So we've all seen Conroe (or Core 2 Duo) numbers from Anandtech as well as other websites. Is AMD in trouble in the short run? What about in the long run?
In the short run (< 0.5 - 1 year), I think there will be several determining factors that decide the tempreture of the water that AMD will be in on the 27th of July. First and foremost is the retail availibility of the Core2 Duo (C2D) processors. So far, most websites feel that Intel will be hard pressed to make C2D processors readily available in retail at launch due to OEMs buying up stock in advance. A few have reported otherwise. The (unconfirmed) postponement of the launch date to the 27th from the 22nd suggests that Intel won't be paper launching the C2D in retail. You don't postpone by 5 days if you can't make your product available easily - you postpone by a longer period of time. The 5 day postponement suggests delivary issues rather than product related issues. If I'm proven wrong, and the 27th turns out to be a paper launch, then AMD will no doubt be in less trouble than they would be otherwise.
The price cuts announced by AMD as well as the rumoured 256kb X2 processors will help sales as well. The K8 architecture is not that cache hungry, atleast when you compare the 512kb models to the 1mb models. So it is very likely that the 256kb part won't be a performance loser. The lower cost of producing 256kb parts should help in that case.
There's also the fact that the 939 platform is very mature with a plethora of boards available with different features. This means that the C2D platform would be competing against the AM2 platform (relatively immature) as well as the 939 platform.
Therefore it seems to me that in the short run, while it is likely that AMD will be swept off it's feet by C2D, there is a definate possibilty that X2 sales won't be hit that hard. In the single core market, there is no C2D processor, so AMD should continue to do well there.
In the long run (> 0.5 - 1 yr), the C2D doesn't sound as daunting to me as the performance numbers would suggest. The K8 architecture itself is sound. The fact that the C2D architecture resembles the K8 architecture (atleast to me from what I've read at Anandtech and X-Bit Labs) reinforces that belief. Improvements to the K8 architecture such as increasing the number of execution units (have 4-5 complex units or 3 complex + 2 simple), widening data paths etc.. should allow it to superscede the C2D architecture. This may take a year or so to implement at worst. Naive as I am (so please feel free to correct me), what needs to be done to the K8 seems akin to what ATI did to the R520 to turn it into the R580 i.e. push numbers up and increase the performance of each unit slightly. This, I feel, is because while the C2D architecture may be revolutionary for Intel, comming after Netburst, it is more evolutionary for AMD and it is the direction that I'd have expected them to take to counter Tejas, had that Netburst design made it to the NewEgg warehouse.
In conclusion, I feel that AMD has a little to be worried about in the short term, but it is sound in the long term. This is the time of the spring of Intel's architecture, wheras it's mid autumn for AMD, meaning that AMD will see a brand new spanking architecture before Intel will (sort of like when AMD's K8 dethroned Netburst after a year or so of mediocre K7 performance). I am not an AMD fanboy, in fact, I don't even think it's possible to favor either Intel over AMD or the other way around because of the length of the product cycle. Let me know what you guys think.
In the short run (< 0.5 - 1 year), I think there will be several determining factors that decide the tempreture of the water that AMD will be in on the 27th of July. First and foremost is the retail availibility of the Core2 Duo (C2D) processors. So far, most websites feel that Intel will be hard pressed to make C2D processors readily available in retail at launch due to OEMs buying up stock in advance. A few have reported otherwise. The (unconfirmed) postponement of the launch date to the 27th from the 22nd suggests that Intel won't be paper launching the C2D in retail. You don't postpone by 5 days if you can't make your product available easily - you postpone by a longer period of time. The 5 day postponement suggests delivary issues rather than product related issues. If I'm proven wrong, and the 27th turns out to be a paper launch, then AMD will no doubt be in less trouble than they would be otherwise.
The price cuts announced by AMD as well as the rumoured 256kb X2 processors will help sales as well. The K8 architecture is not that cache hungry, atleast when you compare the 512kb models to the 1mb models. So it is very likely that the 256kb part won't be a performance loser. The lower cost of producing 256kb parts should help in that case.
There's also the fact that the 939 platform is very mature with a plethora of boards available with different features. This means that the C2D platform would be competing against the AM2 platform (relatively immature) as well as the 939 platform.
Therefore it seems to me that in the short run, while it is likely that AMD will be swept off it's feet by C2D, there is a definate possibilty that X2 sales won't be hit that hard. In the single core market, there is no C2D processor, so AMD should continue to do well there.
In the long run (> 0.5 - 1 yr), the C2D doesn't sound as daunting to me as the performance numbers would suggest. The K8 architecture itself is sound. The fact that the C2D architecture resembles the K8 architecture (atleast to me from what I've read at Anandtech and X-Bit Labs) reinforces that belief. Improvements to the K8 architecture such as increasing the number of execution units (have 4-5 complex units or 3 complex + 2 simple), widening data paths etc.. should allow it to superscede the C2D architecture. This may take a year or so to implement at worst. Naive as I am (so please feel free to correct me), what needs to be done to the K8 seems akin to what ATI did to the R520 to turn it into the R580 i.e. push numbers up and increase the performance of each unit slightly. This, I feel, is because while the C2D architecture may be revolutionary for Intel, comming after Netburst, it is more evolutionary for AMD and it is the direction that I'd have expected them to take to counter Tejas, had that Netburst design made it to the NewEgg warehouse.
In conclusion, I feel that AMD has a little to be worried about in the short term, but it is sound in the long term. This is the time of the spring of Intel's architecture, wheras it's mid autumn for AMD, meaning that AMD will see a brand new spanking architecture before Intel will (sort of like when AMD's K8 dethroned Netburst after a year or so of mediocre K7 performance). I am not an AMD fanboy, in fact, I don't even think it's possible to favor either Intel over AMD or the other way around because of the length of the product cycle. Let me know what you guys think.
.
Facebook
Twitter