you guys would be mad at me then. i have a hexacore that spits 4.0ghz out at 125, and a 260. i'll be upgrading it soon to a fusion power edition mobo w/gtx 460 dual frozr.

i have to say tho, the more performance you get per watt is genius. however... i must say that i'm very happy with my thuban. it might not be a 980x, but i built this computer for the price of that chip.. so go figure, and thats including the thuban, on the very first day it came to retail, 300 bones on newegg. a good investment i think?


to me, its actually all about the psu. not matter how many tdp's your pushing, if your psu isn't gold, you're already in trouble.
how many of you low wattage guys have a 90plus rated psu?
i do, 650 watt earthwatts. i'm pretty sure n owadays tho it's probably rated closer to 85 or so, just by its age.

if you get a build, i'm more th an happy to build it and take my kill a watt to it ;)

by the way, if you need good prices, just send me a message :) i build computers all day, so its just part of my day :)
i'm VERY interested to see where this goes (pressed the subscribe/follow button)

oh yeah, my mips is 4000 Floating Point MIPS per core
thats like 24000, so i think its pretty good..
sounds like i need lessons though..
any chance you can lay that build on us?

Hi Trilith

Yes my PSU is this one http://www.plugloadsolutions.com/psu_reports/...OS%201463_750W_Report.pdf

For details regarding the build see the first post in this thread

The only change to the lean green machine is an update from the Phenom II 905E to the 1055T 95W edition overclocked to 2.99Ghz a core (now 6 of them) - the leaner greener number crunching machine

As for your temperatures they seem ok but motherboard temperature measurements are notoriously inaccurate varying from manufacturer to manufacturer. They are best used to set a benchmark and used to spot when your heatsink is clogged up with dust.

The temperatures are fine when your results make it to valid. If you spot a few errors it is time to break out the dyson.

Get the correct spec heatsink. Apply the thinnest scrapings of Arctic Silver 5 and hoover it out once a blue moon and you will be OK.

I have never damaged a CPU using it for what it is designed - number crunching. I suspect that as they stay nice and toasty and at a stable temperature they probably last longer that way than having their substrates expand and contract with the changing loads and power downs and bootups of those cpu not lucky enough to be involved in a grid project.

But I am just biased

by the way all those load readings are at the power socket so they include the losses (heat) in the psu


Dave

i use tuniq tx-1 performance gel: i live by it. i see an instant temperature increase to 95F, and then it just stays there - all day ;)
and yes, psu's ARE IT man, a good solid 90+ will totally rock your rig, and you'll save a lot of power in the process. :)

Nice one waitingForTheMiracle

This is the the Intel Setup to beat

(Sandy Bridge will be back there's nothing wrong with the CPU it's the supporting chipset and some potential SATA instability - so I hear)

That's a whole load of crunching for not alot of watts

You are right about the benchmark but even with turbo core switched off that's green number crunching

Does the mobo have integrated graphics being Sandy Bridge?

Loving the 22nm die shrink dig

Dave AMD Autumns

Well, the CPU has integrated graphics. The mobos fall in two groups: Those based on the H67 chipset can use the integrated graphics but don't support over/underclocking. Those based on the P67 chipset (like mine) can do overclocking but can't use the integrated graphics.

So that's another option for optimizing Sandy Bridge efficiency: A H67 motherboard would let you skip the discrete graphics, saving a few more watts (and some money, both on the graphics card and by not paying extra for a K model - i.e. overclockable - CPU). Good point

By the time Sandy Bridge is back in the market, a third type of motherboard may be available: The Z68 chipset is supposed to support both integrated graphics and overclocking.

On the HT question, I don't know of any particulars except that HT processors will overclaim on points (because the claim is based on treating virtual cores as real) and appear to give a bit less runtime - I'm averaging 7 days runtime per calendar day with 8 threads running 24/7.