Recently, leaked Intel Partner slides have been floating around the interweb, detailing architectural performance gains moving from the current Sandy Bridge, 2nd Gen Core i series, Intel offering to the upcoming Ivy Bridge, 3rd Gen, processors.
Slides depict an average like-for-like, same clock, same cache, CPU performance gain just shy of 15 per cent. These numbers aren't anything ground breaking, but a fairly significant step forward given that these new figures can be attributed purely to changes in the architecture and improved Turbo Boost technology, admittedly, made largely possible by targeting the design refresh for a much smaller and denser 22nm production, with Turbo Boost likely benefiting from the much improved thermal properties of the new 3D Tri-gate transistor technology being employed.
We do start to wonder, as Turbo Boost becomes a more significant factor in Intel designs, just what effect it may have on benchmarking. As Turbo Boost speed thresholds are raised, the question of performance starts to come down to individual silicon, its thermal and electrical properties, along with the efficiency and ability to maintain such an efficiency, of the attached cooling solution. This is perhaps something to keep a closer eye on in future benchmarking of processors implementing the much more dynamic Intel Turbo Boost 2.0 technology.
What has changed, quite significantly so, is the graphical performance of the on-die GPU, though, don't be misled, the comparison is against the older HD 2000 series of GPU as opposed to the current HD 3000. It's difficult to draw a direct comparison but it's not unreasonable to assume, based on various available benchmarks, that performance of the HD 3000 in these scenarios averages around 30 to 40 per cent greater than that of the HD 2000, making the gains from the current generation HD 3000 to the next generation HD 4000 GPU average around 100 per cent, twice the performance. Not as amazing as the figures try to make out but still pretty amazing.
Of course there's always been and still is plenty of room for improvement in Intel graphics technology, with offerings only becoming notable since the release of the HD series in 2010, still, with the HD 2000 barely holding its own again NVIDIA's ION 2 GPU, targeted at netbooks. The HD 3000 finally broke into the low-end desktop market offering performance in-line with that of low-end, dedicated graphics cards. At twice the performance again, Intel's HD 4000 can look tread comfortably on the toes of current AMD Fusion APU offerings, still relatively low-end, but a market previously occupied only by AMD. We wouldn't expect AMD to go without a fight, however, with the company now focusing on the mobile and low-end segments and with its HD 7000 series architecture looking to be very promising indeed.
Either way, it looks like Ivy Bridge is set to shake-up the market a little; be it treading on the APU toes of AMD or in pushing the performance envelope to its limits once again. Remember that these figures only show like-for-like comparisons, with a large drop from 32nm to 22nm and new transistor technology that claims to be both denser and more power-efficient, Intel should be able to go quite far in providing both improved clock speeds and increased cache sizes.