The Great Moore's Law Compensator: It's a term I coined more than a year ago to describe the process whereby each successive Windows release effectively gobbles up the latest gains in PC hardware performance. The net result is an environment that performs roughly on par with the one you're upgrading from -- despite the fact that the underlying CPU, chip set, RAM, and I/O subsystems are all at least twice as fast as those in your old rig.
As axioms go, the Great Moore's Law Compensator, or TGMLC, has proven to be quite resilient. From DOS-based Windows versions to the great NT kernel transition with XP, the core assertion of TGMLC -- that Windows expands to consume all available hardware -- has been continuously validated. In fact, the only hiccup in this otherwise seamless progression involved Windows Vista. In that instance, Windows outpaced the hardware by a wide margin, causing untold grief for the masses trying to make it perform reasonably well on what were clearly inadequate (by Vista's requirements) systems.
[ If you've already made up your mind to take the plunge, then don't miss this article by InfoWorld's J. Peter Bruzzese: Ready for Windows 7? Here's how to deploy it right. ]
Fortunately, the universe has a way of righting such wrongs -- taking the occasional protruding nail and hammering it down until the entire row looks even again. In the case of Windows, the release of Version 7 -- with its Vista-like system requirements and performance characteristics -- has been projected to serve as a kind of TGMLC "breather": an opportunity for the hardware to finally catch up with the OS, thus returning balance to the Wintel equation. And based on a preliminary review of benchmark data collected by the recently released OfficeBench 7 test script, Windows 7 is indeed living up to its promise of following TGMLC norms.
Take Office 2010 as an example. Under Windows 7, the Office 2010 Beta runs 15 to 20 percent slower than Office 2007 under Vista with SP2, per data collected across several hundred test systems running the latest release of OfficeBench. Assuming some performance tuning between now and the final Office 2010 release, this sizable gap should translate into a cross-version performance delta that falls roughly in line with historical trends in Windows code bloat.
In other words, we'll return to the point where Windows and Office no longer overwhelm the available hardware bandwidth and merely suck up every available cycle while delivering an end-user experience roughly comparable to the one that greeted those first Windows 9x users way back when Bill Clinton was still in his first term and "9/11" was just another fraction. In the world of TGMLC, this constitutes real progress.
Of course, some IT shops will still complain. After all, they're spending their hard-earned budgets to secure the latest and greatest in PC performance. For them to then learn that Windows 7 is not the silver bullet that will solve all of their performance issues, but rather simply the universe restoring balance to the Wintel equation, will be a bitter pill to swallow.
My advice: Take a big sip of water and gulp that puppy down. Because it could have been a lot worse. Microsoft could have pulled another Vista and saddled us all with an OS that's two sizes too big for the hardware it's running on. Windows 7 restores the "fits just right" relationship that TGMLC has long described -- and that's a good thing.
The universe is back in balance. TGMLC is secure. All is well, all is well, and all manner of things are well -- at least until Windows 8.