The Sandy Bridge-E microarchitecture calls for an entirely new socket design, as even the LGA1136 doesn’t provide enough contacts. Instead the new LGA2011 socket will support these latest Intel processors and with it comes the ability to utilize technologies such as quad-channel DDR3 memory.

When compared to previous generation Sandy Bridge processors, the new LGA2011 models have been beefed up. The on-die graphics solution has been dropped and in its place processors such as the flagship model Core i7-3960X boasts a massive 15MB L3 cache, 6-cores, quad-channel DDR3-1600 memory support and will operate with a base clock of 3.3GHz.

Intel will also be releasing a cut down version of the Core i7-3960X called the Core i7-3930K, which will feature a slightly smaller 12MB L3 cache with a base clock of 3.2GHz. In terms of pricing the Core i7-3930K should be over 40% cheaper than the Core i7-3960X, as it has an MSRP of $555 opposed to $990.

Finally there will also be a Core i7-3820, which is a partially unlocked processor with a base clock of 3.6GHz. This model will feature just 4 cores with 8 threads, along with a smaller 10MB L3 cache. When compared to the existing Core i7-2600K, the Core i7-3820 will be 200MHz faster and feature a 2MB larger L3 cache, so it will be interesting to see how those two compare in terms of pricing and performance.

Unfortunately for today’s launch Intel has only provided the Core i7-3960X model, so we will be unable to look at the Core i7-3820 just yet, moreover pricing information has not been released. Instead today we will be looking at what is technically possible from the LGA2011 platform. Even though this is a premium platform, the majority of users that do invest in it will favor the more affordable Core i7-3930K processor.

That said, at $555 US there is no current AMD or Intel processor that will rival it, as the Core i7-980 will be no match at $580 US and the FX-8150 doesn’t even register at $270 US. Before the arrival of Sandy Bridge-E we believed that the Core i7-2600K was the best value high-performance processor, so we will be focusing on how the Core i7-3960X compares.

The memory bandwidth of the Core i7-3960X is surprisingly low given it was tested with quad-channel DDR3-1866 memory. The read throughput reached just 18.1GB/s, while the write performance was limited to 15.1GB/s. This made the Core i7-3960X roughly 9% slower than the Core i5-2500K when measuring read performance and 18% slower for the write performance.

Despite the weaker than expected memory bandwidth performance, we were happy to find that the L2 cache performance of the Core i7-3960X was greater than that or the original Sandy Bridge processors.

The first test from the SPECviewperf v11 benchmark that we looked at was SolidWorks and here you can see that the Core i7-3960X provided a solid performance gain over the Core i7-2600K as well as the AMD FX processors. When compared to the FX-4170, the Core i7-3960X was found to be 15% faster, while it was 37% faster than the Core i7-2600K.

The Core i7-3960X was even more dominant in the SPECviewperf v11 Maya test, delivering 14fps making it 30% faster than the AMD FX-4170 and 41% faster than the Core i7-2600K.

The Core i7-3960X provides the same impressive OpenGL performance in the CINEBENCH R11.5 benchmark when paired with the GeForce GTX 580 as the Sandy Bridge processors. The CPU performance on the other hand was amazing, blowing the older Sandy Bridge processors out of the water. Here the Core i7-3960X was 55% faster than the Core i7-2600K, while it was 75% faster than the AMD FX-8150.

The AMD FX-8150 recently delivered our highest result in the WinRAR test with a throughput of 4287KB/s. The Core i7-3960X was able to beat this with 4473KB/s making it 4% faster, while it was 51% faster when measuring single core performance. When compared to the Core i7-2600K the Core i7-3960X was 23% faster in the multithreading test and 24% faster in the single thread test.

The Core i7-3960X was extremely fast in our Excel 2011 benchmark, taking just 3.74 seconds to complete the task. This made the Core i7-3960X roughly 20% faster than the Core i7-2600K and 41% faster than the AMD FX-8150.

Using our custom WinRAR test we found that the Core i7-3960X took just 92 seconds to compress a 700MB file, while multiple smaller files totaling 400MB’s took 20 seconds. This made the Core i7-3960X slightly faster than the Core i7-2600K and considerably faster than the AMD FX-8150.

The Core i7-3960X was again incredibly fast, this time when testing with Adobe Photoshop CS5. The test took just 15.8 seconds making the Core i7-3960X roughly 20% faster than the Core i7-2600K and 26% faster than the Core i7-965 Extreme Edition.

The Fritz Chess 13 in-game benchmark can utilize all 12 threads and as a result the Core i7-3960X is unbelievably fast, delivering 19131 kilo nodes per second. This made it 45% faster than the Core i7-2600K and 60% faster than the Core i7-965 Extreme Edition.

The Core i7-3960X delivers the highest frame rate we have ever seen when encoding with HandBrake, as it averaged a whopping 158.9fps making it 33% faster than the Core i7-2600K. When compared to the AMD FX-8150 the Core i7-3960X was 40% faster and 51% faster than the old Core i7-965 Extreme Edition.

The x264 HD Benchmark 3.0 results are also amazing as the Core i7-3960X managed to achieve 172.4fps for the first pass test and 53.7fps in the second pass test. This made it 19% faster than the Core i7-2600K for the first pass and 46% faster for the second pass

Finally we have TMPGEnc 4.0 XPress and we were again amazed by the performance possible when using the Core i7-3960X. Here it took just 5 minutes and 32 seconds to complete the conversion, 13% faster than the Core i7-2600K and 34% faster than the AMD FX-8150.

It is worth noting that the Core i7-3960X is so fast here the mechanical hard drive cannot keep up. We went back and performed this test using a 7200-RPM just for interest’s sake and found that the Core i7-3960X was 30 seconds slower as the hard dive had to catch up. The hard drive did not have this kind of impact on the performance of the Core i7-2600K as we were only able to shave a few seconds off when using the SSD, which resulted in the 6 minute and 14 seconds result that you see above.

When it comes to gaming we test using realistic settings, with the exception of anti-aliasing which is disabled. As you can see the Core i7-3960X was slightly slower than the Core i5-2500K and Core i7-2600K, which is not what we were expecting to see. Although this game is primarily GPU dependent we thought that the Core i7-3960X would be able to match the Core i7-2600K. We believe the slightly slower frame rates are caused by the average memory bandwidth performance.

The Core i7-3960X roughly matched the performance of the Sandy Bridge processors when testing with Just Cause 2, though again as you can see this test is primarily GPU bound, as is the case with most games.

Again the Core i7-3960X is only able to match the performance of the older Sandy Bridge processors, this time when testing with The Witcher 2: Assassins of Kings. That said, it was a fraction slower and again we believe this is due to the weaker memory bandwidth performance.

Interestingly the Core i7-3960X was able to provide a little more performance over the Sandy Bridge processors when testing with Crysis 2. Although the margins were minimal we did see a 2fps gain at 1680x1050.

It is safe to say any 32nm processor that features 2.27 billion transistors in a die size a little over 20mm2 is going to be a power hungry sucker and that is exactly what the Core i7-3960X is. Under full load the processor will increase system consumption levels to a little over 250 watts, making it about as power hungry as AMD’s FX-8150.

Furthermore, when left to idle the Intel power saving features only got us down to 104 watts, which is also roughly the same amount of power consumed by the AMD FX-8xxx series. Still, when compared to the LGA1366 processors such as the Core i7-920 and Core i7-965 EE the Core i7-3960X appears quite fuel efficient.

However when compared to the Core i7-2600K the Core i7-3960X does consume a little over 50% more power under load and over 35% more when idle.

Sadly we were limited to 4.4GHz, but even at this frequency there were some pretty decent performance gains to be had.

It is worth mentioning that I also tried overclocking with the Gigabyte G1.Assassin2 but that proved to be even less successful. For whatever reason the board refused to accept the settings we were trying to input, which is a disappointing level of obedience from a military minded product. Despite our best efforts the board would only go to 3.8GHz and again not because of stability reasons, it simply ignored any multiplier above 38x.

Again we find that Dirt 3 is limited by the GPU rather than the CPU and as a result overclocking the Core i7-3960X did almost nothing to improve performance.

Those of you unimpressed by the 172.4fps averaged by the Core i7-3960X in the x264 HD Benchmark 4.0 might be more satisfied with a further 10% performance increase. Now with an average frame rate of 188.9fps, the Core i7-3960X is 30% faster than the Core i7-2600K.

The CINEBENCH R11.5 performance was also greatly improved by the overclock, as the Core i7-3960X was now 12% faster in the OpenGL test and 19% faster in the CPU test.

The content creation claim was only based on the performance seen in Premiere Pro CS5, while the bandwidth was measured in SiSoft Sandra 2001b and the game physics in 3Dmark 11. Our own testing certainly showed that the Core i7-3960X was faster than the Core i7-2600K, though not by quite as much.

Still, when it came to our application and encoding performance tests, the Core i7-3960X was a monster. The Excel and Photoshop gains over the Core i7-2600K were impressive, both programs were around 20% faster. Meanwhile when encoding, impressive gains were also seen and overall the Core i7-3960X was 28% faster than the Core i7-2600K.

Unfortunately while these are impressive gains over an already very fast processor, they are not nearly great enough to justify investing in the LGA2011 platform. Assuming for a moment that the $555 Core i7-3930K delivers almost the same performance as the Core i7-3960X, it still costs 73% more than the Core i7-2600K, and in our application/encoding tests we saw just a 20–30% performance gain.

Still, we are not surprised by this, as the LGA2011 platform was always going to be for extreme users and professionals that require maximum performance regardless of the cost. It was the same situation with the LGA1366 platform when it was first released, the LGA1156 platform made much more sense from a value perspective, especially for gamers.

Furthermore we are not suggesting that the Core i7-3960X or Core i7-3930K are just 20–30% faster than the Core i7-2600K, as in some tests it was much faster. For example in the x264 HD Benchmark 4.0 the Core i7-3960X was 46% faster than the Core i7-2600K for the second pass test, while it was 45% faster in Fritz Chess 13, and 41% faster in the Maya test.

-

-

For gamers there is really very little to see here, the Core i7-3960X for example is no faster than the Core i7-2600K or even the Core i5-2500K processors, and in fact for the most part we found it to be a little slower.

For us the Core i7-3960X delivered as we expected it would, the letdown was the Intel X79 chipset which offers nothing over the Z68. This is going to be a real problem for LGA2011 motherboards, as we expect they will fetch quite a price premium over their LGA1155 counterparts, yet will most likely be equipped with the same features. This is the Achilles' heel of Intel’s flagship platform and it is hard to believe how featureless so many of the LGA2011 motherboards are considering they are designed to be coupled with processors in excess of $500.