Update 15:10 04/07/07:
Having gathered more information, it turns out that the Barcelona benchmarks are performance estimations for Barcelona at 2.6GHz, not 2.3GHz as originally reported. The results are also based on "internal AMD simulations", so the validity of the results is somewhat questionable.
I've left the original story in place below, and there's also a picture of the slide at the bottom too.
---
Amid speculation of even further delays to AMD's next-generation quad-core processors, the company has been talking about Barcelona's performance with its partners.
The Inquirer has managed to get hold of some of these figures, and although they're synthetic and won't give a true indication of application performance, they might give us an idea of what ballpark AMD is expecting to hit.
Despite the comparison being between competing server parts in the $800 USD price bracket, we think it's potentially very relevant because both the Opteron and Xeon processors are based on practically identical architectures to their desktop siblings.
According to the figures, AMD's quad-core Opteron, which is clocked at 2.3GHz, manages to outperform Intel's similarly priced 2.66GHz Xeon 5355 processor by a significant margin.
In SPECint_rate2006, a measure of integer performance, the unreleased Opteron is 21 percent faster than the Xeon. If that wasn't a big enough margin, the Xeon reportedly slips behind the Opteron by 50 percent in SPECfp_rate2006, a floating point benchmark. The latter is certainly an impressive margin and one that Intel will not be able to make up without a significant clock speed boost when it releases its 45nm Penryn processors.
Of course, we're going to remain sceptical until we've seen how well AMD's next-generation processors perform in real applications, but this does whet our appetite a bit while we're waiting for the launch in the fourth quarter of this year. Will you be waiting for some solid performance numbers before you make the plunge? Tell us in our forums.
Having gathered more information, it turns out that the Barcelona benchmarks are performance estimations for Barcelona at 2.6GHz, not 2.3GHz as originally reported. The results are also based on "internal AMD simulations", so the validity of the results is somewhat questionable.
I've left the original story in place below, and there's also a picture of the slide at the bottom too.
---
Amid speculation of even further delays to AMD's next-generation quad-core processors, the company has been talking about Barcelona's performance with its partners.
The Inquirer has managed to get hold of some of these figures, and although they're synthetic and won't give a true indication of application performance, they might give us an idea of what ballpark AMD is expecting to hit.
Despite the comparison being between competing server parts in the $800 USD price bracket, we think it's potentially very relevant because both the Opteron and Xeon processors are based on practically identical architectures to their desktop siblings.
According to the figures, AMD's quad-core Opteron, which is clocked at 2.3GHz, manages to outperform Intel's similarly priced 2.66GHz Xeon 5355 processor by a significant margin.
In SPECint_rate2006, a measure of integer performance, the unreleased Opteron is 21 percent faster than the Xeon. If that wasn't a big enough margin, the Xeon reportedly slips behind the Opteron by 50 percent in SPECfp_rate2006, a floating point benchmark. The latter is certainly an impressive margin and one that Intel will not be able to make up without a significant clock speed boost when it releases its 45nm Penryn processors.
Of course, we're going to remain sceptical until we've seen how well AMD's next-generation processors perform in real applications, but this does whet our appetite a bit while we're waiting for the launch in the fourth quarter of this year. Will you be waiting for some solid performance numbers before you make the plunge? Tell us in our forums.
RELATED ARTICLES
MSI MPG Velox 100R Chassis Review
October 14 2021 | 15:04
Want to comment? Please log in.