
We give you a one page article written with the benchmark monkey in mind! (We have a few around here too!). Core i7 965EE, 940, and 920 paced through four synthetic benchmarks and compared to Core 2 QX9770 and E8500. And we even throw in a little Prescott too, just because it was cold in the office.
All of our readers know that synthetic benchmarks have fallen out of favor with HardOCP over the years because many times those utilities do not point out the true abilities when it comes to the software you will be using to leverage your new hardware. On the other hand though, synthetic benchmarks are surely good tools when it comes to overclocking and tweaking your own setups and can be a great ruler when it comes down to finding out who has the biggest e-pen…, I mean faster system setup.
We still use synthetic benchmarks around HardOCP on a daily basis because there is no quicker way of making sure your hardware is operating as it should be. And on the other hand, there is also no faster way of seeing what sort of real world performance advantages might be held within a new Core i7 processor, as we have in this case. But, don’t for a second think that just because you see a 25% synthetic performance increase that your gaming is going to get that much faster as well.
If you want to know more about the Core i7 / Nehalem processors themselves, we suggest you lend your ear to Intel’s Nehalem designer.
That all said, here are four of our all-time favorite CPU benchmarks.
Sandra's Drystone has been a standby since the late ‘90s. It is quite simple, as are most of the benchmarks here, as it does arithmetic on the cores and measures the workload in Millions of Instructions Per Second, or MIPS.
The i7 965EE surprised even myself in this benchmark. There is a huge “WOW” factor here! Comparing our 965EE to the QX9770, which match up clock for clock, we see the 965EE post an increase of around 50% in this synthetic. When looking at our Core 2 3.16 scores, keep in mind that it is a dual core processor which is clearly shown by the Core 2 3.2GHz processor nearly perfectly doubling its score with the extra two cores.
This Sandra benchmark takes square aim at the processors memory system that it relies on. Our Nehalem configuration is supported by a new on-die memory controller as well as triple channel DDR3. The integrated memory controller will tremendously cut down on latencies of the CPU having to go out to the northbridge, then to attached RAM and back. Instead, our CPU now stays on-die and goes out directly to RAM for memory calls. We also have an additional channel of DDR3 to rely upon compared to the dual channel DDR3 we are used to.
Again, I was surprised. We see the Core i7 965EE processors memory while running at 1600MHz (stock is 1333MHz) over triple the Core 2 processors supporting memory at the same speed and timings but being dual channel rather than triple.
While omitted from the chart our Pentium 4 2.8GHz Prescott with 800MHz dual channel DDR2 scores a whopping 4464MB/s.
We use Hiper Pi around here for single threaded testing that uses “Super Pi” but gets around some of the Vista incompatibilities that some of you may have seen. Hiper Pi can reach into more than one cores but we have used it for our standard single core arithmetic test that is generally used in the enthusiast hardware community.
While not as dramatic as the Sandra results above, clock-for-clock the i7-965EE is outpacing our QX9770 by around 15%. Our Pentium 4 2.8GHz took 47+ seconds…
wPrime has become a favorite of ours in the last year or so. It is the perfect synthetic multi-threaded benchmark that we have found. Version 2.00 brings with it a nice face lift as well as some nice features. Reaching into all 4 physical cores and 4 HyperThreading “cores,” the i7 965 EE pulls in around 22% faster than the QX9770 clock-for-clock. Again we see our dual core 3.16GHz processor pull up the back of the pack. Our Pentium 4 2.8GHz took 91+ seconds.
Any way you slice it, the new Intel Nehalem desktop processor looks like a winner in the land of the benchmark. Even when the benchmarks are not done by Intel PR engineers!
And incredibly, all of this was done inside power envelopes that we are already accustomed to. 130w TDP power envelopes are overstated. Even with our i7-965EE showing to be fully 100% loaded we never saw full system power pull over 280w at the wall, and this was with a GeForce GTX 280 running idle.