I thought it would be fun to make a thread of predictions for 2008 in the CPU market, with the intention of ultimately bumping this thread 12 months down the track to see just how accurate (or not!) we have been.
Intel -
*Will continue to turn in impressive financial results, will take full advantage of AMD's situation with K10, I expect slight growth in profit margins as 45nm ramps.
*Penryn will bring average overclocks into the ~4GHz range for quads, ~4.5GHz for duallies. Will be the enthusiast chip of choice until Nehalem arrives.
*Will introduce a sub $200 quad sometime in Q2 to compete with AMD in price/performance.
*Will introduce a new budget/mainstream lineup in Q3 to replace the current E21x0/E4x00 series.
*Nehalem to bring substantial improvements (~20% IPC average, ~50% in certain bandwith benchmarks) over the Core 2 architecture, but will fall short of being the 'huge leap ahead' (greater than Netburst -> Core 2) that Intel claims.
*Nehalem will be released in the same timeframe (late Q4) as the 45nm K10s to steal the limelight of the launch.
AMD -
*Struggle for profitability will continue. I don't expect them to turn a profit in any quarter this year, though they should stem the bleeding and come close to breaking even by Q4.
*B3 stepping will bring clocks up to 2.8GHz by Q3, but will ultimately fall short of 3.0GHz due to thermal limitations.
*K10 tri-cores to be introduced late Q2. Will end up competing well with high end (3.0GHz+) Wolfdale dual cores in multithreaded performance. Will be priced at around $150.
*K10 dual cores to come in Q3. Will come in mainstream prices ($50 - $150) to compete with Intel's new mainstream 45nm dual cores.
*45nm K10 will be released in late Q4, finally allowing AMD to break the 3.0GHz barrier with K10.
*IPC improvements will bring 45nm K10 close to current Kentsfield levels, but ultimately fall short of Penryn and especially Nehalem.
----------------------------------------------------------------------------------------------------------
In summary, I think Intel will be business as usual in 2008. Their dominant position over AMD will continue. Penryn will prove a success (not hard to guess that one) but the big news is of course Nehalem, which will usher in a new era for Intel post FSB. It may well prove a decisive blow for AMD, as it finally brings Intel on level terms in memory bandwith and scalability, AMD's main technological advantage over Intel to this point.
AMD's outlook for 2008 looks rather dire. There is just nothing on their roadmap that suggests they will be challenging Intel in any way. Like times past, they will inevitably continue to compete with Intel in the mainstream markets, but the high end is out of reach. B3 will be too little too late to really change the competitive landscape. 45nm K10s will help narrow the gap with Penryn, but will be outclassed by Nehalem. 2008 will be a case of 'A day late and a dollar short' for AMD. Let's hope they step up a gear in 2009, for the sake of competition in the CPU market, and their very own survival of course.
Intel -
*Will continue to turn in impressive financial results, will take full advantage of AMD's situation with K10, I expect slight growth in profit margins as 45nm ramps.
*Penryn will bring average overclocks into the ~4GHz range for quads, ~4.5GHz for duallies. Will be the enthusiast chip of choice until Nehalem arrives.
*Will introduce a sub $200 quad sometime in Q2 to compete with AMD in price/performance.
*Will introduce a new budget/mainstream lineup in Q3 to replace the current E21x0/E4x00 series.
*Nehalem to bring substantial improvements (~20% IPC average, ~50% in certain bandwith benchmarks) over the Core 2 architecture, but will fall short of being the 'huge leap ahead' (greater than Netburst -> Core 2) that Intel claims.
*Nehalem will be released in the same timeframe (late Q4) as the 45nm K10s to steal the limelight of the launch.
AMD -
*Struggle for profitability will continue. I don't expect them to turn a profit in any quarter this year, though they should stem the bleeding and come close to breaking even by Q4.
*B3 stepping will bring clocks up to 2.8GHz by Q3, but will ultimately fall short of 3.0GHz due to thermal limitations.
*K10 tri-cores to be introduced late Q2. Will end up competing well with high end (3.0GHz+) Wolfdale dual cores in multithreaded performance. Will be priced at around $150.
*K10 dual cores to come in Q3. Will come in mainstream prices ($50 - $150) to compete with Intel's new mainstream 45nm dual cores.
*45nm K10 will be released in late Q4, finally allowing AMD to break the 3.0GHz barrier with K10.
*IPC improvements will bring 45nm K10 close to current Kentsfield levels, but ultimately fall short of Penryn and especially Nehalem.
----------------------------------------------------------------------------------------------------------
In summary, I think Intel will be business as usual in 2008. Their dominant position over AMD will continue. Penryn will prove a success (not hard to guess that one) but the big news is of course Nehalem, which will usher in a new era for Intel post FSB. It may well prove a decisive blow for AMD, as it finally brings Intel on level terms in memory bandwith and scalability, AMD's main technological advantage over Intel to this point.
AMD's outlook for 2008 looks rather dire. There is just nothing on their roadmap that suggests they will be challenging Intel in any way. Like times past, they will inevitably continue to compete with Intel in the mainstream markets, but the high end is out of reach. B3 will be too little too late to really change the competitive landscape. 45nm K10s will help narrow the gap with Penryn, but will be outclassed by Nehalem. 2008 will be a case of 'A day late and a dollar short' for AMD. Let's hope they step up a gear in 2009, for the sake of competition in the CPU market, and their very own survival of course.