2CPU.com » News » August 2004 » Intel's LGA 775 socket explored

The Inquirer has posted a two part article on Intel's new LGA 775 CPU packaging, and, in traditional Inq style, they are very informative and yet surprisingly entertaining.

EVERYONE KNOWS that Intel's new Socket 775 (LGA775) simply sucks, after all, it has to, could 1,000 people on 100 message boards all be wrong? It is something new and different from Intel, who has none of our best interests at heart, it simply must be a conspiracy. It is fragile, expensive, and does nothing that the old socket would not do. If you look at it cross-eyed, or god forbid, put in a CPU for a fourth time, the board will simply catch fire. Nothing good comes from it, trust the friendly anonymous posters, none of which have an axe to grind, honestly. Put away the tin foil hats people, it isn't any of the above. I read the same things you probably did, and talked to some of the same people you read about, and several more. I have heard just about every conspiracy theory out there, and I noticed something odd, no one asked Intel if they were true. What is even more disturbing is that amidst the armchair quarterbacking, or at least armchair Electrical Engineering, no one bothered to ask the question "Why is LGA775 better than s478"?

Part one can be had here, part two here.

07/30/2004 01:51 PM: Intel's Lindenhurst Woes by Jim_
For the last couple of days, The Inquirer has been telling the sad story of Intel's Lindenhurst. They've gone through the bugs that have the OEM community upset and they even say it's virtually identi...

06/15/2004 03:36 PM: Intel's Nocona to bust out in June by Jim_
C|Net has a short article up this morning discussing Intel's upcoming release of its Nocona processor. The 32/64-bit functionality of Nocona and Opteron are similar. Tom Halfhill, an analyst at Microp...

06/04/2004 02:22 PM: GamePC looks at Intel's Dothan by Jim_
GamePC has taken a look at Intel's latest and greatest, the Dothan! Dothan is more or less an incremental upgrade to the Banias core architecture. Intel has doubled the cache and lowered power consump...

05/27/2004 02:36 PM: Who's 64-bits are better? Intel's or AMD's? by Jim_
The girls over at The Inquirer have posted about an interview with a Microsoft VP where the topic of Intel and AMD's 64-bit implementations was discussed. A WIDE-RANGING interview with Microsoft senio...

04/29/2004 01:06 PM: Intel's next-gen BIOS architecture by Jim_
Thanks to OSNews I found this link to an overview of Intel's next-gen BIOS architecture. Intel recently introduced an innovative new concept in BIOS: the Platform Innovation Framework for Extensible F...

04/22/2004 01:27 AM: Intel's $2 Billion Upgrade by Hooz
Intel is apparently closing down Fab 12 to perform a $2 billion upgrade to the facility. The coolest part is that they are relocating 800 employees while the renovations take place.Intel has built oth...

04/13/2004 01:34 PM: Opteron, Nocona rock Intel's Itanium boat by Jim_
The Inquirer is talking processors today. They love to wax philosophic about Itanium, Opteron and Intel's upcoming Nocona, so why should today be any different? Faced with the threat of low-cost &quot...

04/04/2004 05:33 PM: More on Intel's Processor Ratings by Hooz
We learned a few weeks ago that Intel is planning to implement a processor ratings scheme for their upcoming desktop and mobile processor lines. It appears as though Anandtech has some more info on ho...

03/12/2004 02:59 PM: Modular Chipsets in Intel's Future by Jim_
Tom let me know about an interesting note over at XBit about future chipsets from Intel. Intel...

KalTorak Registered User Posts: 42 Joined: 2004-02-07

#31042 Posted on: 08/04/2004 11:21 PM Hell, it's not MY fault nobody understands this - it's not as tho I haven't explained it before. And although my "NOT SPEAKING FOR INTEL" hat must remain firmly in place, my EE skillz are a fair bit beyond "armchair". Not Speaking For Intel.

Vuke69 Bitpimp Posts: 341 Joined: 2001-03-16

#31043 Posted on: 08/05/2004 01:21 AM There is one thing I dont understand, about all modern mainstream processors. And that is when power requirements increase, why do they keep adding power and ground pins by the hundreds? Arround a third to half of the pins on a modern processor are just for power and ground. Why dont they add a couple of giant lugs like they used to have on some large risc chips. It was either the alphas, mips, ultrasparcs, or power, I cant recall which at this time, but one of them had two big ass power lugs sticking out the top. Unless I am mistaken and those were just brain wave collectors or something. (good thing my brain it safe, in the motherly embrace of my tin foil hat Anyone have a good answer?

jhaislet "Let's Roll" Posts: 3248 Joined: 2002-06-25

#31044 Posted on: 08/05/2004 01:59 AM I agree, since the new P4's are pulling somewhere around ~140w EACH, why not put like 2 of those 3.5mm mini-jack (headphone) style plugs in the bottom/middle part of the CPU? However, I think the small pins are there as part of the power distribution grid on the actual CPU. Each set of pins powers a specific part of the cpu logic. Without the many pins supplying voltage to specific parts, you would probably need more internal power planes, which also btw, would have to be physically larger to carry the increased current. eg. Think if every home in the US was powered directly off a 765,000v transmission wire and the equipment required to step that down for each tap vs the normal way of stepping it off a 7,000-13,000v line to 110v/220v. Mini-ITX Systems: AOpen MP965-DR Core2Duo Mobility T7500 4MB L2, 4GB DDR2 (Vista MCE)| MSI Industrial 945GM2 Celeron M 430 | Intel Little Valley 2 (Firewall) | Via EPIA EN-1500 (Firewall)|

KalTorak Registered User Posts: 42 Joined: 2004-02-07

#31045 Posted on: 08/05/2004 02:10 AM Originally posted by Vuke69 There is one thing I dont understand, about all modern mainstream processors. And that is when power requirements increase, why do they keep adding power and ground pins by the hundreds? Arround a third to half of the pins on a modern processor are just for power and ground. Why dont they add a couple of giant lugs like they used to have on some large risc chips. It was either the alphas, mips, ultrasparcs, or power, I cant recall which at this time, but one of them had two big ass power lugs sticking out the top. Unless I am mistaken and those were just brain wave collectors or something. (good thing my brain it safe, in the motherly embrace of my tin foil hat Anyone have a good answer? Because the power and ground pins aren't simply about power delivery; they're also needed to provide "return paths" for the currents that flow around when the signal pins switch voltage levels. To oversimplify a little bit, currents flow in loops (according to a rule called Kirchoff's Current Law.) The physical size (er, area) of the loop around which the current flows is part of what determines how inductive that current path is. The more inductive that path is, the more voltage that's required to change the amount of current flowing in the path (V = L di/dt). SO, as the loop that the current follows gets bigger, the inductance of the loop gets bigger, as the inductance gets bigger, the "edge rate" of the signal transitions gets slower, and as that edge rate gets slower, it becomes more difficult to speed up the interface (and we all like fast interfaces.) For reasons I'll gloss over here, the loop that's formed by current flowing into or out of a given signal pin will also flow out of or into the nearest ground or power pin (current's returning to the part through that pin, so if it's going into the signal pin, it's going out the power pin, or vice versa.) So, to minimize that loop area, one tries to place ground or power pins near EVERY signal pin, especially high-frequency signal pins (since it's at higher frequencies where these inductances matter.) And that's why ya need lots of ground and power pins, besides the actual bulk delivery of current. Not Speaking For Intel.

Vuke69 Bitpimp Posts: 341 Joined: 2001-03-16

#31046 Posted on: 08/05/2004 02:18 AM Originally posted by jhaislet However, I think the small pins are there as part of the power distribution grid on the actual CPU. Each set of pins powers a specific part of the cpu logic. Without the many pins supplying voltage to specific parts, you would probably need more internal power planes, which also btw, would have to be physically larger to carry the increased current. I was under the assumption that these chips still had power distribution busses on the chip itself, and all the power pins went to the same place anyhow. But after reading your post, and thinking about it for half a second, I realized that you are probably correct. The last processor that i ever had to deal with on an electrical level was a Zilog Z80. That was a 80 pin dip with a mere single Vcc and ground. I guess things have changed quite a bit since then.

KalTorak Registered User Posts: 42 Joined: 2004-02-07

#31047 Posted on: 08/05/2004 02:44 AM Originally posted by Vuke69 I was under the assumption that these chips still had power distribution busses on the chip itself, and all the power pins went to the same place anyhow. Mmm... that's true; power distribution for a given voltage rail on a chip is pretty universal, and even more so with power planes in the chip package. The answer's more what I was saying about high-speed signaling. Not Speaking For Intel.

MichaelMR2 Eve Sucks :) Posts: 1902 Joined: 2001-07-02

#31048 Posted on: 08/05/2004 06:39 PM I like it.. I'm told it's only a matter of time till we have CPU's intergrated into the mobo the other alternative seems to be the whole make everyont on the die thingy Workstation: Q6600@3GHZ|Asus Striker Extreeme|4GB DDR2 800|4 x 15k5 73GIG Raid 0|Adaptec 32105|XFX 8800GTX x 2|1000w Galaxy DXX Servers: 1 x 3.0GIG P4|1GB mem|3ware 7500|11 x 200GB in Raid 5 2 x AMD 252|4GB mem|Areca 1260|16 x 400GB in Raid 5

jonjayal Experienced Duallie Posts: 247 Joined: 2002-01-11

#31049 Posted on: 08/06/2004 01:48 AM the other alternative seems to be the whole make everyont on the die thingy ..as AMD has done with the memory controller on the Opteron 2xP4 Xeon 2.4 @15x160, PC-DL rev 1.05 (1005 BIOS), 512MB Corsair TwinX PC3200, 1x Seagate Cheetah 15K.3, Adaptec 39320D U320 SCSI, Radeon 9600 Pro, Enermax 430W PSU

questionlp the Cowardly Tech Posts: 323 Joined: 2002-02-14

#31050 Posted on: 08/06/2004 08:03 PM Parts III and IV of the Socket 775 article are up: Part III: http://www.theinquirer.net/?article=17681 Part IV: http://www.theinquirer.net/?article=17709 [http://closedsrc.org/] My Rig: 2x 2.4GHz Xeon, 1GB PC2700, Supermicro X5DAL-TG2, ATI AIW 9000, SB Audigy 2 Plat/EX, 2x 80GB 7200.7's, Plextor 40x and 8/20, Pioneer 106D, Antec 1000AMG, Enermax 460W EPS-12V

2CPU.com » News » August 2004 » Intel's LGA 775 socket explored