• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

what is a celeron

R

rcf1987

2[H]4U
Joined
Aug 9, 2004
Messages
3,759
alright I heard from my uncle (whos a cable guy) that a celeron is a pentium that couldnt muster the requirements to be called a pentium 3,4 etc...
 
Sort of.

Basically, take a Pentium 4, drop the front side bus frequency, cut down on L2 cache, and hold back on the core clock frequency, and you get a Celeron. Let's compare the current generation of Celerons to the current Pentium 4s.

Clock frequency
Pentium 4: up to 3.6 GHz
Celeron D: up to 2.8 GHz

Front side bus frequency
Pentium 4: 200 MHz (800 MHz effective)
Celeron D: 133 MHz (533 MHz effective)

L2 cache size
Pentium 4: 1024 kB
Celeron D: 256 kB

These are the only differences. It's been speculated that Pentium 4s that "couldn't cut it" became Celerons. There might be truth to that, but it's also quite clear that Intel has the production capacity to manufacture Pentiums and Celerons side by side without issue.
 
Hey, they're good to put peanut butter on and...


oh, wait


that's celery.

They both have about the same processing power.
 
icthus13 said:
Celerons are also teh suck.
Click to expand...

And in the end - this is about as close to a working definition of a Celeron as you can get. While the information posted by xonik is totally true and very informative, if you can simply remember that Celerons suck and never to own one...you're one up on most of the average computer buyers in the country.
 
A Celeron is basically a lower-end P4/P3

Somebody overclocked a Celeron 2.0ghz I think it was to 3.0ghz and it barely kept up to a 1.6 P4A.
 
Celeron Ds aren't all that bad, but up until now, Celerons had always performed abysmally.
 
Actually, there've been some very good performing Celerons (when overclocked). The 300a and 600s were monster overclockers (most would go from a 66MHz FSB to 100MHz without a hitch).

The problem is that P4 core (NetBurst) requires 2 things to work effectively : a lot of cache & fast FSB. The Celeron formula is to run at a slower FSB & strip out the cache; when applied to a NetBurst core, this results in pretty poor performance.
 
Celeron's weren't always dogs, in the days of 370, they were cheap overclockers... kinda tells the length of time alot of these guys have been into the computer hardware thing.

one thing Xonik didn't mention, no HT for the Celery.

Rumors have said the Mobile Pentium 4's are from the center of the wafer, regular P4's are from just outside that, the Celeron's are on the ragged edge... which isn't really that ragged.
 
AFAIK the only time celerons were "pentiums that couldn't cut it" was the first P2/P3 cores. In the case of the P2 they were chips where some of the cache was actually bad, in the case of P3s, specifically coppermines, there were still some that had less working cache, and also some whose cache couldn't run at full speed (syncronous with CPU clk). As such if you were lucky enough to get a p3 celeron with full speed cache you could expect some incredibly sweet overclocking, the 300As and 600 coppermines were extremely nice as has been mentioned.

But all P4 based celerons are sepperately manufactured cores, and all current product families are manufactured completely sepperate, and most at entirely different fabs from one another. I've never heard anything of the middle/inner/outer track theory, afaik whole wafers are made of the same product, though similar patterns may be observed in speed bining, but with the current fab technologies you're looking at a much larger percentage of stable yield qualities, so even that occurance should be minimal.

Anyhow, just think of most celerons as a Pentium with one nut cut off, or in the case of northwood/prescot based celerons, complete vascectomy.

Remember kids, when a drug dealer offers you a Celeron, just say no.
 
The Pentium 2 never had any on die L2 cache, so there goes theory one.

As for Celeron's being on their own fab, I can't say for certain, but why would the Celery be on its own line?
The die size is the same, the transistors are all there for the L2 and Hyperthreading and such, just disabled. It would cost more to have a seperate fab for a Celeron than it would just to disable certain portions of the chip.
 
Yeah, the FIRST Celerons were simply Pentium II's with NO L2 cache- and they blew. The Celeron A's were the FIRST Pentiums with an on-die cache- the concept was later extended to the second generation Pentium 3's, but the first Pentium 3's were the same as the II's, with external cache.

For the Celeron D's: GO LOOK AT ANAND'S BENCH'S HERE!

You will find that, FOR THE PRICE, the new Celeron D's are not that bad; just keep them in perspective, as they are competing with the Athlon XP right now. I stand by Warrior's comment on the Celeron as a Pentium with at least one testacle missing; but I made it up. Consider the Celeron D as a Pentium 4 with just one missing however; the earlier P4 based Celerons were missing BOTH.
 
Warriorprophet said:
AFAIK the only time celerons were "pentiums that couldn't cut it" was the first P2/P3 cores. In the case of the P2 they were chips where some of the cache was actually bad, in the case of P3s, specifically coppermines, there were still some that had less working cache, and also some whose cache couldn't run at full speed (syncronous with CPU clk).
Click to expand...

Some of the Mendocino core PII's are no doubt Dixons (256k on die L2 mobile PII).with failed cache.
The same goes for Copermine128 P3 celerons. But SRAM either works or it doesn't, there's very little impact on speed in the fab process. (manufacturing can have an impact on static power dissipation, but SRAM is extremely power efficient in comparison to execution paths, up to 10 times less power for the same transistor count)


But all P4 based celerons are sepperately manufactured cores, and all current product families are manufactured completely sepperate, and most at entirely different fabs from one another. I've never heard anything of the middle/inner/outer track theory, afaik whole wafers are made of the same product, though similar patterns may be observed in speed bining, but with the current fab technologies you're looking at a much larger percentage of stable yield qualities, so even that occurance should be minimal.
Click to expand...

All new processes start with comperable failure rates, and as the process matures over the first 12 to 18 months all drop to comperable failure rates at a mature level as well.
90nm didn't start off with more or fewer failures than .25, .18, .13 or .13/300mm have, and it won't end up with more or fewer failures as it hits it's 'stride.'

There most likely were very few .18 P4's rebinned at Celerons because .18 was a very mature process when P4 was introduced.
There are .13 Northwoods that saw service as Celeron. Which is why some have 4 way set associative L2, others have 2 way.
Those with just 2 ass. blocks likely are P4s with more failures in multiple blocks than Intel budgets for, so it can't function as a P4. (and really it wouldn't surprise me if some of the 4 way ass. celerons has 1/2 the lines in each block disabled in addition to 1/2 the blocks from failed P4s)
So enough of the blocks are 'disabled' (6 of 8 in this case) to make it a celeron, make it useful.

One presumes we'll see something similar in some of the Celeron Ds.

This is absolutely not to say Intel doens't produce dedicated Celerons, they most certainly do. Relying on screwups to fuel CPU line isn't a very smart thing to do, but niether is throwing away a die because 9 lines of cache have errors and you only have 8 spares.
 
i had a celeron 433@590, 600@1035, and a 1.3@1.6. im considering getting one of the new celeron d's to thrown in a abit is7 just to toy around with.
 
FreiDOg said:
Some of the Mendocino core PII's are no doubt Dixons (256k on die L2 mobile PII).with failed cache.

****snip*****

This is absolutely not to say Intel doens't produce dedicated Celerons, they most certainly do. Relying on screwups to fuel CPU line isn't a very smart thing to do, but niether is throwing away a die because 9 lines of cache have errors and you only have 8 spares.
Click to expand...
I didn't think about the mobile P2's, completely forgot.

As for the Celeron with its own line, I never said only the bum P4s made it. My understanding was the Celeron was the outside of the wafer, more likely for failure and lower clock speeds, not necesarrily a bum chip. Similar situation to the Thortons vs Bartons using the same fab. Not all Thortons are broken Bartons. Some are 100% functional as a Barton, just snipped, relabeled and sold.

I may be wrong, but that was the way I heard it
 
Celerons have always been based off of the "P" core, they've never been their own core design. They started with the P2-based celeron, then made a P3-based celeron, P4-based celeron, etc. So you can't really just lump them all in the same boat, unless you just want to say "They ain't a P processor." The P2 ones with the overclocked FSB were basically identical in performance to their P2 older brother. Since cores change, they have found different ways to beat it down. I would say there is the biggest gap between the P4 and the P4 Celeron in terms of performance/whats been stripped down. BUT, the P4 celerons (Celeron D) are by no means slouches and I would happily recommend them to severe budget gamers that want an Intel chip, Mom and Dad's machine, etc. As for the Pentium M, its a whole different ball game and the only Celeron like it is (surprise) the Celeron M.
 
IdiotInCharge said:
The Celeron A's were the FIRST Pentiums with an on-die cache- the concept was later extended to the second generation Pentium 3's, but the first Pentium 3's were the same as the II's, with external cache.
Click to expand...

actually, the Pentium Pro was the first with on-die L2 cache.

anyway, Celerons are good chips for people on a strict budget, especially the new "D" chips.
 
0ldman said:
As for the Celeron with its own line, I never said only the bum P4s made it. My understanding was the Celeron was the outside of the wafer, more likely for failure and lower clock speeds, not necesarrily a bum chip. Similar situation to the Thortons vs Bartons using the same fab. Not all Thortons are broken Bartons. Some are 100% functional as a Barton, just snipped, relabeled and sold.

I may be wrong, but that was the way I heard it
Click to expand...

Right, the remark wasn't directed at you.
Just there was alot of talk about Celerons being failed P4s, and thought it was alteast worth mentioning that there are alot celerons that start life with the intention of being celerons.

I don't know about them just being the outside of the waffer, you generally don't see more than 1 die on a waffer, but i'm sure it's possible they start off life with the full cache and inteded as celerons.


felix88 said:
actually, the Pentium Pro was the first with on-die L2 cache.
Click to expand...

well, just to be a smart ass.
PPro used 'next to die cache.' Where the L2 was physically a seperate hunk of silicon from the core, and was attached to packaging next to the core.
http://www.sandpile.org/impl/pics/intel/p6/1m_open.jpg <- good picture of PPro 1mb cache.

Which makes sense, even the 256k chips, the cache was 75% of the transistors, one to many failures in that and you'd have to discard the core along with the cache if they were integrated together.
 
felix88 said:
actually, the Pentium Pro was the first with on-die L2 cache.

anyway, Celerons are good chips for people on a strict budget, especially the new "D" chips.
Click to expand...

It was my understanding that the L2 was On PACKAGE, not on die (may be wrong here though). Not something I care about researching; such chips are outside the realm of enthusiest computing.
 
Sorry but I hate Celerons, and now AMD made a crap processor as well the Sempron, they should of kept the bartons to compete with the celerons. I like intel for other inovations but the celeron, did they pick this out of there asses :D.
 
The Celeron is an incredible product for Intel; it is obviously mainstream/budget, and it has allowed them to penetrate that market and compete with AMD, while keeping the pentium, Xeon and Itanium lines unharmed from image contamination. Pentiums will always be the top of the mainstream, and Xeons will always be server/workstation parts. Such things are important for Chipzilla.
 
Bottom line, you can't have Corvettes without Corsicas... well, ok, you can, but something has to be in the place of the Corsica, Chevette, etc...

You need a value chip to help set the value of the performance chip and vice versa.
 
0ldman said:
Bottom line, you can't have Corvettes without Corsicas... well, ok, you can, but something has to be in the place of the Corsica, Chevette, etc...

You need a value chip to help set the value of the performance chip and vice versa.
Click to expand...

Well said sir.
 
Catsonar said:
Sorry but I hate Celerons, and now AMD made a crap processor as well the Sempron, they should of kept the bartons to compete with the celerons. I like intel for other inovations but the celeron, did they pick this out of there asses :D.
Click to expand...

What do you think the Semprons are? The socket A ones are simply AthlonXP's renamed.
 
Darth_Fluffy said:
512. But they need more oh-so-badly..
Click to expand...

Covington = 0 kb
Mendocino = 128kb
Coppermine = 128kb
Tualatin = 256kb
Willamette = 128kb
Northwood = 128kb
Prescott = 256kb
 
My media server/router/firewall PC has a Tualatin Celeron 1.4gHz, and it still outperforms my roommate's Athlon XP 1800 in most games, and that's with lowly PC133 ram. It also overclocks it to 1.75gHz(at which point it's probably compares to a 2200+). It never goes above 40C. It only recently started having trouble with games. It plays FarCry perfect, but doesn't do so well with Doom 3.

Not all celerons are crap. For the price I paid for that chip almost 4 years ago, all things considered, I'd say it's the best processor I've ever owned.

The P4 Celerons however, are not so good. As mentioned previously, they've been neutered in far to many ways. However, they're still a decent price.
 
pureevilmatt said:
My media server/router/firewall PC has a Tualatin Celeron 1.4gHz, and it still outperforms my roommate's Athlon XP 1800 in most games, and that's with lowly PC133 ram. It also overclocks it to 1.75gHz(at which point it's probably compares to a 2200+). It never goes above 40C. It only recently started having trouble with games. It plays FarCry perfect, but doesn't do so well with Doom 3.

Not all celerons are crap. For the price I paid for that chip almost 4 years ago, all things considered, I'd say it's the best processor I've ever owned.

The P4 Celerons however, are not so good. As mentioned previously, they've been neutered in far to many ways. However, they're still a decent price.
Click to expand...

He wasn't talking about Tualatans. He was talking about Celerons BASED ON Tualatans. Same for the other cores listed.
 
yea, and in the quote he said Tualatin Celeron.

Besides, the only difference between a Celeron and PIII on the Tualatin core was prefetch and fsb... Half the PIII's had 256L2 and 133MHz, the Celeries had 256L2 and 100MHz bus, PIII S had 512K.
 
All Tualatins had full speed, on die L2.

PII had half speed and the Celeron of that generation had on die and just about outran the P2 on occasion. The Celeron and matching Pentium processors have had full speed L2 on die since the PIII, aside from the brief PIII Katmai.
 
OldPueblo said:
Celerons have always been based off of the "P" core, they've never been their own core design. They started with the P2-based celeron, then made a P3-based celeron, P4-based celeron, etc. So you can't really just lump them all in the same boat, unless you just want to say "They ain't a P processor." The P2 ones with the overclocked FSB were basically identical in performance to their P2 older brother. Since cores change, they have found different ways to beat it down. I would say there is the biggest gap between the P4 and the P4 Celeron in terms of performance/whats been stripped down. BUT, the P4 celerons (Celeron D) are by no means slouches and I would happily recommend them to severe budget gamers that want an Intel chip, Mom and Dad's machine, etc. As for the Pentium M, its a whole different ball game and the only Celeron like it is (surprise) the Celeron M.
Click to expand...

are Celerons able to become P4's again, you know by modding or something?
 
Well I know the Tualatin Celerons were fux0ring wicked. I wanted one but for the price/performance an Sis745/AXP was hard to beat.

And the Mobile Celerons are supposed to be hardly slower than the Pentium Ms ....
 
OldPueblo said:
What do you think the Semprons are? The socket A ones are simply AthlonXP's renamed.
Click to expand...


Dude i knew that, but the rename cores have only 256k cache, which really sucks when compared to a barton
 
Catsonar said:
Dude i knew that, but the rename cores have only 256k cache, which really sucks when compared to a barton
Click to expand...
No, not at all. It's obvious that you didn't look at the benchmarks, because you'll only see a 5% difference at best. Usually, it's in the neighborhood of 1%.
 
All depends on the application. It wouldn't be hard to come up with a benchmark that 'shows' machines with more cache to be 100% faster.
 
You must log in or register to reply here.
Back
Top