How long will LGA775 Last?

[USMC2Hard4U]

Is it planned for the Yonah Dual core (and or Desktop variant) comming in 2006 to be LGA775? or will we have a socket change.

 

[theHinge]

I recall seeing somewhere that we shall indeed see a socket change in 2006 - but the article was rather speculative.

 

[Dillusion]

Is it planned for the Yonah Dual core (and or Desktop variant) comming in 2006 to be LGA775? or will we have a socket change.

Not as of yet, LGA775 will be around till late 2006/Q2 '07.

 

[jyi786]

I dont' know. For some reason, LGA775 is starting to remind me of the old Socket 423. The processors all based on those sockets were not as fast as purpoted to be. Perhaps Intel will release a successor to LGA775 that will be smaller, like going from Socket 423 to Socket 478, along with the performance increases. It'll be more than likely going from the 0.09micron to the 0.065micron process.

 

[USMC2Hard4U]

For whatever reason, i am not a fan of the Idea of "contact pads" on the CPU and the Pins on the Mobo... It just seems... I dunno, not right.

I want them to put the Pins back on the CPU and have a ZIF sockey on the Mobo.

 

[ozziegn]

I dont' know. For some reason, LGA775 is starting to remind me of the old Socket 423. The processors all based on those sockets were not as fast as purpoted to be.

I dont know about that.

my 3.0 630 chip has been running @ 4.1GHz (on air cooling) since day one without a lick of problems.

I'd say that 4.1GHz is pretty darn fast.

 

[robberbaron]

I dont' know. For some reason, LGA775 is starting to remind me of the old Socket 423. The processors all based on those sockets were not as fast as purpoted to be.

Explain this please.

 

[Maniacmous]

I can't speak for the man, but I would assume he is refering to the relative lack of performance clock for clock versus the Pentium 3. Namely, I can recall the P4 1.4ghz (and maybe a speed bump up or two) being beaten in a few (at the least) benchmarks by the top P3's at the time. There was a huge disheartening amoung many enthusiasts for Intel at that. Now, as clock speeds ramped up (and the subsequent switch to Socket 478), the clock for clock difference in performance didn't matter so much because there was enough of a raw speed difference to make up for the per clock performance delta, however, he is right, based upon the "faster clocks is better" mentality of the time, the Pentium 4, at least its early iterations, was a disappointment.

Edit: Now, as for the LGA775 not being as fast as it is made out to be...I don't know about that part.

 

[robberbaron]

That's not due to the Socket 423 platform, it's due to the superiority of the P6 core at those speeds.

 

[Eva_Unit_0]

That's not due to the Socket 423 platform, it's due to the superiority of the P6 core at those speeds.

Yeah exactly. It wasn't the switch to 478 that made the performance better...it was the switch to northwood, which could actually hit the clock speeds needed to get netburst chugging, as well as carry enough cache to feed that deep pipeline.

I honestly don't see a reason for intel to introduce a new socket. lga775 was designed to feed the increasing power requirements of netburst...and well, intel is now canning netburst in favor of, ironically, the power efficient p6-on-steroids architecture. I expect to see Intel's power requirements for cpu's drop rather dramatically when they finally can netburst, and as such lga775 will be MORE than sufficient. I personally don't see a reason for a switch until Intel goes to an integrated memory controller, which isn't even slated for the Itanium's for a while, let alone the desktop chips. That's so far off it doesn't even matter.

However, because of the non-integrated memory controller, I expect Intel to keep with the current trend of "new cpu's need new chipsets, regardless of socket type." Dual cores require 945/955, despite the fact that 915/925 carry the same socket. And likewise, when future lga775 chips hit I also expect THEM to require new chipsets as well. So to a certain degree, the similar cpu socket isn't of much comfort...as least not as much as AMD's universal compatibility sockets are.

 

[robberbaron]

An integrated memory controller will require a new socket. It can be done with the same amount of pins, but it will technically be a different socket, a la Socket M2 which happens to have 940 pins.

The boards will need totally different traces to allow for cpu-dram communication and thus it would require intel users to get yet ANOTHER board for their next chip. I don't see it happening for quite a while. If the P6 architecture benefits from low latencies more than the Netburst, then Intel might entertain it, but I wouldn't count on it.

 

[tshontikidis]

For whatever reason, i am not a fan of the Idea of "contact pads" on the CPU and the Pins on the Mobo... It just seems... I dunno, not right.

I want them to put the Pins back on the CPU and have a ZIF sockey on the Mobo.

Could you speculate why you do not like the idea of contact pads and want the old style of pins on the cpu and socket on the mobo.

I was a little spectical about the idea to about a year ago, i had never seen it and i was build a comp and i decied to go with it and i actually really liked it. It was very easy to install and i have not been having any trouble at all with it so far, and i am not seeing any preformance issuses. I have a p4 3.2 i think its the 540 and an abit dura max AA8.

personaly i would rather bend pins on my mobo than my cpu.

 

[FanATIc]

However, because of the non-integrated memory controller, I expect Intel to keep with the current trend of "new cpu's need new chipsets, regardless of socket type." Dual cores require 945/955, despite the fact that 915/925 carry the same socket. And likewise, when future lga775 chips hit I also expect THEM to require new chipsets as well. So to a certain degree, the similar cpu socket isn't of much comfort...as least not as much as AMD's universal compatibility sockets are.

Unfortunetly this is the biggest problem i have with intel. I do not follow the same "omg what a difference" cult many do going to AMD, but for someone who likes to upgrade, i really dislike intel's new chipset requirement every 6 or 8 months. I had a 925X, was planning to go dual core, and the first warning sign that made me dive overboard was the 1066FSB needing 925XE which was identicle in hardware completely to the 925X. So much so that many motherboard manufacturers put in unofficial support on the 925X, but i knew what was ahead. Sure enough.

Its not that motherboards are expensive, its just annoying having to remove such a key part of your computer. I hope they learn soon, otherwise i'll easily be switching over to M2 and riding that for a year or two.