INTEL TO RELEASE A P4 CELERON
- Thread starter ZER0
- Start date
-
- Tags
- CPUs Processors
The initial P4 Celeron processors are based on the previous Willamette 180nm manufacturing process differing only in that it has 128KB Level 2 cache instead of 256KB in the original Pentium 4. Only 400MHz FSB, too. Later, Northwood-based 130nm P3 Celerons will appear with 256KB L2 cache.
The price I saw for 1.7 Celeron Pentium 4 was closer to $80 rather than $90. Not interested in the 1.8GHz version since only those who want to spend the least for P4 system will choose P4 Celeron.
1. They ARE Willamette-based Pentium 4. 180nm instead of 130nm. Closer to the 1st generation P4, not the current generation of Northwood 400 FSB or Northwood 533 FSB. So, about 10-20 percent overclockability?
2. Smaller L2 cache size. Could this mean a little more overclockability, too? I don't know but I hope so.
3. More recent and mature manufacturing step? Since Intel is already shifted its mainstream base to Northwood A and B, it can have the option to let the lower-end P4 Celeron run cooler.
Therefore, I expect around 20 percent overclockability for the 1.7GHz version. Those who will buy P4 Celerons will also buy 845G/845GL/845GLL or P4M266/P4M333 or SiS650/SiS651 motherboards. There will be very few (among them) that provide great overclocking like P4T-, BD-, or TH7-series boards.
2GHz (FSB 100 overclocked to 120) P4 with 128KB L2 for $70-90.
845GL or SiS650/651 for $60 - $100. (OEM prices are reported to have been quoted at as low as $55 according to DigiTimes.)
Searching for the true, the beautiful, and the eternal
The price I saw for 1.7 Celeron Pentium 4 was closer to $80 rather than $90. Not interested in the 1.8GHz version since only those who want to spend the least for P4 system will choose P4 Celeron.
1. They ARE Willamette-based Pentium 4. 180nm instead of 130nm. Closer to the 1st generation P4, not the current generation of Northwood 400 FSB or Northwood 533 FSB. So, about 10-20 percent overclockability?
2. Smaller L2 cache size. Could this mean a little more overclockability, too? I don't know but I hope so.
3. More recent and mature manufacturing step? Since Intel is already shifted its mainstream base to Northwood A and B, it can have the option to let the lower-end P4 Celeron run cooler.
Therefore, I expect around 20 percent overclockability for the 1.7GHz version. Those who will buy P4 Celerons will also buy 845G/845GL/845GLL or P4M266/P4M333 or SiS650/SiS651 motherboards. There will be very few (among them) that provide great overclocking like P4T-, BD-, or TH7-series boards.
2GHz (FSB 100 overclocked to 120) P4 with 128KB L2 for $70-90.
845GL or SiS650/651 for $60 - $100. (OEM prices are reported to have been quoted at as low as $55 according to DigiTimes.)
Searching for the true, the beautiful, and the eternal
Sadly, though, with 1/4th the cache as a NW and half the cache of a Willy, they are going to suck. In fact, I'd expect a 1.0 GHz P3 to whip the snot out of it, let alone any of the Duron line. And since it'll be budget, all those P4 celerys will be based on SDRAM, meaning it'll probably get trounced by a 486...
Another wonderful move in marketing. Even if they oc'd to 10GHz, I wouldn't pay a dime for them. The mini-cache alone would make performance gain from ramped clock speeds non-existant. Not even SSE2 support can save this POS.
-SammyBoy
Another wonderful move in marketing. Even if they oc'd to 10GHz, I wouldn't pay a dime for them. The mini-cache alone would make performance gain from ramped clock speeds non-existant. Not even SSE2 support can save this POS.
-SammyBoy
Yeah, everything I've heard confirms what you said.
The only reason for buying one would be to get a cheap processor and then upgrade that. But with the change to a 133(533) bus, that's not really possible.
<font color=blue>Hi mom!</font color=blue>
The only reason for buying one would be to get a cheap processor and then upgrade that. But with the change to a 133(533) bus, that's not really possible.
<font color=blue>Hi mom!</font color=blue>
Be reasonable.
Only two major differences between Willamette 256KB 400 FSB and Northwood 512KB 400 FSB.
1. 256KB more L2 cache memory.
2. Enhanced manufacturing process from 180nm to 130nm thus running cooler and a lot more overclockable.
How much performance difference is there between unoverclocked P4 1.8 and unoverclocked P4 1.8A? How much will there be between unoverclocked P4 1.8 and unoverclocked P4 1.8 Celeron? For most applications, Duron 850MHz processor is not that slow compared to Athlon Thunderbird 850MHz. Will those who wanna spend very little use applications that primarily require more L2 cache?
BTW, Intel has just updated its Pentium 4 page, too: http://www.intel.com/design/Pentium4/prodbref/index.htm#streaming
Searching for the true, the beautiful, and the eternal
Only two major differences between Willamette 256KB 400 FSB and Northwood 512KB 400 FSB.
1. 256KB more L2 cache memory.
2. Enhanced manufacturing process from 180nm to 130nm thus running cooler and a lot more overclockable.
How much performance difference is there between unoverclocked P4 1.8 and unoverclocked P4 1.8A? How much will there be between unoverclocked P4 1.8 and unoverclocked P4 1.8 Celeron? For most applications, Duron 850MHz processor is not that slow compared to Athlon Thunderbird 850MHz. Will those who wanna spend very little use applications that primarily require more L2 cache?
BTW, Intel has just updated its Pentium 4 page, too: http://www.intel.com/design/Pentium4/prodbref/index.htm#streaming
Searching for the true, the beautiful, and the eternal
Looking at it from the standpoint of the budget market, I'd have to say that this hopelessly castrated P4 will be unable to compete, performance-wise. Granted, we all know how little people care about performance when getting a budget system, but the fact remains that it will be even more horribly underpowered than the Willy was with RDRAM, and this will be a little brother of Willy with SDRAM (since the chances of anyone getting this system and DDR or RDRAM is nil).
It just makes my life harder, in that I am building budget systems for college student, where money is everything. I have had college students tell me that they want an Intel, not an AMD, and I'm forced to go against my better judgement and supply them with a cost-effective Intel-based system, which is hard to do when I know that for the same price I probably can supply them with a better performing AMD-based system. Now that they are moving away from from the P3 Celeron (L2 is 256KB) to this ill-concieved P4 Celeron (L2 is 128KB), my job is even more difficult. I know that the Willys up to 1.5GHz were getting challenged in MP3 encoding and DivX encoding (hobbies of many college students) by a Duron 1.2GHz, so a P4 Celeron 1.7GHz and 1.8GHz are going to be about that level or lower, and initially cost more ($63 for a 1.3 Duron, ~$80 for a Celery 1.7GHz). It just pains me to do that when I'm trying to give people the best updrage path and the best bang for their buck. I have no problem building a P3 Celeron system, as it's very competitive with the Duron, both in price and performance, but now Intel has removed itself from the budget line completely, at least for builders with a conscious.
-SammyBoy
It just makes my life harder, in that I am building budget systems for college student, where money is everything. I have had college students tell me that they want an Intel, not an AMD, and I'm forced to go against my better judgement and supply them with a cost-effective Intel-based system, which is hard to do when I know that for the same price I probably can supply them with a better performing AMD-based system. Now that they are moving away from from the P3 Celeron (L2 is 256KB) to this ill-concieved P4 Celeron (L2 is 128KB), my job is even more difficult. I know that the Willys up to 1.5GHz were getting challenged in MP3 encoding and DivX encoding (hobbies of many college students) by a Duron 1.2GHz, so a P4 Celeron 1.7GHz and 1.8GHz are going to be about that level or lower, and initially cost more ($63 for a 1.3 Duron, ~$80 for a Celery 1.7GHz). It just pains me to do that when I'm trying to give people the best updrage path and the best bang for their buck. I have no problem building a P3 Celeron system, as it's very competitive with the Duron, both in price and performance, but now Intel has removed itself from the budget line completely, at least for builders with a conscious.
-SammyBoy
eden
Champion
- Oct 3, 2001
- 26,549
- 0
- 50,780
You know what IMO? Put all budget line building of PCs (below 999$, or 799$ if US money matters), restricted to AMD. You should not always suffer more for people who will end up suffering with such a future cheap processor. Restrict budget line to AMD if it isn't a P3 Cellery, that way Durons will give you a more proud feel.
--
Thunderbirds in wintertime, Northwoods in summertime! :lol:
--
Thunderbirds in wintertime, Northwoods in summertime! :lol:
slvr_phoenix
Splendid
- Dec 31, 2007
- 6,223
- 1
- 25,780
Yeah, but the customer is always right. If they want a PoS, then they're going to get a PoS. If you refuse to build a Wilty (Celeron Willamette) then depending on just how much of a friend they are, they might storm off in a huff and go pick one up from someone else. (Or they might even pick one up at Radio Shaft/Worst Buy/Short Circuit City/etc.)
Plus if they go somewhere else, besides feeling generally bad for how screwed over you know they are about to get if they go OEM, you also won't be making any money. As much as I hate the concept of the Wilty, if they were adamant about getting one, I'd sell it to them with an added clause of "Don't say that I didn't warn you."
<pre><font color=orange><b>du hast den Sweater verkehrt an</b></font color=orange>
Oh my!</pre><p>
Plus if they go somewhere else, besides feeling generally bad for how screwed over you know they are about to get if they go OEM, you also won't be making any money. As much as I hate the concept of the Wilty, if they were adamant about getting one, I'd sell it to them with an added clause of "Don't say that I didn't warn you."
<pre><font color=orange><b>du hast den Sweater verkehrt an</b></font color=orange>
Oh my!</pre><p>
lhgpoobaa
Illustrious
- Dec 31, 2007
- 14,462
- 1
- 40,780
works for me LOL
im amazed at how suprised peeps are about this cpu commin out. ive known about it for at least 3 munfs!
and i bet they will keep the 400fsb for quite some time.
one good side though, when the p4 cellery moves to the 0.13 process u could stick your 400fsb cellery into a 533 fsb socket and get an extra 33% performance out of it.
doubt u could do that with the 0.18's though... they run too close to the 0.18 micron mhz limit.
Today on Toms: Trisexual hamsters, anal applications of peanut butter and Marrage councilling!
im amazed at how suprised peeps are about this cpu commin out. ive known about it for at least 3 munfs!
and i bet they will keep the 400fsb for quite some time.
one good side though, when the p4 cellery moves to the 0.13 process u could stick your 400fsb cellery into a 533 fsb socket and get an extra 33% performance out of it.
doubt u could do that with the 0.18's though... they run too close to the 0.18 micron mhz limit.
Today on Toms: Trisexual hamsters, anal applications of peanut butter and Marrage councilling!
slvr_phoenix
Splendid
- Dec 31, 2007
- 6,223
- 1
- 25,780
Wilty is an abbreviation/nickname for Willamette. It also however carries a derogitory tone (implying a feeble/starved/diseased state, as in this flower has wilted), which any Celeron version of a Willamette surely deserves.
Surely any Willamette with half the cache is starved at the very least, if not also capable of only feeble performance. And certainly the idea for using the Willamette core for a Celeron instead of the Northwood core was diseased.
So my opinion is that Wilty is a very good nickname for the P4 Celeron based on the Willamette core with half of Willy's cache.
<pre><font color=orange><b>du hast den Sweater verkehrt an</b></font color=orange>
Oh my!</pre><p>
Surely any Willamette with half the cache is starved at the very least, if not also capable of only feeble performance. And certainly the idea for using the Willamette core for a Celeron instead of the Northwood core was diseased.
So my opinion is that Wilty is a very good nickname for the P4 Celeron based on the Willamette core with half of Willy's cache.
<pre><font color=orange><b>du hast den Sweater verkehrt an</b></font color=orange>
Oh my!</pre><p>
slvr_phoenix
Splendid
- Dec 31, 2007
- 6,223
- 1
- 25,780
<font color=green>AMD</font color=green>
AXP = 133/266 MHz FSB, 256KB L2 3D Now! <i>and</i> SSE
Duron = 100/200 MHz FSB, 64KB L2, <i>only</i> 3D Now!
So the Duron has 1 step lower FSB and 1/4 of the L2 cache.
<font color=blue>Intel</font color=blue>
Northwood = 133/533 MHz FSB, 512KB L2
Wilty (Willamette Celeron) = 100/400MHz FSB, 128KB L2
So the Wilty has 1 step lower FSB and 1/4 of the L2 cache.
Looking at the exact specs, it seems pretty clear to me that AMD and Intel have virtually identical treatment of their budget processors. If you want to get technical, AMD's budget is even more stripped than Intel's, as AMD's budget doesn't support the SSE instruction set which their high-end processor line does. At least the Wilty supports all of the instruction sets that the Woody does.
<pre><font color=orange><b>du hast den Sweater verkehrt an</b></font color=orange>
Oh my!</pre><p>
slvr_phoenix
Splendid
- Dec 31, 2007
- 6,223
- 1
- 25,780
Well I'll be damned.
**looks around**
Yep. This is hell alright. A SCSI card that won't detect our APEX detector. A million different SCSI cables laying around and yet <i>none</i> that can connect an old external hard drive from our SGI box to the card to test that the card even works. Strangely enough, no jumper anywhere to connect the prongs on an internal SCSI hard drive to be a single device on the chain to test that the card even works. It's one of <i>those</i> days. We have a ton of hardware, and yet somehow, we can't get any of it to play nicely together just to test a single bloody SCSI card.
Damn it, I want to know if my program, Spikey, proves that the onboard CPU and RAM of the new SCSI card will keep the data stream from crashing when it loses synchronization because of a heavy CPU load!
After all, it isn't every day that you are actually told to write a program which intentionally spikes the CPU to 100% every 10 seconds. I want to know if this deranged program actually did something useful! Argh! To think that I might have given it a smiley face icon for nothing!
He he he he he.
Seriously though, back to the actual discusion instead of my insane ranting, I thought that the Duron did have SSE. I wasn't really sure though, so I checked AMD's website on the Duron:
<A HREF="http://www.amd.com/us-en/Processors/ProductInformation/0,,30_118_1200_1202^1938,00.html" target="_new">http://www.amd.com/us-en/Processors/ProductInformation/0,,30_118_1200_1202^1938,00.html</A>
They mentioned 3D Now! and MMX, but not SSE. AMD didn't mention SSE anywhere on their Duron product description, but yet they <i>did</i> on their AXP product description. (Not that I needed to look up the AXP since I already knew for certain that the AXP did support SSE, but what can I say, I wanted to make extra sure.) So I figured AMD must know better than me, and by them not mentioning SSE for the Duron, it did leave me wondering if I had just thought wrong or what. So in the end, I decided AMD must be right and made my statement accordingly to their data.
Still, I'll take your word for it and just admit that I'm wrong since I hadn't felt entirely sure about that statement in the first place anyway. Plus if both you and FatBurger endorse the Duron having SSE, then that's a pretty good endorsement. You both are two of the more 'seasoned' posters around here and often have pretty darn good data.
In the end though, really it has little bearing on my point anyway, that AMD treats their Duron <i>exactly</i> the same way that Intel is planning for the Wilty. So I still stand by my statement that AMD is <i>just</i> as guilty of hacking the performance their budget line as Intel.
To be honest, it disgusts me that either company would hack their 'budget' line <i>that</i> much just to keep it from performing as well as their professional line. But hey, you get what you pay for.
<pre><font color=orange><b>du hast den Sweater verkehrt an</b></font color=orange>
Oh my!</pre><p>
**looks around**
Yep. This is hell alright. A SCSI card that won't detect our APEX detector. A million different SCSI cables laying around and yet <i>none</i> that can connect an old external hard drive from our SGI box to the card to test that the card even works. Strangely enough, no jumper anywhere to connect the prongs on an internal SCSI hard drive to be a single device on the chain to test that the card even works. It's one of <i>those</i> days. We have a ton of hardware, and yet somehow, we can't get any of it to play nicely together just to test a single bloody SCSI card.
Damn it, I want to know if my program, Spikey, proves that the onboard CPU and RAM of the new SCSI card will keep the data stream from crashing when it loses synchronization because of a heavy CPU load!
After all, it isn't every day that you are actually told to write a program which intentionally spikes the CPU to 100% every 10 seconds. I want to know if this deranged program actually did something useful! Argh! To think that I might have given it a smiley face icon for nothing!
He he he he he.
Seriously though, back to the actual discusion instead of my insane ranting, I thought that the Duron did have SSE. I wasn't really sure though, so I checked AMD's website on the Duron:
<A HREF="http://www.amd.com/us-en/Processors/ProductInformation/0,,30_118_1200_1202^1938,00.html" target="_new">http://www.amd.com/us-en/Processors/ProductInformation/0,,30_118_1200_1202^1938,00.html</A>
They mentioned 3D Now! and MMX, but not SSE. AMD didn't mention SSE anywhere on their Duron product description, but yet they <i>did</i> on their AXP product description. (Not that I needed to look up the AXP since I already knew for certain that the AXP did support SSE, but what can I say, I wanted to make extra sure.) So I figured AMD must know better than me, and by them not mentioning SSE for the Duron, it did leave me wondering if I had just thought wrong or what. So in the end, I decided AMD must be right and made my statement accordingly to their data.
Still, I'll take your word for it and just admit that I'm wrong since I hadn't felt entirely sure about that statement in the first place anyway. Plus if both you and FatBurger endorse the Duron having SSE, then that's a pretty good endorsement. You both are two of the more 'seasoned' posters around here and often have pretty darn good data.
In the end though, really it has little bearing on my point anyway, that AMD treats their Duron <i>exactly</i> the same way that Intel is planning for the Wilty. So I still stand by my statement that AMD is <i>just</i> as guilty of hacking the performance their budget line as Intel.
To be honest, it disgusts me that either company would hack their 'budget' line <i>that</i> much just to keep it from performing as well as their professional line. But hey, you get what you pay for.
<pre><font color=orange><b>du hast den Sweater verkehrt an</b></font color=orange>
Oh my!</pre><p>
Read the first THG article about the 1.2 I think it was, the first Duron on the Morgan core. I remember the writer mentioning that SSE might be a reason to buy the Duron over the Athlon (since the XP wasn't out at that point).
<font color=blue>Hi mom!</font color=blue>
<font color=blue>Hi mom!</font color=blue>
Silver, you cannot compare raw numbers and expect it to be accurate, the 1/4th cache will hurt the p4 MUUUCH more than the duron, AND the lower fsb will hurt the p4 MUUUCH more than the duron, in fact I am very very confident in claiming that when the p4 celly comes out, its benchmarks will be SIGNIFIGANTLY slower compared to the northwood than the duron is to the axp.
(PS the duron DOES have sse, and hardware prefetch.)
<A HREF="http://www.amdzone.com/articleview.cfm?articleid=796" target="_new">http://www.amdzone.com/articleview.cfm?articleid=796</A>
"Physically you can see that the die is shaped differently. The actual size of the die is only 6 square millimeters more. The larger size is due to the 180,000 additional transistors the Morgan design requires.
So what is in this extra die space? AMD has added three features aimed at improving upon performance that will be familiar to you if you know of the Athlon 4 or Athlon MP. 3DNow! Professional, SSE, and data pre-fetch. 3DNow! Professional is 52 additional instructions tacked onto 3DNow! Enhanced. SSE is the SIMD that is currently included in the PIII. Data pre-fetch is defined by AMD as follows:"
If I wanted to get technical I would say your last point was wrong ;-).
The Cash Left In My Pocket,The BEST Benchmark 
Facebook
Twitter
Instagram