From a performance standpoint, the P9X79 WS doesn't have much that we would have expected to significantly differentiate itself from our first look at the P9X79 family apart from the extra multi-GPU bandwidth- the core components hold mostly the same with some extra tuning and validation to ensure reliability. We found that the P9X79 WS's results laid almost exactly over the top of the P9X79 Deluxe at +/- ~1% average deviations from the results shown below. It was actually a surprisingly-consistend result, literally straddling (some better, some worse) the performance of the P9X79 Deluxe by a tiny (read: statistically insignificant) margin across the board. NIC tests were performed at a high level to confirm low CPU usage, and when paired with a high-end router, provided great Gigabit throughput.
For the storage tests, we ran our (bare) Secondary SSD, a Patriot Pyro SE 60GB through three separate benchmarks, AS SSD, ATTO, and CrystalDiskMark. For the USB tests, CrystalDiskMark was used to measure the transfer speed of our Seagate GoFlex Desktop 2.0TB USB 3.0 hard drive.
We see a roughly 4% advantage in 6GB/s testing from the Gigabyte X79-UD3, and nearly identical performance in 3GB/s speeds. In terms of absolute values, and the rough error margin of our tests, these results would seldomly be recognized in normal usage, and even in the most intense applications would be difficult to tell apart. We see the Deluxe version perform slightly better across the board, but nearly identically to the P9X79. The additional Marvell SATA ports on the P9X79 Deluxe do quite well in read, but is the slowest in write speeds, which is probably why these are ideally labelled as the "SSD Caching" ports, being read-optimized.
We also see a slight advantage for the Deluxe over the base P9X79, but the Gigabyte X79-UD3 bests them both with an 11% write and a 10.3% read advantage. The speeds with USB 3.0 Boost slightly edge those of Gigabyte's "boosted" USB 3.0 speeds, but for all practical purposes, they're essentially the same.
Our computation benchmarks are meant less to be a comparative test, and more of a validation test that nothing strange is going on. We don't expect any non-trivial differences in these tests, and it's only when there are differences that we become alarmed.
We see nearly identical performance between all three boards in the computational benchmarks, which is as expected, and is a validation that each board does not have any gaping flaws or computational disruptions. We do see a larger overall performance boost during overclocking for the P9X79 boards, but, again, we see very slim differences which we could not call appreciable or statistically significant enough to draw a particular advantage toward any of the boards.
We do see a roughly 6.7% advantage in the P9X79 Deluxe over the X79-UD3, with an edge of 3.8% over the P9X79 in copy speeds. Read and write speeds are nearly identical for both the P9X79 boards, and carry a 3.3% advantage over the X79-UD3.
ASUS has fully-included all overclocking options in their AISuiteII as well as in the BIOS. This includes their advanced DIGI+ power control, phase control, flexible fan speed profiles, memory options, as well as the standard overclocking settings. Using manual settings, we were able to push the our i7-3930K up to just over 4.5GHz at an imposed voltage limit of 1.45V, which we used as a baseline for the AutoTune settings found from the TPU switch, the OCTuner in the BIOS, and the AutoTune settings in AISuiteII. We should also note that the overclocking behavior was nearly identical between the P9X79, the P9X79 Deluxe, and the P9X79-E WS, which verifies ASUS' claim about the base specs being fairly equal and was also expected because it appears that our RAM was holding us back. All in all, the Deluxe model has more overall low-level "tweakability," but unless you're an enthusiast overclocker, I can't say with certainty you'll notice a big difference.
The Digi+ power control is one of ASUS' proudest innovations, but the average user won't understand how to use it. But for the power users out there, it helps open up another world of possibilities for tweaking and optimization. Digi+ really deserves a dedicated article to do it all justice, so we will point you to ASUS' site for further details. For X79, ASUS has extended the capability of their 8th-generation power control design to include control over the CPU, VRAM, and also DRAM. Again, the power controls are available in both AISuite II and the BIOS, for convenience and flexibility.
The "Fast" setting in AISuite, the TPU Switch, the "Performance" Quick Profile in the BIOS, and the BIOS OCTuner all lead to the same end overclocking result. Each of these settings automatically set a clock speed, perform stability tests, and iterate until they settle on a final clock. In each of our tests, these settings all lead to a 3.9GHz clock speed, which is a nice boost, but also very safe, even with rather poor cooling equipment. The beauty of these settings is that they're dynamic, so if you're running the stock cooler and its clogged with dust, if the stability tests see the chip is running hot, it'll choose a lower clock speed. These settings usually only took a minute to converge on a setting, and it was also rather aggressive on the base clock speed, bumping it up to 103.00 in all cases.
The "Extreme" auto-overclock option is found in AISuiteII, and performs the same iterative process as described above, with a clock setting>stability test>repeat until it settles on an aggressive but safe overclock. In general, the iterative process likes to boost the base clock to 103.00 right off the bat, and then cycling the multiplier. We got a very aggressive 4.532 GHz, but it did take it up to 1.456V, which is actually only slightly higher than our pseudo artificial voltage limit we imposed to avoid damaging our expensive chip. The process worked flawlessly, and it really is great even for extreme overclockers who plan on pushing it hard themselves, as the AutoTune will iterate and at least give you a good starting point. The peak core temperature under load was at 88°C, which is a bit high for long-term usage, but overall we got a great result. When the dust has settled, you're rewarded with the image shown below: