Yummy, I'll be putting those in the new line of computers down at the Chop Shop. Cooling's gonna be a b**ch though.
Time to invest in copper methinks. "Buy your own copper mine TODAY! It's more cost effective than saving up to buy a heatsink big enough.."
You're gonna need a hell of a lot of copper to cool those... dual preshotts... Toasty.. from the look of that report (and prior reading) they dont perform that great either.... Seems a bit pointless imo...
Compared to how much it costs to develop a new core, it's cheaper and still adds performance until they can roll out 65nm process - which we hope will be better than the 90nm.
Yea, but why not either launch a new (decent) mobo for the Dothans, or tweak them to run in desktops?
Because Intel cant do that lest they want to save face. Intel is built on PR; if their face image is suddenly twisted when they backpedal to a different processor than they were flaunting before because the one they put their efforst into isnt that good, their image would be heavily tarnished in the eyes of their biggest purchaser; business.
THG says that dual pressys will only run 13% hotter than a single. Thats pretty impressive. But then again its THG.
I cant wait for dual core Xeon's .... 4 processor - 8 logical processors I think I'll hold off upgrading anything until then.
So, that's why you need an all copper heatsink the size of a brick with a pelt? (so was the early reference design according to the inq). 13% of 65C is still 73C, and at 80C that's 90.4C! (Yes, i finally found my calculator earlier and im not afraid to use it!) What bothers me though, is the possible assignment of single tasks, like SETI to one physical processor but having the computer free todo other things like word process/play music/surf net. But windows is a pain in the posterior because it picks up logical processors, which means you will have to try and force SETI (or whatever process) to 2? logical processors? Or just one, and then itll sort itself out for one physical??
when calculating with temperatures you need to calculate from 0K (-273C) not from 0C ( +273K) if it is true for power plants it must be true for PC's made that mistake in college a while back and got my teacher quite angry ...
Good God. Two Dual-core xeons with SLI. Can I hear an "overkill"? I also wish they would punch out a 915 chipset for Dothans, but considering they have DFI and a couple other companies doing the leg work for them, they can focus on other things. Personally, I think they should bring back the Northwood.
Just been reading Anandtech and it looks like the dual core desktop P4s will NOT have HT, but the server (xeons) will. But you're still looking at 2/4 physical, 4/8 logical processors trying to access 1 memory bus. The huge hunk of cache strapped onto Xeons helps but will possibly make them more expensive (?) Gotta admit the Dual core Xeons *do* look kick ass though: Same with the single cored Prescott revisions due out soon too - cant scale the clock speed? So increase the cache. 2M on new prescotts makes it nearly P4EE size, but L2 is faster than L3 so may make up for it somewhere. Still costs more money to produce on silicon and transistor costs and it is still a prescott where the P4EE cores are 130nm Galatin (northwood veriation) based. AMDs dual cores will also share a single memory bus (same qty of pins on the cpus) but their HTT tech is much more scaleable as any number of CPUs can have their own memory assigned to them - chipset independence. Also, dual core P4s will require a 945/955 chipset, whilst AMDs dual cores will be backward compatible with all the latest chipsets (so ive read). Again, time to invest in AMD as making CPUs with less silicon (less cache) is more cost effective.
I won't mind buying a new mobo, to be honest. I think 2 mb cache on a processor is stupid, because nothing will be able to take advantage of it. Plus, you'll usually never have enough instuctions to have to keep it that local (ie: on the chip). Seems foolish to me. But thank you much for the added information.
