News
 

Bookmark and Share

(5) 

Both Advanced Micro Devices and Intel Corp. are working hard these days to create low-power high-performance microprocessors for servers as power consumption and heat dissipation are becoming major problems of modern datacenters. According to Neal Nelson, an independent computer performance consultant, AMD’s Opteron HE “Shanghai” platforms are more efficient compared to low-power Intel Xeon 45nm platforms.

“The AMD-based server beat the Intel based server in our tests. We processed millions of transactions against a real database and we measured the total throughput and the total power used by the servers,” said Neal Nelson, president of the consulting firm.

The servers were tested with three different transaction workloads: a calculation intensive workload, a disk I/O intensive workload and the "idle" state when the server was powered up and waiting for transactions to arrive. The entire test suite was run twice, first with 4GB and then with 16GB of main memory.

The servers used in this test had virtually identical configurations. The only notable differences were that the Intel server had an 8% faster clock frequency with 2.5GHz versus 2.31GHz and the Intel server used fully buffered memory modules (FB-DIMMs). It should be noted that Intel is about to roll-out Xeon processors based on the Nehalem micro-architecture with integrated memory controller. Therefore, the results will be mostly outdated in several months time.

For the calculation intensive workload the Opteron reported 1% to 6.3% higher throughput.

The disk intensive throughput tests ranged from 1.1% favoring the Xeon to 5.0% favoring the Opteron.

The power consumption, however, favored the Opteron in every test with a range of 13.1% to 20.8%.

When the power readings were normalized by throughput the Opteron advantage ranged from 21.8 to 26.8%.

The net effect of architectural differences between AMD Opteron and Intel Xeon platforms (the main difference is integrated memory controller in AMD chip and external memory controller that supports only FB-DIMM in case of Intel Xeon) is that an currently available Intel-based server can draw more power at the platform-level when compared to an AMD based server that delivers comparable throughput. A server's total throughput and total power consumption are much more important than whether some individual computer chip consumes more or less power.

The Nelson test results should not be confused with power usage test results from the Standard Performance Evaluation Corporation (SPEC).  The SPECPower test was created by a committee of engineers from computer manufacturers (including Intel).  SPEC ssues a disclaimer which specifically states that the test results may not predict real world experiences, according to the consulting firm.

Tags: AMD, , Intel, , Opteron, Shanghai, 45nm

Discussion

Comments currently: 5
Discussion started: 03/20/09 04:25:32 PM
Latest comment: 03/22/09 04:16:51 PM
Expand all threads | Collapse all threads

[1-2]

1. 

Nice work of AMD's second iteration of their 45 nm Opterons, but for the sake of truthfullness and completeness, let's see how they match up against Intel's due-to-arrive-in-March/April Nehalem based 2-way Xeon LV's when these roll out of the fabs. Going from FSB -> QPI, memory controller off-die -> on-die and FB-DIMM DDR2 -> ECC REG DDR3 should show a different picture to say the least.
0 0 [Posted by: sanity  | Date: 03/20/09 04:25:32 PM]
Reply
- collapse thread

 
That will not be provide a fair advantage. The Opteron based on the Shanghai is designed to compete against quad core Xeon processors. They were not designed to compete against Xeon DP processors. AMD's Bulldozer core will provide a fair advantage with Intel's Nehalem core processors. Unfortunately, the Bulldozer core will be out in 2011. As an AMD fan, I would be surprise if the Shanghai based Opteron comes close to the performance as Xeon DP.

What AMD does have that Intel does not have is an easy way to upgrade processors with out taking apart the server or getting new hardware.

The Shanghai is the only 45 nm fab processor from AMD for their server processor line up. While the Deneb is the only 45 nm fab processor for the desktop line up.
0 0 [Posted by: jmurbank  | Date: 03/20/09 06:30:34 PM]
Reply
 
My bad; I thought AMD already went to 45 nm with Phenom 1 and its server counterpart. But anyway, this comparison still looks iffy to me. Shanghai has Harpertown to compete with now allright, but as already mentioned above Gainestown is on its way, along with the architectual changes for that platform. Most important aspect in terms of the electricity bill is getting rid of those friggin' hot FB-DIMMs; the others give a nice speed boost and place Intel's platform technically on a roughly equal footing with AMD's. Which brings up the question of AMD having a higher performance per watt ratio being due to Xeon vs Opteron, or FB-DIMM vs ECC REG DDR2 and FSB vs HT. Hence the question in my first post.



BTW: Xeon is just the brand name for intel x86(-64) server CPU's regardless of number of cores. And what about Magny-Cours? That 12-core beast seems to get little press coverage lately. I for one am pretty interested to see these babies come into action.
0 0 [Posted by: sanity  | Date: 03/20/09 09:06:28 PM]
Reply

2. 
That a l@me article. compared CPU, but checked disk throughput!
How northbridge can inpact CPU speed so much? Linux with EXT2 lol
Can't see what memory peases used in tests, what it's speed?
Comparing CPU without any CPU calcualtion tests!
Looks like it is made by AMD fanboy...
0 0 [Posted by: deepm  | Date: 03/22/09 01:37:22 AM]
Reply
- collapse thread

 
There is plenty of evidence that AMD Opteron based on the Shanghai core consumes less power than present Xeon processors. A site like anandtech.com has this information. One of Xeon's problem is FB-DIMM and the other is Core 2 microarchitecture is not designed for servers. FB-DIMM uses more power compared to traditional DDR2 ECC memory. Also FB-DIMM has more latency compared to traditional DDR2 ECC memory. Having the memory controller built into the processor improves performance because shorter wires can be used and do not have to wait a long time to go through different hardware layers.

AMD is the best for servers while Intel is the best for both desktops and notebooks at this time. AMD makes it easier to upgrade the server with out being force to dismantle all hardware just to use the latest processor.
0 0 [Posted by: jmurbank  | Date: 03/22/09 04:16:51 PM]
Reply

[1-2]

Add your Comment




Related news

Latest News

Wednesday, August 27, 2014

9:09 pm | Samsung Begins to Produce 2.13GHz 64GB DDR4 Memory Modules. Samsung Uses TSV DRAMs for 64GB DDR4 RDIMMs

Tuesday, August 26, 2014

6:41 pm | AMD Quietly Reveals Third Iteration of GCN Architecture with Tonga GPU. AMD Unleashes Radeon R9 285 Graphics Cards, Tonga Graphics Processor

Monday, August 25, 2014

6:05 pm | Chinese Inspur to Sell Mission-Critical Servers with AMD Software, Power 8 Processors. IBM to Enter Chinese Big Data Market with the Help from Inspur

Sunday, August 24, 2014

6:12 pm | Former X-Bit Labs Editor Aims to Wed Tabletop Games with Mobile Platforms. Game Master Wants to Become a New World of Warcraft

Thursday, August 21, 2014

10:59 pm | Khronos Group to Follow DirectX 12 with Cross-Platform Low-Level API. Khronos Unveils Next-Generation OpenGL Initiative

10:33 pm | Avexir Readies 3.40GHz DDR4 Memory Modules. DDR4 Could Hit 3.40GHz This Year

12:10 pm | AMD to Lower Prices of A-Series APUs for Back-to-School Season. New Prices of AMD A-Series APUs Revealed