Sign in with
Sign up | Sign in
Your question

Core i9

Last response: in CPUs
Share
August 7, 2009 10:14:16 PM

Any pricing estimates. That socket 1366 is really tempting. Wiki said $999 but I don't trust them.

More about : core

August 7, 2009 10:33:03 PM

As its replacing a current 999$ chip, at least that much
a c 99 à CPUs
August 7, 2009 11:02:55 PM

nonxcarbonx said:
Any pricing estimates. That socket 1366 is really tempting. Wiki said $999 but I don't trust them.


It's on Intel's roadmap as an Extreme Edition part, which means it will cost at least $999. I have seen Extreme Editions go for up to $1599 (Core 2 Extreme QX9775), but generally they are $999-1199. I'd expect the new i5 and i7 parts to cost about the same as the currently-shipping parts: ~$550 for the i7 870, $260-285 for the i7 860, and ~$180 for the i5 750.
Related resources
a c 99 à CPUs
August 7, 2009 11:10:26 PM

Quote:
A grand isn't that steep. Start saving now :) 


A grand for an x86 CPU is an absolute ripoff unless it is an server CPU that works in four-socket or better motherboards. If you want a lot of cores, you can get two six-core Opteron 2427s for $898, and you can buy them today. The Core i7 920 is really the only thing in a single LGA1366 socket that is worth buying, and it costs nowhere near a grand.
August 7, 2009 11:10:50 PM

Not alot of upgrade scenarios for i7 currently, everyones looking to the server chips
a b à CPUs
August 8, 2009 4:11:13 AM

I would bet everything on Intel, in that scenario.
August 8, 2009 8:06:06 AM

exactly how many of you are considering to buy this for $1000

I would pay around $500 for a 2.6+ six core
not any slower for $500+
August 8, 2009 12:14:57 PM

^ well it's an EE part at between 2.93GHz and 3.6GHz so I think it's worth it considering it'll probably go over 5GHz on air.
August 8, 2009 1:15:32 PM

The new duals, at least the ES ones dont show hitting 5Ghz on air, hows a hex going to do it?
August 8, 2009 1:19:39 PM

^ well the i7 975 can do it, and the hexacores start at a higher clock speed along with being 32nm so have an even higher chance of reaching 5GHz +.
August 8, 2009 1:31:51 PM

How many 975s can do it?
What Im saying is, the TDP is higher in a smaller area, too much heat concentrated there using a little more power, regardless of node size. If you use x amount of power, you have x amount of heat, and that heat is concentrated in the die size
August 8, 2009 1:33:39 PM

Id also like to see the links on the 5Ghz 975, cause if its the one Im thinking of, no one will ever get a chip like that, unless thru odd luck, and the TDP isnt the same as stock 975s, just like the TWKR chip, it had high leakage
a c 99 à CPUs
August 10, 2009 9:15:58 PM

Quote:
How much you willing to bet that gulftown would stomp the *** out of an 8 core Amd server?


It would most likely be a function of the clock speeds of the two CPUs and the kind of loads you are running. At best, the Core i7 is something like 25% faster core-for-core, clock-for-clock than the current 45 nm K10 (mainly due to HyperThreading in heavily multi-threaded applications). If you're an idiot and running mainly single-threaded applications on such a machine, the Gulftown's turbo mode will kick in and make it considerably higher-clocked and thus faster than a dual Shanghai. (But if you wanted to run single-threaded applications, why do you want a six-core machine instead of a higher-clocked and less-expensive 2/3/4-core machine?) If you are running heavily-threaded programs on both machines, they will be roughly equal clock-for-clock, with the greater number of cores in the dual Shanghai compensating for the advantage HyperThreading gives the Gulftown.

However, your statement has no bearing in reality. First, I said you could buy two six-core Opterons for the price of one Gulftown (12 cores in total.) A dual Istanbul machine would be a ton faster than a single Gulftown clock-for-clock in any kind of multithreaded situation. The Gulftown would need to run at a >1 GHz higher clock speed than the Istanbuls to be faster in multithreaded situations. Sure, that may very well be true but you forget that you can get a six-core Istanbul today while the Gulftown isn't supposed to debut until Q2 2010. The current DDR2-using six-core Istanbuls will be replaced by faster DDR3-using variants in Q1 2010 and by 32 nm K11-based six- and eight-core units in ~Q1 2011. A pair of eight-core K10s would wipe the floor with a six-core Gulftown in anything multithreaded, and the 32 nm K11s ought to be both higher-clocked and faster per-clock. Servers are built for multithreaded prowess- expecting a single-socket desktop CPU to be faster in such a situation is ridiculous.
a c 99 à CPUs
August 10, 2009 9:24:31 PM

JAYDEEJOHN said:
How many 975s can do it?
What Im saying is, the TDP is higher in a smaller area, too much heat concentrated there using a little more power, regardless of node size. If you use x amount of power, you have x amount of heat, and that heat is concentrated in the die size


The Phenom II X4 has a die size within a few mm^2 of the Core i7, yet the Phenom IIs have overclocked well over 6 GHz, let alone 5 GHz. Your "heat density limits Core i7 overclockability" hypothesis doesn't seem to stand up very well. My best guess is that AMD's 45 nm process is just more suited to extreme overclocking than Intel's is. The overclockability varies by process a bit- Intel's 65 nm tolerated a lot of volts but the 45 nm doesn't do that very well. The original 65 nm Phenoms didn't overclock all that well, but the Phenom IIs overclock like mad.
August 11, 2009 1:14:24 PM

That wasnt my point. With turbo etc, the i7s have a higher overall usage and run hotter, and are slightly bigger, so yes it holds up. But what I was refering to was the claim of a air cooled 975 at 5Ghz, where it was one of the best oceers, working together with a Intel employee who had procured a special cherry chip, not for sale, much like the TWKR chips we see, and to assume a hex core would dial up to 5Ghz is quite a stretch of the imagination, regardless of just sheer physics
August 11, 2009 4:34:53 PM

JAYDEEJOHN said:
That wasnt my point. With turbo etc, the i7s have a higher overall usage and run hotter, and are slightly bigger, so yes it holds up. But what I was refering to was the claim of a air cooled 975 at 5Ghz, where it was one of the best oceers, working together with a Intel employee who had procured a special cherry chip, not for sale, much like the TWKR chips we see, and to assume a hex core would dial up to 5Ghz is quite a stretch of the imagination, regardless of just sheer physics


Intel wouldn't allow that, it would make their octo cores somehow obsolete

Everyone knows that all the companies are holding back on us

for a fact

apple could have put 10mp cameras on their iphones
but then what profit does it leave in the future
people will buy 2 then 4 then 6 right....
-apple representative

Intel is no doubt holding back just for a profit
August 11, 2009 5:01:12 PM

MU_Engineer said:
The Phenom II X4 has a die size within a few mm^2 of the Core i7, yet the Phenom IIs have overclocked well over 6 GHz, let alone 5 GHz.


the reason for this is because the i7's don't do well in the extreme cold of LN2 benching, whereas the phenom's do, so of course the phenom will reach further overclocks, but the i7 overclocks quite a bit further than the phenom when it comes to air cooling.
a b à CPUs
August 11, 2009 7:03:13 PM

So AMD is upto par with Intel in some ways tight? @helloworld
November 10, 2009 4:32:01 PM

Any chance you can put an I9 in something like a Rampage II Extreme?
a b à CPUs
November 10, 2009 6:50:27 PM

OK massive Necro, who looks up a thread over threee months old?
!