in some apps, faster in other apps, up to 70%

I don't think 40-70% faster than core 2 quad is possible, but 40-70% faster than K8 is somewhat doable.

Dunno bout you but I cant wait for some benches, I want to get a new workstation and want to know if its worth waiting

---

40-70% faster than the COMPETITION. He's talking about Intel Quad-Core :stick:

Quote:

---

Originally Posted by LOE

I think he isn't, if you try thinking harder you will get it.

Here is the original:



Now let me break it down a bit:

The question
CRN: What will be the performance of Barcelona vs. Intel's quad-core?

The answer
RICHARD: It's in the speculation stage.




He says GAIN - GAINING is adding to what you already have. That means they expect 40-70% improvement compared to K8

If he indeed wanted to say 40-70% better than kentsfield, he would say they expect 40-70% performance LEAD, or ADVANTAGE

---

Maybe you're right :(

Quote:

---

Originally Posted by oldblue

You've got to love a reporter who asks direct questions:

Those performance gains of 40 to 70 percent are probably rev. G quad-core vs. rev. F, which is what they've been saying all along.

---

A direct and clear question, a direct a clear response and still you manage to take the wrong conclusion... The 40 to 70 percent is against the Intel Quadcore, not against Intel dual core, not against a imaginary AMD rev.F quadcore.

On topic, it's hard to believe the 40 to 70 percent to be true... personally i dont' believe it.

Quote:

---

Originally Posted by LOE

He says GAIN - GAINING is adding to what you already have.

---

What you already have, is a Intel quadcore in the market, a Intel quadcore that was the subject of the question. :fact:

"In the speculation stage"... well if it's 40 to 70% faster than the K8, we just need to do the math, to know how much faster it will be against c2d, so where's the speculation?

Quote:

---

Originally Posted by LOE

You are wrong, the answer to the question is the first part of his sentence, the "It's in the speculation stage"

---

lol go se my edit... daam you're fast :)

I'm going to have to go with LOE on this one, 40-70% over Kentsfield is insane!:eek:

Quote:

---

Originally Posted by LOE

In a way he is saying "dunno how much faster K10 will be over kentsfield, but we know we improve our performance by 40-70%"

---

And how it's possible not to know how much faster K10 will be over kenstfield, when we go K8 numbers, we got kent numbers and we now know K10 will "gain" 40% to 70% over K8? Where's the speculation?

Quote:

---

Originally Posted by Ruiz

"In spite of all the hype and hoopla, there is no really such thing that Intel has leapfrogged AMD. It's quite the contrary. As a matter of fact, despite all the perceptions of Intel closing the gap, half the time they do a little bit better and the other half we do. And all of that will end with the introduction of Barcelona because it's such a significant jump in performance and quality. I'm confident we will lead in that respect."

---

So he is not considering the 10-50% advantage of Intel as a leapfrog, but yet as Intel closing the gap, but he considers the 20%-50% advantage of K10 as a significant jump that will end the performance crown switching from one side to the other: "half the time they do a little bit better and the other half we do."

Well i only see AMD staying with the performance crown for a while with a 40% to 70% "gain" over Intel quadcore... but hey maybe he also didn't said what he meant here, maybe we should ignore what he says and try to figure out what he really means :)
Propaganda is a :banana::banana::banana::banana::banana:.



And english is not my native language also. :rolleyes:

LOE might be right here. The way I look at is simple. AMD knows the hard numbers. Saying 40%-70% seems to be "throwing a bone" to me. When AMD obtains the performance crown you would hear more solid numbers. Using 40%-70% is not solid numbers. And, the 30% disparity is not technical data IMO. If it is, something is wrong with the data. To lose 30% performance just doesn't make whole lot of sense. Is it going from 32 bit to 64 bit? Running 3dmark06 vs BF2, what would cause a 30% loss in performance in a next gen cpu?

if it were gains over K8, then we'd have to know if we're talking about core to core gain based on enhancements, or dual-core to quad-core plus enhancements gain, i'd hope for the first type...

Some new info on Barcelona:
http://www.hkepc.com/bbs/attachment....0&noupdate=yes
http://babelfish.altavista.com/babel...%26endtime%3d0

Lol thats one funny translation!
So first it was K8L and then K10 but the hidden real name is chinese red army edition =)

The problem with the new K10s or whatever they're going to be called isn't going to be the amount of their baseline advantage over c2d or kentsfiled. The issue is going to be how well do they clock?

A 2.2ghz chip that I can only clock to 3ghz under SS Phase (a la X2 3800+) won't really be any better for me than my current c2d clocked with not much trouble to 4.1ghz.

Quote:

---

Originally Posted by DoubleZero

A direct and clear question, a direct a clear response and still you manage to take the wrong conclusion... The 40 to 70 percent is against the Intel Quadcore, not against Intel dual core, not against a imaginary AMD rev.F quadcore.

---

I don't think it's quite as clear, as has been discussed on this thread. I tend to agree with LOE's interpretation. These 40% and 70% numbers have been used before by AMD, at AMD's Virtual IT experience and in this press release:

Simulations conducted in AMD laboratories indicate that certain database applications will see performance improvements up to 70 percent and certain floating point applications will experience performance gains of up to 40 percent over platforms powered by current dual-core AMD Opteron processors.

They've always been used before relative to a rev. F dual core. I wrote a post about it when this press release came out.

As for the "speculation" part, I think the speculation is for performance vs. Clovertown. I don't think AMD does knows exactly what Intel will be releasing this summer. There has been talk of a 3.0 GHz Clovertown and improved chipsets.

The slightly easier to see graphic

http://www.hkepc.com/bbs/attachment....0&noupdate=yes

I hope we see some CRAZY numbers... hehehe :-)

~Mike

Motiv,look five posts above...:rolleyes:

Quote:

---

Originally Posted by LOE

You don't need that much cooling for a 3Ghz 3800+

Also that clocks are for QUAD cores, dual cores should go atleast 400 Mhz higher

If amd are beeing hones and a 2.3 Ghz K10 is 40% faster than a 2.66 Ghz kentsfield that would mean you need at least 4500 Mhz core 2 to match 3000 Mhz K10

---

yep - but the question still remains as to how will they clock. Hitting 3.2 -3.6 on good air or water is pretty normal for a c2d. will it be the same for Agena/Barcelona? If yes, then I'll very happily go back to AMD as I suspect will many others. But A64 chips were never the speed demons like intel chips were/are so we'll have to wait and see how the new AMDs do.

All I have to say is that if nVidia can surprise everyone and make the Geforce 8 series better than a lot of people expected...why can't AMD?

Quote:

---

Originally Posted by Eastcoasthandle

Saying 40%-70% seems to be "throwing a bone" to me. When AMD obtains the performance crown you would hear more solid numbers. Using 40%-70% is not solid numbers. And, the 30% disparity is not technical data IMO.

---

so you're ignoring the fact that different cores like different types of workloads? what one cpu might do really well in, the other might be so-so, and vice versa? ;)

If K10 is faster than what Intel is offering at Christmas time 07, then I'll switch back to AMD. Hopefully, they'll clock upwards of 3.2 to 3.6ghz on air or water by then! :)

Quote:

---

Originally Posted by freeloader

If K10 is faster than what Intel is offering at Christmas time 07, then I'll switch back to AMD. Hopefully, they'll clock upwards of 3.2 to 3.6ghz on air or water by then! :)

---

If I knew what the world was gonna be like in 9 months then I would be filthy stinking rich. The future holds many surprises and secrets, patience is key at times

Quote:

---

Originally Posted by nn_step

If I knew what the world was gonna be like in 9 months then I would be filthy stinking rich. The future holds many surprises and secrets, patience is key at times

---

QFT...

Speculation is fun, as long as it doesn't cost you anything.

Dont Forget Nehalem and be sure that intel is not Blellum!!

The thing I find interesting is that these are 95W parts. Sureley, if they felt the need, they could have bumped it up to 125W and raised the clock, if they felt they couldn't compete at 2.3 Ghz. And what is RDDR2?



Btw, here's an english version.

http://www.hkepc.com/bbs/itnews.php?...me=0&endtime=0

RDDR2=Registered DDR2

so where are those people that said april 22nd release now?

Quote:

---

Originally Posted by PallMall

they are talking about floating points 40% increase, 70% in other apps

---

I think it's the other way round, with 70% in SSE4 intensive apps and/or 70% in highly synchronized 8-core systems (exact context wasn't clear).

But whatever it is, we have seen the list of integer performance relevant improvements. Even though some people rate it higher than my bet (which is 5% better than K8L for non-compression, non-encoding integer work such as compilation, per core at same clockspeed), nobody has come forward with giving 40% a shot, much less 70%.

Also, keep in mind that the highest clockspeed at launch time is 2.5 GHz. Core2, which is 20% or more faster than K8, comes in 3.0 GHz today, and you can bet that Intel is piling up 3.33 GHz capable dies to throw into some EE when the new AMD launches.

Quote:

---

Originally Posted by uOpt

I think it's the other way round, with 70% in SSE4 intensive apps and/or 70% in highly synchronized 8-core systems (exact context wasn't clear).

---

Read my posts earlier in this thread. The 70% is for OLTP, and the 40% is for specfp_rate2000. They made this clear in their virtual IT experience.

As far as I know, Barcelona will not support SSE4.

Quote:

---

Originally Posted by uOpt

Also, keep in mind that the highest clockspeed at launch time is 2.5 GHz. Core2, which is 20% or more faster than K8, comes in 3.0 GHz today, and you can bet that Intel is piling up 3.33 GHz capable dies to throw into some EE when the new AMD launches.

---

Intel does not have a 3.0 GHz quad-core available, although they plan on releasing one at some point. You're comparing AMD quad-core clocks to Intel dual-core clocks.

@oldblue: Barcelona will have SSE4A.

Quote:

---

Originally Posted by w0mbat

@oldblue: Barcelona will have SSE4A.

---

SSE4 is different. SSE4 is planned for introduction with Penryn.

It would be nice if they made the names more clear.