No, I wasn’t seeing double writing the title, nor are you reading it. There has been two separate attempts to overclock dual core processors, with air, no less. XBit Labs tried overclocking a PentiumD; PC Perspective overclocked an Athlon 64 X2.
XBit Labs used a Zalman CNPS7700Cu to cool (well, try to cool would be closer to the truth) the D, while PC P used the standard AMD FX cooler.
To make a long story short, the D couldn’t do much more than 3.2GHz without throttling part or all of itself, while the X2 could do a bit over 2.6GHz before hitting the wall.
Given that the X2 results works out to the equivalent of roughly a Prescott 4.0-4.1GHz, this is no contest. AMD wins, big. If you want performance without much fuss, there is and will be only one choice until Intel tries again at 65nm.
Two Different Bottlenecks
We mentioned the other day that the bottleneck for Prescotts seemed to be simply heat, while it seemed to be more than that for something else for Hammers.
The testing done here seems to confirm that. XBit Labs’ overclocking problems seem to be purely heat-related. Prescotts are certainly individually capable of a lot more than 3.2GHz by themselves. The X2, on the other hand, got pretty close to what looks like the wall for individual Rev E Hammers.
We seem to have the rather odd situation where one processor does considerably better under normal environmental conditions, while the other at least improves much more under extreme cooling conditions.
Since serious freezing costs much more than air, again, big advantage AMD.
But then they throw it all away.
The Real AMD Bottleneck: AMD
It’s clear that the X2 is better. Too bad it costs too much.
Sorry, but I’m not going to pay over $500 for this muffler. Not going to happen, period, end of story.
Would I lay out around $300 or a bit more for a 2Ghz version of this? In a flash.
That’s one lost sale, and sorry AMDroids, the vast majority of AMD fans are going to vote the same way for the same reason. Actually, on the whole, they’re probably not as willing to lay as much out as I am.
So will most people, fan or not, buying OE< machines.
So they’ll just buy a single, or maybe not buy anything at all until socket M2 comes along, and AMD will have even dual-core sons of Semprons around.
I grant you Pentium Ds aren’t likely to become big hits among DIYers, but they’ll sell many times more Dell boxes than AMD will overall.
The future of desktop computing is at least dual core. It’s new territory, a place where Intel isn’t entrenched, a place where AMD could make the first impression among many that, yes, we are better than Intel. Let Intel spend its money convincing people they need dual cores, and let AMD focus on saying “But we have the better ones.” They could even charge a little more than Intel.
But nooooooo. It’s better to forfeit that just to make a couple hundred dollars more on a relative handful of chips.
This is why AMD will never, ever, ever be a serious threat to Intel as an independent company. They can never rise to the occasion. They either can’t see much beyond their noses, or (more likely) they can’t deliver in the clutch.
Yes, the product is good, but the company is a loser.