Sign in with
Sign up | Sign in
Your question

AMD Partners Demonstrate "Barcelona"

Last response: in CPUs
Share
June 6, 2007 8:02:12 AM

This is anandtech.. sorry if it has been posted before

Link!

Quote:
One vendor demonstrated Barcelona to DailyTech running at 1.6GHz. Current AMD Barcelona samples are not scaling too well. AMD partners confirmed the highest running, POST and OS capable, Barcelona processor is 2.0 GHz.
June 6, 2007 8:09:45 AM

Quote:
This is anandtech.. sorry if it has been posted before

Link!

One vendor demonstrated Barcelona to DailyTech running at 1.6GHz. Current AMD Barcelona samples are not scaling too well. AMD partners confirmed the highest running, POST and OS capable, Barcelona processor is 2.0 GHz.
I like the post below the article.

Quote:
I guess that puts the recently rumored supposed 500 MHz gain from a supposed miracle stepping in perspective.
:lol:  :lol: 
June 6, 2007 8:13:33 AM

Quote:
This is anandtech.. sorry if it has been posted before

Link!

One vendor demonstrated Barcelona to DailyTech running at 1.6GHz. Current AMD Barcelona samples are not scaling too well. AMD partners confirmed the highest running, POST and OS capable, Barcelona processor is 2.0 GHz.

....at 2.0Ghz?

Not to be a flamming fanboy here, but isn't 2.0Ghz a little too...pathetic, for a chip that AMD solely relies on?

Well if this is true, it confirms with Jack's prediction about AMD's 65nm SOI process being inferior in terms of scaling, and heat reduction.

It will also concur with the rumor about defects in the current bin, that AMD needs to do some major changes in the architecture, which caused the delay.

I thought Fuad and theINQ had hyped the B0 stepping, which could reach 3.0Ghz? Perhaps they got the news from Sharidouche after all. Talking about a 3.0Ghz running chip at 1.1V. :lol: 

Nonetheless, a great find. If this is true, AMD is probably in another round of beating.
Related resources
June 6, 2007 8:28:57 AM

As a comparison Intel has already demonstrated a quad core 3.33Ghz/1333 FSB Yorkfield CPU.
June 6, 2007 8:37:38 AM

Quote:
This is anandtech.. sorry if it has been posted before

Link!

One vendor demonstrated Barcelona to DailyTech running at 1.6GHz. Current AMD Barcelona samples are not scaling too well. AMD partners confirmed the highest running, POST and OS capable, Barcelona processor is 2.0 GHz.

....at 2.0Ghz?

Not to be a flamming fanboy here, but isn't 2.0Ghz a little too...pathetic, for a chip that AMD solely relies on?

Well if this is true, it confirms with Jack's prediction about AMD's 65nm SOI process being inferior in terms of scaling, and heat reduction.

It will also concur with the rumor about defects in the current bin, that AMD needs to do some major changes in the architecture, which caused the delay.

I thought Fuad and theINQ had hyped the B0 stepping, which could reach 3.0Ghz? Perhaps they got the news from Sharidouche after all. Talking about a 3.0Ghz running chip at 1.1V. :lol: 

Nonetheless, a great find. If this is true, AMD is probably in another round of beating.

Well, this jives with all the rest of the rumors that have been coming out lately regarding K10. Sadly, you're probably right about AMD taking another round of beating.

The worst of it is, without a clear performance edge, AMD won't be able to dictate prices. Net profit margin will continue to come under strong pressure from Intel.
June 6, 2007 8:48:53 AM

Quote:

The worst of it is, without a clear performance edge, AMD won't be able to dictate prices. Net profit margin will continue to come under strong pressure from Intel.


I don't want it to be able to dictate prices. 8) I also don't like Intel to come out a clear winner so that it won't enjoy a monopoly.
June 6, 2007 8:50:40 AM

Quote:
is barcelona a 4way proc? conroe is a 4 way proc.
whats penryn?
i see this tag of drop in replacement proc on these k10 parts isnt penryn also drop in replacement but also 45nm and high k dialectric?


I think what they meant was drop in replacement AFTER architectural changes.. Penryn is the same architecture as Core2...Or am i wrong about this one?
June 6, 2007 9:00:54 AM

Quote:

The worst of it is, without a clear performance edge, AMD won't be able to dictate prices. Net profit margin will continue to come under strong pressure from Intel.


I don't want it to be able to dictate prices. 8) I also don't like Intel to come out a clear winner so that it won't enjoy a monopoly.I agree....these rock-bottom prices that we're enjoying, are awesome...but they can't and won't last forever. AMD especially, needs to start making more
$/chip or they're in big trouble. They have the ATI loan to take care of,lots of R&D for the AMD and ATI units, building the new fab in NY(they may just end up shelving that plan), and IIRC..they still have to convert Fab 30 to 38. They need moola, and fasttttt!! :?
June 6, 2007 9:01:21 AM

just as dual cores from intel are faster than ther quad cores, we can hope for somewhat decent clocks for dual-cores...maybe...i hope :?
June 6, 2007 9:04:43 AM

Quote:
is barcelona a 4way proc? conroe is a 4 way proc.
whats penryn?
i see this tag of drop in replacement proc on these k10 parts isnt penryn also drop in replacement but also 45nm and high k dialectric?


I think what they meant was drop in replacement AFTER architectural changes.. Penryn is the same architecture as Core2...Or am i wrong about this one?I think Penryn is said to work on most 965P chipset boards(sort of like C2D works on 975 boards), but P35 is the intended chipset for Penryn. :?
June 6, 2007 9:05:21 AM

I personally think this was a directive from AMD to avoid any issues with systems going down during the show.
Having been an exhibitor at a few large shows in my lifetime (Cebit in Germany) a show environment is not very kind to hardware. With alot
of heat, vibration, wandering fingers, etc, etc.
I remember back when the original A64 was released is was only ever running at 800Mhz at trade shows before the official release date.
June 6, 2007 9:05:51 AM

Quote:
no penryn is different it has a high-k dialectric which means higher proc speeds and less heat and less power consumption.

But the architecture is still the same..It is still 2 dual cores on one die right?
June 6, 2007 9:09:16 AM

Quote:
no penryn is different it has a high-k dialectric which means higher proc speeds and less heat and less power consumption.

But the architecture is still the same..It is still 2 dual cores on one die right?Yes, with some arch. changes to improve IPC(SSE4, Radix-16, etc.).
June 6, 2007 10:05:38 AM

This explains all the delays and other bits and pieces we've heard. To only be hitting 1.6-2.0 GHZ one month from launch spells trouble.

Hopefully they can boost the clocks to something credible. Doesn't say much for the overclockability of the design does it? 8O

No wonder AMD refuses to release any benchmarks on real silicon.
June 6, 2007 10:12:10 AM

This could well be the first real benchmark on the Barcelona...

Quote:
We had the opportunity to benchmark the AMD Barcelona, native quad-core on an early stepping. We only had a few minutes to test the chip, but we were able to run a quick Cinebench before we were instructed to leave.


This sounds like a: "you distract that guy and I'll quickly run this benchmark I've got on my memory stick here" sort of job :twisted:

Quote:
Cinebench completed the default benchmark in 27 seconds for the 1.6 GHz K10; 17 seconds for the Intel Xeon X3220. The Kentsfield Xeon was 58% faster with a 50% higher clock frequency for Cinebench.


Make of it what you will.

Lyngvi.
June 6, 2007 10:53:52 AM

Maybe they are having heat issues with the upgraded integrated memory controller. It's running at a different (higher) voltage from the cores and it's a more complicated design that needs to supply four cores, each having two dedicated cache buffers and one shared buffer for all four cores. There is going to be a lot of data shuffled back and forth between caches and cores aswell as between cores and memory banks. If i got the anandtech article right (link at the bottom) this is all regulated by the ondie memory controller. On the intel side this resides inside the northbridge chipset wich is running pretty hot on the new boards wich IMO is due mostly to the memory controller and not so much the integrated various other functions of the chipset. This may well be another 2900 with too many options stuffed in to tight a package to make it work efficiently... or i'm very wrong and probably be BBQ'd for it :p  please be gentile?
Anyways, here is mentioned linkage

http://www.anandtech.com/showdoc.aspx?i=2939&p=7
June 6, 2007 5:11:43 PM

Quote:
This could well be the first real benchmark on the Barcelona...

We had the opportunity to benchmark the AMD Barcelona, native quad-core on an early stepping. We only had a few minutes to test the chip, but we were able to run a quick Cinebench before we were instructed to leave.


This sounds like a: "you distract that guy and I'll quickly run this benchmark I've got on my memory stick here" sort of job :twisted:

Quote:
Cinebench completed the default benchmark in 27 seconds for the 1.6 GHz K10; 17 seconds for the Intel Xeon X3220. The Kentsfield Xeon was 58% faster with a 50% higher clock frequency for Cinebench.


Make of it what you will.

Lyngvi.

That iceberg isn't a problem...... HMS TITANIC
a b à CPUs
June 6, 2007 6:22:59 PM

Quote:
its 2 cores on one die but its revolutionary
its the first of its kind
it has metal gate transistors to start with


Not to quibble but I think the revolutionary part is the "high-K" metal gate. The original MOSFETs from the late '70s had metal (aluminum I believe) gates. In fact, the acronym stands for Metal Oxide Semiconductor Field Effect Transistor, with the first 3 describing the order of the gate structure materials. Later versions went with polysilicon gates since for one thing those are self-aligning, but still were called MOSFETs for some reason.

If Jack would come back he could verify the above :) 
a b à CPUs
June 6, 2007 6:40:54 PM

Quote:
That iceberg isn't a problem...... HMS TITANIC


From the other THG thread on the LegitReviews Cinebench10 bench:

Quote:
"AMD wouldn't comment on the record on what their upcoming Barcelona processor would score, but people in the know said that AMD is getting ~16,000 with a pair of their Barcelona processors on Cinebench 10. If this proves to be true and you keep in mind that the 45nm Penryn V8 system is underclocked right now and scoring ~23,000 on the same benchmark it means things could get interesting really soon."


Funny how Sharikou never mentions any of this in his blog. Instead he loves to quote one person's bad experience with C2D as evidence that Intel is gonna "BK in 2Q 08" :roll:
June 6, 2007 7:16:25 PM

Quote:
This is anandtech.. sorry if it has been posted before

Link!

One vendor demonstrated Barcelona to DailyTech running at 1.6GHz. Current AMD Barcelona samples are not scaling too well. AMD partners confirmed the highest running, POST and OS capable, Barcelona processor is 2.0 GHz.

....at 2.0Ghz?

Not to be a flamming fanboy here, but isn't 2.0Ghz a little too...pathetic, for a chip that AMD solely relies on?

Well if this is true, it confirms with Jack's prediction about AMD's 65nm SOI process being inferior in terms of scaling, and heat reduction.

It will also concur with the rumor about defects in the current bin, that AMD needs to do some major changes in the architecture, which caused the delay.

I thought Fuad and theINQ had hyped the B0 stepping, which could reach 3.0Ghz? Perhaps they got the news from Sharidouche after all. Talking about a 3.0Ghz running chip at 1.1V. :lol: 

Nonetheless, a great find. If this is true, AMD is probably in another round of beating.

What delay? AMD announced Q3 availability, and its not even Q3 yet, let alone the last day of Q3. Until Oct 1, AMD is still on track, as they projected.
June 6, 2007 7:33:06 PM

Quote:
I think what they meant was drop in replacement AFTER architectural changes.. Penryn is the same architecture as Core2...Or am i wrong about this one?

no penryn is different it has a high-k dialectric which means higher proc speeds and less heat and less power consumption.

No, Penryn is a modified C2D Uarch, with more cache, new SSE extensions and a few other minor changes. Nothing significant, however it will require a chipset capable of supporting the changes/power requirements and a bios update, meaning it will not be a drop in replacement for any old board.

The High K dialectric to which you refer is not related to/dependant on the Uarch. It is new materials/implementation methodologies in the manufacturing design process at the smaller 45nm node. While it is independent of the Uarch, however, any Uarch, C2D, Netburst, whatever, will be dependant upon the new process to function successfully at the smaller node. [/quote]
June 6, 2007 7:37:33 PM

Quote:


No, Penryn is a modified C2D Uarch, with more cache, new SSE extensions and a few other minor changes. Nothing significant,


+10% per clock is insignificant? C2D only has +20% on K8 :roll:
June 6, 2007 7:56:41 PM

Quote:


No, Penryn is a modified C2D Uarch, with more cache, new SSE extensions and a few other minor changes. Nothing significant,


+10% per clock is insignificant? C2D only has +20% on K8 :roll:

I wasnt speaking in terms of perfromance gains or losses. Only changes to th Uarch. You do understand the percentage performance gain or loss is independant of the percentage of change to the design, yes?

Ex: You put a 350 CFM carb on a 454 and cut 12mph off of it quarter mile time of 108. Thats a 10% performance loss, for a single component swap. Conversely, you could swap a cam and gain perfromance. The carburator and cam are relatively insignificant to the overall "whole" of the car.
June 7, 2007 9:12:01 AM

Quote:


No, Penryn is a modified C2D Uarch, with more cache, new SSE extensions and a few other minor changes. Nothing significant,


+10% per clock is insignificant? C2D only has +20% on K8 :roll:

I wasnt speaking in terms of perfromance gains or losses. Only changes to th Uarch. You do understand the percentage performance gain or loss is independant of the percentage of change to the design, yes?

Ex: You put a 350 CFM carb on a 454 and cut 12mph off of it quarter mile time of 108. Thats a 10% performance loss, for a single component swap. Conversely, you could swap a cam and gain perfromance. The carburator and cam are relatively insignificant to the overall "whole" of the car.

But, without a carburator and/or cam the car is just scrap iron.
June 7, 2007 9:57:50 AM

Quote:


No, Penryn is a modified C2D Uarch, with more cache, new SSE extensions and a few other minor changes. Nothing significant,


+10% per clock is insignificant? C2D only has +20% on K8 :roll:

I wasnt speaking in terms of perfromance gains or losses. Only changes to th Uarch. You do understand the percentage performance gain or loss is independant of the percentage of change to the design, yes?

Ex: You put a 350 CFM carb on a 454 and cut 12mph off of it quarter mile time of 108. Thats a 10% performance loss, for a single component swap. Conversely, you could swap a cam and gain perfromance. The carburator and cam are relatively insignificant to the overall "whole" of the car.

But, without a carburator and/or cam the car is just scrap iron.Unless it's a fuel-injected two-stroke. :tongue: :D  :D 
June 7, 2007 12:14:31 PM

Quote:


No, Penryn is a modified C2D Uarch, with more cache, new SSE extensions and a few other minor changes. Nothing significant,


+10% per clock is insignificant? C2D only has +20% on K8 :roll:

I wasnt speaking in terms of perfromance gains or losses. Only changes to th Uarch. You do understand the percentage performance gain or loss is independant of the percentage of change to the design, yes?

Ex: You put a 350 CFM carb on a 454 and cut 12mph off of it quarter mile time of 108. Thats a 10% performance loss, for a single component swap. Conversely, you could swap a cam and gain perfromance. The carburator and cam are relatively insignificant to the overall "whole" of the car.

But, without a carburator and/or cam the car is just scrap iron.Unless it's a fuel-injected two-stroke. :tongue: :D  :D 

Sums up Barcelona's Computex showing quite well. 8O :twisted:
A two-stroke :lol:  :lol:  :lol: 
!