Intel Reveals Core i3, i5, i7 CPU Naming System

Lynnfield, Clarksfield, Arrandale, Clarkdale – all codenames for upcoming Intel chips based on the latest architecture and all very confusing.

ZoomToday we know that the “Core” family for the performance segment goes mainly from Core 2 Duo to Core 2 Quad all the way to the top with Core i7. But with Core 2 on the way out to be replaced by new CPUs later this year, Intel needs to sort out its naming and branding conventions to that consumers will be able to figure out what they’re buying.

Intel corporate communications manager Bill Calder wrote in a blog post, “Today the Intel Core brand has a mind boggling array of derivatives (such as Core2 Duo and Core 2 Quad, etc). Over time those will go away and in its place will be a simplified family of Core processors spanning multiple levels: Intel Core i3 processor, Intel Core i5 processor, and Intel Core i7 processors.”

“Core i3 and Core i5 are new modifiers and join the previously announced Intel Core i7 to round out the family structure. It is important to note that these are not brands but modifiers to the Intel Core brand that signal different features and benefits,” Calder added.

The new Core line will be naturally be position from bottom to top Core i3, Core i5 and Core i7. The odd-numbers convention makes sense with the possibility that some consumers may confused Core 2 Duo and Quad with 2 and 4.

The desktop processors codenamed Lynnfield, which are due this fall, will marketed as both Core i5 or Core i7, depending upon the feature set and capability. Interestingly, all Clarksfield mobile chips will have the Core i7 name.

The lower-cost line will remain mostly unchanged with the Celeron being the entry point, the Pentium for basic computing and the Atom doing what it’s been doing to the netbook and MID segment.

“For PC purchasing, think in terms of good-better-best with Celeron being good, Pentium better, and the Intel Core family representing the best we have to offer,” said Calder. “This will be an evolutionary process taking place over time, and we acknowledge that multiple brands will be in the market next year including older ones, as we make the transition. But overall this is a good thing, designed to make it easier and more rational over the long run.

The WinFast PxVC1100 Video Transcoding Card: Worth The Price?

Table of contents

• 1 – Introduction
• 2 – The WinFast PxVC1100
• 3 – Installation And Use
• 4 – Test Systems And Benchmarks
• 5 – Encoding Performance Benchmarks

• 6 – Image Quality Comparison
• 7 – CPU And Power Usage Benchmarks
• 8 – Conclusion: Encoding Performance Per Dollar

Sony, Toshiba, and IBM probably didn't put much thought into using the Cell microprocessor as part of a Windows-based PC. Those of you unfamiliar with the Cell Broadband Engine Architecture likely know of its primary role in the consumer space as the CPU platform in the PlayStation 3 game console. You may also have heard about the U.S. Customs Enforcement Cyber Crimes Center using the PS3 running Linux to crack the passwords of suspected child pornographers. Indeed, the Cell processor is a very powerful component, and every Sony PS3 is rated for somewhere around 204 GFLOPS (single-precision float).

Admittedly, a gigaflop, which represents one billion floating point operations per second, is not a perfect metric, as these processors have different strengths and weaknesses. However, the number of gigaflops does give you an idea about the Cell's compute performance. When this article was written, the world's second-fastest supercomputer, the IBM Roadrunner, was using a combination of 12,960 Cell processors and 6,912 Opteron processors.

The trio of Sony, Toshiba, and IBM co-designed the Cell broadband engine architecture. Toshiba has taken the liberty of modifying the Cell processor into something a little more appropriate for multimedia processing by retaining half of the Cell's eight synergistic processor elements (SPEs) and adding video encoding and decoding hardware, it created the SpursEngine processor:

Consuming a mere 10-20W, the SpursEngine is intended to fill the role of a co-processor dedicated to 3D and video processing. With four SPEs running at 1.5 GHz, the SpursEngine is good for about 50 GFLOPS, and capable of 8/16/32-bit integer and single/double precision floating-point calculations. The processor has one megabyte of local memory, 256KB of which are dedicated to each of the four SPEs.

I know what you're thinking: it sounds great, but what can it do for me? Well, Leadtek asked the same question and responded by placing the SpursEngine processor on a video transcoding board called its WinFast PxVC1100.

We look at this SpursEngine-equipped card to show you just what it can do. However, more importantly, we're going to compare its performance to a number of CPUs across the price spectrum to see if your hard-earned dollars are best spent on the WinFast PxVC1100 or if they are better served upgrading your host processor instead.

HP Puts 1000 Cores in a Single Rack

Want to get the most processing power possible in your data center to run cloud computing and Web 2.0 apps? HP is introducing the ProLiant BL2x220c G5 server blade today, which doubles the processing density by putting two servers into each half-height blade. Using Intel Xeon 5400 quad-core processors, you can put up to 1024 cores and 2 terabytes of RAM in 128 servers in a 42U c-Class rack – that’s 12.3 teraflops in eight square feet.

Fitting two servers into a single blade means leaving some things out, Iain Stephen, HP’s vice president for industry standard servers, said.

“The memory, the drives, the processors and the heatsinks are the things that take up the space in a server,” Stephen said. “We stripped off the things customers tell us they don’t value: we stripped off the hot pluggable storage; we stripped off the storage redundancy; we reduced the memory footprint and we have a smaller number of DIMM sockets. When a customer looks at the 220c, they may think it’s underspecified. But on the connectivity side we’ve enhanced things; we can add InfiniBand or high-speed Ethernet. It’s a balance. If I already boot from network-attached storage, if I’m running an app where four DIMMs are sufficient, if I can compromise on local storage and the memory footprint – then I get the processor density."

HP expects customers to use the dual gigabit Ethernet network interface card and optional x8 PCI-Express mezzanine socket allowing 4x double data rate InfiniBand to connect to storage arrays like HP’s petabyte-scale ExDS9100 instead of using storage inside the rack. Making room for more processors creates a much more efficient system, Stephen said. “You have to balance the amount of processing per square foot and the power requirements. You can either drive towards ultimate density or the ultimate in efficiency, but with the 220c you get three times the density of a 1U rack,” Stephen said. “We use the same power supplies, the same fans and the same chassis, but we double up the density and hopefully get 60% better performance per watt.”

Specifically, HP claims a 60% performance-per-watt advantage over a cluster of Dell PowerEdge 1955 servers. In HP’s own tests using the SPECjbb2005 benchmark to measure business operations per second, the BL2x220c delivers 1,582.73 bops/watt compared to 958.86 bops/watt with PowerEdge.

The approach does share similarities with IBM’s iDataPlex system for Web 2.0 computing, Stephen said. “There are only so many things you can flex in the x86 architecture,” Stephen said. “You can flex the I/O, the processing, the number of sockets and cores, the memory – these are the core technologies. We’re flexing the same number of things but the way we deliver the balance is slightly different to the way IBM delivers iDataPlex.”

HP isn’t adding water cooling or other extreme measures to the new blades. Instead, it relies on the c-Class chassis’ features, such as the ability to turn off four or five of its six power supplies to deliver power at 90% efficiency. The c-Class chassis also uses 10 Active Cool Fans, based on the design of jet engines for radio-controlled model aircraft. The fans run at up to 166 miles per hour and are more efficient than the fans in each server. Initially, HP is offering only Intel Xeon 5400 quad-core or Xeon 5200 dual-core processors. HP will have AMD’s quad-core Barcelona processors in other ProLiant servers while Stephen said there could be a dual-server Barcelona blade if there’s demand. “Intel has had a performance advantage since May last year so the majority demand is for Intel processors,” Stephen said. “If Barcelona delivers - and I think it probably will - I expect customer demand to split between the two again.”

Although most customers for the BL2x220c blades will be businesses running cloud computing and Web 2.0 applications or high-performance computing systems where fitting in thousands of servers at a time is critical, with prices starting at $6,349 Stephen predicted the blades will appeal to some smaller companies. “There will be small business customers who look at this and say ‘we’re already using a small storage network, and these are ideal to use as file and print servers, Web servers and application servers,’” Stephen said. But one of the first systems will go to special effects company WETA Digital for use on films like James Cameron’s “Avatar,” “Neon Genesis Evangelion” and the “Halo” adaptation.

Zoom