1. Ah! The missing 2.93GHz part I complained about in a previous post has appeared as a lovely "2.93Gz", errr...
2. ---Sorry, the rest of this post includes opinion, paranoia, speculation and a rant---
3. I hate to say this, but this looks a lot like a 'confusion tactic' just for Intel to protect their sales until they have a chance to catch up with AMD in performance again. Yes, the numbers are vaguely in line with increasing performance, but notice most of the families are near the top of their ranges (of 100) already. An optimist would say they've got something new in the labs which will displace the current generation of stuff, and they're just giving the old stuff these numbers to shift it.
4. Regardless, can anyone tell my why this isn't stupid?:
What ARE they playing at? Surely one of them could at least be labelled '351' or '350A' to stop people having to figure out why two identically-named parts have totally different performance characteristics?!
I've always liked BMW's naming system and I think Intel could have done worse than this.
It clearly separates the different classes, which is the part that I like, but inside those classes is a preety big mess.
I'd have to agree that Intel needs to drop a few products if it really wants to dumb things down, because this is gonna get preety hard to decipher, even for us. Almost as bad as it is now with sSpec numbers.
I guess the price will be the last part of the puzzle that will help people decide what PCU to take home.
Enough whining already. No, really, stop it. It seems like almost all of the comments on this move by Intel have been a bunch of babys crying because somebody took away their empty bottle and gave them a different one. This is not odd, nonsensical, crap, etc. It makes perfect sense. Intel has ridden the more MHz (or GHz) equals a faster processor myth for far too long now. I work for a major computer retailer and trust me this makes perfect sense.
Put yourself in my shoes for a moment. Say your grandmother walks into the store (assuming she's not already dead and knows about as much about computers as my grandmother), how are going to quickly and easily explain to her that a P4 2.6 GHz is 10-25% faster than a Celeron 2.6 GHz processor. Now imagine that instead of someone like your grandmother who will take you at your word when it comes to anything computer related it is instead a complete stranger. This has not hurt Intel too much in the past since the prevailing thought among the general public is that there is something seriously wrong with Celeron processors, but now they have thrown the Pentium M into the mix. Try telling someone you've never met who knows nothing about computers, except that more MHz/GHz = a faster machine, that a 1.6 GHz Pentium M is as fast or faster than a 50% higher clocked 2.4 GHz P4 and that you think it is worth their money to pay top dollar for the thing too. Good luck.
So, quit your whining. This is not a bad/dumb move on Intel's part. Your not seeing it for what it is. It is an intentional dumbing down. It is not for you, the enthusiast. Intel knows good and well that when you, the enthusiast, go to buy a P4, PM, or recommend a Celeron for a friend that you are going to still be looking at clock speed, bus speed, and cache. This move is for the 90% of people who are the "everybody else." Where the real money is. The people who don't know what a Gigahert is and quite frankly don't care.
This is just fscking odd... BTW, if the Pentium Ms are being labelled this way, it could mean desktop P-M (it, and the AMD K8 will kill NetBurst, IMNSHO (not so humble)). BTW, where are Celeron-M numbers?
This makes me want to find some CPU where they don't give you this crap, though. Intel CPUs (except for XScales, which aren't x86) are now all screwed up, AMD uses performance ratings (why not logos saying THIS AMD PROCESSOR BEAT THIS INTEL PROCESSOR IN THIS MAGAZINE/SITE'S REVIEW?), Transmeta boards are almost non-existant, and VIA C3s aren't very good performers (1GHz is it for clock, and the only good part is performance per watt).
Understatement is a trillion times more effective than exaggeration, too. :P
I went to their site on the naming conventions, and that only served to heighten the confusion. It seems as if it'd be easier if Intel just went ahead and discontinued all their 533 MHz bus speeds, phased out anything under 1 MB of cache, and just stuck with their GHz markings.
This is the most retarded naming system I could imagine... hell, I couldn't have even imagined this... it's so off the wall... there's no order to any of it. Intel may start using the model numbers... but you can bet enthusiasts will call a 3.6 Ghz Pentium 4 800 Mhz bus just that... not a Pentium 4 560.
I have a SERIOUS problem of Intel not indicating the difference between the 533hmhz and 800 speed FSB models, when it has been PROVEN that their is a performance difference.
Im so ashamed to be running a p4 now, the sooner AMD brings out the 939pin Athlons, the happier me and many other tech heads will be.
Someone mind telling me what's going on with the Pentium-M ULV chips? They're giving the 1.1ghz/1MB PM-ULV a rating of 718, putting it ahead of the 1.5ghz/2MB PM, puting the 1.4ghz/2MB PM-LV ahead of the 1.6ghz/2MB PM, and other wierd things. The last digit makes sense(8=ULV, 3=LV, 0/5=normal), but the whole PM rankings seem sketchy at best.
The Pentium 3.8 and 4.0 are still 800 MHz. But these are supposed to be socket T with new mobos supporting DDR2. The only production DDR2 I've heard of is 533 which if run synchronously would quad pump at 1066MHz. So if the table is true, you'll have to underclock the DDR2 at DDR400 or asynchronously-not the best for performance. Else, socket T/Grantsdale/Alderwood is all a myth and Intel is sticking to Sckt478/i8x5.
Am I the only one who's noticed the peculiar gap in the Celeron lineup - i.e. where's the 2.93GHz part? I think Intel skipped that one because nobody wants to buy something that's "just under" 3GHz :)
16 - But do you actually trust the sales people to know what they're talking about? I figure the chance of getting an honest, intelligent salesperson is about 50-50 these days, and only 25-75 if you go to a large store like Frys, CompUSA, Best Buy, etc. With these model numbers from both camps, computer salespeople are likely to become as bad as car salespeople. They'll just say whatever it takes to close the deal. :p
Sure, technophiles like most of the AT readers (looking away from CRAM) can be trusted to know what's good and what's not. For the common man, though, I don't know that it really matters. Computer performance really only matters now if you want to play games, do 3D modelling, or work on video content. For almost everything else, 2 GHz is currently fast enough and will remain so for the next two years at least. For businesses, even 1 GHz is usually fast enough.
Ignorance is bliss, they say, and I think there are a lot of people out there that prefer ignorance when it comes to computers.
The sales people are just going to quote the actual megahertz speed anyway, no matter the name or model no., as they do now for a-64/XP often at the buyer's request.
I was told the 533 and 800 at 2.8 will have the same product number. However, i think there is also some debate as to whether or not both will actual show up.
Cramitpal: Pentium M is by far the better notebook chip. It is clocked slower, but puts out a better punch per MHz. Its also generally cooler- but just a little more expensive. Pick your battles wisely.
It's all a con game to defraud consumers. Even the mainstream media has FINALLY realized Intel has obsolete product now that A64/Opteron is availabe in the laptop-desktop-server segments.
Intel has an over-heating, over-priced, under-performing piece of crap they have applied more bandaids to that can't compete. Anyone with a clue would recognize PressedCrap as a defective design that should have been scrapped years ago. Maybe after another large Class Action Suit Intel will stop peddling this crap to naive consumers without a clue??? In the meantime, Intel can make the buying process really dishonest and confusing by giving HUGE financial incentives to talking heads, sales staff (sic) at big box stores, using buy-offs to enterprise, etc. This is the ONLY means Intel has to sell their non-competitive and defective CPU designs that don't even do 64 bit or improve on prior model designs.
I immediately see something that I really don't like in Intel's names (assuming that AT has them correct: Notice that there are *two* 520 parts, but one has 533 FSB and the other has 800 FSB? I seem to recall the bump in FSB speed resulting in about a 10% performance increase for the P4 platform. At least before we have "2.8B" and "2.8C"; now we just have stupid names.
I suppose that Intel is just learning from AMD on this one. When Intel released 3.0C and 3.2C which were *clearly* superior to any Athlon XP processor, what did AMD do? Oh, we'll just take the 2.08 GHz 2800+, change the multiplier and FSB speed, get it up to 2.1 GHz, and call it a 3000+! And for the 3200+, we'll raise the CPU speed to an astounding 2.2 GHz. If 100 MHz equals 200+ in PR rating, shouldn't the 2.2 GHz XP be a 4400+? But it really didn't matter - it was all marketing.
So now, if AMD releases their socket 939 chips with dual-channel RAM and they can ramp up to 3 GHz (4800+?), it won't even matter. Intel can just claim that the new 780 CPU is the fastest chip out there. Benchmarks will tell the truth, but for the people that don't care much, they'll just buy whatever they can afford and assume it's a good CPU. That's their problem, though. :)
This naming scheme is not for the "tech user", who has deep understanding, but for the "average Joe" who would easily understand your first scentence, but be "dazzed and confused" by the techno-jargon of the rest of your post.
There are so many types of processors and processor capabilities these days--simplification is needed for the marketplace.
Moreover, a bench mark on a processor alone is near useless, other factors are seriously important in determining system performance, such as, memory type and speed, memory controller type and buss speed, GPU type and graphics memory type and buss speed, hard drive type and speed, mother board capabilities and BIOS controls, and many other ancillary components, such as DVD/CDROM type and speed, etc...
The complexities of the CPU with the complexities of the other system components is just too overwhelming for the average Joe.
Indeed, if one equates processors to car engines it is simple to compare horsepower, RPM, and gas mileage, but who buys a car based soley on its engine disregarding the rest of the chassis?
It is now the same with processors and a computer system.
I think one of the more valid points that it brought up is the right now as it stands there are certain high-end 500’s that out perform the lower 700’s. Now I don’t know about you, but I would find this confusing. The first digit has nothing to do with performance, but rather, its all blue crystals
Am I the only one doesn't see a problem with the number progression? 2.8 - 520 3.0 - 530 3.2 - 540 3.4 - 550 3.6 - 560 3.8 - 570 4.0 - 580 4.2 - 590 4.4 - 600 4.6 - 610 4.8 - 620 5.0 - 630
Looks like Intel is following BMW's naming system: 3-series, 5-series and 7-series for low range, mid range and high range respectively!
Note to ceefka: I bet you're the only one on earth that calculated 760-550=210! They are just names, not numbers for use in mathematics! And Intel name the EE a higher number than Prescott because benchmarks simply showed that EE is perfoming better than an equally clocked Prescott, regardless of how many nm is the fabrication process! Who knows, maybe the next 790 is a 90nm part? It doesn't matter, as long as the performance is relative to the name! Remember AMD's Palomino 1700+ and Thoroughbred 1700+? The have the same clockspeed (1.47GHz) but different fabrication process (Palomino is 180nm while Tbred is 130nm); however their performance are almost the same so they share the same name! (but their not the same when you use them; Tbred 1700+ runs much cooler than Palomino 1700+ at the same speed; and most important, Tbred 1700+ can easily be overclocked to 2++GHz while Palomino 1700+ will cook you a nice pancake if you ever try to do so...) The same principle will apply to current and future Intel processors as well.
Note to ZapZilla: it is difficult to define and compare the "absolute performance" of cars, but for CPUs we have something called benchmark to judge the "absolute performance"! Generally, a comparison of the average for ALL "tested" benchmarks will tell that one CPU is better than another (unless the average performance of two CPU is so close that we can say these two CPUs is on par with each other).
The logic in this name game is to provide the average Joe with a way to easily compare processors within a company's product line, while, making it much harder to compare processors between companies.
This has the effect of moving away from "absolute performance" and moving to "relative performance".
Yet, relative to what?
The car industry offers a good model for comparison.
Consumer decision #1: Should one buy Luxury, Mid, or Economy class?
Consumer decision #2: What will the main use be: Off-road, City, or Highway? (Games, Buisness, or Net surfing?)
Consumer decision #3: What company suits one's style: Rolles-Royce, Hummer, or Kia? (Intel, AMD, or Transmeta?)
Defining and comparing the "absolute performance" of processors is more and more like defining and comparing the "absolute performance" of cars -- very difficult and subjective.
It is a good way to simplify the ever growing number of options and choices.
Who can see the logic in this name game? Am I the only one who doesn't get it? Does a Pentium 4 3.4GHz 90nm perform 210 "bleeps" less than a Pentium 4 EE 3.4GHz 130nm (760 -/- 550 = 210)? Why lower numbers for 90nm CPU's?
It looks like result of some formula : pipeline stages / GHz * IPC * FSB / nm. Ehm no, this doesn't add.
Since Intel also says MHz is not everything, then what in their opinion does make the difference? I agree that they needed to make some kind of change because the GHz numbers are likely to go through the roof next year, especially if IPC stays on 6. Or not Intel?
"Comparable" AMD and Intel processors trade #1 scores in benchmarks when put head to head. So time will tell what good a 720 or so is. Aren't y'all looking forward to the benchmarks on these?
There's no timeline on these roadmaps. ??? That doesn't make it a roadmap, does it?
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
33 Comments
Back to Article
stephenbrooks - Friday, April 16, 2004 - link
1. Ah! The missing 2.93GHz part I complained about in a previous post has appeared as a lovely "2.93Gz", errr...2. ---Sorry, the rest of this post includes opinion, paranoia, speculation and a rant---
3. I hate to say this, but this looks a lot like a 'confusion tactic' just for Intel to protect their sales until they have a chance to catch up with AMD in performance again. Yes, the numbers are vaguely in line with increasing performance, but notice most of the families are near the top of their ranges (of 100) already. An optimist would say they've got something new in the labs which will displace the current generation of stuff, and they're just giving the old stuff these numbers to shift it.
4. Regardless, can anyone tell my why this isn't stupid?:
Celeron M 1.3GHz 90nm 400MHz 1MB = 350
Celeron 3.2GHz 90nm 533MHz 256KB = 350
What ARE they playing at? Surely one of them could at least be labelled '351' or '350A' to stop people having to figure out why two identically-named parts have totally different performance characteristics?!
AtaStrumf - Friday, April 9, 2004 - link
I've always liked BMW's naming system and I think Intel could have done worse than this.It clearly separates the different classes, which is the part that I like, but inside those classes is a preety big mess.
I'd have to agree that Intel needs to drop a few products if it really wants to dumb things down, because this is gonna get preety hard to decipher, even for us. Almost as bad as it is now with sSpec numbers.
I guess the price will be the last part of the puzzle that will help people decide what PCU to take home.
fezzik1620 - Friday, April 9, 2004 - link
Thank you for your gentle reproof, DoubleParadoxx. Yes, I mispoke.DoubleParadoxx - Friday, April 9, 2004 - link
"The people who don't know what a Gigahert is and quite frankly don't care."Clearly you dont know. Hertz is the unit for 1 cycle/second, there is no such thing as a hert.
fezzik1620 - Friday, April 9, 2004 - link
Enough whining already. No, really, stop it. It seems like almost all of the comments on this move by Intel have been a bunch of babys crying because somebody took away their empty bottle and gave them a different one. This is not odd, nonsensical, crap, etc. It makes perfect sense. Intel has ridden the more MHz (or GHz) equals a faster processor myth for far too long now. I work for a major computer retailer and trust me this makes perfect sense.Put yourself in my shoes for a moment. Say your grandmother walks into the store (assuming she's not already dead and knows about as much about computers as my grandmother), how are going to quickly and easily explain to her that a P4 2.6 GHz is 10-25% faster than a Celeron 2.6 GHz processor. Now imagine that instead of someone like your grandmother who will take you at your word when it comes to anything computer related it is instead a complete stranger. This has not hurt Intel too much in the past since the prevailing thought among the general public is that there is something seriously wrong with Celeron processors, but now they have thrown the Pentium M into the mix. Try telling someone you've never met who knows nothing about computers, except that more MHz/GHz = a faster machine, that a 1.6 GHz Pentium M is as fast or faster than a 50% higher clocked 2.4 GHz P4 and that you think it is worth their money to pay top dollar for the thing too. Good luck.
So, quit your whining. This is not a bad/dumb move on Intel's part. Your not seeing it for what it is. It is an intentional dumbing down. It is not for you, the enthusiast. Intel knows good and well that when you, the enthusiast, go to buy a P4, PM, or recommend a Celeron for a friend that you are going to still be looking at clock speed, bus speed, and cache. This move is for the 90% of people who are the "everybody else." Where the real money is. The people who don't know what a Gigahert is and quite frankly don't care.
bhtooefr - Thursday, April 8, 2004 - link
This is just fscking odd... BTW, if the Pentium Ms are being labelled this way, it could mean desktop P-M (it, and the AMD K8 will kill NetBurst, IMNSHO (not so humble)). BTW, where are Celeron-M numbers?This makes me want to find some CPU where they don't give you this crap, though. Intel CPUs (except for XScales, which aren't x86) are now all screwed up, AMD uses performance ratings (why not logos saying THIS AMD PROCESSOR BEAT THIS INTEL PROCESSOR IN THIS MAGAZINE/SITE'S REVIEW?), Transmeta boards are almost non-existant, and VIA C3s aren't very good performers (1GHz is it for clock, and the only good part is performance per watt).
Praeludium - Thursday, April 8, 2004 - link
I'm confused.Understatement is a trillion times more effective than exaggeration, too. :P
I went to their site on the naming conventions, and that only served to heighten the confusion. It seems as if it'd be easier if Intel just went ahead and discontinued all their 533 MHz bus speeds, phased out anything under 1 MB of cache, and just stuck with their GHz markings.
Jeff7181 - Thursday, April 8, 2004 - link
This is the most retarded naming system I could imagine... hell, I couldn't have even imagined this... it's so off the wall... there's no order to any of it.Intel may start using the model numbers... but you can bet enthusiasts will call a 3.6 Ghz Pentium 4 800 Mhz bus just that... not a Pentium 4 560.
Icewind - Thursday, April 8, 2004 - link
I have a SERIOUS problem of Intel not indicating the difference between the 533hmhz and 800 speed FSB models, when it has been PROVEN that their is a performance difference.Im so ashamed to be running a p4 now, the sooner AMD brings out the 939pin Athlons, the happier me and many other tech heads will be.
KristopherKubicki - Thursday, April 8, 2004 - link
Virge, I dont think 8=ULV and 3=LV . Look at the pentium Ms.
ViRGE - Thursday, April 8, 2004 - link
Someone mind telling me what's going on with the Pentium-M ULV chips? They're giving the 1.1ghz/1MB PM-ULV a rating of 718, putting it ahead of the 1.5ghz/2MB PM, puting the 1.4ghz/2MB PM-LV ahead of the 1.6ghz/2MB PM, and other wierd things. The last digit makes sense(8=ULV, 3=LV, 0/5=normal), but the whole PM rankings seem sketchy at best.KristopherKubicki - Thursday, April 8, 2004 - link
Pumpkinierre, latency will suffer but thats the plan. The FSB is definitely 800MHz with DDR2-533 and DDR2-667 eventually.Kristopher
Pumpkinierre - Thursday, April 8, 2004 - link
On the new naming posting:The Pentium 3.8 and 4.0 are still 800 MHz. But these are supposed to be socket T with new mobos supporting DDR2. The only production DDR2 I've heard of is 533 which if run synchronously would quad pump at 1066MHz. So if the table is true, you'll have to underclock the DDR2 at DDR400 or asynchronously-not the best for performance. Else, socket T/Grantsdale/Alderwood is all a myth and Intel is sticking to Sckt478/i8x5.
amdfanboy - Thursday, April 1, 2004 - link
All I can say is why ?stephenbrooks - Thursday, April 1, 2004 - link
Am I the only one who's noticed the peculiar gap in the Celeron lineup - i.e. where's the 2.93GHz part? I think Intel skipped that one because nobody wants to buy something that's "just under" 3GHz :)THEonlyRightguard - Wednesday, March 31, 2004 - link
well ,instead of the xp 2800+, will they say the athlon xp 760+ ? it seems to me its a way to try and discredit comparison between amd and intel.TrogdorJW - Wednesday, March 31, 2004 - link
16 - But do you actually trust the sales people to know what they're talking about? I figure the chance of getting an honest, intelligent salesperson is about 50-50 these days, and only 25-75 if you go to a large store like Frys, CompUSA, Best Buy, etc. With these model numbers from both camps, computer salespeople are likely to become as bad as car salespeople. They'll just say whatever it takes to close the deal. :pSure, technophiles like most of the AT readers (looking away from CRAM) can be trusted to know what's good and what's not. For the common man, though, I don't know that it really matters. Computer performance really only matters now if you want to play games, do 3D modelling, or work on video content. For almost everything else, 2 GHz is currently fast enough and will remain so for the next two years at least. For businesses, even 1 GHz is usually fast enough.
Ignorance is bliss, they say, and I think there are a lot of people out there that prefer ignorance when it comes to computers.
Pumpkinierre - Wednesday, March 31, 2004 - link
The sales people are just going to quote the actual megahertz speed anyway, no matter the name or model no., as they do now for a-64/XP often at the buyer's request.KristopherKubicki - Wednesday, March 31, 2004 - link
SDA: A man can dream,... :-pCheck out what i wrote a few weeks ago:
http://www.anandtech.com/cpu/showdoc.html?i=1994
Kristopher
SDA - Wednesday, March 31, 2004 - link
Heh, "pick your battles wisely," says KristopherKubicki to CRAMITPAL. Sorry, Kris, but I don't think this is one you can take down with logic.What's AMD going to do about this, I wonder...? There'll be nothing to have Joe Consumer compare the XXXX+ to!
KristopherKubicki - Wednesday, March 31, 2004 - link
I was told the 533 and 800 at 2.8 will have the same product number. However, i think there is also some debate as to whether or not both will actual show up.Cramitpal: Pentium M is by far the better notebook chip. It is clocked slower, but puts out a better punch per MHz. Its also generally cooler- but just a little more expensive. Pick your battles wisely.
Kristopher
CRAMITPAL - Tuesday, March 30, 2004 - link
It's all a con game to defraud consumers. Even the mainstream media has FINALLY realized Intel has obsolete product now that A64/Opteron is availabe in the laptop-desktop-server segments.Intel has an over-heating, over-priced, under-performing piece of crap they have applied more bandaids to that can't compete. Anyone with a clue would recognize PressedCrap as a defective design that should have been scrapped years ago. Maybe after another large Class Action Suit Intel will stop peddling this crap to naive consumers without a clue??? In the meantime, Intel can make the buying process really dishonest and confusing by giving HUGE financial incentives to talking heads, sales staff (sic) at big box stores, using buy-offs to enterprise, etc. This is the ONLY means Intel has to sell their non-competitive and defective CPU designs that don't even do 64 bit or improve on prior model designs.
TrogdorJW - Tuesday, March 30, 2004 - link
I immediately see something that I really don't like in Intel's names (assuming that AT has them correct: Notice that there are *two* 520 parts, but one has 533 FSB and the other has 800 FSB? I seem to recall the bump in FSB speed resulting in about a 10% performance increase for the P4 platform. At least before we have "2.8B" and "2.8C"; now we just have stupid names.I suppose that Intel is just learning from AMD on this one. When Intel released 3.0C and 3.2C which were *clearly* superior to any Athlon XP processor, what did AMD do? Oh, we'll just take the 2.08 GHz 2800+, change the multiplier and FSB speed, get it up to 2.1 GHz, and call it a 3000+! And for the 3200+, we'll raise the CPU speed to an astounding 2.2 GHz. If 100 MHz equals 200+ in PR rating, shouldn't the 2.2 GHz XP be a 4400+? But it really didn't matter - it was all marketing.
So now, if AMD releases their socket 939 chips with dual-channel RAM and they can ramp up to 3 GHz (4800+?), it won't even matter. Intel can just claim that the new 780 CPU is the fastest chip out there. Benchmarks will tell the truth, but for the people that don't care much, they'll just buy whatever they can afford and assume it's a good CPU. That's their problem, though. :)
ZapZilla - Tuesday, March 30, 2004 - link
dragonfx:This naming scheme is not for the "tech user", who has deep understanding, but for the "average Joe" who would easily understand your first scentence, but be "dazzed and confused" by the techno-jargon of the rest of your post.
There are so many types of processors and processor capabilities these days--simplification is needed for the marketplace.
Moreover, a bench mark on a processor alone is near useless, other factors are seriously important in determining system performance, such as, memory type and speed, memory controller type and buss speed, GPU type and graphics memory type and buss speed, hard drive type and speed, mother board capabilities and BIOS controls, and many other ancillary components, such as DVD/CDROM type and speed, etc...
The complexities of the CPU with the complexities of the other system components is just too overwhelming for the average Joe.
Indeed, if one equates processors to car engines it is simple to compare horsepower, RPM, and gas mileage, but who buys a car based soley on its engine disregarding the rest of the chassis?
It is now the same with processors and a computer system.
tfranzese - Tuesday, March 30, 2004 - link
I like how they label their 533 and 800 FSB processors the same, because there's no difference between the two chips performance. :/mkruer - Tuesday, March 30, 2004 - link
I think one of the more valid points that it brought up is the right now as it stands there are certain high-end 500’s that out perform the lower 700’s. Now I don’t know about you, but I would find this confusing. The first digit has nothing to do with performance, but rather, its all blue crystalsSKiller - Tuesday, March 30, 2004 - link
Am I the only one doesn't see a problem with the number progression?2.8 - 520
3.0 - 530
3.2 - 540
3.4 - 550
3.6 - 560
3.8 - 570
4.0 - 580
4.2 - 590
4.4 - 600
4.6 - 610
4.8 - 620
5.0 - 630
etc... who says they *have* to stop at 600?
dragonfx - Tuesday, March 30, 2004 - link
Looks like Intel is following BMW's naming system: 3-series, 5-series and 7-series for low range, mid range and high range respectively!Note to ceefka: I bet you're the only one on earth that calculated 760-550=210! They are just names, not numbers for use in mathematics! And Intel name the EE a higher number than Prescott because benchmarks simply showed that EE is perfoming better than an equally clocked Prescott, regardless of how many nm is the fabrication process! Who knows, maybe the next 790 is a 90nm part? It doesn't matter, as long as the performance is relative to the name! Remember AMD's Palomino 1700+ and Thoroughbred 1700+? The have the same clockspeed (1.47GHz) but different fabrication process (Palomino is 180nm while Tbred is 130nm); however their performance are almost the same so they share the same name! (but their not the same when you use them; Tbred 1700+ runs much cooler than Palomino 1700+ at the same speed; and most important, Tbred 1700+ can easily be overclocked to 2++GHz while Palomino 1700+ will cook you a nice pancake if you ever try to do so...) The same principle will apply to current and future Intel processors as well.
Note to ZapZilla: it is difficult to define and compare the "absolute performance" of cars, but for CPUs we have something called benchmark to judge the "absolute performance"! Generally, a comparison of the average for ALL "tested" benchmarks will tell that one CPU is better than another (unless the average performance of two CPU is so close that we can say these two CPUs is on par with each other).
ZapZilla - Tuesday, March 30, 2004 - link
The politics of naming.The logic in this name game is to provide the average Joe with a way to easily compare processors within a company's product line, while, making it much harder to compare processors between companies.
This has the effect of moving away from "absolute performance" and moving to "relative performance".
Yet, relative to what?
The car industry offers a good model for comparison.
Consumer decision #1: Should one buy Luxury, Mid, or Economy class?
Consumer decision #2: What will the main use be: Off-road, City, or Highway? (Games, Buisness, or Net surfing?)
Consumer decision #3: What company suits one's style: Rolles-Royce, Hummer, or Kia? (Intel, AMD, or Transmeta?)
Defining and comparing the "absolute performance" of processors is more and more like defining and comparing the "absolute performance" of cars -- very difficult and subjective.
It is a good way to simplify the ever growing number of options and choices.
ZapZilla - Tuesday, March 30, 2004 - link
ceefka - Tuesday, March 30, 2004 - link
Who can see the logic in this name game? Am I the only one who doesn't get it? Does a Pentium 4 3.4GHz 90nm perform 210 "bleeps" less than a Pentium 4 EE 3.4GHz 130nm (760 -/- 550 = 210)? Why lower numbers for 90nm CPU's?It looks like result of some formula : pipeline stages / GHz * IPC * FSB / nm. Ehm no, this doesn't add.
Since Intel also says MHz is not everything, then what in their opinion does make the difference? I agree that they needed to make some kind of change because the GHz numbers are likely to go through the roof next year, especially if IPC stays on 6. Or not Intel?
"Comparable" AMD and Intel processors trade #1 scores in benchmarks when put head to head. So time will tell what good a 720 or so is. Aren't y'all looking forward to the benchmarks on these?
There's no timeline on these roadmaps. ??? That doesn't make it a roadmap, does it?
Ecmaster76 - Tuesday, March 30, 2004 - link
Maybe intel never plans on passing 4.6Ghz any time soon.Pumpkinierre - Tuesday, March 30, 2004 - link
I imagine the quesion marks are a deep rethink on Intel's behalf.