Chip major Intel announced the launch of quad-core processors on Thursday, September 6 in Hyderabad, designed for servers running multiple applications. The Director-Sales, Intel South Asia, Mr R. Ravichandran, said these Intel Xeon multi-processor servers would now form the backbone of enterprises delivering new levels of performance. Mr Ravichandran said that these processors have the capability to handle multiple tasks and are akin to four brains on a single chip.
The six new Quad-Core Xeon 7300 series processors can deliver more than twice the performance and more than three times the performance per watt over the company's previous generation dual-core products. These would fit well into environments with multiple servers and large databases, and prove to be ideal for business intelligence.
Mr Ravichadran said that a large part of the work for the quad-core processors, such as design, validation and testing was handled by engineers based in the Digital Enterprise group of Intel India at Bangalore in a record time of 15 months.
At the top end of the GHz scale is the 130-watt 2.93GHz model. 80-watt models sit in the middle, while 50-watt 1.86GHz variants are designed for blades and other high density environments. "The quad-core multi-processors in the 7300 series have shown enterprises, data centers and small and medium businesses (SMBs) can save 85 percent of space, reduce energy bill by 70 percent and yet improve performance by about 90 percent," Ravichandran said.
Servers based on Xeon 7300 series processors are expected to be announced by more than 50 system manufacturers, including HCL, Wipro, Dell, Egenera, Fujitsu, Fujitsu-Siemens, Hitachi, HP, IBM, NEC, Sun, Supermicro, and Unisys.
Let's take a look on the technology and specs used by these new CPUs.
First we need to talk about how multi-core CPUs are built. There are two approaches. On the first one, called "multi-chip", each core is made by different chips that are put together in a single package. Pentium D is a good example of a dual-core CPU that uses this approach. On CPUs based on this technology when one of the CPU cores wants to talk with the other, it needs to go outside the CPU package, i.e. use the front side bus. Also the L2 memory cache is separated, thus when one core wants to access data stored on the cache of the other core, it need to access it thru the front side bus. Of course using the front side bus isn't the best option, as the CPU runs internally faster than then front side bus.
The second approach is called "monolithic". Here the cores are already manufactured in only one chip. This is the technology used by Core 2 Duo CPUs, for example. When the CPUs want to talk to each other they don't need to use the front side bus, as they were already built stuck together. Also, under this technology the memory cache is shared between the two cores. Performance wise this is the best option. This is the basic difference between Dual Core and Core 2 Duo CPUs
The new quad-core CPUs from Intel will mix both technologies. They will have two monolithic dual-core chips installed together in a single package. This means that cores 1 and 2 share the same memory cache, the same happening with cores 3 and 4. If cores 1 or 2 needs to talk with core 3 or 4 or to access the other cache, they will need to this externally, i.e. using the front side bus - which on the first models will be of 1,066 MHz. We drew a diagram illustrating this on Figure.
Intel's approach for their first quad-core CPUs
These processors will run with a 1,066 MHz external bus, 8 MB total L2 cache (4 MB x 2) and will come with its over clocking protecting unlock, following the tradition of Intel's Extreme series. The other specs continue to be the same as the rest of the Core 2 Duo family. Pricing of the new quad-core processors depends on the speeds, features, and amounts ordered, and ranges between Rs 34,830 and Rs 93,618 per CPU.
Related Online Articles:
No comment yet. Be the first to post a comment.