I know that :) but still my point is that it's not impossible with current tech. If they wanted it, they could do it.Quote:
Originally Posted by Hornet331
Printable View
I know that :) but still my point is that it's not impossible with current tech. If they wanted it, they could do it.Quote:
Originally Posted by Hornet331
Its not technical impossible. But its is economic impossible so to say.Quote:
Originally Posted by Katanai
I keep waiting for a 4Ghz Quad too and it sh1ts me to tears that it ain't out yet or likely probably before 32nm.Quote:
Originally Posted by Katanai
I dont think 32nm will give you 4Ghz. Unless there are radical architecture changes. Aka Netburst 2.0Quote:
Originally Posted by Chad Boga
So will they require new FB-DIMMs based on DDR3?
Any word on pricing on that?
Will there be an equivalent to the 5100 chipset that uses regular registered memory?
Turbo mode in servers seems hard to validate though. I thought Turbo mode was a dynamic feature, and the amount of clock increase is unique to each chip? How could you validate systems with such a dynamic variable?Quote:
Originally Posted by Jacky
The power management chip the size of a 486 is the key.Quote:
Originally Posted by ryboto
Regular DDR3.Quote:
Originally Posted by uOpt
For Nehalem, at least Bloomfield/Gainestown, they are pretty much red-headed twins. 'X58' (Tylersburg, notice the -burg ending, meaning workstation/server) is the same chipset being used for both high-end desktop, workstation and dual-socket server boards. The only difference between Bloomfield and Gainestown is 1 or 2 QPI links.
'Serious' boards will be much closer to their desktop cousins this time around, using regular DDR3 (although ECC is recommended if you use it for anything that matters) and the same socket.
what he said, exact same platform as the core i7 stuff. Only difference is on X58 one of the QPI links on the chipset has been disabled. The same difference between Bloomfield and Gainestown.Quote:
Originally Posted by bowman
That also means if a motherboard manufacturer were so inclined, they could add all of the exact same tweaking options that any of the high end X58 motherboards have out now.
from Tom's a while back
http://www.tomshardware.com/news/int...2009,6384.html
Nice! Looks like the takeover of the server Market is well under way. If this depression we're in ever turns around I might have to get my hands on Intel stock, as it's about to go to the moon.
Shame one will have to go down to the 2.13GHz EP SKU in order to be able to stomach a dual-socket setup.. (scratch that, that's a dual-core apparently or one with a gimped L3, so the $300+ one then) I wonder what the baseclock/multiplier relation is there. Probably 133MHz as well and a very low (for Nehalem) multiplier?Quote:
Originally Posted by Blauhung
You might take a look at the new Opterons...Quote:
Originally Posted by T_Flight
I'm interested in how those workstation processors will work in X58 boards.
Same as i7..Quote:
Originally Posted by Bobsama
Q9450 -> X3350
Q9550 -> X3360
Q9650 -> X3370
i7 965 -> W3570 (except not unlocked multiplier)
i7 940 -> W3540
i7 920 -> W3520
They're just rebranded.
That's just ridiculous if true, a 1600$ CPU without unlocked multi? I know most workstation users do not overclock anyway, but it would be strange nonetheless.Quote:
Originally Posted by bowman
ryboto, turbo-mode is not that hard to validate, but HT is a btich probably.
I'm sure Intel could release and may yet do, i7 @ 3.6Ghz, so I don't know why you would think 32nm wouldn't be able to.Quote:
Originally Posted by Shintai
Perhaps not a 6 core Westmere, but just as there will be Dual Core Nelly's, I would suspect we will have 4 core Westmere's.
Well, no, a $1000 locked CPU. 3xxx are single-socket. 5xxx are the dual-socket puppies.. For reference, go and look at what the 3.2GHz Harpertowns cost right now. :DQuote:
Originally Posted by Jacky
The 2 socket chips don't have the "max" TDP anymore, which means no turbo mode?
Not that any servers use it anyway, lol.
Don't think of all dual socket boards as servers..Quote:
Originally Posted by Macadamia
Think workstations and some of them can clock pretty well, even on stock volts on air!:D
Yes I know, but I have a feeling that this will be hacky. :p:Quote:
Originally Posted by Movieman
I might be interested in 2 of the E55XXs if
i) They can be used on Skulltrail
ii) They rock in Maxwell Render (we'll know soon)
iii) My brother actually lives up to his promise and pays me ~250 smackers per month :rofl:
So the Gainestown does the rendering, and the Deneb works as the creative (mostly gaming) unit. Harmony and paychecks for both!
Only Tylersburg-based 'Skulltrail 2', obviously. When/if that arrives.Quote:
Originally Posted by Macadamia
In any case it should be easier for ASUS etc this time to make a proper enthusiast dualsocket board, given that it's the same chip.
My only fear is that they limit ST to the $999 XE CPU, and Xeons are barred. Although if they don't, they have me as an immediate customer.
I can wait, my money's not arriving that soon anyway.
Regular DP Xeons worked on the first Skulltrail, so why not this one?Quote:
Originally Posted by Macadamia
Anyway, whatever ASUS and co. cook up in terms of Tylersburg-DP will probably be a better dualsocket enthusiast platform than Intel reference designs so I wouldn't worry about it.
Aren't all the 3 and 3.2ghz Harpers unlocked? X5482 as in eXtreme (=unlocked)?Quote:
Originally Posted by bowman
They could also release all of their CPUs either undervolted or overclocked, giving us 4ghz 45nm penryn dual cores, but Intel is probably not keen on losing another billion in profits...Quote:
Originally Posted by Chad Boga
What for? We know nehalem will destroy them in a lot of benchmarks, but we do not know the margin of victory yet. A single socket Nehalem scores ~80% of a 2P shanghai set-up in specfp_rate & int.Quote:
Originally Posted by nullface