Some Windows benchmark comparisons between the two would be interesting. You can install new drivers straight from Nvidia (this took a little inf adjustment last time I tried it, not sure if that's still true) and see what the chip is capable of using their unmodified drivers.
You may also be able to make some adjustments with Kepler "over-clocking" tools (EVGA Precision, MSI Afterburner). The GPU should still self-throttle for thermal protection, so these tools won't actually harm the chip.
It's not known, but there's speculation based on Intel's demonstrated Broadwell power savings: They showed a Broadwell chip using only 70% of the power of a Haswell chip at exactly the same processing speed while at 100% load. They're talking about it in these terms because the current focus is all about mobility, not speed.
The first thing you might ask when you see this is, "what if I don't care about that 30% reduction in power consumption?" The answer is that you could produce a faster chip at the same TDP (how much faster is currently unknown). For example, with a flat 30% power reduction (which is what they showed) a 65W TDP chip becomes a 45W chip (which is 15" rMBP territory). Intel's 4770R is the 65W TDP equivalent of the 4770K, and that chip could run in the rMBP if you built it with Broadwell's process.
How much faster is a 4770R than the chips in this year's rMBP? I couldn't find any benchmarks actually, since it's a very odd low TDP desktop chip, but if anyone can find a comparison we might have a good idea of how much performance gain we could be looking at.
Alternatively you could stick with the already impressive performance of the 2.6 15" rMBP (3338 Geekbench 3 single core) but get 30% more battery life across the board (even at 100% load on all 4 cores). Combine that with IGZO and people without regular access to outlets might find this very exciting.
Last edited:
