Intel Takes Quad Core To the Desktop 191
Rob writes to mention a Computer Business Review Online article about Intel's official launch of the Kentsfield chipset. Their Quad Core offering, Intel is claiming, is up to 80% faster than the dual-core Conroe released this past July. From the article: "Kentsfield, a 2.66GHz chip with a 1066MHz front-side bus, is more for computational-heavy usage, including digital content creation, engineering analysis, such as CAD, and actuarial and other financial applications. Steve Smith, director of operations for Intel digital enterprise group, claimed rendering is 58% faster for users building digital content creation systems, for video, photo editing or digital audio. In other words, Kentsfield is for high-end desktops or workstations only. For the average office worker who uses their PC for general productivity apps, such as communications and garden-variety computing, Smith recommended the Core 2 Duo from 'a price point and performance perspective.'"
Not native Quad core (Score:3, Informative)
http://xstremehardware.co.uk/index.php?option=com
Re:overkill (Score:5, Informative)
Other than jumping between cores to improve heat dissipation, how do you propose to make a highly serially-dependant algorithm run on more than one core at a time? Until computers can actually make programmers redundant by writing their own code given a high-level English description of the task (and even then, you'll still have some proveably-serial code), multithreading will remain at the whim of the programmers, not the scheduler.
also why dont they just make dual-core processors faster!
For the same reason they stopped the MHz-wars and moved to a core-war in the first place... Making each core faster has started to hit physical limits (power draw and heat dissipation, electron migration in progressively smaller transistors, clock speeds limited by the speed of light across the width of the chip, etc). Make no mistake, the speed will keep creeping up over time, but the end of 18-month speed doubling ended a few years ago. Major new improvements will either involve radical new technologies (and no, spintronics and diamond substrates will only yield incremental improvements) such as quantum, or what we see now, the move toward massive parallelism.
seems the only way we are going to get ahead in the field
Gaming, while interesting, does not drive research into the highest end of computing.
Benchmarks! (Score:3, Informative)
Intel's right. On games it doesn't do any better. On video though, well, lets just say I know some architecture majors who would have loved these in their lab several years ago, when 1 frame took 10 minutes to render. And they had 300 frame videos to do.
Re:640 (Score:2, Informative)
Re:Not native Quad core (Score:2, Informative)
I recently met with an Intel rep and they are very much pushing their new core architecture. Quad-core this year, Octo-core next.. Core count is the next clock speed. However one of it matters until the software manufacturers can take advantage of it, and very few server applications can at this point, let alone games.
Reference: http://www.itwire.com.au/content/view/7120/53/ [itwire.com.au]
Re:What about other parts of the computer? (Score:4, Informative)
Say hello to AMD, HyperTransport and integrated memory-controllers. Each CPU has it's own bank of RAM, and Each CPU is directly (well, 8-socket system needs one intermediate jump) connected to the other CPU's, and they can access the RAM connected to the other CPU's as well. So if you have dual-socket system, each socket has it's own RAM-bank, with 128bit bus between the CPU and the RAM, and the CPU can access the RAM attached to the other CPU as well. So as the number of CPU's goes up, the memory-bandwidth goes up as well.
This tech has been used since 2003 in the AMD's x86-64 CPU's. In the future AMD will have systems where you can plug co-processors and vid-cards to HyperTransport-sockets, alloweing them to directly communicate with the CPU's.
Re:Not native Quad core (Score:3, Informative)
There are three sides to this: Intel's, AMD's, and the truth.
Re:What about other parts of the computer? (Score:3, Informative)
Try connecting 4 of these chips together using only 3 HyperTransport links per core, with a single-hop memory latency, and allow for one link to external I/O. Can't be done. There are two hops required for the core that handles I/O, which is not a good thing when you consider how important I/O links are in a server.
Try connecting 8 sockets using only 3 HyperTransport links, and allow 2 connections minimum for external I/O - now most of your connections are two hops or more.
K8L attempts to solve these problems in two ways: [realworldtech.com]
1. K8L adds a fourth HyperTransport link, which allows easy single-hop 4-socket systems (and allows all 4 sockets to interface with external I/O, if desired).
2. K8L allows the HyperTransport links on each socket to be split from 4 16-bit links to 8 8-bit links, to allow single-hop memory latency on 8-socket configurations. Combined with the faster bus speeds of HyperTransport 3.0, that's plenty of bandwidth to feed 32 cores. And of course, there's potential for 16-socket configurations (with only 2-hop memory latency, depending on whether AMD decides to support this gluelessly).
Meanwhile, even with the massive caches and Dual Independent Bus architecture, Intel's 4-core chips are going to reach saturation at 4 sockets.
Re:number of processors (Score:3, Informative)
That page's primary purpose is to clarify their policy for server software (probably in response to Oracle defining a "processor" as a core), but they also mention Windows XP:
Windows XP Professional can support up to two processors regardless of the number of cores on the processor.
If Microsoft changed their definition of "a processor" for Windows Vista, believe me, we would have heard all about it by now. Slashdot would have been slashdotted by all the comments.