A few years ago there were a number of us that did some benching for a particular company and we tested many setups without going to script cheats etc. ACPI had no apparent negative effect whatsoever on actual performance scores. That was my personal experience.
I just g00gled a number of search terms related to ACPI and the majority of problems have seemed with a Linux O/S and little if anything actually related to Windows.
Knowing that there is much seeming good from ACPI, I would and have always utilized it on any of my computers; certainly over the last 4 years at least. It appears that ACPI and it being recognized accurately by the Windows O/S seems a bonus in managing IRQ's and the like.
I don't put much stock in what are likely what many call "old wives tales" and it is likely in this age of computing that ACPI effecting real performance is more a rumored thing rather than a validated situation. If I really needed to know the outcome of disabling ACPI, I would try it for myself since the results will likely vary from chipset to chipset version and drivers and Windows O/S. It would take two installations of Windows since you could test with the default ACPI disabled and then install Windows with ACPI enabled and of course if it were somehow faster with ACPI enabled then you would not want to RE-install with it off and that would be a single installation actually. Luck man either way.