What's left for LGA775?

S

sethk

2[H]4U
Joined
May 3, 2005
Messages
2,133
So we already know that the next chipset from intel, the X58 will be for Nehalem. There will be Nehalem introductions in the second half of this year, but clearly, LGA775 is not dead yet.

I'm wondering:
a) What new LGA775 CPUs will we see between now and when LGA775 goes EOL (end of life)?
b) What new chipsets will we see from Intel and others before this platform is no longer enthusiast worthy?
 
colinstu said:
http://en.wikipedia.org/wiki/List_of_future_Intel_Core_2_microprocessors
http://en.wikipedia.org/wiki/List_of_future_Intel_Celeron_microprocessors

Not much at all, if you want the best LGA775 chip, either the QX9650 or QX9770.

No new highend 775 chipsets that I am aware of either.
Click to expand...

However, for that same reason, if you're looking LGA775 (which still makes sense for price and bang-for-buck reasons) on the desktop, think quad-core, not dual-core.

Why quad-core? Simple: it still comes down to *capital preservation*, and quad-core lets you go further before upgrades become necessary (unless you are banking on overclocking above 3.6 GHz). Nehalem intros will by and large be on the high-end (not the mainstream where most of the buying action is).
 
I agree, quad core makes sense at this point, but I'd really like to crack 4ghz on air without crazy voltages on a new build, and I don't see it happening with the Q9450, plus even if that was possible, the FSB walls on most motherboards with quads is going to be hit. Maybe a Q9550 and a new stepping along the lines of the B3->G0 respin except for the 45nm generation, and a more mature P45 board/bios might make it possible.

Otherwise I'll just go cheap with a Q6600 and P45 or X48 and take a low to mid 3ghz range overclock.

I'd consider a QX9650 but it seems late in the game for a $1000 Penryn CPU - even one with an unlocked multiplier. I'd rather wait for a Bloomfield and see what those $s buy me, it will probably be good for a couple of years.
 
sethk said:
I agree, quad core makes sense at this point, but I'd really like to crack 4ghz on air without crazy voltages on a new build, and I don't see it happening with the Q9450, plus even if that was possible, the FSB walls on most motherboards with quads is going to be hit. Maybe a Q9550 and a new stepping along the lines of the B3->G0 respin except for the 45nm generation, and a more mature P45 board/bios might make it possible.

Otherwise I'll just go cheap with a Q6600 and P45 or X48 and take a low to mid 3ghz range overclock.

I'd consider a QX9650 but it seems late in the game for a $1000 Penryn CPU - even one with an unlocked multiplier. I'd rather wait for a Bloomfield and see what those $s buy me, it will probably be good for a couple of years.
Click to expand...

Instead, I'd go for the less-expensive (even today) Q9450 if you're going to go Penryn at all in the short term; however, given the versatility of HD4850/4870 for graphics (and especially at these unheard-of price points), I wouldn't exactly kick the still-available Kentsfield (Q6600) to the curb either, especially if you can find a decent P45-based motherboard (P45 is a solid midrange chipset that can swallow both flavors of quads without locking you into DDR3, which is still expensive; this is also why I'm not that excited about X48).
 
You can sometimes find the QX9650 on ebay for $400-800, just imagine what'd they cost when nehalems start rolling out. Plus they could've been OC'd crazy, but hey - it's cheaper :D
 
Theres the 6 core chips (dunnigton?) coming out, but I think those are for s604 (quad socket), not s775. For s775? Nothing is coming out, AFAIK.
 
Those 6 core chips aren't even for socket 771, they are for socket 604! (They can only run in quad socket 604 systems).
 
Frankly, it doesn't matter. The CPU has reached the point where it's significantly eclipsed it's performance profile. Games and software aren't going to be CPU driven. The next big advancement will be something like NVidia's GPGPU system, where the computer can harness the power of the graphics card as well as the CPU. To be honest, I'll probably skip Nehalem and wait for the next next big thing. By then, there might be true multithreaded software and operating systems to take advantage of the hardware.
 
I wouldn't discount Nehalem just because of nVidia's activities. GPGPU, if it's succesful, probably won't see real mainstream action until Nehalem has been phased out, which means the performance you get from a Nehalem configuration could still have utility for as long as it's current. More tantalizing still is if GPGPU is an immediate smash hit; Nehalem + GPGPU could = the one-two punch to end all short-term hardware longevity issues. Especially near its introduction, when the prices are extremely competitive. If things really go the way you seem to foresee, hardware prices will rise to compensate for the fact that everyday people don't need to upgrade as frequently as they used to. That is, of course, unless software developers take advantage of more's law skipping a beat and churn out high-demand, high performance programs to continue the symbiotic relationship of hardware and software developers. Still, though, we don't even have an efficient way to have programs dynamically handle any given number of CPU cores, so the thought of having an efficient and effective way of coding programs for an unknown number of CPU cores along with an unknown number of GPU's is still unlikely at least for the next couple of hardware cycles. That is unless nVidia has some sneaky way of integrating GPU processing into the normal CPU functions; to be honest I haven't read up on the subject nearly as much as I probably should have by now.

I think that counts as a ramble.
 
What's with the "CPU doesn't matter" crowd? If you think CPUs don't matter, why post in a n intel CPU forum? Stop thread-crapping. I care about CPU speed - yes I am very interesting in GPUs and GPGPU, but I believe it will be a LONG LONG time before GPGPU displaces the CPU in importance and I have a lot of apps that I'd like to run fast in the meanwhile that actually require the x86 / x86-64 instruction set.

To get back on topic, I think the Q9650 is an interesting chip with a 9x multiplier. The used QX9650 path is also an interesting option at the right price, but I'm afraid of getting an abused chip with a very short lifespan. To be honest I'm really surprised at the low, low projected price on the 2.6ghz Bloomfield chip - I wonder if only the "Extreme" chip will allow overclocking, that is the only thing that explains the low price. Otherwise, almost all new sales in the ~200-300 enthusiast price range will start going to that chip. If it is overclockable, it would probably bleed a lot of sales from the Penryn quads, since there would be less of an incentive to purchase Penryns unless people are locked into their current mobos. Intel must be very confident in their ability to produce Bloomfield chips with good yeilds, which would be an excellent sign for us.

I guess I'll pick up Q6600 or Q9450 to tide me over for the next few months and save the killer rig for Bloomfield.
 
Indeed. a 3.6-4 Ghz quad will tide me over till Nehalem/Bloomfield becomes more affordable. Hell i'll only jump on the new chipset when DDR3 don't have "RAPE" all over them, along with CPU prices around 200-300 for an octa core.. which hopefully will be in 1.5-2 yrs
 
Yeah DDR3 is still an open question for me. 4GB worth of decent CAS 5 1066 DDR2 is still ~$100 on newegg, whereas a good deal on say 4GB worth of CAS 7 1333Mhz DDR3 is ~$200, so 2x the price. That's better than the 4x differential we saw not too long ago (actually a lot of DDR3 is still that much more expensive) but I bet the difference will shrink considerably as higher grades of DDR3 are released and manufacturing ramps up further towards the end of the year.
 
I am still going to build a system later on with a E8400. Hardware moves to fast till the point where i dont care about the best. I just want a system thats good at gaming and can run pretty much all games on high, and yes you dont need the most expensive stuff to do that.
 
Bloomfield is going to be massively expensive.
I'm going to sell my Striker and 9650 when they come out ,but I highly doubt I'll get little return as colinstu stated. These chips don't depreciate fast, and not everyone can afford a 1500 dollar Bloomfield, since Intel has STATED that the Extreme version of Nehalem is due out this year and the server parts; the q9450-esque nehalems aren't due out till next year!
 
It would fit that they release nehalem to the pampered 4-5 months early like they did with the penryns last year.
I mean let us enjoy our purchase :(
 
The problem with that is, unlike Penryn-supporting X38 and P35 chipsets, which had the luxury of compatibility with the existing 65nm duals and quads (at launch), the new X58 chipset doesnt have a sea of older CPUs it supports.

Intel won't be able to ramp X58 production if only a few people buy a $999 cpu and the chipset can't be used with the last generation. They need to start selling the platform, the socket and of course the chipset. If the ONLY chip that uses these boards costs >= $999, lets just say that third party boards would be... scarce.

Brand new sockets are tricky and require good pricing to allow for rapid proliferation. This is why I believe that there may actually be some semblance of a mainstream right off the bat.
 
sethk said:
The problem with that is, unlike Penryn-supporting X38 and P35 chipsets, which had the luxury of compatibility with the existing 65nm duals and quads (at launch), the new X58 chipset doesnt have a sea of older CPUs it supports.

Intel won't be able to ramp X58 production if only a few people buy a $999 cpu and the chipset can't be used with the last generation. They need to start selling the platform, the socket and of course the chipset. If the ONLY chip that uses these boards costs >= $999, lets just say that third party boards would be... scarce.

Brand new sockets are tricky and require good pricing to allow for rapid proliferation. This is why I believe that there may actually be some semblance of a mainstream right off the bat.
Click to expand...

Something tells me the 2.66GHz Bloomfield CPU won't cost $999. Assuming that the performance increase from Penryn to Bloomfield is reasonable, intel will have plenty of buyers...
 
whats left for lga775? .... only taking over the world ... that pretty mcuh all they havn't done
 
You must log in or register to reply here.
Back
Top