Barcelona model numbers and clockspeeds

zedfckg · Feb 5, 2007

Blog site Tracking AMD reveals that according to several sources these will be the first Barcelona-based Opteron model numbers and clockspeeds to launch

Check out more at http://trackingamd.blogspot.com/2007/02/barcelona-model-numbers-revealed.html

carlhungis · Feb 5, 2007

I see a red x

Can you list them?

CaptRobertApril · Feb 5, 2007

I see a red x

That's the image! AMD has X'd out Barcelona entirely!

BTW, Carl! It's me! Your long lost brother! Well! :lol:

CaptRobertApril · Feb 5, 2007

Here's the chart.

I know frequency aint worth squat no mo' but DAMN THOSE ARE LOW FREQUENCIES!!!!

Wombat2 · Feb 5, 2007

Here's the chart.

I know frequency aint worth squat no mo' but DAMN THOSE ARE LOW FREQUENCIES!!!!

This is the thing I've been pointing out all along. Penryn clocks are going to be 3.8 GHZ and upwards.

Barcelona will give AMD, at best, a small performance lead over Conroe for, at best, 6 months. From Q4 2007 onwards Barcelona is going to eat Penryn dust.

Conroe is ~30-40% faster than K8 clock-for-clock.
AMD claims K8L = 1.8x K8

Lets do some math:
3 GHz K8 performance index: 1.0

Penryn @ 3.8 Ghz: 1.0 * 1.4 * (3.8/3) = 1.8
K8l @ 2.4 GHZ: 1.0 * 1.8 * (2.4/3) = 1.44

My prediction: Penryn > K8L by 25%.

Bear in mind that the 1.8X multiplier for K8L is actually just the best case claimed FP performance by AMD.

AMD fanbois pinning all their hopes on Barcelona are in for a nasty surprise :lol:

CaptRobertApril · Feb 5, 2007

that'd be funny if they did, everyone has been waiting forever for k8l now, but anyway, barcelona's the core, k8l the arch, so even if they kill barcelona, their still gonna use k8l over k8

Some guys on that site were saying that some of this might be Bucharest. I dunno. I thought I was on a CPU forum, not a Travel forum!

This is the thing I've been pointing out all along. Penryn clocks are going to be 3.8 GHZ and upwards.

I dunno what kind of magic AMD thinks that it can build into a 2.4GHz chip that's gonna let it catch one that's topping 4GHz, but I'm willing to keep my mind at least somewhat open to the possibility that they will pull it off.

CaptRobertApril · Feb 5, 2007

Also, what's the deal with the HT speed difference between F and AM2+? 3200MT/s? And howcum F only supports 667MHz DDR2?

CaptRobertApril · Feb 5, 2007

you mean budapest?

Damn I always get those two cities confused. It's a damn good thing I know the difference between Romania and Hungary!

LordPope · Feb 5, 2007

Here's the chart.

I know frequency aint worth squat no mo' but DAMN THOSE ARE LOW FREQUENCIES!!!!

This is the thing I've been pointing out all along. Penryn clocks are going to be 3.8 GHZ and upwards.

Barcelona will give AMD, at best, a small performance lead over Conroe for, at best, 6 months. From Q4 2007 onwards Barcelona is going to eat Penryn dust.

Conroe is ~30-40% faster than K8 clock-for-clock.
AMD claims K8L = 1.8x K8

Lets do some math:
3 GHz K8 performance index: 1.0

Penryn @ 3.8 Ghz: 1.0 * 1.4 * (3.8/3) = 1.8
K8l @ 2.4 GHZ: 1.0 * 1.8 * (2.4/3) = 1.44

My prediction: Penryn > K8L by 25%.

Bear in mind that the 1.8X multiplier for K8L is actually just the best case claimed FP performance by AMD.

AMD fanbois pinning all their hopes on Barcelona are in for a nasty surprise :lol:

Links to where you got scores to make these absolute conclusions

only people who deal in absolutes are the SITH and INTEL FAN BOIS

LordPope · Feb 5, 2007

Here's the chart.

I know frequency aint worth squat no mo' but DAMN THOSE ARE LOW FREQUENCIES!!!!

This is the thing I've been pointing out all along. Penryn clocks are going to be 3.8 GHZ and upwards.

Barcelona will give AMD, at best, a small performance lead over Conroe for, at best, 6 months. From Q4 2007 onwards Barcelona is going to eat Penryn dust.

Conroe is ~30-40% faster than K8 clock-for-clock.
AMD claims K8L = 1.8x K8

Lets do some math:
3 GHz K8 performance index: 1.0

Penryn @ 3.8 Ghz: 1.0 * 1.4 * (3.8/3) = 1.8
K8l @ 2.4 GHZ: 1.0 * 1.8 * (2.4/3) = 1.44

My prediction: Penryn > K8L by 25%.

Bear in mind that the 1.8X multiplier for K8L is actually just the best case claimed FP performance by AMD.

AMD fanbois pinning all their hopes on Barcelona are in for a nasty surprise :lol:

Links to where you got scores to make an absolute conclusions

only people who deal in absolutes are the SITH and INTEL FAN BOIS

Haven't you learned your lesson yet? :? I hope that barcelona is a killer, but we all know that intel has far more resources to deal with, and their 45nm will kill amd in price wise, and their performance should be amazing due to the high gate and k thing

learned what lesson.... I learned alot of lessons with the FDIV bug, the Camino Chipset, The p3 1.13 gHZ RECALL and the Preschott debacle ... SHOW ME WHAT U CAN DO ...dont just tell me...

I heard this same crap on every single INTEL release since the P3 was invented...AMD is dead...

How do you know INTEL wont have massive yield problems with the BRAND NEW technology... YOU DONT KNOW.... You can GUESS but YOU DONT KNOW... So don't talk about lessons... you may buy INTEL's lines like its the gospel... i like to see retail silicon first.

gOJDO · Feb 5, 2007

This is the thing I've been pointing out all along. Penryn clocks are going to be 3.8 GHZ and upwards.

Nope!
Penryn will be released with 20% higher clocks. So, expect the eXtreme Edition dualcores clocked at 3.66GHz and quadcores at 3.33GHz for server and 3.2 or 3.47 for desktop.

Barcelona will give AMD, at best, a small performance lead over Conroe for, at best, 6 months. From Q4 2007 onwards Barcelona is going to eat Penryn dust.

Expect Penryn to be available in Q1 2008.

Conroe is ~30-40% faster than K8 clock-for-clock.

This is not true! But it is 20% faster clock for clock!
IMO clock for clock, Penryn will be 25% faster than K8, only because it will have 50% more cache and 25% faster FSB.

AMD claims K8L = 1.8x K8

That is FPU performance, not overall CPU performance.

Lets do some math:
3 GHz K8 performance index: 1.0

Penryn @ 3.8 Ghz: 1.0 * 1.4 * (3.8/3) = 1.8
K8l @ 2.4 GHZ: 1.0 * 1.8 * (2.4/3) = 1.44

My prediction: Penryn > K8L by 25%.

Bear in mind that the 1.8X multiplier for K8L is actually just the best case claimed FP performance by AMD.

Your math is plain wrong and your prediction has no sense:
1. There would be no Penryn @ 3.8GHz. The 45nm shrink of Conroe, successor will have FSB1333(quad pumped 333MHz). So the difference between the frequency of different models is 333MHz: 3.0GHz, 3.33GHz, 3.66GHz, 4GHz and so on. Also I don't think that Intel will release quadcore Penryn clocked higher than 3.47GHz(for desktop) or 3.33GHz(for server)!
2. Clock for clock, Conroe is not 40% faster than K8, nor Wolfdale would be!
3. The K8L dualcores will be clocked up to 2.9GHz, while the quadcores up to 2.5GHz
4. 1.4 * 1.8 = 1.773; 1.773 / 1.44 = 1.231; According to your prediction and the math Penryn > K8L by 23.1%

According to AMD claims, in various workloads the K8L(quadcore 2.5GHz) will be 40% faster than Clovertown(quadcore 2.66GHz). So, according to AMD claims:
2.66GHz Clovertown performance index: 1.0
3.33GHz Yorkfield performance: 1.04(the extra L2) * 1.2 = 1.248
2.5GHz K8L: 1.4

1.4 / 1.248 = 1.22 or 2.5GHz Barcelona to be 22% faster than 3.33GHz Yorkfield! I think that AMD claims are Science Fiction, unless they consider RAM bandwidth as "various workloads", which is not CPU performance.

IMO, K8L will perform similar like Penryn, clock for clock. It will beat Penryn in FP, but will loose in AL, the difference in SSE performance will be +-5%. The highest clocked Penryn(3.66/3.2GHz DC/QC) will wipe the floor with highest clocked K8L(2.9/2.5GHz DC/QC).

AMD fanbois pinning all their hopes on Barcelona are in for a nasty surprise :lol:

Yes. But most of them know that K8L will lost the battle against Penryn but don't want to admit it forth.

Wombat2 · Feb 5, 2007

Do a search for "penryn launch" on www.dailytech.com that gives the 3.8+ GHZ clocks. Everything else is from AMD itself.

3.8 GHZ clocks is hardly surprising given the extreme overlocks possible on Conroe.

Sorry to be the bearer of bad news Mr. AMD fanboi :lol:

ajfink · Feb 5, 2007

This is the thing I've been pointing out all along. Penryn clocks are going to be 3.8 GHZ and upwards.

Nope!
Penryn will be released with 20% higher clocks. So, expect the eXtreme Edition dualcores clocked at 3.66GHz and quadcores at 3.33GHz.

Barcelona will give AMD, at best, a small performance lead over Conroe for, at best, 6 months. From Q4 2007 onwards Barcelona is going to eat Penryn dust.

Expect Penryn to be available in Q1 2008.

Conroe is ~30-40% faster than K8 clock-for-clock.

This is not true! But it is 20% faster clock for clock!
IMO clock for clock, Penryn will be 25% faster than K8, only because it will have 50% more cache and 25% faster FSB.

AMD claims K8L = 1.8x K8

That is FPU performance, not overall CPU performance.

Lets do some math:
3 GHz K8 performance index: 1.0

Penryn @ 3.8 Ghz: 1.0 * 1.4 * (3.8/3) = 1.8
K8l @ 2.4 GHZ: 1.0 * 1.8 * (2.4/3) = 1.44

My prediction: Penryn > K8L by 25%.

Bear in mind that the 1.8X multiplier for K8L is actually just the best case claimed FP performance by AMD.

Your math is plain wrong and your prediction has no sense:
1. There would be no Penryn @ 3.8GHz. The 45nm shrink of Conroe, successor will have FSB1333(quad pumped 333MHz). So the difference between the frequency of different models is 333MHz: 3.0GHz, 3.33GHz, 3.66GHz, 4GHz and so on. Also I don't think that Intel will release quadcore Penryn clocked higher than 3.33GHz, although its FSB is clocked at 1066MHz!
2. Clock for clock, Conroe is not 40% faster than K8, nor Wolfdale would be!
3. The K8L dualcores will be clocked up to 2.9GHz, while the quadcores up to 2.5GHz
4. 1.4 * 1.8 = 1.773; 1.773 / 1.44 = 1.231; According to your prediction and the math Penryn > K8L by 23.1%

According to AMD claims, in various workloads the K8L(quadcore 2.5GHz) will be 40% faster than Clovertown(quadcore 2.66GHz). So, according to AMD claims:
2.66GHz Clovertown performance index: 1.0
3.33GHz Yorkfield performance: 1.04(the extra L2) * 1.2 = 1.248
2.5GHz Barcelona: 1.4

1.4 / 1.248 = 1.22 or 2.5GHz Barcelona to be 22% faster than 3.33GHz Yorkfield! I think that AMD claims are Science Fiction, unless they consider RAM bandwidth as "various workloads", which is not CPU performance.

IMO, K8L will perform similar like Penryn, clock for clock. It will beat Penryn in FP, but will loose in AL, the difference in SSE performance will be +-5%. The highest clocked Penryn(3.66/3.33GHz DC/QC) will wipe the floor with highest clocked K8L(2.9/2.5GHz DC/QC).

AMD fanbois pinning all their hopes on Barcelona are in for a nasty surprise :lol:

Yes. But most of them know that K8L will lost the battle against Penryn but don't want to admit it forth.

I dunno - I fully expect 3Ghz K8L parts at least by the time Penryn is going to be released. It's going to be a fight.

LordPope · Feb 5, 2007

go read this son

http://www.overclockers.com/tips01102/

Ed at overclockers is known for bashing AMD...and even he thinks Penryn will be delayed and not clocking faster than curent c2d's in its initial form..

he smells delay due to yield issues

so before you call someone a fanboi... make sure you have FACTS backing you up ...AND U DO NOT

dasickninja · Feb 5, 2007

Its an opinion. Without hard facts. Both Barcelona and Penryn will be on time. They'll both perform well. Why don't we just quit the bitching and just wait and see which one beats the other. Besides.. they're aimed at different markets.

LordPope · Feb 5, 2007

ninja u are the cool hand of reason

but u see why this guys got me a lil heated...

like u said,,,its all speculation and opinion....

They are acting like INTEL's word and whatever website they quote is fact.... if i offer a different opinion, i am an AMD FAN BOI

where are thier facts...where is a reputable score of PENRYN at 3.8 Ghz?

Wombat2 · Feb 5, 2007

Well aside from you flinging alot of faeces ... we are essentially in agreement.

Using AMDs 1.8x multiplier claim:

3 GHz K8 performance index: 1.0

Dual core:
Penryn @ 3.7 Ghz: 1.0 * 1.2 * (3.7/3) = 1.48
K8l @ 2.9 GHZ: 1.0 * 1.8 * (2.9/3) = 1.74

K8L > Penryn 17%.

Notes:
Assumes K8L = 1.8x K8 overall which is obviously bunk.
Assumes no benefit for Penryn due to larger caches etc

Quad core:
Penryn @ 3.3 Ghz: 1.0 * 1.2 * (3.3/3) = 1.32
K8l @ 2.5 GHZ: 1.0 * 1.8 * (2.5/3) = 1.5

(A) K8L > Penryn 14%

Notes:
Assumes K8L = 1.8x K8 overall which is obviously bunk.
Assumes no benefit for Penryn due to larger caches etc

Using AMDs 40% quad core claim (Godjo analysis):

K8L(quadcore 2.5GHz) = 1.4x Clovertown(quadcore 2.66GHz).

2.66GHz Clovertown performance index: 1.0
3.33GHz Yorkfield performance: 1.04(the extra L2) * 1.2 = 1.248
2.5GHz K8L: 1.4

(B) 2.5GHz Barcelona > 3.33GHz Yorkfield 22%

Conclusion: (A) vs. (B) AMDs performance claims roughly consistent.

Real world though there will be < 10% difference, if that...

Ultimately though, Intel will be using a 45n nm process vs. AMDs 65 nm process. Guess who is going to be able to undercut their competitors pricing?

As regards overclockability I would certainly put my money on Penryn :lol: .

gOJDO · Feb 5, 2007

Here are some thoughts about Penryn from LostCircuits:
http://www.lostcircuits.com/cpu/intel_penryn/

We estimate that the highest production speedgrades will be released at 3.6 GHz, which can be achieved at 900-950 mV core voltage. The cache size has little impact on the idle power consumption of the Pentium M - because of the implementation of selective clock gating - the nominal idle power consumption is expected to be lower than 10 W (SpeedStep enabled). The TDP will most likely be around 35W (Core2 Duo) and 70W for the quad core version that, like in the case of Conroe, combines 2 dies on a single package.

ajfink · Feb 5, 2007

No, AMD isn't claiming a performance lead over K8, Wombat, they're claiming 1.8 and 1.4 over CLOVERTOWN.

Wombat2 · Feb 5, 2007

MAKE THE BAD MAN STOP. I WANT MY MOMMY.
\

<= Serf Pope

Hint:

(A) Visit a few Intel overclocking forums to establish Conroe's overclockability. Oh ... wait ... my bad. AMD fanbois think overclocking means +10% clock speeds :lol:

(B) Re-read the articles describing 45 nm transistor technologies (which AMD/IBM are madly scrambling to catch up with now) ... you may have to turn off your AMD fanboi Reality Distortion Field (RDF) to get the gist of the artcle.

(C) A + B makes 3.3-3.7 GHZ launch clocks seem conservative.

Wombat2 · Feb 5, 2007

No, AMD isn't claiming a performance lead over K8, Wombat, they're claiming 1.8 and 1.4 over CLOVERTOWN.

http://www.zdnetasia.com/news/hardware/0,39042972,61984216,00.htm

"At a given clock frequency, a Barcelona core outperforms a current Opteron core by a factor of 1.8."

Is english your first language? If so I suggest you sue your teachers
:lol:

Wombat2 · Feb 5, 2007

I think what will steal the public from intel is if amd manages to somehow get ht out of the way for overclocking, and get it's cpu to clock past 4ghz like conroes can.

This wont ever happen.

Conroe can clock well because it is a brand new design, designed from the outset to scale to very high speeds. Evidence: Incredible Conroe overclockability. Penryn is pretty much a dumb shrink with fancier transistors.

Barcelona on the otherhand is an evolutionary design based on K8L which is an overclocking dog.

While Barcelona may overclock better than K8, expecting it to suddenly do the equivalent of 100% overclocks is silly.

gOJDO · Feb 5, 2007

Here is the interview with Randy Allen, where he speaks about Barcelona and says that it will outperform Clovertown by 40% in various workloads:
http://www.youtube.com/watch?v=G_n3wvsfq4Y

Wombat2 · Feb 5, 2007

Come to think of it ... having equivalent performance for Penryn and K8L is probably the best possible outcome for consumers.

If neither AMD nor Intel can compete on performance they will have to compete on price 8)

tamalero · Feb 5, 2007

Here's the chart.

I know frequency aint worth squat no mo' but DAMN THOSE ARE LOW FREQUENCIES!!!!

This is the thing I've been pointing out all along. Penryn clocks are going to be 3.8 GHZ and upwards.

Barcelona will give AMD, at best, a small performance lead over Conroe for, at best, 6 months. From Q4 2007 onwards Barcelona is going to eat Penryn dust.

Conroe is ~30-40% faster than K8 clock-for-clock.
AMD claims K8L = 1.8x K8

Lets do some math:
3 GHz K8 performance index: 1.0

Penryn @ 3.8 Ghz: 1.0 * 1.4 * (3.8/3) = 1.8
K8l @ 2.4 GHZ: 1.0 * 1.8 * (2.4/3) = 1.44

My prediction: Penryn > K8L by 25%.

Bear in mind that the 1.8X multiplier for K8L is actually just the best case claimed FP performance by AMD.

AMD fanbois pinning all their hopes on Barcelona are in for a nasty surprise :lol:

actually, they said 40%faster than intel's version..
so around 10%faster per core against intel's core technology

so id say if penryn really clocks that fast, then obviusly intel will...
but like they said before(people like jack ), a shrink doesnt mean a huge impulse of speed , nor frecuency.

you mean budapest?

Anyway, if ocing on k8l sucks, I'm switching to intel, even if they do get a 40% lead, if they still have the 3.2ghz barrier, they're screwed for when teh k+high gate design comes in to double the efficiency of the 45nm proccessors

Didnt actually Jack and others talked about this metal gate thingie?
I remember reading about intel NOT USING high-K nor metal gate until 32nm

Barcelona model numbers and clockspeeds

Distinguished

Distinguished

Distinguished

Distinguished

Distinguished

Distinguished

Distinguished

Distinguished

Distinguished

Distinguished

Distinguished

Distinguished

Distinguished

Distinguished

Splendid

Distinguished

Distinguished

Distinguished

Distinguished

Distinguished

Distinguished

Distinguished

Distinguished

Distinguished

Distinguished

Share this page