Socket 775 just won't die... Celeron E3900 on deck

J

Joe Average

Ad Blocker - Banned
Joined
Apr 6, 2008
Messages
15,459
http://www.techpowerup.com/102396/Intel_Plans_3.40_GHz_Celeron_E3900_for_2010.html

Hadn't seen this one mentioned, and considering how cheap Celerons typically end up, this could be a stupendous processor for HTPCs as long as the HTPC sticks to that task I suppose, and not attempt to become a monster gaming machine, etc.

Will this be the last hurrah for Celerons, and for Socket 775 CPUs? Only time will tell...
 
I just wonder why Intel can't/won't produce a 4ghz chip. 333X12 = 4ghz. They pretend they cant make it, we pretend we dont overclock them to 4ghz. This will be 4.525ghz at 267fsb. Nice!!
 
gwarren007 said:
I just wonder why Intel can't/won't produce a 4ghz chip. 333X12 = 4ghz. They pretend they cant make it, we pretend we dont overclock them to 4ghz. This will be 4.525ghz at 267fsb. Nice!!
Click to expand...

Because Intel can and will damnit.
 
They probably don't care enough to improve or push the binning process, and why would they, even before i7 came out they had 0 reasons to (from a practical business standpoint)... 775 will probably last thru the end of the year as a budget alternative, I imagine after that there'd be enough budget i5 parts around to stick a fork in it... 'Specially given all the efficiency advantages on laptops and such.
 
Why are saying "s775 won't die?" Of course it's not going to die any time soon, there is nothing to replace it for the lowend/mainstream.
 
Thats_Fantastic said:
Well, actually very soon the i5 and maybe i3
Click to expand...

until those solutions can get down to 299 and 399 for a complete system, intel will keep stamping out inexpensive 775 sockets and cpus. if this happens by the end of the year, I would be shocked.
Posted via [H] Mobile Device
 
The topic was meant to incite interest in reading about the Celeron E3900... geez people, go out for a walk or something, don't take everything quite so seriously. :)
 
I dunno, I remember s754 dying pretty quick after s939 was out... Never had any dual-cores did it? And even s939 died relatively quickly compared to s775's lifespan imo (due in part to the transition from AGP to PCI-E)... It did have it's share of X2s and whatnot tho, but getting a decent GPU upgrade was a PITA after a while.

I think S775 will carry on for a while with exactly this kinda part tho... For those budget $350-400 desktop systems you see everywhere. It'll be a couple of months before they can have an entire i5 system (not just mobo/CPU/RAM) for that kinda money, as a poster above mentioned. It'll be really close tho... I mean, Atom-based systems border on $300, and Intel's gonna have the remaining s775 lineup pinched in between $350 and $500 basically.

That's on the desktop end tho, and we all know laptops make the majority of sales these days... I have no idea how the price scales over there but I imagine the transition will be lower still unless manufacturers and consumers are both really excited about some of the i5 advantages in efficiency/performance.
 
Impulse said:
I dunno, I remember s754 dying pretty quick after s939 was out... Never had any dual-cores did it? And even s939 died relatively quickly compared to s775's lifespan imo (due in part to the transition from AGP to PCI-E)... It did have it's share of X2s and whatnot tho, but getting a decent GPU upgrade was a PITA after a while.
Click to expand...

Apparently you never used an Nforce 4 board with its 16x pci-e? I had a skt 939 Opty 165 running on an nforce 4 with an 8800GTX until about a year ago. The thing was a beast and overclocked to 2.8ghz handled most everything pretty well. On the flip side it used more power than my Q6600:eek:

One difference between the 775 death and the 939 death is that it doesn't look like we'll see any cpu price gouging. When 939 was giving the death rattle people were paying 80 to 100 bucks for an x2 that cost $40 the year before.
 
No I never used a s939 w/PCI-E, obviously I even forgot they existed... But by the time S939 boards w/PCI-E came out AM2 boards were also out no? Anyone that bought s939 when it actually launched (or close to it) bought a strictly PCI/AGP mobo, KT800/NF3, 'least that's what remember, maybe I'm mistaken. I just know I didn't bite on s754 in order to wait for s939 then got screwed by the move to PCI-E. /shrug
 
Impulse said:
No I never used a s939 w/PCI-E, obviously I even forgot they existed... But by the time S939 boards w/PCI-E came out AM2 boards were also out no? Anyone that bought s939 when it actually launched (or close to it) bought a strictly PCI/AGP mobo, KT800/NF3, 'least that's what remember, maybe I'm mistaken. I just know I didn't bite on s754 in order to wait for s939 then got screwed by the move to PCI-E. /shrug
Click to expand...

no they weren't. When I built my s939 nforce4 system it was the best besides s940.

still have it just never use it.
 
gwarren007 said:
I just wonder why Intel can't/won't produce a 4ghz chip. 333X12 = 4ghz.
Click to expand...
Because it's difficult to make a 4GHz chip that fits within a reasonable thermal envelope. Intel would have to raise the TDP of a chip like that, meaning they'd have to include a more substantial stock heatsink as well. There would also be a relatively small number of dies that are actually capable of running at that speed while still remaining within the VID spec.
Impulse said:
No I never used a s939 w/PCI-E, obviously I even forgot they existed... But by the time S939 boards w/PCI-E came out AM2 boards were also out no? Anyone that bought s939 when it actually launched (or close to it) bought a strictly PCI/AGP mobo, KT800/NF3, 'least that's what remember, maybe I'm mistaken. I just know I didn't bite on s754 in order to wait for s939 then got screwed by the move to PCI-E. /shrug
Click to expand...
S939 was a PCI-E platform. There were few AGP boards for it, and most used legacy chipsets or ones made by VIA.
 
Zero82z said:
Because it's difficult to make a 4GHz chip that fits within a reasonable thermal envelope. Intel would have to raise the TDP of a chip like that, meaning they'd have to include a more substantial stock heatsink as well. There would also be a relatively small number of dies that are actually capable of running at that speed while still remaining within the VID spec.

S939 was a PCI-E platform. There were few AGP boards for it, and most used legacy chipsets or ones made by VIA.
Click to expand...

Yup nforce4 featured socket 939 + PCIE + sata.
 
Zero82z said:
Because it's difficult to make a 4GHz chip that fits within a reasonable thermal envelope. Intel would have to raise the TDP of a chip like that, meaning they'd have to include a more substantial stock heatsink as well. There would also be a relatively small number of dies that are actually capable of running at that speed while still remaining within the VID spec.
Click to expand...
Good points. That is a reasonable explanation for not taking the C2D to 4GHz. I think a better question is, not why not 4Ghz, but rather, why not more than 3.3 GHz? 3.5 to 3.8 is easily reachable. I think it's because they've moved on to the next architecture for the top performing product tier. Much the same way the Pentium3 was not advanced when the Pentium4 came out, despite the fact the Tualatin core demonstrated the older architecture could outpeform it. I'm not suggesting a C2D (or even a C2Q) can outperform an i7, but at approaching 4ghz, they can certainly be competitive at all but bandwidth intensive situations. If X is your top performing product, for top dollar, you kind of have to maintain that performance gap between it, and the next cheaper thing, or you're not going to sell much X.
 
Xaero_toast said:
I think a better question is, not why not 4Ghz, but rather, why not more than 3.3 GHz? 3.5 to 3.8 is easily reachable.
Click to expand...
For pretty much the exact same reasons. The fact is that only the best dies are capable of reaching those kinds of speeds on stock voltages, and that means that it would be expensive to produce those kinds of chips, and with the newer i5 and i7 chips available, people wouldn't be buying the older C2D and C2Q CPUs, even if they are clocked very high.
Xaero_toast said:
Much the same way the Pentium3 was not advanced when the Pentium4 came out, despite the fact the Tualatin core demonstrated the older architecture could outpeform it.
Click to expand...
There are a number of reasons why Intel transitioned to the Netburst architecture, but the major one was clockspeeds. The fastest PIII produced was the 1.4GHz Tualatin, and that CPU was right up at the limit of what the architecture was capable of. Even overclockers weren't able to get PIIIs much higher than that without extreme cooling. Clock for clock Pentium III CPUs were faster than P4s, but that means very little when comparing a 1.4GHz PIII against a P4 running 1GHz faster. Intel eventually did bring back the PIII in a heavily-modified form as the Pentium M, but it took major changes to make that happen, as well as a die shrink before they could start to reach moderately high clockspeeds (the fastest 130nm Pentium M ran at only 1.7GHz).

The situation right now is pretty much the opposite. Intel has an architecture that performs better clock for clock than the C2D architecture, and doesn't have that significant of a clock speed deficit. Faced with that, they have no real incentive to work at improving their old architecture and ramp it to higher clock speeds when they can just focus on their new products instead and obtain better results.
 
You must log in or register to reply here.
Back
Top