Cedar Mill based Celeron is an OC king

Funny how the Celerons now have the 512k L2 cache just like the Northwoods. I might look into one of those if they hit 5 Ghz ;).
 
vibe said:
LOL. It's 65nm. That cpu will cost more than two vince amd 3000's.
http://www.matbe.com/divers/zoom.php?img=000000024186.png

Any 90nm athlon64 @ 3000mhz core can do sub-28 seconds:
26sec: EIJI AMD Athlon64 3700+ 2200 CABGE 0512XPMW 1.67V 3030MHz (275MHz)

27sec: Opteron146 2000 CABYE 0540FPMW 1.55V 3035MHz (304MHz)

Yeah , but that Celeron is 120$ while your 3Ghz A64s are 150+ and have a slim chance of hitting 3. ( 2.6 99% of the time )
 
Yeah, 3 Ghz is a stretch on an A64 unless you have a good core. That Celeron supposedly did 5Ghz on the stock cooler :eek: . If that's true, there's room for improvement with good cooling.

Honestly, with 512k L2 cache, I might pick one up for myself just to play with.
 
Note it's probably an engineering sample too.
Production line CPUs rarely perform like the first (released for review) samples.
No way in heck Intel is gonna release a cpu for $120 that has 3.5 stock and 5.0 easy OC.
It would instantly kill the resale value of every Intel cpu sitting on every oem's shelf.
It cost them a fortune to retool to 65nm, they have to pass that cost along.

It does get me interested in 65nm though, imagine AMD on 65nm!
 
I hope Intel understands that if they have the heart of the enthusiast they will really start competing hard on AMD, and price is #1. I bought my Sempron because it was $65 and overclocked to a proc worth much more. Let's see if this Deleron can hit a price point where people like me would buy it.
 
vibe said:
No way in heck Intel is gonna release a cpu for $120 that has 3.5 stock and 5.0 easy OC.
It would instantly kill the resale value of every Intel cpu sitting on every oem's shelf.

How many people do you think overclock their cpus?
 
vibe said:
Note it's probably an engineering sample too.
Production line CPUs rarely perform like the first (released for review) samples.
No way in heck Intel is gonna release a cpu for $120 that has 3.5 stock and 5.0 easy OC.
It would instantly kill the resale value of every Intel cpu sitting on every oem's shelf.
It cost them a fortune to retool to 65nm, they have to pass that cost along.

It does get me interested in 65nm though, imagine AMD on 65nm!

Actually quite the contrary. Intel is just getting ready to crank up the 65nm production. So the process is not mature yet, after a few new steppings the overclocks usually do better.

The new Celeron D looks damn fine!! 512Kb of cache should make it perform well.
And at 5Ghz its at 800FSB. So these will be nice to play with.
 
Also note that 5Ghz was hit with an unlocked multi - wont be quite so easy to get that high without good ram and a good board, as the production line chips will almost certainly be locked.
And while it could do a 1M SuperPi without crashing, that doesnt mean that its stable. Leave it running prime for a day, and it'll probably error out.
 
RavenD said:
Also note that 5Ghz was hit with an unlocked multi - wont be quite so easy to get that high without good ram and a good board, as the production line chips will almost certainly be locked.
And while it could do a 1M SuperPi without crashing, that doesnt mean that its stable. Leave it running prime for a day, and it'll probably error out.

once again... quite the contrary.

its stock multiplier is 25 (it is an ES, but all newer Intel chips with EIST, have unlocked lower multipliers so they can drop down to the 14x multiplier when its not being used)

25 x 133 = 3.3Ghz one of the stock speeds of the new Celeron D (533FSB)

25 x 200 = 5Ghz (800FSB = stock FSB speed on Pentium 4's! )
 
RavenD said:
And while it could do a 1M SuperPi without crashing, that doesnt mean that its stable. Leave it running prime for a day, and it'll probably error out.

probably true - but the possibility of having another 300a story is just too sweet, especially as it is now taking place at ten times the speed :D

any info on when these will be out? haven't really looked into intel processors for a while to be honest, but at <130$ this will be a lot of fun ;)

edit: heck, the more i think about it it indeed is quite amazing. 50°C /w stock HSF and 5 GHz - holy shit! 10 of these would be one hell of a cheap renderfarm, really.
 
6Ghz maybe with good cooling? Makes me think of the old XP Barton days when CPU's were $80 :D.
 
hell, now i want one... just to play with

might have to hold off though... im buying a Yonah laptop when they come out.
 
wizzackr said:
but the possibility of having another 300a story is just too sweet, especially as it is now taking place at ten times the speed

That was the first thing that I thought of while reading the original post. I just hope that upon release these Celerons manage to be this good.
 
savantu said:
http://www.matbe.com/actualites/11983/intel-celeron-cedar-mill/

According to the test above a 5Ghz/800FSB Celeron 356 ( 3.33Ghz/533/512kb L2 ) managed to score the same time in Super Pi 1M ( 28s ) as a 3Ghz FX53.

All that with the stock intel cooler.Base freq full load temps ( burn in ) are 50C.

I must say I'm impressed and looking forward for Cedar Mill Pentium 4s....

Sweeeeetttttttt!

Donnie27
 
vibe said:
Note it's probably an engineering sample too.
Production line CPUs rarely perform like the first (released for review) samples.
No way in heck Intel is gonna release a cpu for $120 that has 3.5 stock and 5.0 easy OC.
It would instantly kill the resale value of every Intel cpu sitting on every oem's shelf.
It cost them a fortune to retool to 65nm, they have to pass that cost along.

It does get me interested in 65nm though, imagine AMD on 65nm!


AMD is on track for this summer for 65nm. I'm assuming they are getting IBM to help them with that also.

Ninja edit: Yep. http://www.amdboard.com/65nm_120605.html
 
Looks like 300a All Over Again :D :D(btw I am an amd Fan-Boy- but I would buy a cheapo intel motherboard and the celly chip just for its asomeness)
 
The super pi score is nice, though not earth shattering, you can get in the same range with an overclocked Venice, San Diego or Prescott 2M core today. Another budget oriented chip that overclocks its way into the performance arena is always welcome though.

The 1.456V Vcore is at the same time a bit worring, and encouraging.
Either we're going to see a really lousy votlage drop on the 65nm Netbursts, or that chip needed some major overvolting to hit 5ghz.

At the same time, if this was done on air cooling, that bodes very well for lower power and heat on the intel desktops.
 
This is nice... it figures Netburst would finally get its legs just as its being discontinued...

If Intel had kept it up I bet we would have seen 10ghz on 45 nm process.
 
FreiDOg said:
The super pi score is nice, though not earth shattering, you can get in the same range with an overclocked Venice, San Diego or Prescott 2M core today. Another budget oriented chip that overclocks its way into the performance arena is always welcome though.

The 1.456V Vcore is at the same time a bit worring, and encouraging.
Either we're going to see a really lousy votlage drop on the 65nm Netbursts, or that chip needed some major overvolting to hit 5ghz.

At the same time, if this was done on air cooling, that bodes very well for lower power and heat on the intel desktops.

Honestly I couldn't care less about SuperPi scores (this coming from a dothan user too). All that matters if it's cheap and fast. And I'm betting a Celeron D w/ 512KB L2, 800FSB, and a 5Ghz clock is pretty fast. BTW they didn't overvolt the CPU if you compare screenshots of the first and second (actually it's a bit lower in the second). I'd be interested to see how well the CPU undervolts.

I wonder too if intel may add HT support to some Cedar Mill Celeron D's? (wishful thinking I bet)
 
vibe said:
Note it's probably an engineering sample too.
Production line CPUs rarely perform like the first (released for review) samples.
No way in heck Intel is gonna release a cpu for $120 that has 3.5 stock and 5.0 easy OC.
It would instantly kill the resale value of every Intel cpu sitting on every oem's shelf.
It cost them a fortune to retool to 65nm, they have to pass that cost along.

It does get me interested in 65nm though, imagine AMD on 65nm!

Intel ES CPU's are usually overclocking like crap with FSB.... The only thing that saves them is the unlocked multi. The 200MHz FSB on the new celly will most likely be a given with the release core, and at $130 (retail price for "high-end" Celly's) it's gonna be a sweat deal.
Also, AMD on 65nm would be nice but according to this http://www.overclockers.com/tips00885/ , it's not gonna happen any time soon... All this, while Intel is (and has been) making 65nm CPU's on daily basis and sending them out to OEMs. The Dark side is in full-force comeback mode for 2006 and beyond... :eek:

Mike
 
mzs_biteme said:
The Dark side is in full-force comeback mode for 2006 and beyond... :eek:

no shit - and all there is in AMDs roadmaps to counter this is: a gaping hole. i am really anxious to see what they'll come up with against conroe
 
chrisf6969 said:
Actually quite the contrary. Intel is just getting ready to crank up the 65nm production. So the process is not mature yet, after a few new steppings the overclocks usually do better.
.....

Intel cranked up 65nm this summer and starting with early october it has gone full steam on 65nm.Intel's 65nm ramp was the fastest in company history and yields are at 90nm level right now which is simply incredible.

There are 2 fabs right now churning 300mm waffers with 65nm cpus and another 2 will join the party in early 2k6.
 
Let me get this straight. So Intel has had 65nm production above satisfactory levels for months now, but they're building up stock levels for the demand once its officially released to consumers? :)
 
perplex said:
Let me get this straight. So Intel has had 65nm production above satisfactory levels for months now, but they're building up stock levels for the demand once its officially released to consumers? :)

Actually they stampped out 65nm for Validation last year. Settled on the shipping models and then went into prodution this summer. The press releases are at their site. Chipsets and motherboards are the hold up or so say the rumors. An un-named beta tester told me, "Intel said they'd rather take a hit or get a black eye from yet another new chipset, than have the new processors perform poorly on an old one." It seems like they at least know they suck for doing this.

The 65nm Celeron should have all features except VT but that's just speculation on my part. My 2.6C served me well for almost 3 years, I have an AMD system to tide me over until something like this or Cedar Mill becomes commonplace. Being a Loyal _______ user is kind of silly IMHO. From everything I've seen online, AMD might not have a working 65nm Processor until early 2nd Quarter 2007.

The last thing I'd like to see them do is maybe clean up the latency on the L2. Going to 1MB from 512K caused a hit, and going from 1MB to 2MB caused yet another 17% hit.

Donnie27
 
vibe said:
LOL. It's 65nm. That cpu will cost more than two vince amd 3000's.
http://www.matbe.com/divers/zoom.php?img=000000024186.png

Why would it cost more? If its produced on 65nm, it would cost less to manufacture than a 90nm counterpart. Same thing for Yonah vs Dothan. A DualCore Yonah will cost less to produce than a single core Dothan. AFAIK, the 90nm Prescott costs $40 per chip. Anything above $40 can be construed as profit and money towards R&D/Q&A/Advertising/Packaging.

In any event, its probably going to retail around the $100 mark, just like previous high tier Celerons.

vibe said:
Note it's probably an engineering sample too.
Production line CPUs rarely perform like the first (released for review) samples.

Actually, on the contrary, pre-production ES's are generally poor performers compared to retail parts. The simple explanation is that pre-production ES's are manufactured on an immature process, and process technology and revisions only get better with time.

vibe said:
Note it's probably an engineering sample too.
No way in heck Intel is gonna release a cpu for $120 that has 3.5 stock and 5.0 easy OC.
It would instantly kill the resale value of every Intel cpu sitting on every oem's shelf.
It cost them a fortune to retool to 65nm, they have to pass that cost along.

Considering the target audience that buys Celerons are value oriented businesses and home PC's, it doesn't really matter if it clocks high. The cost to retool, R&D, Q&A are all factored into the pricing formula. The highest profit margins are on the high-end chips, though.

And also, Celerons are chips that have defective L2's (thus the disabling). If they were not sold as Celerons, they'd be thrown out because they couldn't make the grade as full fledged P4's.
 
well, given that Anandtech could overclock the Presler to 4.5 GHz on stock cooling, i'm not really supprised that the Celeron can hit 5 GHz. It's got an 8th of the cache, and is missing a core...

also, with the 65nm process, it is stable and mature enough to run multiple chip generations on it. According to tomshardware, Intel will be using the 65nm processor for a lot of processors with the last one being a quad core processor. If Intel can do that well on the 65nm process, you know that that they've spent a good few billion researching it...

also, my friend works for Unisys who works heavily with Intel. (they supply servers for dell so they kinda have to be close with Intel. Anyway, he was saying that Intel does not really care as mu ch about the enthusiast market then they do the server market. Some of the high end servers cost at least half a million dollars as oppoed to 1k for the enthusiast market although they seem to be moving toward the direction of satisifying enthusiasts.
 
Duke3d87 said:
well, given that Anandtech could overclock the Presler to 4.5 GHz on stock cooling, i'm not really supprised that the Celeron can hit 5 GHz. It's got an 8th of the cache, and is missing a core...

also, with the 65nm process, it is stable and mature enough to run multiple chip generations on it. According to tomshardware, Intel will be using the 65nm processor for a lot of processors with the last one being a quad core processor. If Intel can do that well on the 65nm process, you know that that they've spent a good few billion researching it...

also, my friend works for Unisys who works heavily with Intel. (they supply servers for dell so they kinda have to be close with Intel. Anyway, he was saying that Intel does not really care as mu ch about the enthusiast market then they do the server market. Some of the high end servers cost at least half a million dollars as oppoed to 1k for the enthusiast market although they seem to be moving toward the direction of satisifying enthusiasts.

But it seems AMD are increasing market share every year in the server market?
 
But it seems AMD are increasing market share every year in the server market?

That's because AMD has a better server processor right now. That could all change within one year, depending on AMD's execution. If history is any indicator of AMD's execution, they will drop the ball as usual.
 
perplex said:
Let me get this straight. So Intel has had 65nm production above satisfactory levels for months now, but they're building up stock levels for the demand once its officially released to consumers? :)

Intel's inventory went up 250 million $ in Q3 , altough according to Andy Bryant ( Intel CFO ) they couldn't meet supply and their fabs were running full steam for the past 9 months. ( Intel has 16 fabs , 5 ( or 6 now ;) ) of which are 300 mm and 8 200mm ).

That extra inventory could be 65nm production stockpilled for the December 27th launch.

Mind you , that is about 2 million cpus so by the release date if you add another 2 months (nov and dec ) they probably have 5 million ready 65nm cpus.

Btw , it is estimated that Intel amd AMD produced around 15 million dual cores in 2k5 , with Intel having around 13 of that....

As Andy said "it's always better to have excess inventory than not being able to respect contracts "
 
A Pentium D 950 @ 4.8GHz on air speaks for itself. 65nm is the real deal, and these 65nm chips should keep us happy until Conroe arrives.
 
It makes sense that Intel might be wanting to sell off excess 90nm inventory before releasing the 65nm parts as well. Once the 65nm processors hit the streets no one in their right mind would buy one of the 90nm ones unless there are some major major price cuts.
 
Back
Top