WTF? $266 quad core? Also, 45 nm question

scryer_360

Distinguished
Jan 13, 2007
564
0
18,980
Someone posted an article about how quad core prices will soon plummet. The $266 is not what we will buy it for obviously, as that is wholesale value of 1000 units. But I expect $310 maybe for one of these upgrade dual core units at 3 gigahertz, no more than $400 for the best quad core of the current C2D generation.

What the hell? I just bought a E4300 with the intention to make it 2.4 gigahertz, now the chips that once did that for $310 stock will be what, $200 at most?

But the big question is this: will the new 45nm chips launching later this year and early next fit an LGA775 socket? If they have an integrated memory controller, I cannot see how they would work with the current FSB setup. Or am I off on a limb and LGA775 will do just fine?
 

kamel5547

Distinguished
Jan 4, 2006
585
0
18,990
Someone posted an article about how quad core prices will soon plummet. The $266 is not what we will buy it for obviously, as that is wholesale value of 1000 units. But I expect $310 maybe for one of these upgrade dual core units at 3 gigahertz, no more than $400 for the best quad core of the current C2D generation.

What the hell? I just bought a E4300 with the intention to make it 2.4 gigahertz, now the chips that once did that for $310 stock will be what, $200 at most?

But the big question is this: will the new 45nm chips launching later this year and early next fit an LGA775 socket? If they have an integrated memory controller, I cannot see how they would work with the current FSB setup. Or am I off on a limb and LGA775 will do just fine?

IThe integrated memory controller does not kick in until Nehalem, the initial 45nm are just a shrink. There *shouldn't* be any reason LGA775 that handles Core2 won't handle the new chips.
 

kbrasee

Distinguished
Sep 3, 2006
36
0
18,530
Yes, the CPU slot will need to be 45nm x 45nm. These things are getting small enough where it will be difficult to insert the CPU.
 

Mex

Distinguished
Feb 17, 2005
479
0
18,780
Yes, the CPU slot will need to be 45nm x 45nm. These things are getting small enough where it will be difficult to insert the CPU.
Uh...no. It refers to the spacing and size of the transistors on the die. For reference, the Core 2 Duo has 291 million transistors on a die that is 143 sq millemeters in size, about the area of the nail on my thumb. Does that tell you the size of nanometers? I can't explain what the measurement specifically denotes because it confuses me as well (Ask Jack). The shrink will not affect the size of the package (what you put in the socket) unless Intel decides to create a new socket, which isn't happening until the introduction of Nehalem.

call me Captain Obvious, but won't 45 nm require a whole new chipset, given the fact that the processor will be smaller?
The processors won't neccesarily need a new chipset, althought there will be new chipsets. The main issue is the VRM circuit on the motherboard; however, the rumor is that Intel ran 45nm processors on existing motherboards.
 

cb62fcni

Distinguished
Jul 15, 2006
921
0
18,980
Yes, the CPU slot will need to be 45nm x 45nm. These things are getting small enough where it will be difficult to insert the CPU.

Sarcastic comment of the month award winner there.

For Capitan Obviouso, no, shrinking the process doesn't neccessarily mean the processor itself gets smaller. Just the internal components. 775 has been through a number of process shrinks already. The reason C2D needed new ones had nothing to do with the nm of the chips, but rather the voltages that were required.
 

cb62fcni

Distinguished
Jul 15, 2006
921
0
18,980
Oops, beat me to it.


Sure, lots of C2D mobos out there now can run at 1.6Ghz FSB, but most of the low-end boards can't. It'll be interesting to see what the solution is.
 

Mex

Distinguished
Feb 17, 2005
479
0
18,780
The rumors that Penryn would run at a 1600MHz FSB originated from the Inquirer. Most boards are only officially certified to run up to 1333MHz at this point. If Penryn was 1600MHz, the boards would need to be capable of that speed (mostly taken care of), but they also have to advertise that they're capable of that speed, or else the average Joe will automatically assume that it won't work.

It smells like BS to me, but it's too early to tell.
 

cb62fcni

Distinguished
Jul 15, 2006
921
0
18,980
Well, Anand ran with it, so I'm assuming it's fairly credible. It would also help explain how Intel is going to pull off this rumored 20% performance increase.
 

Mex

Distinguished
Feb 17, 2005
479
0
18,780
Really...thanks for the link. This is the final hour of the front side bus since CSI is finally ready, so I guess Intel decided to go for broke.
 

epsilon84

Distinguished
Oct 24, 2006
1,689
0
19,780
Isn't penryn now supposed to be 1600Mhz FSB?

That may very well be a limiting factor on the current gen 775 boards.

Considering most P965 boards are capable of 500MHz (2000MHz) FSBs I don't see that as a big problem. The main problem would be overclocking these since the FSB is so high to start with. Hopefully there will also be 333MHz (1333MHz) FSB parts available, which is still not ideal from an overclockers perspective but it's the lesser of two evils we're talking here. :twisted:
 

cb62fcni

Distinguished
Jul 15, 2006
921
0
18,980
Well, a lot of 965 boards and most of the 650/680 parts can easily hit around 500, but many of the low-end 965's and all the 945's cannot. Even the 975X's will be kind of pushing it. I'd guess that the 1600FSB is their high-end part, but I haven't yet heard of them releasing 1333 penryns. It would be smart for them to release both 1333 and 1066 parts aimed at low to mid-range buyers, and this would help widen mobo compatability too. A 1066 penryn would likely be an overclocker's wet dream.
 

jeff_2087

Distinguished
Feb 18, 2007
823
0
18,980
Yes, the CPU slot will need to be 45nm x 45nm. These things are getting small enough where it will be difficult to insert the CPU.
Uh...no. It refers to the spacing and size of the transistors on the die. For reference, the Core 2 Duo has 291 million transistors on a die that is 143 sq millemeters in size, about the area of the nail on my thumb. Does that tell you the size of nanometers? I can't explain what the measurement specifically denotes because it confuses me as well (Ask Jack). The shrink will not affect the size of the package (what you put in the socket) unless Intel decides to create a new socket, which isn't happening until the introduction of Nehalem.

:lol: Hook, line, and sinker.
 

darkstar782

Distinguished
Dec 24, 2005
1,375
0
19,280
400MHz, or 1600MT/s FSB will be for Xeons.

It will be used with Dual Independent Buses (DIB) for two CPU systems.

The desktop ones will be 333MHz or 1333MT/s.
 

dobby

Distinguished
May 24, 2006
1,026
0
19,280
45nm is refering to the sizw of the transistors, 45nm is 5 atomic layers thick - that means 400 transistors on a redblood cell - and they are tiny. 775 has a huge array of manufacture size already, i duobt they will change it, you may require a new or udated chipset, but the LGA has life left to breath
 

kbrasee

Distinguished
Sep 3, 2006
36
0
18,530
I'm already having a tough enough time getting my 90nm processor in, I think with 45nm I'm going to have to move to smaller tweezers.
 

Vash-HT

Distinguished
Jun 30, 2006
104
0
18,680
Yes, the CPU slot will need to be 45nm x 45nm. These things are getting small enough where it will be difficult to insert the CPU.
Uh...no. It refers to the spacing and size of the transistors on the die. For reference, the Core 2 Duo has 291 million transistors on a die that is 143 sq millemeters in size, about the area of the nail on my thumb. Does that tell you the size of nanometers? I can't explain what the measurement specifically denotes because it confuses me as well (Ask Jack). The shrink will not affect the size of the package (what you put in the socket) unless Intel decides to create a new socket, which isn't happening until the introduction of Nehalem.

call me Captain Obvious, but won't 45 nm require a whole new chipset, given the fact that the processor will be smaller?
The processors won't neccesarily need a new chipset, althought there will be new chipsets. The main issue is the VRM circuit on the motherboard; however, the rumor is that Intel ran 45nm processors on existing motherboards.

Actually, 45nm refers to the gate width of the transistors that the cpu is using. The gate is getting so thin that new methods need to be found to shrink it any further, which is the whole deal with metal gates we saw a little while ago. The processor may be smaller, but that doesnt mean the final package it comes in will be any smaller. There have been a few platforms I can think of off the top of my head where process changes have taken place and required just a bios update. I don't know if Penryn will be the same though, because they are doing more than jsut shrinking the process for that.
 

Mex

Distinguished
Feb 17, 2005
479
0
18,780
Yes, the CPU slot will need to be 45nm x 45nm. These things are getting small enough where it will be difficult to insert the CPU.
Damn, I should've seen right through that one...

90% of me thought that was wrong.
10% of me thought that was too obvious to be wrong.

I can't figure out why I don't listen to that 10% of me more often. :cry:
 

fishboi

Distinguished
Apr 25, 2006
1,415
0
19,280
Maybe not so. Seems like the P5W DH isnt the overclocker of choice:
http://www.behardware.com/articles/651-2/intel-core-2-quad-q6600.html

"Unlike with the QX6700, the Q6600’s multiplying coefficient is blocked for increases. Some will say that this isn't a problem as it is possible to increase the FSB. This is true, but unfortunately Kentsfield processors like the QX6700 and Q6600 have some trouble with high FSBs and some motherboards for a reason unknown to us.

With the ASUSTeK P5W-DH Deluxe and the bios 1707, with a stable configuration we were unable to reach over a FSB of 333 MHz. This was 3 GHz for the Q6600 and 3.33 GHz for the QX6700. Even at 3.1 GHz, the Q6600 had errors when several sessions of Prime 95 were launched simultaneously. If, however, we used the multiplying coefficient, the QX6700 was stable from 3.46 GHz."