Intel's marketing - No Celeron

[m25]

Just because you know that a long pipeline performs better with more cache in no way validates your point. In fact, you can find a LOT of articles where the celeron performs very well against the P4/PD. Why don't you start here? Again, not that it has anything to do with my OP, the smaller cache of the Celeron 4/D was offset by the fact that it could be better overclocked (a similar trait of the E6300/E6400, though again, not to the extent of the C2D)

Well, not exactly; While Celerons have always offered a decent office performance, most of the time, it's been just that. I ave tried it myself and it hurts like hell to OC a celeron to 2.67GHz and still have it perform like a 2.0GHz P4. The only great celerons ever (IMO) have been Tualatins.

NEVAR FORGET THE Celeron 300A!
Nope, I just said a decent celeron, the 300A was superb because it could be easily turned into a top performer, on the other hand, Tualatins were more Duron/Sempron like regarding performance.

 

[mitch074]

Celeron got a bad rap because it was, indeed, very often very crippled at factory settings. Those cripplings were:
- much less L2 cache
- toned down FSB.

Historically, the Celeron appeared as a L2 cache-less Pentium 2/300 (at the time, L2 cache was held outside the core), there was no Celeron for Pentium/586 and /MMX variants.

The Celeron 300A was a boon due to its VERY easy overclocking: a chip tailored for 500 MHz, with a multiplier set at 4.5, integrated L2 cache which, while 4 times smaller than its P2 counterpart, was much faster, made a 450 MHz perform slightly better in some cases than a much more expensive P2/450. But it was an exception (the 333A had a much lower overclock success due to it being much more borderline).

Later Celys (P3 cores with half of L2 disabled) were usually stuck with a 66 MHz FSB on cores that didn't allow an easy 50% overclock, and even worse, at a time where the FSB was switching to 133 MHz (the laptop versions were decent performers due to FSB at 100 MHz) - their crap reputation started then, moreover AMD CPUs were cheaper and more efficient at the time (the Duron for example could have its multiplier unlocked and FSB raised, and K7 cores didn't mind low L2 caches as much as they minded low latency RAM - feeding a high FSB and fast RAM to a Duron would reach performances very close to the same clocked Athlon).

When Intel switched to P4, the Celys left were designed to upgrade older Socket 370 mobos - they were nothing more than relabelled P3 with somewhat smaller L2 cache, but the new production process, correct L2 cache and 100 MHz FSB made them better overclockers, and the generally better performing P3 architecture made them much better than much pricier, higher clocked P4s (a 1.4 GHz Cely would trounce a 1.8 GHz P4).

For the P4 (which was a crapola architecture all the way), the Celys were even worse performers: the only things going the P4 way were its big L2 cache and memory throughput, and those being crippled on the P4 Celys, those chips were even worse than VIA's (they performed no better and sucked up more juice). Overclocking them usually just... failed, the low L2 cache failing to keep the deep pipeline fed, extra MHz were usually like giving marmalade to pigs.

 

[Crashman]

Well if my E6400 is a "CELERON" according to Whizzard9992.

Then it must be the BEST CELERON EVER (sorry for capitals) that intel has ever made!!!

No, the BEST CELERON EVER MADE was the 300A. The NEXT BEST was the 1.0 Tualatin.

 

[m25]

Well if my E6400 is a "CELERON" according to Whizzard9992.

Then it must be the BEST CELERON EVER (sorry for capitals) that intel has ever made!!!

No, the BEST CELERON EVER MADE was the 300A. The NEXT BEST was the 1.0 Tualatin.
...Then you've got 20-30 empty rows, then comes the CedarMill CeleronD series 😀

 

[Whizzard9992]

Thanks everyone for the quality replies. For those that I've disagreed with, (485438573945 or w/e your name is 😉), I'd rather not get into a back-and-forth online because really, no one ever wins, no matter who is right.

Thanks to JMecc and joefriday for the quality replies, and the other person who set me straight on Intel bringing back the celeron names on the roadmap (JoeFriday, you too).

I'll just say this: The Celerons were pentium chips that had faults in them; usually cache. If that was the case, they'd blow some fuses on the silicon to disable the bad cache, and to disable some features (usually speed-step and stuff like that).

The lower FSB was purely marketing, because if the chips couldn't handle the higer FSB, they couldn't be overclocked. We know this is not the case at all. The lack of features often didn't affect performance, except disabling a single core (as on the Celeron D) obviously does.

I'm sorry I flubbed up ONE paragraph where I didn't mention that the Celeron D had a die disabled.

To say a Celeron is a 'crippled P4' is accurate, for the most part, but performance-wise, you're only really taking a hit in the cache by going Celeron (Except Celeron D, before someone bites my head off again).

And, for the record, I did often state tha the celeron could compete with its full-cached counterpart when overclocked, not clock-for-clock which is more the case with the C2D. I wasn't trying to get into the intricacies of the Celeron, and I could honestly care less. That wasn't the topic.

All I was saying was this: Intel went from "Disable half the cache to Celeron" to "Disable half the cache to lower model." That plus marketing the NetBurst chips really turned things around for Intel, and I just thought some people might be able to take a step back and appreciate that Intel did more than just release a kick-ass chip. They actually made some good marketing decisions where they really failed before.

For everyone who wants to get into a Celeron debate; Celerons are pretty damn similar to the full-cache chips because they WERE full-cache chips. Again, some Celerons have NO faults, but the fuses are blown to meet demand. I was just stating an opinion that people hate celerons not because it's a celeron, but because just about every mainstram celeron MACHINE has substandard parts.

 

[dezonio2]

SUP, any one know if Intel might introduce a new processor under the Pentium 5 name? i like that name, "PENTIUM 5"...I'm sure it would make any AMD fan boy run and cry in a corner

L8er

Been there, done that. Pentium was code named P5, so that's the "Pentium 5" :-D


to the best of my knowledge intel has completly dropped pentium and celeron names for the future.... in the words of one of the regional marketing managers for intel, "pentium is dead, gone, you will never see another new processor labeled that way"

made me kind of sad 😀

 

[4745454b]

Thanks everyone for the quality replies. For those that I've disagreed with, (485438573945 or w/e your name is 😉), I'd rather not get into a back-and-forth online because really, no one ever wins, no matter who is right.

I agree. More so when the other person chooses to ignore all evidence. I know that celerons are disabled P4 chips. This is hardly news to anyone. I don't know how more clear I can try to refute your "argument" that celerons are good chips just paired with bad parts. Celerons are not good chips. They are steaming piles of crap. I've already told you why. Other people have agreed, and you still continue to say I'm wrong...

And, for the record, I did often state tha the celeron could compete with its full-cached counterpart when overclocked, not clock-for-clock which is more the case with the C2D. I wasn't trying to get into the intricacies of the Celeron, and I could honestly care less. That wasn't the topic.

It is on topic if your trying to explain WHY the chip is crap!!!!!!!! You said, celeron good, just paired with crap parts. That was the point I got out of your first post. I have spent way to much time trying to convince you, by showing the details of the chip, that celerons are crap. Again, you have either not read, or not understood my posts. Celerons are bad performers because of what Intel did to them while still in the factory. Slower cache + smaller bus = crap netburst chip. If I could draw the picture I would have...

All I was saying was this: Intel went from "Disable half the cache to Celeron" to "Disable half the cache to lower model." That plus marketing the NetBurst chips really turned things around for Intel, and I just thought some people might be able to take a step back and appreciate that Intel did more than just release a kick-ass chip. They actually made some good marketing decisions where they really failed before.

THE 6300/6400S ARE NOT CELERONS! Having half the cache does not automatically make a chip a celeron. Why do you refuse to listen to me? What parts of what I'm saying hurts you so much you shut down and stop paying attention? For some reason I feel that if someone else came in and told you this you would believe them. Several things have to happen inorder for a chip to be a celeron, half the cache is only one of them.

For everyone who wants to get into a Celeron debate; Celerons are pretty damn similar to the full-cache chips because they WERE full-cache chips. Again, some Celerons have NO faults, but the fuses are blown to meet demand. I was just stating an opinion that people hate celerons not because it's a celeron, but because just about every mainstram celeron MACHINE has substandard parts.

Again, incorrect. The machine might have substandard parts, but that includes the empty pipeline celeron. Lower bus limiting memory transfers, along with a smaller place to hold memory information = CRAPPY CHIP!

(you still haven't shown one idea why/how I'm wrong. I believe that you believe in your idea so much that you won't listento reason[/quote], and this should be finished. My final post this is then to you sir.)

 

[joefriday]

For light duty use and media encoding, Celerons offer tremendous value. They are only crappy chips if you decide to use them for applications in which they perform poorly. If that's the case, I can hardly see the reasoning in slighting the CPU when the problem obviously lies in the user's inability to select the proper hardware for the job.

And for Whizzard9992: Celeron Ds are not Pentium Ds with one core disabled, they are Prescott Pentium 4s. Officially, no defective Smithfields have ever been made into Celeron Ds.

 

[1Tanker]

For light duty use and media encoding, Celerons offer tremendous value. They are only crappy chips if you decide to use them for applications in which they perform poorly. If that's the case, I can hardly see the reasoning in slighting the CPU when the problem obviously lies in the user's inability to select the proper hardware for the job.

And for Whizzard9992: Celeron Ds are not Pentium Ds with one core disabled, they are Prescott Pentium 4s. Officially, no defective Smithfields have ever been made into Celeron Ds.

Don't know if you've seen this.....4GHz O/C. 8O Not quite benchable....but amazing just the same.

http://www.xtremesystems.org/forums/showthread.php?t=126315

 

[joefriday]

That's a nice thread. Those guys over at extreme go well above the average overclock. I like to think of their systems as Top Fuel dragsters. They're incredibly quick and built specifically to do one thing, but they aren't very practical at all, and just like Top Fuel dragsters, they'd probably blow up if they had to run any longer than what it takes to perform.

However, I've been hearing plenty of good things about the newer Prescott Celeron Ds. It seems that they are much more likely to hit 4GHz+ with just stock cooling and a voltage bump than the earlier releases. I would have absolutely no doubts about overclocking one of the Celeron D 326s to 3.8Ghz/800fsb and doing it on the stock fan. A 326 at 3.8GHz and a 356 at 5GHz were actually two CPU setups I was considering back in July for my Dell upgrade before settling on the undervolted D 805.

 

[m25]

Quite correct; Intel has always intentionally kept Celerons far from reaching Pentiums and this has almost always made their performance pure CRAP. They could have given the first coppermines a 100MHz FSB, the P4 Celerons a 256K L2 etc bu they just didn't want to do so and I don't know why, especially while during all this period, all benchmarks showed Durons and later Semprons, badly slaughtering the Celerons... At the end, it was the same sh!tty marketing that kept Netburst alive for so long; just two sides of the same coin :roll:

To joefriday:
You could have had better performance than a 3.8GHz 326 if you had sent an AM2 Sempron 2800+ to 2.4GHz. I have said it hundreds of times; in percentage, it's the same OC, however, the GHz crappy myth is still alive and healthy and most people get fascinated by the 4GHz barrier, however, nobody has managed do show a true performance benchmark of a 4GHz celeron vs say a 2.5-2.6 Sempron.

 

[4745454b]

Quite correct; Intel has always intentionally kept Celerons far from reaching Pentiums and this has almost always made their performance pure CRAP. They could have given the first coppermines a 100MHz FSB, the P4 Celerons a 256K L2 etc bu they just didn't want to do so and I don't know why, especially while during all this period, all benchmarks showed Durons and later Semprons, badly slaughtering the Celerons... At the end, it was the same sh!tty marketing that kept Netburst alive for so long; just two sides of the same coin :roll:

They didn't because the performance would be to close to the P3/P4. IIRC, the celerons went for about half of the P4s. If you could get ~80% of the performance (stock) by spending only 50% of the money, you'd be an idiot to buy the P4. At the time, Intel didn't have much to worry about AMD. Few people knew about them, and they still had ~20% market share. Intel can't afford that today, as many more people know of AMD, and their market share has been increasing.

 

[1Tanker]

That's a nice thread. Those guys over at extreme go well above the average overclock. I like to think of their systems as Top Fuel dragsters. They're incredibly quick and built specifically to do one thing, but they aren't very practical at all, and just like Top Fuel dragsters, they'd probably blow up if they had to run any longer than what it takes to perform.

However, I've been hearing plenty of good things about the newer Prescott Celeron Ds. It seems that they are much more likely to hit 4GHz+ with just stock cooling and a voltage bump than the earlier releases. I would have absolutely no doubts about overclocking one of the Celeron D 326s to 3.8Ghz/800fsb and doing it on the stock fan. A 326 at 3.8GHz and a 356 at 5GHz were actually two CPU setups I was considering back in July for my Dell upgrade before settling on the undervolted D 805.

Yeah, i know. Still interesting what they can get out of the Si. Guys in Strong-Man competitions are not your average weightlifter/bodybuilder either...but it's still incredible to see what they can lift/push...etc. 😀

 

[joefriday]

You do know that mobile Celerons on the Northwood core received 256KB L2 cache? Since they were manufactured for socket 478, they work on most socket 478 desktop boards, run cool, and outperform the Prescott Celeron Ds, barring applications that have SSE3 optimizations of course.

 

[joefriday]

I hear ya. There's plenty of "Wow!" factor over there.

 

[elbert]

Intel has replaced the Celerons price point with P4D's so really no need to have them. Intel also planed a C2D to replace Celerons which are the E4300 and E4400's comming the middle of 07.

 

[Criminal89]

This is all I have to say to Whizzard9992 :

"A man is his own easiest dupe, for what he wishes to be true he generally believes to be true."
-Demosthenes

Plain and simple... Netburst based Celerons are crap any way you slice them.

 

[m25]

You do know that mobile Celerons on the Northwood core received 256KB L2 cache? Since they were manufactured for socket 478, they work on most socket 478 desktop boards, run cool, and outperform the Prescott Celeron Ds, barring applications that have SSE3 optimizations of course.

To tell the truth; I didn't know it and it's pretty interesting; Northwood was still a decent core and had they sticked with it; it would have competed more decently against K8; would have reached the same stock 3.8GHz of Prescrap on 65nm and as you certainly know, Northwood had a higher performance/clock than prescott.
...And Northwood only had 512K L2; I just wonder what a 65nm Northwood with 2M L2 would perform like...

 

[joefriday]

To joefriday:
You could have had better performance than a 3.8GHz 326 if you had sent an AM2 Sempron 2800+ to 2.4GHz. I have said it hundreds of times; in percentage, it's the same OC, however, the GHz crappy myth is still alive and healthy and most people get fascinated by the 4GHz barrier, however, nobody has managed do show a true performance benchmark of a 4GHz celeron vs say a 2.5-2.6 Sempron.

True, but the point of the upgrade was to be Conroe compatible, hence the Intel-only options, and the fact that I get to reuse my pc2700 DDR ram was a major deciding factor as well. Besides, I use my CPU cycles for encoding work, which is something where even a 2.4Ghz Sempron has a tough time competing with a 3.8GHz Celeron D.

 

[cwj717]

Later Celys (P3 cores with half of L2 disabled) were usually stuck with a 66 MHz FSB on cores that didn't allow an easy 50% overclock, and even worse, at a time where the FSB was switching to 133 MHz (the laptop versions were decent performers due to FSB at 100 MHz)

This isn't completely correct... since starting with the 800MHz celeron, they had a 100MHz bus. It was the 766MHz and lower celerons that had a 66MHz bus. Although, this could be taken as an advantage... since this made it a good upgrade for some older systems that didn't support the higher FSB.

 

[m25]

You're probably true' however, gaming and professional software go to the Sempron and the worst thing is that I haven't run into a Celeron/CeleronD review newer than of 2003 :roll:

 

[cwj717]

I personally think the original comparison was fair. Some of you people need to calm down and stop nitpicking 😉. In my experience the only celerons that were complete crap, were the 128KB cache and 400MHz FSB celerons... they had terrible performance because of this. But the newer celerons are pretty good, if you overclock them and pair them with a dual channel DDR board.

 

[m25]

...not exactly; their deep pipeline hurts theirgaming and overall arithmetics and makes them terrible multitaskers even when OC-ed.

 

[cwj717]

...not exactly; their deep pipeline hurts theirgaming and overall arithmetics and makes them terrible multitaskers even when OC-ed.

Tell me something i don't know, lol. but really that applies to the P4s and PDs as well. I also wasn't comparing them on a clock for clock basis... more price/performance 😉 As for the "arithmetics" statement... i could provide some sandra and drystone whetstone comparisons for you, if you would like.

 

[heartview]

I'm just waiting for the Celeron 2 Extreme to come out...


Or not.