Intel made a big splash at the IDF held in Beijing, a lot of novelties brought forth, and some light shed on previously known plans. The first of these include the performance Penryn
will bring to us, humble users, compared to the current Conroe-based products. The company isn't wasting any time; they are, in lack of a better word, afraid of something, and that something I believe to be AMD. But how is that possible? I mean, from the "stories" I've heard so far, AMD's new line of processors ISN'T going to overtake Intel's offerings.
And even if that were true, Intel is already taking precautions by giving out information about Penryn, Nehalem, and other products; they are expecting to increase the FSB on Nehalem-based processors so that they would be a match for AMD's Barcelona-based processors. They're publishing numbers with synthetic tests that show their products are better, faster, eat less power, give away less heat, and so on.
Patrick P. Gelsinger, the senior vice president and co-general manager of Intel Corporation's Digital Enterprise Group (DEG) mentioned at IDF that for desktop PCs, "a 15 percent increase in imaging-related applications is expected, 25 percent for 3D rendering, more than 40 percent for gaming and more than 40 percent faster video encoding with Intel SSE4 optimized video encoders."
The tests were done by using 45nm-based quad-core processor running at 3.33GHz with a 1333MHz FSB and 12MB (2x6MB) L2 cache, against an Intel Core 2 Extreme QX6800 processor (2.93GHz, 1066MHz FSB 2x4MB L2 cache). Of course that when you look at the numbers they look impressive to the naked eye, this being that kind of marketing "magic" PRs use to make you see something that isn't there. The difference is notable, but, at least, it's spread unevenly, you get 15 percent for imaging, 25 percent for 3D, 40 percent for gaming and 40 percent for video encoding. The numbers sure sound well adjusted to the consumer's needs, a lot more people are interested in gaming and video encoding than the ones that deal with Maya, 3D Studio Max, POV-Ray or Autocad and Photoshop.
Not to mention that the platform used deals with quad-core applications, and the software support for multi-core processors is still limited, but let us not digress. The point I was trying to make is that the difference has something to do with the new SSE4 instructions, 47 instructions to be more exact, the remaining 7 will most likely be used on Nehalem-based processors, the rest of the performance gap being filled by the difference in the processing frequency and FSB. Not that I don't believe that there will be a performance increase from the Penryn processors, but for this test, I would have liked it if they had overclocked the quad-core QX6800 to the same frequency, and/or FSB, depending on the multiplier of the Penryn processor, and then see what the performance gap between the processors really is.
Penryn is just another remake of the Conroe core; it is built on a smaller manufacturing process and allows for less power consumption and heat dissipation, which means that it would perform better than Conroe-based processors, but not by much, given the fact that we are talking about the same architecture. On the High-Performance Computing (HPC) segment, as well as on workstation systems, a 45 percent in bandwidth intensive applications is expected, with a 25 percent increase "for servers using Java."
This line of tests was conducted by using Xeon processors built on the 45nm process node, with a 1600MHz FSB for the workstation and HPC tests and a 1333MHz FSB on the server tests. Against these models, Intel pitched quad-core Xeon X5355 processors (2.66GHz, 2x4MB L2 cache). No need to comment on those results now, do I?
I'd just like to finish with the words of Justin R. Rattner, Intel's chief technology officer: "Welcome to the era of multicore, an era in which all of our computing capabilities will multiply our own personal capabilities (...) Intel is delivering a breadth of multicore processors worldwide and a product roadmap providing the incredible performance boost and energy efficiency needed to put the consumer more in control of the information age."