Slashdot Mirror
Canterwood Motherboards Refined
YingYang writes "With Intel's i875P (otherwise known as Canterwood) chipset launch a couple of
weeks ago, we were shown what an 800MHz System Bus can do for performance of the
Pentium 4. At the time however, there were few Taiwanese OEM motherboards out and test-beds used to showcase the new chipset and throttled-up P4, were based on Intel designed motherboards. Now however, the Canterwoods are beginning to flow out of Taiwan and vendors like
Abit and Asus have put together boards with a ton of integrated features and performance, that reminds us of the days of the 'BX,' when Intel chipsets were the only way to fly. Check out
this Abit/Asus Canterwood head to head comparison at HotHardware."
Anyone know why they are sticking the IDE sockets on their side?
Mouse powered Chips, Open source Processors and Lego
This is the chip that kills intel's god awful dependency on rambus, about time. VIA did release a chipset that supported ddr quite some time back IIRC, but the performance was god awful so it was 'overlooked'. It appears as if this one beats out rambus on performance which is a "good thing"(tm)
Not only does Intel dominate the processor market, also Intel's chipsets run many of the best motherboards. I'm starting to wonder if Intel will ever see any real competition. We all know AMD, although more bang for your buck in many instances, will never be the dominate chip maker at this rate. Anyone see any upcoming companies that will challenge intel's deathgrip on the #1 position in this industry?
The Springdale 865 chipset has started to show up in motherboards. The price in nearly $100 less. And the 2 chipsets are nearly identical as quoted from this article
.13-micron process. But only those components that pass Intel's stringent requirements, including optimum timing (Intel calls this Performance Acceleration Technology, or PAT), are qualified as 875P. Intel has different requirements for those components that will qualify as Springdale."
"In many respects, the 875P is identical to Intel's forthcoming Springdale chipset, which will launch next month for the mainstream PC market. Both have an 800MHz FSB and offer support for dual-channel DDR400 memory, Serial ATA, AGP 8X, Gigabit Ethernet, and Intel's own Hyper-Threading technology. In fact, both chipsets are manufactured using the same
http://www.kubuntu.org/
This could very well put the desktop performance lead firmly into Intel's hands.
The 3.0 GHz P4 with 200 MHz FSB and dual channel DDR400 should handily beat the Athlon XP 3200+, and it will likely be priced less initially since Intel expects to introduce a 3.2 GHz part at the same time. That part will have a large performance margin over anything AMD has or will introduce this year.
I don't think they will get this technology into the Xeons soon enough to fight off Opteron, though, or even to take the performance lead for x86 servers.
Intel's BX chipset was for the PII but works on the PIII as well. I wonder if more PII or PIII processors were mated with this chipset over the years? I've got four BX chipset boards now (one Intel, one Tyan and two Asus) running processors from PIII 850 Coppermine to 1.3 Gig Tualatin Celerons.
"My mother works for Microsoft now. A whole other cult."
Ok, I've got a couple BX board still running .... That chipset was fast and performed far beyond what Intel was hoping for. My Abit BE6-II reached amazing FSB overclocking while remaining stable, and my SuperMicro SBU continues to push content in production, having the CPU updated from a PII 300 to a PIII 800. The Abit was over designed, allowing me to go from a PII 266, PII 400, PIII 500, to finally a Celeron 566@952. The chip had legs.
Contrast this to the Intel boards, however. They were soo bloody afraid of someone running the CPU faster than the spec, that they tended to not handle the additional voltages or clock multipliers. Intel designed motherboard is not an asset in my book.
As for the chipset itself, it will be some time before it proves it's salt. I got burned badly with the i840 chipset debacle and stranded with the GX chipset. The i840 was what drove me to AMD on the workstation side.
+++ UGUCAUCGUAUUUCU
Actually the performance wasn't so bad. The problem was that there were legal issues concerning the VIA's right to make such chipsets. This led to few large OEM mobo makers to use the chips, making it hardly worth investing in.
Now that the issues are settled though, you can expect to see more competition from VIA's new P4 chipset that will support 800Mhz FSB CPU's.
-- taking over the world, we are.
That only works if people actually buy the competitive products that are supposed to keep Intel honest.
One guy who got deservedly lambasted on Usenet years ago said, "I like having AMD and Cyrix around, since they keep Intel's prices down for me." AMD and Cyrix aren't in the business so others can buy Intel's chips more cheaply. For all of our innovation, we're in a really STUPID industry, because for practically its entire history, computing has simply moved from one monopoly to another. From IBM to Microsoft, and now it's apparent that Microsoft is on the wane, because Intel is emerging as the One Dominator. Only for intervals where one monopoly is falling and another rising do we have real competition. I know it's been Wintel for over a decade, but really Microsoft has been in the driver's seat. That's changing, right now.
The living have better things to do than to continue hating the dead.
But, the new chipset provides wider pathways for everything on the system bus. your IDE channels have wider paths for data transfer. your PCI bus is wider. most current technology has a limit of only transfering around ~120 mb/sec across the whole PCI bus (which included the IDE channels).
AGP was brought around to provide faster access to the CPU/northbridge than PCI.
AMDs HyperTransport technology (which my motherboard has) widens the Bus paths between the south bridge and north bridge to 4 bits (from a 1 bit path and hypertransport can scale to 64 bit wide paths!). now all of my PCI, IDE, SATA, Onboard Audio paths are 4 bit wide capable of going at almost 800 mb/sec
This is what is going to put Intel back as the performance market leader... until that hypertransport starts getting into the 8 and 16 bit wide bus pathways....
I follow the SDK and GDN principles.. Spelling Dont Kount, Grammer Dont Neither
This ignorant statement gets repeated (With, of course, ever-increasing frame rates plugged into it) so often that I feel compelled to blow it away here before it gets voted up by some mod bastards.
The "fps" rating is an average. As the number of polygons and lights in a scene changes over time, so does the time to render a frame. You might have 150fps average but get, say, 20 fps when you walk into a big open room and find eight or nine people blowing each other away with assorted lighting effects and so on occurring.
Also, more bandwidth means you can sustain higher fill rates which in turn means you can run higher resolutions. While this is much less of an issue now between CPU and video card because of onboard T&L which we have enjoyed greatly since nvidia brought it to PC gaming with the GeForce 256, pushing textures still depends on bus bandwidth. Anything memory-intensive does as well, so doing video encoding (a much more common practice than you think; witness all the DVD copy software out there) is highly dependent on memory bandwidth. Since intel is not using an integrated memory controller, this makes a big difference. AMD doesn't need as much FSB speed in the Athlon 64 and Opteron as intel needs in P4, because they have an integrated memory controller.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
Wow, that's almost 2% faster for twice the price! What a deal.
And only 10x more power then 2% of the users can figure out what to do with.
On the other hand, it's almost enough to actually run XML based Java applications.
- Adam L. Beberg - The Cosm Project - http://www.mithral.com/