Intel Z68 Sandy Bridge versus Z77 Ivy Bridge chipset comparison

Posted Dec 6 2011 in
  • CPU
  • sandybridgeversusivybridge1

    Z68 Sandy Bridge, 1Q11
    X79 Sandy Bridge-E, 4Q11 (costly CPUs for enthusiasts)
    Z77 Ivy Bridge, 2Q12

    Two of the better very recent articles on the new Z77 Ivy Bridge chipset:
    Intel's Ivy Bridge vs. Sandy Bridge Benchmarks Leaked
    Intel’s Next Processors Detail Leak: Ivy Bridge Benchmarks, Specs, and Release Date

    Here's the net of the differences, of course, this is just my personal opinion, based on preliminary, public information. I have somewhat unique requirements, with a focus on efficiency and virtualization features for affordable lab use. I admittedly am ignoring the Xeon family of server class CPUs at the moment, which are generally more expensive, but are also more likely have more DIMM and CPU sockets for more expandability, and be more appropriate for production workload.

    Z68 versus Z77:

    The good:

    • Core i7 2600 on Z68 versus equivalent Core i7 3770 on Z77 gives roughly a 15% boost in performance
    • Drop from 95 watts to 77 watts is a nice, appreciated move in the right direction, especially always-on systems, like home servers
    • Building in the USB 3.0 chipset, a good thing, finally, standardized performance expectations and drivers (I've found ASMedia always outperform NEC, for example)

    The bad:

    • Much like the Z68 Intel SATA ports can't be set to passthrough to VMs on the Z68, it's seems quite possible that Ivy Bridge also won't allow passthrough of just some USB 3.0 ports to work
    • Performance boost of ~15% is not enough to make me regret going with Sandy Bridge (Z68) chipsets and the 2600k/2600 for 2 of my system upgrades earlier this year:  gZilla for gaming, vZilla for virtualization/backup system
    • Despite some X79 boards having 8 DIMMs, Ivy Bridge might only have 4 DIMM slots, seemingly confirmed here
    • It's not out until 2Q2012

    *There's a great summary regarding 1155/2011 CPU sockets and compatibility here, which appears to be accurate