No one will argue that Intel Skulltrail platform can set the performance in certain applications to a totally new level thanks to 8 processor cores. However, as our tests showed, there are still very few applications like that. Among them are the ones that can not only split the workload between 8 parallel cores efficiently enough but also do not require high memory subsystem performance. These are first of all final rendering applications as well as some video editing and encoding tasks. As for the majority of typical desktop applications, Intel Skulltrail doesn’t look very attractive there at all, falling behind single-processor systems with a quad-core CPU and unbuffered memory.
I would like to specifically draw your attention to the fact that although Intel Skulltrail is being promoted as an extreme gaming platform, its gaming performance is quite disappointing. Most contemporary games cannot load eight processor cores of the Skulltrail platform with work. And slow memory subsystem using FB-DIMM DDR2 modules doesn’t let it even catch up with the single-processor gaming systems.
In fact, the only advantage of the new Intel Skulltrail platform from the gaming standpoint is compatibility with both: SLI and Crossfire technologies. However, it will hardly satisfy those users who will spend a few thousand dollars on it. Especially, since it supports only one version of SLI: with two graphics cards.
However, it is not only the performance in games and other applications that may shake the attractiveness of Intel Skulltrail platform. Even its unbelievably high price of $1500 per Core 2 Extreme QX9775 CPU and $650 for the mainboard is not that upsetting. We have actually expected all of this. It is much worse that Skulltrail doesn’t make the impression of a solution that has been finalized and ready to go into mass production. Now that we have taken a really close look at this system we tend to believe that we still dealt with an engineering prototype, which is quite far from a mass solution.
And the first thing that pushes us towards this conclusion is the memory subsystem. We believe it is Intel’s major strategic miscalculation that they left the Skulltrail memory subsystem in its “server” representation without changing anything. FB-DIMM DDR2 memory modules not only heat up a lot during work and slow the whole system down, they also cannot be overclocked. The Skulltrail mainboard BIOS offers no options for clocking the memory at frequencies other than what SPD states.
The system’s power appetite is also a cause for concern. Two powerful CPUs and FB-DIMM modules push Skulltrail power consumption far beyond that of a system with a similarly performing CPU and faster memory. Besides, Enhanced Intel SpeedStep technology doesn’t work on Skulltrail, which makes this platform completely uneconomical in idle mode.
Skulltrail has no noise-reducing technologies. The mainboard it uses doesn’t allow adjusting any fan rotation speeds, and the chipset fan howls so badly that you will hardly be able to use this system at home.
All in all, we can probably regard Intel Skulltrail as yet another demonstration of Intel’s technological superiority. Maybe, if AMD didn’t give up their dual-socket FASN8 enthusiast platform, Skulltrail could have become much more user friendly. However, now that there is simply no competition, it looks like Intel engineers didn’t want to invest too much effort in adapting server components for desktop use. As a result, the platform we saw today impresses us mostly with its specifications and looks, but doesn’t prove up to our expectations in action.