AMD goes PCIe, onboard that is.

[Aggressor Prime]

Well, it looks like AMD will be integrating another component on their already powerful, low-latency CPUs: the PCIe controller. I just hope these new Socket 1207s will support SLI/Crossfire. If they can't do both, they better go SLI.

[Mr. Perfect]

S1207 is the M2 socket, yes or no?

[Edit]Nevermind, it's the Opteron socket that gets this, not the desktop. So it means jack ★■◆● to gamers.

[DCrazy]

PC-on-a-chip baybee!

[Aggressor Prime]

That article seems anti-AMD.

Anyway, Opterons aren't too bad for gaming. If games start utilizing multiple cores, as the PS3 will promote, then multi-CPU systems will become popular among gamers. Also, there are the Opteron 100s.

[fliptw]

No, your just anti-sane.

That article is stupid bashing based in limited info.

AMD only stated on standardizing on a single socket format only for socket A chips - never made any promises about anything after that(not to mention that adding functionality to a chip demands more lines).

Tho, is anyone doing anything to reduce potential accidents with heat sink installs?

[Aggressor Prime]

My last post talked about the Opteron 100s utilizing this technology.

Apparently, the 100s will continue with the Athlon 64s as they have started and migrate to Socket M2 as they have migrated already to Socket 939.

[Mr. Perfect]

Have you priced an Opteron lately? They're considerably more expensive and as such, not all that atractive for a gamer rig. Also, if you'd read the article, even if the 100s migrate to M2 they would not have the on-board PCIe. Only the 1270 pin Socket F Opterons will have it. That's what all the extra pins are for in that socket.

Tho, is anyone doing anything to reduce potential accidents with heat sink installs?

If you mean chipped cores, all Athlon 64s have heatspreaders.

[Admiral LSD]

As long as gamers remain stupid enough to shell out for technologies with effective life cycles of 6 months or less then this technology will end up in a fair few gaming PCs, particularly if AMD keep basing the Athlon FX line on the Opterons.

[woodchip]

As long as gamers remain stupid enough to shell out for technologies with effective life cycles of 6 months or less then this technology will end up in a fair few gaming PCs, particularly if AMD keep basing the Athlon FX line on the Opterons.

Since "ALL" comp. tech. is old after 6 months, just what is the "Intelligent" gamer going to buy?

[Krom]

As long as gamers remain stupid enough to shell out for technologies with effective life cycles of 6 months or less then this technology will end up in a fair few gaming PCs, particularly if AMD keep basing the Athlon FX line on the Opterons.

The Athlon FX line uses the same cores as the normal Athlon 64s now, and has been for some time. They use 939 pin sockets and do not require registered DDR memory like Opterons do. Past that, the Athlon 64, Opteron, and Athlon FX are all using the same execution unit.

[Ferno]

That article does have a point. AMD did screw up by having two different sockets. So a person like me goes: 'well if I get a board with socket 754, I can't get anything in the 939 formfactor'.

now it's: 'I can either get a chip without the integrated PCIe controller and stick with what I already bought, or I can get the one with the integrated PCIe controller. But that means I have to get a new motherboard, and new memory'.

It's one thing to make a mistake and learn from it, but it's quite another to make essentially the same mistake again. that's defined as 'stupidity'

[fliptw]

AMD stuck with socket A while improving performance for the XP line - Intel went from 423 to 478 during that time. Neither added any functionality that demanded extra pins.

AMD didn't do what intel did, which they promised to do -not changing socket format for no good reason.


That being said, how exactly are you going to get the PCI controller onto the die without adding more lines(pins)? You can't. You are moving the PCI controller off the motherboard onto the die, and thusly moving the lines that connected the controller to the many PCI devices that populate the board.

Insanity is doing the same thing over and over again and expecting different results.

[Aggressor Prime]

For those who don't want any change whatsoever:
How could AMD be doing the wrong thing by integrating a PCIe controller? If an onboard memory controller can take down Intel, why not an onboard PCIe controller as well.

For those who want all product lines to change to "Socket F" and "Socket F - 1 pin":
AMD has always been there to not make a mistake. If they shifted all of their products to an onboard PCIe controller, and something goes wrong, then where do they turn? The server market is like an investment to them. They test the new technology. AMD went this route with the first K8s. They might as well take the same route with the last of the K8s. I'm sure when the K9s come out in 2007, everything will have PCIe controllers.

[Admiral LSD]

AMD has always been there to not make a mistake.

LOL, if that's the case then why did they release at least three CPU lines (K6, K6-2 and K6-3)with the same basic flaw: poor floating point performance?

[MD-2389]

Dude, its pentium3, what do you expect? He just switched from masturbating to intel.com to amd.com.

[Ferno]

For those who don't want any change whatsoever:
How could AMD be doing the wrong thing by integrating a PCIe controller? If an onboard memory controller can take down Intel, why not an onboard PCIe controller as well.

oy...

this has nothing to do with AMD integrating the PCIe controller onto the die. This has to do with AMD putting the PCIe controller onto only the OPTERON which requires a differet socket, instead of putting the controller on ALL A64's.

[Aggressor Prime]

Uh, please find and read the comment posted to you. I'm sure if you read my whole post you will find it.