No Barcelona systems until late Q4/early Q108??

[terrace215]

http://www.tgdaily.com/content/view/31562/137/

While we are hearing from some sources that Barcelona is on track and will make an early debut in Q3, some server vendors have told TG Daily that they do not expect volume availability until Q4 and therefore tell customers that systems will not ship until late Q4 2007 or early Q1 2008.

That would actually mesh pretty well with what Hector said in early March about first Barc shipments "towards the end of summer".

other excerpts:

Meanwhile, Intel is preparing its 50-series of Core 2 Duo processors running on FSB1333 and DDR3-800 memory platforms. Information is scarce, but reliable sources are indicating that these processors will be seeing a “substantial” speed increase. One industry source mentioned to TG Daily that benchmarks comparing Intel Bearlake-X enthusiast platforms compared to AMD X2/Nvidia 680 platforms indicate that Intel will be going after AMD in a “vicious” way.

and:

Toni Duboise, senior analyst for desktop computing at Current Analysis, today confirmed to TG Daily that AMD has seen a dramatic drop in its U.S. market share for desktop retail sales. After a virtual tie in market shares in the third quarter of 2005, AMD was able to quickly expand its presence in U.S. retail to a stunning 77% in Q1 2006. Following a negligible drop to 76% in Q2 2006, AMD has been suffering a rapid decline in this segment – due to Intel’s aggressive pricing of Pentium 4 and Pentium D processors as well as the introduction of the Core 2 Duo. In Q3 2006, AMD held 66% and 54% in Q4. Current Analysis estimates that AMD achieved only 43% in Q1 2007.

[dinos22]

This Is Bad!

[turtle]

Not completely unexpected on AMD's part. Sounds similar to what they said at the Teraflop Q&A. I imagine Q4 seems more reasonable than Q108 though.

Regarding Intel's "substantial" speed increase, that seems obvious. If the 1600FSB is true, which it very well may be, it's not difficult to do the math based on current offerings. It also would make more sense to do than 1333 if they can, as it will actually show differential between the products where-as parts clocked closer to Conroe/Allendale offerings may not. For example, a 333x10 = 3.33ghz C3D/Q (Q)X7800 may leave people with high-end and/or overclocked Allendale/Conroe's saying "Do I really need that, or can I wait for the next big thing?"

Take the X6800 or QX6800 which are 10x266. A Wolfdale/Yorkfield@1600fsb following the C2D/Q structure would result in 4ghz (10x400)...Which isn't completely unreasonable given what current Intel CPUs can do and what the 45nm shrink will bring to the table. Heck, that even leaves room for little brothers above 3ghz, with perhaps the Allendale kin in the 2.4-2.8 range. With Intel's parts clocking like they do, it wouldn't make sense to keep clocking them as low as they are comparatively to what they can achieve. This could remedy that, and also allow them to charge more for this performance. Crappy for overclockers, but makes sense for the market if yeilds are good, which I have little doubt they will be.

[erwinz]

yaiks.. Intel really means business this time.. mm.. hope AMD is not caught with his pants down and have a secret up his sleeves.. :D

[perkam]

It'll have financial repurcussions (q4 07 sales will suck)...but by mid '08 AMD should be on equal ground with Intel.

Perkam

[nn_step]

It'll have financial repurcussions (q4 07 sales will suck)...but by mid '08 AMD should be on equal ground with Intel.

Perkam

you underestimate AMD

[VulgarHandle]

key word here is 'volume'

*sniff sniff* i smell FUD

[EternityZX9]

Oh for :banana::banana::banana::banana:s sake it's coming out THIS YEAR.

The meaning to words matters the most and the one word that matters most here is the word: "volume".

"Volume" meaning that the MAJORITY of the production cycle or peak for the Barcelona and Agena processors will not occur until '08. However, that doesn't mean that no chips will come out this year!

Servers and desktop chips alike WILL sell out fast (more so than Conroe did). Why? Because Intel is a manufacturing company and can produce product faster than AMD. I'm not sure of the statistics (if there are any), but I'm pretty confident in saying that AMD will release FEWER Barcelona / Agena processors than Intel did when they released their Conroe’s on the first day of launch - simply because of manufacturing power.

If the Barcelona / Agena processors are anywhere near as powerful as AMD is touting, then with less of them released, they are going to sell like hot cakes and the supply will vanish. Making them much harder to find than those Conroe’s everyone was looking for last summer. This is why server companies are predicting people that these systems will not be fully available until Q4 '07 or Q1 '08. And frankly, with how long it took people to get a decently priced Conroe last summer (minus price gouging), I wouldn't be surprised if it took until Q4 of this year to see a non-price gouged Agena (of course those with the money and are willing to pay will get what they want).

This thought makes me shudder - I don't want to even begin thinking about the price gouging that will most likely occur when these processors come out.

And honestly, I think this whole "wahhhh the sky is falling, AMD is losing the battle to Intel!" crap is just that - crap! This is how business works, Intel has a great product out there at the moment. AMD does not (please don't argue about minute details and "competitive" products, yes I know AMD has them, but nothing that surpasses a great Conroe system at the moment) that's the bottom line. But, that doesn't mean that this will not change. The power crown will pass back and forth - that's the ebb and flow of this business. I would like to believe that people could understand this, but I guess the simple principles of supply and demand and of business are beyond some people's grasp.

:slapass:

[zir_blazer]

yaiks.. Intel really means business this time.. mm.. hope AMD is not caught with his pants down and have a secret up his sleeves.. :D
That is exactly what didn't happened, I was also very enthusiast about the fact that after 3 years and lots of rumors, it was pretty probable that AMD already had a K8 replacement almost ready (As 2002 and 2003 news pointed out that K9 Engineering Samples were to be ready for 2005 and production in 2006). But they didn't had anything, the Conroe arrived and mopped the floor with them and is still doing so. If they had something, they should already have released it because Intel is demolishing they profits right now.

AMD future depends quite heavily on K8L/K10 readiness, they are a solid choice in the low end market with the A64X2 3600+ Brisbane, but that is where it ends (Anyone that can buy a Core 2 Duo would consider a high end A64X2? Obviously no, C2D average overclock push them fairly ahead). And with Pentium Exxx series for the low end not very far away, now that Intel almost get rid of the NetBurst based Processor inventory, they would be able to dominate the low end segment with solid Conroe derivates, too.
Besides all this, what will be what AMD launches at K8L/K10 release? A 999 U$D Athlon 64 FX for Socket M2 or F for the 4x4 platform while the mainstream segment must wait a few more months before general availability? That wouldn't work either. At this rate, Penryn will be available before.

[red]

you underestimate AMD

From the ways things have been turning out, I'd have to say you overestimate AMD:p:

[turtle]

lol...Sigged. :)

[terrace215]

Oh for :banana::banana::banana::banana:s sake it's coming out THIS YEAR.

The meaning to words matters the most and the one word that matters most here is the word: "volume".

"Volume" meaning that the MAJORITY of the production cycle or peak for the Barcelona and Agena processors will not occur until '08. However, that doesn't mean that no chips will come out this year!

Well, to start with, no one is claiming the chips will not ship this year.

But your definition of "volume" is not correct either. An OEM would use the term to mean "more than a trickle", really. That is, paper launch with a few wafers' worth of parts doesn't count. But they don't wait until maximum output to start producing systems. That's crazy.

Hector himself said shipments *of chips* would start "toward the end of summer". That is consistent with initial shipments at end of Q3, which does indeed imply "volume shipments" in Q4.

So that would fit with the server OEM timing of systems a month or so after volume chip shipments.

[terrace215]

It'll have financial repurcussions (q4 07 sales will suck)...but by mid '08 AMD should be on equal ground with Intel.

Perkam

To me, mid 2008 will feature:

AMD ramping 65nm Barcelona toward crossover with 65nm + 90nm K8.
Intel ramping 45nm Penryn ("Core2+") toward crossover with 65nm Core2.

AMD will be trying to have 45nm wafer starts in H208, though my opinion is this will be delayed.

Intel will be preparing to launch Nehalem ("Core3") in H208.

And therefore, I don't see AMD on equal ground. For, aside from parity on some specialized FP benchmarks, I expect Barcelona, et. al. will trail Penryn products by a 15-20% performance margin, largely due to Penryn clocking higher.

It'll be interesting to watch it all unfold, though.

[gOJDO]

I smell another AMD paper launch

[safan80]

you underestimate AMD

and your overestimating AMD. From a management point of view AMD is doing everything wrong that they can. You can't let the competition run away with the market and expect sales to stay at their current rate or grow when you have no new ground breaking products being released. The only thing AMD can expect with their current products is for sales to drop.

[awdrifter]

Damn, AMD must really be in trouble. THe R600 and K10 keeps getting delayed and delayed.

[virtualrain]

lol...Sigged. :)

LOL Indeed! :p:


you underestimate AMD


From the ways things have been turning out, I'd have to say you overestimate AMD.

Fact is, I gotta agree with red!

[dinos22]

lol...Sigged. :)

^^^^ good call ^^^^ :D :ROTF:

[OBR]

Yes this info is true, i have the same about Q1/08 directly from AMD representative ...

[turtle]

I don't see AMD on equal ground. For, aside from parity on some specialized FP benchmarks, I expect Barcelona, et. al. will trail Penryn products by a 15-20% performance margin, largely due to Penryn clocking higher.

It'll be interesting to watch it all unfold, though.

Oh, so you mean a return to the good ol' days? :D

[oldblue]

http://www.tgdaily.com/content/view/31562/137/

While we are hearing from some sources that Barcelona is on track and will make an early debut in Q3, some server vendors have told TG Daily that they do not expect volume availability until Q4 and therefore tell customers that systems will not ship until late Q4 2007 or early Q1 2008.

The different stories might be due to some customers (HP, Dell) sucking up most of the early shipments.

[verndewd]

I for one am a firm believer that if AMD could have avoided 65nm altogether they would have or at least delayed it till am2+/AM3.I have felt for some time that they knew 65nm wouldnt perform well until am3. and planning brings that too close to 45nm.
So with their IBM manufacturing guru calling 65nm old hat,it means 65nm was a reluctant acceptance to offer something to the 65nm race.

[wedfighter]

looks they're on trouble since purchasing ATi and being left behind by Intel...
they've to be wake up ASAP if they don't wanna loose their fanatic consumer

[verndewd]

looks they're on trouble since purchasing ATi and being left behind by Intel...
they've to be wake up ASAP if they don't wanna loose their fanatic consumer

Maybe ,but likely not.

[alpha0ne]

Once AMD is awarded ~ 5 Squillion bucks paid by Intel for their anticompetitive practices AMD should be in a far better position :D

[verndewd]

Once AMD is awarded ~ 5 Squillion bucks paid by Intel for their anticompetitive practices AMD should be in a far better position :D

:eek: :rolleyes: at least pay for the cost of unmoved channel inventory at price premiums that were lost due to the bribery.:fact:

[perkam]

I wouldn't mind waiting for 45NM K10 :slobber:

Perkam

[w0mbat]

I wouldn't mind waiting for 45NM K10 :slobber:

Perkam

Travel to Shanghai :banana:

[SparkyJJO]

To me, mid 2008 will feature:

AMD ramping 65nm Barcelona toward crossover with 65nm + 90nm K8.
Intel ramping 45nm Penryn ("Core2+") toward crossover with 65nm Core2.

AMD will be trying to have 45nm wafer starts in H208, though my opinion is this will be delayed.

Intel will be preparing to launch Nehalem ("Core3") in H208.

And therefore, I don't see AMD on equal ground. For, aside from parity on some specialized FP benchmarks, I expect Barcelona, et. al. will trail Penryn products by a 15-20% performance margin, largely due to Penryn clocking higher.

It'll be interesting to watch it all unfold, though.

Shrinking the process does not equal more performance. Just because Penryn might be 45nm and K10 65nm doesn't guarantee that Penryn will be better performing. Size alone won't make a difference. Example: Intel had 65nm P4 and PD chips, AMD had 90nm K8 chips, who won? ;)

Now intel would be stupid to just do a die shrink only. But where is this "15-20%" performance lead number coming from? Pulling it from the air? Penryn clocking higher, OK, P4 vs K8 again... hmmm.... Yeah I know, not netburst anymore, but it has been AMD's way to pack better performance in a slower clockspeed chip, wouldn't surprise me if they did it again. No idea how this will all turn out though, so instead of pulling numbers from nowhere let's just wait and see.

[Shintai]

Now intel would be stupid to just do a die shrink only. But where is this "15-20%" performance lead number coming from? Pulling it from the air? Penryn clocking higher, OK, P4 vs K8 again... hmmm.... Yeah I know, not netburst anymore, but it has been AMD's way to pack better performance in a slower clockspeed chip, wouldn't surprise me if they did it again. No idea how this will all turn out though, so instead of pulling numbers from nowhere let's just wait and see.

Did you miss the info with Penryn? Its far from a die shrink. It contains alot of updates. Specially on the SSE front. Intel itself says it would outperform conroe 10-20% clock for clock.

[rozzyroz]

Shrinking the process does not equal more performance. Just because Penryn might be 45nm and K10 65nm doesn't guarantee that Penryn will be better performing. Size alone won't make a difference. Example: Intel had 65nm P4 and PD chips, AMD had 90nm K8 chips, who won? ;)

Now intel would be stupid to just do a die shrink only. But where is this "15-20%" performance lead number coming from? Pulling it from the air? Penryn clocking higher, OK, P4 vs K8 again... hmmm.... Yeah I know, not netburst anymore, but it has been AMD's way to pack better performance in a slower clockspeed chip, wouldn't surprise me if they did it again. No idea how this will all turn out though, so instead of pulling numbers from nowhere let's just wait and see.

here are some good reads that will put some light on the penryn for ya i think.
http://www.theinquirer.net/default.aspx?article=38568
http://www.firingsquad.com/hardware/intel_penryn_sneak_peek/
http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=2955

they basically state that the penryn will have 20% faster switching from the 45nm process, 6mb of shared l2 cache, faster fsb, more sse4 instructions and a few more goodies. they also confirm that it will have a ipc and will be at speeds in excess of 3ghz. I heard that there wil be 20% faster gaming performance, but they didnt go into saying if it was a clock for clock measurement. I would assume so, because with penryn clocking higher, 20% would seem like a small boost to me.

we wil have to wait and see.

[terrace215]

No idea how this will all turn out though, so instead of pulling numbers from nowhere let's just wait and see.

Nowhere?

On general purpose apps & games, Barc is supposed to be ~10% faster than Core2, *clock for clock*.

Penryn's core improvements, together with the new chipsets, should easily push this to parity *clock-for-clock*.

But Penryn will easily make ~3.4GHz to Barc's roadmapped 2.6GHz. That's 30% higher clock.

I stand by my 15-20% higher performance view. Your mileage may vary.

[SparkyJJO]

here are some good reads that will put some light on the penryn for ya i think.
http://www.theinquirer.net/default.aspx?article=38568
http://www.firingsquad.com/hardware/intel_penryn_sneak_peek/
http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=2955

they basically state that the penryn will have 20% faster switching from the 45nm process, 6mb of shared l2 cache, faster fsb, more sse4 instructions and a few more goodies. they also confirm that it will have a ipc and will be at speeds in excess of 3ghz. I heard that there wil be 20% faster gaming performance, but they didnt go into saying if it was a clock for clock measurement. I would assume so, because with penryn clocking higher, 20% would seem like a small boost to me.

we wil have to wait and see.

*me puts foot in mouth*
Not gonna read that inq link - I've boycotted them ;) - but the others I will.

Only point I have is, people say "Penryn is 45nm and AMD is 65nm, Penryn is better" but half the time don't mention anything else, hence my comment about size isn't everything. I'll be honest, I haven't kept up with either CPU specs, just going by what I've been hearing, and most I've heard was size and not much else.

I'm still gonna root for AMD though, I don't like Intel much :p: Hopefully AMD gets their act together with their CPUs and GPUs and such.

I'm sorry for being an idiot.... don't hurt me.... :(

[frankR]

What about Agena? Most of us will be running AM2 Agena CPUs. Does this push it back too?

[MAS]

Did you miss the info with Penryn? Its far from a die shrink. It contains alot of updates. Specially on the SSE front. Intel itself says it would outperform conroe 10-20% clock for clock.

larger cache plus SSE4 - alot of updates :D
do you yourself beleive in these 10-20 %? The great advantage of penryn is its ocability

[verndewd]

Did you miss the info with Penryn? Its far from a die shrink. It contains alot of updates. Specially on the SSE front. Intel itself says it would outperform conroe 10-20% clock for clock.

And they moved high k up from nehalem to penryn and demonstrated it. :slobber: Thats pretty significant.

[verndewd]

larger cache plus SSE4 - alot of updates :D
do you yourself beleive in these 10-20 %? The great advantage of penryn is its ocability

preliminary benches are saying just that about performance. I have heard rumors the oc cieling is hitting a cap. We wont know till we see it though.

[Shintai]

larger cache plus SSE4 - alot of updates :D
do you yourself beleive in these 10-20 %? The great advantage of penryn is its ocability

But with K10 we believe anything, even 46% :rolleyes:
Penryn gives up to 40% in AMD "estimated" bench type.

Its easy to boost Penryn, due to the wast unused resources in Conroe.

[duploxxx]

Yes this info is true, i have the same about Q1/08 directly from AMD representative ...

right, and my NDA from 2 days ago says still on track... and delivery in August.

i'll have my test system end of june, already confirmed

Penryn is November btw......

[duploxxx]

Nowhere?

On general purpose apps & games, Barc is supposed to be ~10% faster than Core2, *clock for clock*.

Penryn's core improvements, together with the new chipsets, should easily push this to parity *clock-for-clock*.

But Penryn will easily make ~3.4GHz to Barc's roadmapped 2.6GHz. That's 30% higher clock.

I stand by my 15-20% higher performance view. Your mileage may vary.


any links to that?

i'll give a hint, at the moment they do not confirm up to 3,5 ghz, they only state 3.0 and 3.2 and they ask oem to keep in mind that a it is possible a raise in tdp to 150W!

wathever they say we will see, the same about the SOI-ge wasn't that also 20% faster ?????

[turtle]

any links to that?

i'll give a hint, at the moment they do not confirm up to 3,5 ghz, they only state 3.0 and 3.2 and they ask oem to keep in mind that a it is possible a raise in tdp to 150W!

wathever they say we will see, the same about the SOI-ge wasn't that also 20% faster ?????

3 and 3.2?

That doesn't compute to me unless they mix 266/333...Which would be odd but possible. I did hear that 3.43/3.73 YorkfieldXE rumour a while back that stated that...I just hoped it was bogus.

3 and 3.3, or 2.8 and 3.2 seems to make more sense, depending on the bus (333/400).

I would surely hope for faster bins though, considering Wolfdale will be a smaller die than Conroe by approx 25% (actually closer to 30%) and apparently have the same TDP. If a Yorkfield is two wolfdales, it too should have greater headroom than Kentsfield by a substantial margin...Clocks that low just don't make sense to me.

[saaya]

some guys at amd deserve a big :slapass:
the plans and strategies were great, the execution was awfull

[terrace215]

any links to that?

i'll give a hint, at the moment they do not confirm up to 3,5 ghz, they only state 3.0 and 3.2 and they ask oem to keep in mind that a it is possible a raise in tdp to 150W!

wathever they say we will see, the same about the SOI-ge wasn't that also 20% faster ?????

150W TDP is for the 65nm QX6800 shipping now, and that is supposed to fall in Q3, pre-Penryn, even.

The AMD clock speeds that are roadmapped already take into account the SiGe process that Barc is due to launch on.

Those speeds (for QC): 2.3GHz @ 95W. 2.5GHz @ 120W (launched about 3 months after 95W parts). Bumps of 100MHz to each in Q208.

Hector himself said "first shipments toward the end of summer" in early March, so that should be Aug/Sept. If you are getting a part in June, it is probably a final pre-prod sample.

[red]

http://processorfinder.intel.com/details.aspx?sSpec=SL9UK
http://www.techreport.com/reviews/2007q2/core2-qx6800/index.x?pg=13
It's 130W, I'm guessing they pulled that off by charging $1199 so that they could buy time to cherry pick the best dies.

[theteamaqua]

wow QX6800 costs $1199?? thats gonan gouge to like $1500 on some greedy sites

well Q6600 is gonna drop to $266 in Q3 isnt it?? this is bad news as intel might not drop the price for kentsfield at all ...

[theteamaqua]

but then again i think what tis trying to say is that mass availibility @ Q4 07or Q1 08 ... that doesnt mean we cant buy it @ Q3, right??

the desktop part

[red]

http://www.dailytech.com/article.aspx?newsid=6488
QX6800 gets cut in Q3.. But considering how it's debut recently was unexpected, stranger things could happen.

"but then again i think what tis trying to say is that mass availibility @ Q4 07or Q1 08 ... that doesnt mean we cant buy it @ Q3, right??"
:(?

[Jimmer411]

This might be my call to intel :( Either way I want my damn Quadcore.