Avneesh Saxena, vice president of enterprise storage, systems and workstations, IDC Asia-Pacific, added that although AMD has a head-start over Intel in releasing its dual-core server products first, the company's success in making inroads into the market will depend on several factors.
"AMD's ability to gain share is linked to their OEM vendor partnerships, their existing momentum (in terms of market and mind share) and customer acceptance of dual-core (technology)," he explained.
He noted that it will take some time for the broader mainstream market to fully comprehend the benefits of dual-core architecture and adopt the technology.
Software licensing issues will also have to be ironed out. HP's Chu said that corporations will be mindful about how much more they will have to fork out, in terms of software licenses, if they move to a dual-core platform.
He added that some ISVs charge by the number of processors and may choose to up their fees for a dual-core system. While based on a single-chip architecture, dual-core machines are said to run like a multi-processor system.
It may also take some time before such issues are resolved as application vendors are just now considering multi-core processors in their pricing schemes.
For Teo Chin Seng, CIO of Singapore Technologies Engineering (ST Engineering), an upgrade will depend on the ability of a dual-core platform to offer better performance at a lower price. The question, he said, should be whether such chips can produce low-cost server systems.
"As end-users, we look for better price performance, so it's not about the CPU cost but the total cost of a system that uses the chip," he explained. "If it provides better price performance, we will certainly look at using dual-core processors in our servers."
Teo added that several factors affect the overall performance of a system including other chipsets, memory and buses. "Together they determine the effective speed or computing power. Having a faster core without more efficient memories or chipset may not increase the computation power of a server," he noted.
"How fast the server is for a given price is the key for us," he said, adding that his organization measures the processing speed of a server by running an application over the system.
Teo said ST Engineering will "go for it" if the server can support a particular software based on faster performance and at a better price. "Whether it is single, dual or multi-core, the (eventual) performance is more critical."
According to IDC's Saxena, early adopters will likely be companies that depend on high-performance desktops and workstations and run applications in a clustered server environment.
"Blade (servers) is definitely one such area where the compute capacity per footprint will shoot up with dual-core, and the benefits AMD brings from a power consumption perspective will make the offering more compelling," he said.
To gain further traction in the market, AMD will need to build a stronger relationship with its OEMs or create new alliances with second-tier server vendors and ties with local ISVs, he noted. These would increase its sales avenues, he said.
HP's Chu said that AMD is "very proficient" in running most 32-bit applications, and this is what mainstream customers typically look at.
"Opteron consumes low power and less heat, which would otherwise give raise to higher electricity bills. Heat and performance make a good value proposition…that's an important criterion for customers, which is why mainstream customers look at AMD."
But Chu agreed with Intel's Wu that applications will need to be "dual-core aware". Otherwise, while the software may still run, it may not do so at the optimized performance that a dual-core platform would provide for, he noted.
Play video
There are currently no comments for this post.