INTEL - X79
the KING IS HERE! (Maybe!)
T
he Intel X79 (Patsburg) platform
and 3000 series of CPUs have
finally arrived. A better-kept
secret than the original Sandy Bridge
launch, we finally have an enthusiast
platform to replace the very long
lasting X58 which has been with us
for a whole 3 years. While Tylersburg
brought with it something completely
new and unprecedented IPC in many
of our favourite benchmarks, this
time we are looking at an evolution
of what we already had for almost a
year. It’s essentially the same CPU if
you will, just more of it and that isn’t
necessarily a bad thing. (Yes like a
King is a Prince, just more of him
yes? Er... No!)
As always, we are only interested
in the fastest of the recently released
CPUs. After all this is an expensive
platform and we are thinking the ones
who will invest in it are the ones who
most likely skipped LGA 1155 and
will move from their X58 systems to
X79 machines or rather add it to their
benchmark platforms(for gamers, not
necessarily overclockers).
A quick rundown of the new CPU/
Platform has us where X58 collides
with P/Z67 to make X79. Essentially
better than both on paper and largely
in practice as well. For your hardearned dollars, the CPU under
investigation here will offer you a new
Turbo Boost version (2.0), 40 lanes of
PCI-Express 3.0 connectivity, 15MB of
L3 cache, a much improved or at least
far more flexible four channel IMC,
AVX, AES, SSE 4.1 and 4.2 instruction
sets, native SATA 6Gbps support and
a new socket. We have had 12 threads
before on the 980 and 990X CPUs
so those aren’t new. We’ve also had
the use of the new instruction sets
via the LGA 1155 socket CPUs, but
for the first time these are married
10 The OverClocker Issue 17 | 2011
into the X79 platform to bring us in
essence the best of both. Let’s keep
in mind however that in some ways
the platform and CPUs are inheriting
the strange properties of the original
Sandy Bridge CPUs as well.
Before we get to the numbers,
we will briefly detail our experience
with the platform and in particular
the CPU. Great news for all is that
as far as overclocking goes it isn’t
necessarily worse than any 2600K
out there. Much like in the beginning
though, finding a sample capable of
5.6Ghz let alone 5.7GHz will prove
excruciatingly difficult. While this
wasn’t as big a problem for the
LGA1155 CPUs, we are now dealing
with a CPU that is at present just
short of being three times the price
so naturally the binning process is
significantly more expensive. Our
particular CPU is only capable of
5.1GHz at the most, regardless of
cooling or tweaking. We would have
previously suspected better results
possible on a more proficient board,
but the hard wall we experienced
leaves us with very little hope as to
how a different board could alleviate
this limitation by any meaningful
measure.
Heat will be a problem as well as
one would expect, the CPU is much
warmer than the 2600K at load, but
given the very high Turbo frequency,
this isn’t much at all. In fact whatever
cooling mechanism you had for the
2600K would easily suffice for the
3960X, barring the Intel reference
cooler. The new Asetek designed LCS
in particular is good, able to keep
this CPU functioning even at a fiery
4.8GHz with six threads. %?e???????)??????????1
L?5???????????)???`????????????????????????)??????
AU??????????????QA?((0