AMD Nods at Shorter Design Cycles, More Synthesized Designs
by Anand Lal Shimpi on February 2, 2012 1:32 PM EST- Posted in
- CPUs
- AMD
- Trade Shows
- AMD FAD 2012
The presentations we've seen from AMD thus far today haven't been very specific in terms of numbers, but there's a lot of reading between the lines that we can do here. AMD already mentioned that it's no longer interested in pursuing bleeding edge process technologies, but its CTO expanded on that idea by mentioning that AMD will transition to shorter design cycles. We got a hint of this transition with the annual Brazos/Llano cadence, but now it's official AMD policy to have shorter design cycles. These shorter design cycles will leverage lower amounts of custom block design and lean more on easily synthesizable architectures. The tradeoff is obviously performance but you do get better time to market. As was the case with Brazos however, if you can bring the right combination of technologies to market at the right time, the tradeoff is worth it.
Facebook
Twitter
RSS
7 Comments
View All Comments
frozentundra123456 - Thursday, February 2, 2012 - link
In other words, they can't compete in the high end and have to lower the bar to get stuff out on time.umbrel - Thursday, February 2, 2012 - link
more like, competing in the high end is too stressing so they will go at an easir but actually faster pace. Missing release dates hurts much more than not having the best process tech.They already have low margins, not pushing for the latest process and keeping the architecture as modular as posible might actually increase their margins. And by meeting the right dates (mathcing refresh cycles and big sale days) they should be able to increase market share.
jaydee - Thursday, February 2, 2012 - link
As opposed to chasing a shrinking market that they are already far behind in... Do you disagree with this strategy? You do realize that there is far more volume in the mid-low end CPU's than the high-end, right?fic2 - Thursday, February 2, 2012 - link
If they actually start releasing on schedule instead of delay after delay and then release a "who cares" product it will probably be a lot better for everyone. The Brazos example given is a good example of what to do. Bulldozer is, of course, an example of what not to do. AMD would have been better off just doing a die shrink of Phenom.frozentundra123456 - Thursday, February 2, 2012 - link
There may be more volume, but the margins are paper thin. And no, I dont necessarily disagree with this strategy.More to the point, I just question how effective and timely they can be when switching to a new strategy when they have not even been able to compete in what was once their area of expertise.
Yes, they may not have to compete directly aginst Intel, but they will be competing against an established ARM landscape, and intel is also trying to get into the field.
Beenthere - Thursday, February 2, 2012 - link
All the great ideas in the world are worthless if GloFo and TSMC can't deliver chips. AMD needs to fix design/engineering issues within AMD to get better performing CPU/APU/GPU products to market while helping/Forcing GloFo and TSMC to get their production issues sorted out NOW and to stay ahead of the process production curve as this is killing AMD sales and reputation.fic2 - Thursday, February 2, 2012 - link
My understanding was that TSMC was ahead in their manufacturing which allowed Brazos to come out earlier than expected - something close to a quarter early. GloFo is another story, but I don't think that AMD helped much there with the Bulldozer design and bugs.