On Monday 13 March 2006 07:37, Denis wrote: > Iain, > > So are you saying that a P4 is actually faster with HT disabled, or > simply that you don't have much to gain by using HT?
I have a laptop and a desktop the sport p4's with HT. I don't see any difference whether HT is turned on or turned off. From what I see, if your software isn't taylored to take advantage of HT, there's an actual penalty having it... > I've had a > dual-Xeon machine with 4 logical processors running Gentoo for a > couple years now, and I like being able to run 4 threads of > simulations at the same time. The threads, when started nearly > simultaneously, seem to finish the same jobs in nearly the same amount > of time... I've been told before that it's simply because the HT > makes both the actual processors less efficient... Either way - I > haven't had any complaint about HT, but if there is conclusive > evidence that HT drastically takes away from true P4 power, then I'd > definitely like to know about it. > Think of a P4 with HT as an ordinary P4 with two inputs and only one processing unit. If the thread on one input wants all the cpu power, it gets it and the other input stalls and starves. It's easy to see by running top on a busy P4 with HT computer. I can remember something I heard while growing up, you don't get something for nothing... Jerry -- [email protected] mailing list
