Why on earth are their chipsets/CPUs named after Texas small towns? Our corpoate office is in Bulverde, TX and we have a store in Conroe, TX. Whats next? The Huntsville? Austin? Oh I know. The Midland.
- Creathir
Creathir, you know its a code-name that otherwise won't be revealed anywhere but over computer forums, so it doesn't matter.
And I can sort of see that 5x performance/watt increase as valid. Since they are comparing against Northwood for 5x figure, and against Smithfield is like 4-4.5.
Let's look at Pentium M. It already performs more than 30% faster per clock than Pentium 4. Conroe was said to be many that it will be 30% faster than Yonah, and Yonah is advancement per clock over Dothan, current Pentium M.
Since its 65W, against Pentium D's 130W, there is advantage of two already. The 2-2.25 reduces to 50-70% advantage after factoring out Conroe vs. Yonah. Get what I mean??
...Apple is moving to Intel. I'll be the low-power, high-performance requirements for the Powerbooks is a *HUGE* factor - and this 64-bit laptop version is why Apple is interested. IBM just couldn't deliver on the power/performance ratio on the laptop side...
Power per Watt as we all know is a measurement pertaining to notebooks. Their slides show that desktop purchases will reach zero by the end of the decade (look at the graph and extrapolate 55% to 47% in 1 year) so they are betting on the notebook avenue for future revenue increases. From what I am seeing in this article they are betting VERY heavily on portability. They attribute the Centrino package with quadrupling access points, which in turn increased notebook purchases. I personally think they are nuts. They are giving up on the desktop, basically cedeing this ground to AMD, and moving into a market where they have an advantage.
actually if you watched the presentation, which you can't (live webcast here in the offices).
this is the mobility group, as there focus is on mobility, each group has their own mission/goals, so of course their slides are geared towards that focus.
looking at pictures, with no words is a big difference. if you've watched or been at idf today, you will see how the focus is to make each group (desktop/server/mobile) on basically one architecture, sans itanic.
0 at the end of the decade, i didn't know the graph went that far. there are plenty of desktop products coming, don't make assumptions so soon.
OMG
can u even read a graph
yes desktop purchases are going to reach zero by the end of the decade
and by the end of the next decade they will reach -50%
/end sarcasm
its called a plateau -- and the reason why intel is shifting is that there is more money to be made on mobile products
rotf at the -50% at the end of the next decade...
Until the laptops would be just as powerful, just as cheap, just as fast (hard drive speed wise), have dual DVI, accept professional video cards, and even more, be just as big and heavy as desktops to avoid "shoplifting" them from schools, libraries and so on,
there WILL be a market for desktop computers.
In fact, I do like the Mac Mini and the SFF designs and I would welcome a full powered SFF. I won't get a laptop as I still have a place for a desktop with it's larger ergonomic keyboard and dual LCDs. What can I say? I like the promises, but will I be left at the alter?
Shens to no desktop purchases in the next 5 years. I personally plan on having a PC desktop in the next 5 years, and so does most of AT, i'm sure. Schools, businesses, etc., all will have desktops. So yeah.
Anyone reminded of the IDF in 2000, when tejas was supposed to have us to 10GHz by now...Basically, I'll believe it when I see it.
Are those future wattage estimates for processor alone, or total system expenditure? Because if it's CPU alone, 65 watts on the desktop is good - but it doesn't seem too mindblowing.
I don't know, I think he whole 'per watt' angle is a good way to go, even on the desktop. A lot of people feel this will usher in an era of stylish, small, and quiet desktop PC's. As long as I can still buld my own, I'm all for it. Now then, on the other hand it's true that just because a chip has superior performance per watt, doesn't mean it has superior performance period.
It's interesting that for desktop CPU's they use "Integer" performance, while for servers they use "Relative TPC-C", i.e. a real-world database measurement. This could be nothing, but given that in the past the trend has been to push "media" performance, which generally means floating-point SIMD stuff, it's maybe a significant change. By which I mean: I would guess that this means 5x performance increase per watt doing some rather special, probably very heavy SIMD, integer work.
As someone else said, it's odd that they are moving to "per watt" on desktop: it makes sense for laptops, and maybe for servers (heat and power are important) but on the desktop? It almost smacks of a new marketting angle: the new chips won't actually be measurably quicker than current ones, but will use a significant amount less power. Maybe good for business users, but not so great for the Anandtech market!
Does anyone else find this marketing direction(atleast for desktops and servers, it makes sense for laptops) is really stupid?
Also i cant trust any diagram that depicts the Prescott as providing more performance per watt than the Northwood because anyone who knows anything about the Prescotts abnormal power requirements knows thats blatantly not true and also makes me highly skeptical about that 5x performance per watt number that Intel is giving
It doesn't matter. Intel can tell the true believers that it will be 5x as powerful and they'll believe it and buy it before bothering to verify the claims. Then if Intel doesn't live up to the promise, the fanbase will come up with excuses as to why that's no big deal. Me, I'd rather see real numbers from any CPU from Intel or AMD first, but you can already find folks swearing on their family's lives what the new Intel chips will be able to do, what the new ATI gpu will do, etc. It's a sickness, really.
"On electrical cost savings alone, PC users will save $1 billion per year for every 100M computers."
Ten whole dollars per year? Whoopdee doo!
Yes, lower power is better overall, but it's not the Alpha and Omega of processor value to me, especially in my hobbyist desktop. It just can't be "too hot" or draw "too much power." Hope Intel doesn't lose sight of this, although figuring out where to draw that line can be quite hard to do.
It occurred to me that they may be trying to capitalize on the U.S. fuel pseudocrisis. People will probably buy anything right now if you just tell them it's more efficient. Although, if you do the math (1.0 billion divided by 0.1 billion), Intel is only claiming to save people $10 per PC per year.
Ok the word is not to build 32 bit apps anymore. Who, aside from the stupid or desperate is going to buy Yonah when 4 months later you'd be able to get a 64 bit improved Merom? Can you just upgrade and drop a Merom into an appropriately prepared Yonah laptop? I doubt it, or they'd have been singing that song at the IDF.
Yonah is just late. If it had been here Sep'ish and there was still 6-9mo till a 64bit version, it would have made some sense. If the Napa chipset had been built to accomodate both Yonah and an upgrade to Merom, it would have made sense.
Intel is either lying about the real schedules involved here, or they are desperately throwing out chip designs as fast as possible with little idea of planning.
And most people are using (and will be using for a long time) 32-bit applications. Microsoft Office, IE, Outlook, FireFox, iTunes, etc. This is very basic stuff that simply doesn't benefit from 64-bit addressing. Windows XP x64 edition consequently offers no benefits to most users (Tomshardware just did a x64 edition test, their conclusion is that it is useless, PC World concluded the same).
For the vast majority of consumers, Yonah will be just fine. Most people will never know the difference.
In a try by Microsoft to find reasons to force the upgrade of operating systems, taking another slide of the people's money.
Too bad ( :) ) that they have big problems convincing people to upgrade from Office XP (or sometime Office 2000) to their latest greatest Office 2003.
Here's to hoping that these lower power chips will overclock better than the current P4 chips (at least without resorting to cascading phase change or liquid N2).
If Conroe uses 65 watts and is about 5x as faster per watt than the current p4 3.8 which uses 135 watts (just over 2x 65) and if Conroe is dual core (another factor of two) then each core should be about 20% faster than a p4 3.8 GHZ to get up to a total of 5x better speed per watt. If this is really true, it is very exciting.
I'm a bit worried about those 64-bit windows. Where did x86-64 go??? Is AMD in a bit of trouble here? Is the end of licensing agreement in sight? Find out on the next episode of $50 Athlon 64!!!
It looks like 5x per watt than Northwood judging from that graph (highest TDP for that is 89 watts at 3.4 GHz) which is probably the worst performance/watt of any Northwood.
Using this figure, 5/(89/65) = 3.65x faster than a 3.4 GHz Northwood or 82% faster per core than a 3.4GHz Northwood.
If we assume it was a 1.8 GHz Northwood, then its TDP is 47W
Using this figure, 5/(47/65) = 6.9x faster than a 1.8 GHz Northwood or 145% faster per core than a 1.8 GHz Northwood.
Either way, that seems very optimistic, although it means that we should see a very sizeable jump in spead even in single threaded apps
The slide shows the handtop as 0.5 watts and the subnotebook at 5.0 watts, but your text says the handtop is 5.0 watts and that 0.5 is something else. Interesting article otherwise though, I REALLY look forward to the new generation.
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.
Facebook
Twitter
RSS
34 Comments
Back to Article
Griswold - Wednesday, August 24, 2005 - link
The IDF has been renamed to IMF - Intel Marketing Forum.creathir - Tuesday, August 23, 2005 - link
Why on earth are their chipsets/CPUs named after Texas small towns? Our corpoate office is in Bulverde, TX and we have a store in Conroe, TX. Whats next? The Huntsville? Austin? Oh I know. The Midland.- Creathir
IntelUser2000 - Thursday, August 25, 2005 - link
Creathir, you know its a code-name that otherwise won't be revealed anywhere but over computer forums, so it doesn't matter.And I can sort of see that 5x performance/watt increase as valid. Since they are comparing against Northwood for 5x figure, and against Smithfield is like 4-4.5.
Let's look at Pentium M. It already performs more than 30% faster per clock than Pentium 4. Conroe was said to be many that it will be 30% faster than Yonah, and Yonah is advancement per clock over Dothan, current Pentium M.
Since its 65W, against Pentium D's 130W, there is advantage of two already. The 2-2.25 reduces to 50-70% advantage after factoring out Conroe vs. Yonah. Get what I mean??
achbed - Tuesday, August 23, 2005 - link
...Apple is moving to Intel. I'll be the low-power, high-performance requirements for the Powerbooks is a *HUGE* factor - and this 64-bit laptop version is why Apple is interested. IBM just couldn't deliver on the power/performance ratio on the laptop side...bupkus - Tuesday, August 23, 2005 - link
10X performance/watt = 10GHz processorsATI move over, Intel has the vaporware podium now.
I was going to buy a dual core A64, but I'm going to wait now.
bldckstark - Tuesday, August 23, 2005 - link
Power per Watt as we all know is a measurement pertaining to notebooks. Their slides show that desktop purchases will reach zero by the end of the decade (look at the graph and extrapolate 55% to 47% in 1 year) so they are betting on the notebook avenue for future revenue increases. From what I am seeing in this article they are betting VERY heavily on portability. They attribute the Centrino package with quadrupling access points, which in turn increased notebook purchases. I personally think they are nuts. They are giving up on the desktop, basically cedeing this ground to AMD, and moving into a market where they have an advantage.AdvanS13 - Tuesday, August 23, 2005 - link
actually if you watched the presentation, which you can't (live webcast here in the offices).this is the mobility group, as there focus is on mobility, each group has their own mission/goals, so of course their slides are geared towards that focus.
looking at pictures, with no words is a big difference. if you've watched or been at idf today, you will see how the focus is to make each group (desktop/server/mobile) on basically one architecture, sans itanic.
0 at the end of the decade, i didn't know the graph went that far. there are plenty of desktop products coming, don't make assumptions so soon.
xsilver - Tuesday, August 23, 2005 - link
OMGcan u even read a graph
yes desktop purchases are going to reach zero by the end of the decade
and by the end of the next decade they will reach -50%
/end sarcasm
its called a plateau -- and the reason why intel is shifting is that there is more money to be made on mobile products
Calin - Wednesday, August 24, 2005 - link
rotf at the -50% at the end of the next decade...Until the laptops would be just as powerful, just as cheap, just as fast (hard drive speed wise), have dual DVI, accept professional video cards, and even more, be just as big and heavy as desktops to avoid "shoplifting" them from schools, libraries and so on,
there WILL be a market for desktop computers.
bupkus - Tuesday, August 23, 2005 - link
In fact, I do like the Mac Mini and the SFF designs and I would welcome a full powered SFF. I won't get a laptop as I still have a place for a desktop with it's larger ergonomic keyboard and dual LCDs. What can I say? I like the promises, but will I be left at the alter?Leper Messiah - Tuesday, August 23, 2005 - link
Shens to no desktop purchases in the next 5 years. I personally plan on having a PC desktop in the next 5 years, and so does most of AT, i'm sure. Schools, businesses, etc., all will have desktops. So yeah.Anyone reminded of the IDF in 2000, when tejas was supposed to have us to 10GHz by now...Basically, I'll believe it when I see it.
bob661 - Tuesday, August 23, 2005 - link
I don't think they're giving that market away. I think Intel is saying that there won't be much money to be made in the desktop market after a while.
xbdestroya - Tuesday, August 23, 2005 - link
Are those future wattage estimates for processor alone, or total system expenditure? Because if it's CPU alone, 65 watts on the desktop is good - but it doesn't seem too mindblowing.xbdestroya - Tuesday, August 23, 2005 - link
I don't know, I think he whole 'per watt' angle is a good way to go, even on the desktop. A lot of people feel this will usher in an era of stylish, small, and quiet desktop PC's. As long as I can still buld my own, I'm all for it. Now then, on the other hand it's true that just because a chip has superior performance per watt, doesn't mean it has superior performance period.Matthew Daws - Tuesday, August 23, 2005 - link
It's interesting that for desktop CPU's they use "Integer" performance, while for servers they use "Relative TPC-C", i.e. a real-world database measurement. This could be nothing, but given that in the past the trend has been to push "media" performance, which generally means floating-point SIMD stuff, it's maybe a significant change. By which I mean: I would guess that this means 5x performance increase per watt doing some rather special, probably very heavy SIMD, integer work.As someone else said, it's odd that they are moving to "per watt" on desktop: it makes sense for laptops, and maybe for servers (heat and power are important) but on the desktop? It almost smacks of a new marketting angle: the new chips won't actually be measurably quicker than current ones, but will use a significant amount less power. Maybe good for business users, but not so great for the Anandtech market!
Jep4444 - Tuesday, August 23, 2005 - link
Does anyone else find this marketing direction(atleast for desktops and servers, it makes sense for laptops) is really stupid?Also i cant trust any diagram that depicts the Prescott as providing more performance per watt than the Northwood because anyone who knows anything about the Prescotts abnormal power requirements knows thats blatantly not true and also makes me highly skeptical about that 5x performance per watt number that Intel is giving
fishbits - Tuesday, August 23, 2005 - link
It doesn't matter. Intel can tell the true believers that it will be 5x as powerful and they'll believe it and buy it before bothering to verify the claims. Then if Intel doesn't live up to the promise, the fanbase will come up with excuses as to why that's no big deal. Me, I'd rather see real numbers from any CPU from Intel or AMD first, but you can already find folks swearing on their family's lives what the new Intel chips will be able to do, what the new ATI gpu will do, etc. It's a sickness, really."On electrical cost savings alone, PC users will save $1 billion per year for every 100M computers."
Ten whole dollars per year? Whoopdee doo!
Yes, lower power is better overall, but it's not the Alpha and Omega of processor value to me, especially in my hobbyist desktop. It just can't be "too hot" or draw "too much power." Hope Intel doesn't lose sight of this, although figuring out where to draw that line can be quite hard to do.
nullpointerus - Tuesday, August 23, 2005 - link
It occurred to me that they may be trying to capitalize on the U.S. fuel pseudocrisis. People will probably buy anything right now if you just tell them it's more efficient. Although, if you do the math (1.0 billion divided by 0.1 billion), Intel is only claiming to save people $10 per PC per year.erinlegault - Tuesday, August 23, 2005 - link
I can't believe anything Intel says at these IDF's.bob661 - Tuesday, August 23, 2005 - link
I just hope this "new" architecture performs better than what's available now.cHodAXUK - Tuesday, August 23, 2005 - link
Substitute 'new micro-architecture' with 're-jigged P3 core' and it would be much nearer to the truth.stateofbeasley - Tuesday, August 23, 2005 - link
Let me guess: you think K8 is a 're-jigged K7 core' too :roll;Anemone - Tuesday, August 23, 2005 - link
Ok the word is not to build 32 bit apps anymore. Who, aside from the stupid or desperate is going to buy Yonah when 4 months later you'd be able to get a 64 bit improved Merom? Can you just upgrade and drop a Merom into an appropriately prepared Yonah laptop? I doubt it, or they'd have been singing that song at the IDF.Yonah is just late. If it had been here Sep'ish and there was still 6-9mo till a 64bit version, it would have made some sense. If the Napa chipset had been built to accomodate both Yonah and an upgrade to Merom, it would have made sense.
Intel is either lying about the real schedules involved here, or they are desperately throwing out chip designs as fast as possible with little idea of planning.
Who knows..
stateofbeasley - Tuesday, August 23, 2005 - link
And most people are using (and will be using for a long time) 32-bit applications. Microsoft Office, IE, Outlook, FireFox, iTunes, etc. This is very basic stuff that simply doesn't benefit from 64-bit addressing. Windows XP x64 edition consequently offers no benefits to most users (Tomshardware just did a x64 edition test, their conclusion is that it is useless, PC World concluded the same).For the vast majority of consumers, Yonah will be just fine. Most people will never know the difference.
bob661 - Tuesday, August 23, 2005 - link
Office is going to 64 bit as well as those other apps you mentioned. This will happen sooner rather than later.Calin - Wednesday, August 24, 2005 - link
In a try by Microsoft to find reasons to force the upgrade of operating systems, taking another slide of the people's money.Too bad ( :) ) that they have big problems convincing people to upgrade from Office XP (or sometime Office 2000) to their latest greatest Office 2003.
Doormat - Tuesday, August 23, 2005 - link
Here's to hoping that these lower power chips will overclock better than the current P4 chips (at least without resorting to cascading phase change or liquid N2).Den - Tuesday, August 23, 2005 - link
If Conroe uses 65 watts and is about 5x as faster per watt than the current p4 3.8 which uses 135 watts (just over 2x 65) and if Conroe is dual core (another factor of two) then each core should be about 20% faster than a p4 3.8 GHZ to get up to a total of 5x better speed per watt. If this is really true, it is very exciting.Calin - Wednesday, August 24, 2005 - link
I lived under the impression that Northwood was more efficient per watt than Prescottmino - Wednesday, August 24, 2005 - link
It is.AtaStrumf - Tuesday, August 23, 2005 - link
Or you can just get an Athlon 64 now ;-)I'm a bit worried about those 64-bit windows. Where did x86-64 go??? Is AMD in a bit of trouble here? Is the end of licensing agreement in sight? Find out on the next episode of $50 Athlon 64!!!
RaynorWolfcastle - Tuesday, August 23, 2005 - link
It looks like 5x per watt than Northwood judging from that graph (highest TDP for that is 89 watts at 3.4 GHz) which is probably the worst performance/watt of any Northwood.Using this figure, 5/(89/65) = 3.65x faster than a 3.4 GHz Northwood or 82% faster per core than a 3.4GHz Northwood.
If we assume it was a 1.8 GHz Northwood, then its TDP is 47W
Using this figure, 5/(47/65) = 6.9x faster than a 1.8 GHz Northwood or 145% faster per core than a 1.8 GHz Northwood.
Either way, that seems very optimistic, although it means that we should see a very sizeable jump in spead even in single threaded apps
Den - Tuesday, August 23, 2005 - link
The slide shows the handtop as 0.5 watts and the subnotebook at 5.0 watts, but your text says the handtop is 5.0 watts and that 0.5 is something else. Interesting article otherwise though, I REALLY look forward to the new generation.Avalon - Tuesday, August 23, 2005 - link
Sounds like a great direction for Intel. I'm very interested in the Conroe chip.