Falcon Northwest Mach V


Falcon Northwest Mach V

It’s no secret that we’ve had nothing but headaches with overclocked quad-core Intel systems this year. The cause of the problems—be it heat, over-overclocking, or other—doesn’t really matter. Frankly, we don’t care. These systems are being sold to consumers who don’t want to know the shape of the piston heads in their engines—they just want to be slapped back into the seat when they step on the gas.

Which is what the Falcon Northwest Mach V does so well. But then, what else would you expect from a quad-core CPU running at an amazing 3.73GHz? With four cores, you’ve got roughly 15,000MHz under the hood. This unbelievable overclock isn’t all Falcon’s doing, the Mach V sports Intel’s new limited-edition quad Core 2 Extreme QX6800, which ups the ante to 2.93GHz.

Would the Mach V explode like so many other systems we’ve seen recently? When we fired up our benchmarks to find out, we had just one issue. In our Nero Recode 2.0 test, the machine hard-locked during the import process. This could have been a result of shipping issues. The rig was jostled enough in transport that we had to jiggle the SATA cables to get it to initially boot. We wondered if the hard landing was responsible for the hard-lock since the machine completed the test flawlessly on subsequent runs. We discovered no further stability problems with the rest of our standard benchmarks.

As you might expect, performance was quite amazing with a quad-core ticking along at 3.73GHz. When we pulled up our benchmark spreadsheet and entered the Mach V’s numbers, we discovered the Mach V set new records in Premiere Pro, Photoshop CS2, Recode 2, and Quake 4. And the rigs the Mach V beat aren’t a sad-sack collection of Pentium 4s and Athlon 64s, mind you—this system bested a collection of quad-core Core 2s equipped with dual GeForce 8800 GTX cards. The only benchmarks it didn’t clean house in were FEAR (the ABS 3.47GHz quad holds that record) and SYSmark2004 SE, which didn’t even run. Normally, we’d blame the overclocking, but this old benchmark has become so flaky with modern hardware that we can’t hold PC vendors to blame for its foibles.

What is impressive is that the Mach V’s hardware assortment is mostly the same as the other systems’. It uses an EVGA nForce 680i board, two 150GB WD Raptors, a 750GB Seagate Barracuda drive, a pair of GeForce 8800 GTX cards in SLI, and an X-Fi Fatal1ty card. It doesn’t use all the same parts though. Falcon upped the RAM to a curious 4GB and sidesteps the thorny “Vista drivers suck” issue by dual-booting XP Pro and Vista Ultimate. Of course, we’ve already noted that the Falcon has the new 2.93GHz quad core, which is so rare now that buying it from a PC vendor may be the only way to get it for a while.

Normally, with four record-setting benchmark scores, this would be the end of a happy story, but the stability issues with recent OC’d quads have us worried enough that we’re running additional tests on all overclocked quad machines. The first is a real-world encoding test using ProShow Gold 3.0, which pegs all four cores during an encode. On one of the flakier quad cores we’ve seen, the encoder crashed the machine within minutes, yet the Mach V made it through the test with no issues.

Our second test is a bit more controversial. Another PC vendor approached us with general concerns about overclocking their quad-core rigs and provided us with a script they developed internally to test their systems. The script launches four instances of the Prime95 burn-in test, which maxes out a CPU core by searching for Mersenne prime numbers. We’re normally reticent to run vendor-created tests on machines we review, but our curiosity about quad-core stability trumped our other concerns. With four sessions of Prime95 going, the Mach V was rebooting spontaneously within 10 minutes.

That put us in a tough position. The Mach V aced all of our normal benchmarks, plus our multithreaded ProShow test—almost all of which use real-world workloads. Prime95 uses real math, but is the workload realistic for normal users? Probably not. All our standard tests show that the Mach V is a stable machine under normal operating loads but can be brought down under certain circumstances. This is better than the other quad cores we tested, which crashed running our normal benchmarks. Still, it’s not ideal.

And then there’s the price. At $9,900, the Mach V is one of the most expensive machines we’ve ever reviewed—it’s more expensive than some of our Dream Machines. However, it outperforms all of the overclocked quad-core boxes we’ve reviewed to date in most benchmarks, and it’s more stable. We are concerned that system vendors are pushing clock speeds too far. And, as fast as this rig is, its price makes it a difficult pill to swallow for all but the wealthiest enthusiast.


Fastest PC we’ve ever tested.


High-pitched component
whine under certain loads.




+ Add a Comment


I am not a super guru in computers but I like a pretty fast PC. I bought an AMD X4 640 recently and it does wonders for my applications. I also put a 5770 radeon in my PC and I don't have gaming problems either. The only problem's I've been having is getting games running into Linux because of the exe file extension which is not recognized in Linux.



The Falcon Northwest Mach-V brought up a controversy of what constitutes a reasonable application/program to test stability of overclocked rigs. The vendor you got a modified Prime95 program is only using one tool most overclockers use, of which I use Super-Pi (light test), 3DMark05 and 06 (heavy test), and then Orthos (super stress test)... if all these pass, you're good to go. 'Real world' is subjective instead of objective, which is 'this rig is 100% stable no matter WHAT you throw at it.' I for one think you should make Orthos and 3DMark06 permanent tests for all overclocked pc's that you review, otherwise stable goes only so far as using Windows calculator or playing Freecell...


PS: Falcon Northwest needs to sell the ICON cases seperately... I want one without paying ten grand.



We're going to add some sort of stability test in the next rev of benchmarks. We've used things like SuperPi and Prime95 before, but the script we recently started using is showing better results than either of those apps. I'll see if I can get Gordon to post instructions for using it.

Log in to MaximumPC directly or log in using Facebook

Forgot your username or password?
Click here for help.

Login with Facebook
Log in using Facebook to share comments and articles easily with your Facebook feed.