I thought I ought to give you guys a quick update on what's cooking in Damage Labs. You've probably seen my Skulltrail review by now, but if you didn't look closely, you may have missed something that I included in it: full results for the Phenom 9600 both with and without the TLB erratum workaround enabled. Ever since we first published our article about the performance hit caused by the workaround, I had intended to follow up with a full set of results for the entirety of our CPU test suite. The Skulltrail review has that data, and you can see that the workaround's performance impact is substantial in a great many applications, including games. Flip through the review and check out the numbers, if you'd like. I think the workaround's true impact is worth keeping in mind, especially since I'd expect the vast majority of Phenom-based systems from PC makers to be shipping with the TLB workaround enabled, per AMD's recommendation. Like many of you, I'd much rather be talking about happier subjects, but this one affects consumers in a very concrete way. I think we're obligated to point that out.
This is our first review with power consumption testing using production a production Phenom with Cool'n'Quiet enabled, as well. The Phenom/790FX combo looks pretty good on idle and peak power, but its performance in Cinebench isn't snappy enough to allow a good showing in our power efficiency measurement. I suppose those things are no surprise given its relatively low clock speed.
Another upcoming write-up will be the Phenom 9600 Black Edition. I've tested it, overclocked it, and have results ready. I've just been trying to find time to write this one up.
Also on the test bench: the Phenom 9500 and the Core 2 Duo E8500. The Penryn-based E8500 is intriguing because it's priced the same as the 65nm Core 2 Quad Q6600. The addition of this chip to Intel's lineup, combined with an unexpected wait for 45nm quads, really raises the stakes in the battle between two fast cores versus four slower ones. Testing is underway...