




|
Via:
AMD | News Archive
| Tags:
AMD,
Intel,
core i7,
Nehalem,
Atom,
Core i5,
Sandy Bridge,
Thuban,
Shanghai,
APU,
Bulldozer,
Bobcat,
Magny-Cours,
Ontario,
Hot Chips,
CPU module,
CPU+GPU,
Brazos
|
|
"The longer Bulldozer is delayed, the greater the chance that it'll debut into the teeth of new six-core and eight-core Sandy Bridge products". I like that line. But who knows about hardware. The kind of people who keep up with this kind of hardware news also care about quality. Maybe it's not too comparable, but as far as I know, even though the gtx 460 was really late, there's been a great response to it. So if AMD releases a great product, it could still be successful despite stiff competition. |
|
Sackyhack, Here's how I'd answer that. Shanghai/Deneb/Phenom II is a good core. Considering the fact that it's an evolutionary step forward from Phenom, it's an excellent core--it solved the heat and power consumption problems that plagued Phenom I and it's a much more attractive product. Problem is, it's never been performance-competitive against Intel's Core i3/i5/i7 products. AMD has countered with price cuts, aggressive positioning (ie--pushing six-core chips into enthusiast price ranges), and attempted to build a more attractive motherboard/GPU platform. Of course, it also spun off its foundries in a tacit admission that it couldn't afford to invest the necessary capital to stay competitive with Intel (we'll leave the reason why that's so for another time.) In the situation you describe, NVIDIA started out as the clear top dog. Its pockets are deeper, its present in more markets, and its the leader in high-margin areas. NV may have taken a beating over the past 18 months, but it's survived just fine. AMD's position is more precarious. Don't misunderstand me--if Bobcat hits its targets, AMD should be alright (relatively speaking), but Bulldozer needs a strong debut. Ever since it became clear that Phenom I wouldn't match Core 2 (and, later, that Phenom II wouldn't match Core i7), the mantra has been "Wait for Bulldozer." If Bulldozer doesn't improve things with a combination of a 32nm die shrink and a brand-new architecture, AMD will have a hard time explaining when things *will* improve. |
|
"Ontario (codename: Bobcat) " What? Wait. Both are codenames. Bobcat is the CPU core, Ontario is the APU containing two Bobcat cores and one GPU. "AMD is billing Ontario as a "sub one-watt capable core"" To be corrected. |
|
Arte, Fixed that. Thanks for pointing it out. |
|
I think that the 32nm die will help a lot. Even if Bulldozer is not comparable to Sandy Bridge, It will still be a large step up, even from the current six cores. I think that 8-core bulldozer looks pretty BA, and also like to hear it will drop into current AM3 sockets. If i'm not mistaken, Sandy Bridge will be moving to a new socket, making upgrades much more expensive for Intel users. |
|
Bulldozer may yet churn up some new ground if it's released at a really attractive price point. Yep, as usual, Intel's offering are faster and that's great if you can afford to buy them. But AMD's proclivity to talk to us 'cheapskates' through our wallets amplifies their points quite well. I can justify spending less to get good performance, but I can never justify spending more than what I have. |
|
Bobcat looks very promising. Although they haven't disclosed what kind of GPU will go with it yet, this could very well be their ticket to profiting next year. |
|
Don't forget AMD's bulldozer will only run on am3+ motherboards, will support improved turbo-boost, rumored to have better hyper threading then Intel. |
|
*taps on microphone* Pardon me folks, this on? Ok good. Just to make sure this is clear,
BULLDOZER AND BOBCAT DON'T USE HYPER-THREADING. |
|
Don't just leave 'em hanging Joel ... "Hyperthreading" is Intel's trademarked brand of its version of simultaneous multithreading. No chips but Intel's will have "hyperthreading". Bulldozer will do a type of cluster based mutithreading, but as far as I know they haven't coined a marketing name for it. |
|
Wasp, They haven't. In fact, they *don't* refer to it as CMT (I've always understood that to mean "Chip Multithreading," but as a "third way." Whether or not it's actually a third way is debatable. I'm reasonably certain that Hyper-Threading isn't just Intel's brand-name--it's also their own unique implementation of SMT. In other words, AMD *could* implement Hyper-Threading with the appropriate license (but hasn't) and AMD could implement an SMT solution that doesn't use Hyper-Threading's approach at all. |