I apologize for my ignorance on the deeper technical issues of CPU's... I don't quite understand what the difference between 45 and 65nm really is other than transistors that are closer together, making them faster and more efficient.
Interestingly, the apple website claims to have 65nm processors in their MB's
"The Intel Core 2 Duo uses Intel?s groundbreaking 65-nanometer process ? resulting in transistors so small, you could fit a hundred inside a single human cell."
What exactly does this mean in comparison to the 65nm CPU's you guys are talking about? Will the new one allow for more cache and a higher BUS speed as well as a faster CPU?
I apologize for my ignorance on the deeper technical issues of CPU's... I don't quite understand what the difference between 45 and 65nm really is other than transistors that are closer together, making them faster and more efficient.
Interestingly, the apple website claims to have 65nm processors in their MB's
"The Intel Core 2 Duo uses Intel?s groundbreaking 65-nanometer process ? resulting in transistors so small, you could fit a hundred inside a single human cell."
What exactly does this mean in comparison to the 65nm CPU's you guys are talking about? Will the new one allow for more cache and a higher BUS speed as well as a faster CPU?
Bus speed? Not necessarily. But the key is that Penryn both makes for more room on a single chip and cuts down on the electricity leaks that generate heat and waste power, so it ultimately runs cooler than a 65nm chip but still offers you more cache or more instruction sets.
[LIST=1][*]In a same speed comparison between say Merom and Penryn, Penryn will win because of the extra L2 cache, SSE4, and some other minor micro-architecture changes that haven't been disclosed as yet.
Noob question: Wouldn't a 45nm of same speed as a 65nm process faster due to being smaller in size (less space = goes from A to B faster)? Even without the extra L2 cache, SSE4, and some other minor micro-architecture changes?
Noob question: Wouldn't a 45nm of same speed as a 65nm process faster due to being smaller in size (less space = goes from A to B faster)? Even without the extra L2 cache, SSE4, and some other minor micro-architecture changes?
The same clock, different process of the same chip design would not mean a faster chip. In the end, the signal would not be moving any faster because the clock itself is used to control the speed of the signal. Silly transportation analogy: think of the clock as a traffic signal grid. No matter how fast the signal can move, it is going to be held up at the next light.
What the the shorter signal delays allow is faster clocks speeds. That would be equivalent to increasing the switching speed of traffic signals.
Have been reading up on Barcelona, AMDs Quad core chip due out this summer. AMD is really pumoing it up with claims of a 40% faster performance over Clovertown. That would be pretty incredible. Wonder if this will prove to be true.
The same clock, different process of the same chip design would not mean a faster chip. In the end, the signal would not be moving any faster because the clock itself is used to control the speed of the signal. Silly transportation analogy: think of the clock as a traffic signal grid. No matter how fast the signal can move, it is going to be held up at the next light.
What the the shorter signal delays allow is faster clocks speeds. That would be equivalent to increasing the switching speed of traffic signals.
I appreciate the layman's analogy. Thanks for taking the time to explain that.
Have been reading up on Barcelona, AMDs Quad core chip due out this summer. AMD is really pumoing it up with claims of a 40% faster performance over Clovertown. That would be pretty incredible. Wonder if this will prove to be true.
Yes, it would be increadible... to the point that it stretches credibility.
I have no doubt that in some test case it will out perform Clovertown by 40%, but that case might be so obscure or narrow or contrived as to be irrelevant. And about the time it arrives AMD will have to compete with Penryn which ought to outperform Clovertown by a similar margin, at least. Intel has a substantial lead in term of process, and core design. And I don't say this because I'm an Intel or AMD fanboy, I say it because they do and they do because they have lots of money.
I have no doubt that in some test case it will out perform Clovertown by 40%, but that case might be so obscure or narrow or contrived as to be irrelevant. And about the time it arrives AMD will have to compete with Penryn which ought to outperform Clovertown by a similar margin, at least.
I agree, I would not be surprised if the AMD chip simply catches up with what Intel will have available by the time the AMD chip is released.
I'm no fanboy either. If AMD is able to produce such a fine chip I would be in favor of Apple using it in the Mac Pro. Looks like an ideal match, if it lives up to AMD's promise. This little chip war between AMD and Intel is really going to benefit users.
Intel was following AMD for almost last two year, now its AMD term to catch up with Intel. end consumers getting great product. AMD and its survival is very crucial to the CPU industrty and to the extend PC industry...
hope Apple will get into AMD Chips down the road (like HP, IBM, Dell) ... Still OS X is the heart of Mac (Not the CPU), even though their focus now on Intel low power CPUs ...
who knows with good AMD cpus Macs may be cheaper (for consumer) and still profitable (for APPLE)
It's an article about PENRYN shipping earlier than expected, much earlier. I'm curious on what everyone thinks about it, because, if true, it means some amazing Mac news in the next few months that we didn't think possible. I hope there's some truth to it, because I want to spend some freaking money!!!!
There have been several reports that intel ha been pushing Penryn ahead of its schedule. So it sounds as if they are attempting to get it out as quickly as possible.
Imagine if the next MacBooks have Leopard, Santa Rosa, Penryrn, Intel GMA X3000, and Robson. That would be quite an update.
There have been several reports that intel ha been pushing Penryn ahead of its schedule. So it sounds as if they are attempting to get it out as quickly as possible.
Imagine if the next MacBooks have Leopard, Santa Rosa, Penryrn, Intel GMA X3000, and Robson. That would be quite an update.
Doesn't Santa Rosa include GMA X3000 and Robson, so you can just say Leopard + Santa Rosa + Penryn.
Penryn will probably be released on the server first at speeds higher than the current Woodcrests and Clovertowns. That's where Intel is most exposed wrt AMD's Barcelona. There is no need to rush Penryn for the mobile and desktop segments.
Yes, it would be increadible... to the point that it stretches credibility.
I have no doubt that in some test case it will out perform Clovertown by 40%, but that case might be so obscure or narrow or contrived as to be irrelevant. And about the time it arrives AMD will have to compete with Penryn which ought to outperform Clovertown by a similar margin, at least. Intel has a substantial lead in term of process, and core design. And I don't say this because I'm an Intel or AMD fanboy, I say it because they do and they do because they have lots of money.
AMD still has the better cpu to cpu link, cpu to chipset links and the chip sets have alot more pci-e lanes then Intels ones
AMD still has the better cpu to cpu link, cpu to chipset links and the chip sets have alot more pci-e lanes then Intels ones
Yes, but it hasn't done them a lot of good. Its only a matter of time before Intel equals or exceed them on the inter-chip links. And really, the number of PCIe lanes on the Intel chipsets is sufficient for most purposes so the extras are just wasteful in those cases. For the edge cases Intel will have a solution eventually (if not already), but they hardly matter for the bulk of the sales.
Yes, it would be increadible... to the point that it stretches credibility.
I have no doubt that in some test case it will out perform Clovertown by 40%, but that case might be so obscure or narrow or contrived as to be irrelevant. And about the time it arrives AMD will have to compete with Penryn which ought to outperform Clovertown by a similar margin, at least. Intel has a substantial lead in term of process, and core design. And I don't say this because I'm an Intel or AMD fanboy, I say it because they do and they do because they have lots of money.
Quote:
Originally Posted by backtomac
I'm no fanboy either. If AMD is able to produce such a fine chip I would be in favor of Apple using it in the Mac Pro. Looks like an ideal match, if it lives up to AMD's promise. This little chip war between AMD and Intel is really going to benefit users.
Quote:
Originally Posted by backtomac
Have been reading up on Barcelona, AMDs Quad core chip due out this summer. AMD is really pumoing it up with claims of a 40% faster performance over Clovertown. That would be pretty incredible. Wonder if this will prove to be true.
AMD has a history of talking but not doing the walking.
The problem is that Barcelona is set to be released at a maximum clock speed of 2.30Ghz.
Current Core 2 Duos easily overclock to nearly 4.0Ghz and Penryn is conservatively clocked in that respect.
An Intel 3.0GHz Penryn will walk all over the AMD 2.30Ghz Barcelona.
Doesn't Santa Rosa include GMA X3000 and Robson, so you can just say Leopard + Santa Rosa + Penryn.
Penryn will probably be released on the server first at speeds higher than the current Woodcrests and Clovertowns. That's where Intel is most exposed wrt AMD's Barcelona. There is no need to rush Penryn for the mobile and desktop segments.
I rarely see new procs hitting servers before Desktop/Workstations. Servers always require a tad bit more care. If your Workstation crashes you reboot and hope you haven't lost data. If your Server crashes services don't get completed and many people can be affected. Santa Rosa will contain two IGP the X3000 and the GMA3000 which is the sucky one. We cannot assume that x3000 is with every Crestline chipset
Quote:
Originally Posted by Joe_the_dragon
AMD still has the better cpu to cpu link, cpu to chipset links and the chip sets have alot more pci-e lanes then Intels ones
So what? Core2 cache coherency isn't bad and the dynamic partitioning of shared cache will be back with 6MB of L2 in Penryn. Right now Intel is cleaning AMD's clock. In the end overall compute speed rules the day and AMD is getting a bit old. Barcelona better get a speed boost because they will not catch Intel and if Nehalem ships on time it could be game over.
Doesn't Santa Rosa include GMA X3000 and Robson, so you can just say Leopard + Santa Rosa + Penryn.
Penryn will probably be released on the server first at speeds higher than the current Woodcrests and Clovertowns. That's where Intel is most exposed wrt AMD's Barcelona. There is no need to rush Penryn for the mobile and desktop segments.
I thought santa rosa and penryn were different platforms. How can they come at the same time
Comments
Interestingly, the apple website claims to have 65nm processors in their MB's
"The Intel Core 2 Duo uses Intel?s groundbreaking 65-nanometer process ? resulting in transistors so small, you could fit a hundred inside a single human cell."
What exactly does this mean in comparison to the 65nm CPU's you guys are talking about? Will the new one allow for more cache and a higher BUS speed as well as a faster CPU?
I apologize for my ignorance on the deeper technical issues of CPU's... I don't quite understand what the difference between 45 and 65nm really is other than transistors that are closer together, making them faster and more efficient.
Interestingly, the apple website claims to have 65nm processors in their MB's
"The Intel Core 2 Duo uses Intel?s groundbreaking 65-nanometer process ? resulting in transistors so small, you could fit a hundred inside a single human cell."
What exactly does this mean in comparison to the 65nm CPU's you guys are talking about? Will the new one allow for more cache and a higher BUS speed as well as a faster CPU?
Bus speed? Not necessarily. But the key is that Penryn both makes for more room on a single chip and cuts down on the electricity leaks that generate heat and waste power, so it ultimately runs cooler than a 65nm chip but still offers you more cache or more instruction sets.
It a combination of the two factors.
[LIST=1][*]In a same speed comparison between say Merom and Penryn, Penryn will win because of the extra L2 cache, SSE4, and some other minor micro-architecture changes that haven't been disclosed as yet.
Noob question: Wouldn't a 45nm of same speed as a 65nm process faster due to being smaller in size (less space = goes from A to B faster)? Even without the extra L2 cache, SSE4, and some other minor micro-architecture changes?
Noob question: Wouldn't a 45nm of same speed as a 65nm process faster due to being smaller in size (less space = goes from A to B faster)? Even without the extra L2 cache, SSE4, and some other minor micro-architecture changes?
The same clock, different process of the same chip design would not mean a faster chip. In the end, the signal would not be moving any faster because the clock itself is used to control the speed of the signal. Silly transportation analogy: think of the clock as a traffic signal grid. No matter how fast the signal can move, it is going to be held up at the next light.
What the the shorter signal delays allow is faster clocks speeds. That would be equivalent to increasing the switching speed of traffic signals.
The same clock, different process of the same chip design would not mean a faster chip. In the end, the signal would not be moving any faster because the clock itself is used to control the speed of the signal. Silly transportation analogy: think of the clock as a traffic signal grid. No matter how fast the signal can move, it is going to be held up at the next light.
What the the shorter signal delays allow is faster clocks speeds. That would be equivalent to increasing the switching speed of traffic signals.
I appreciate the layman's analogy. Thanks for taking the time to explain that.
You can learn a whole lot around here, you know?
Have been reading up on Barcelona, AMDs Quad core chip due out this summer. AMD is really pumoing it up with claims of a 40% faster performance over Clovertown. That would be pretty incredible. Wonder if this will prove to be true.
Yes, it would be increadible... to the point that it stretches credibility.
I have no doubt that in some test case it will out perform Clovertown by 40%, but that case might be so obscure or narrow or contrived as to be irrelevant. And about the time it arrives AMD will have to compete with Penryn which ought to outperform Clovertown by a similar margin, at least. Intel has a substantial lead in term of process, and core design. And I don't say this because I'm an Intel or AMD fanboy, I say it because they do and they do because they have lots of money.
I have no doubt that in some test case it will out perform Clovertown by 40%, but that case might be so obscure or narrow or contrived as to be irrelevant. And about the time it arrives AMD will have to compete with Penryn which ought to outperform Clovertown by a similar margin, at least.
I agree, I would not be surprised if the AMD chip simply catches up with what Intel will have available by the time the AMD chip is released.
hope Apple will get into AMD Chips down the road (like HP, IBM, Dell) ... Still OS X is the heart of Mac (Not the CPU), even though their focus now on Intel low power CPUs ...
who knows with good AMD cpus Macs may be cheaper (for consumer) and still profitable (for APPLE)
http://www.theinquirer.net/default.aspx?article=37562
It's an article about PENRYN shipping earlier than expected, much earlier. I'm curious on what everyone thinks about it, because, if true, it means some amazing Mac news in the next few months that we didn't think possible. I hope there's some truth to it, because I want to spend some freaking money!!!!
Imagine if the next MacBooks have Leopard, Santa Rosa, Penryrn, Intel GMA X3000, and Robson. That would be quite an update.
There have been several reports that intel ha been pushing Penryn ahead of its schedule. So it sounds as if they are attempting to get it out as quickly as possible.
Imagine if the next MacBooks have Leopard, Santa Rosa, Penryrn, Intel GMA X3000, and Robson. That would be quite an update.
Doesn't Santa Rosa include GMA X3000 and Robson, so you can just say Leopard + Santa Rosa + Penryn.
Penryn will probably be released on the server first at speeds higher than the current Woodcrests and Clovertowns. That's where Intel is most exposed wrt AMD's Barcelona. There is no need to rush Penryn for the mobile and desktop segments.
Yes, it would be increadible... to the point that it stretches credibility.
I have no doubt that in some test case it will out perform Clovertown by 40%, but that case might be so obscure or narrow or contrived as to be irrelevant. And about the time it arrives AMD will have to compete with Penryn which ought to outperform Clovertown by a similar margin, at least. Intel has a substantial lead in term of process, and core design. And I don't say this because I'm an Intel or AMD fanboy, I say it because they do and they do because they have lots of money.
AMD still has the better cpu to cpu link, cpu to chipset links and the chip sets have alot more pci-e lanes then Intels ones
AMD still has the better cpu to cpu link, cpu to chipset links and the chip sets have alot more pci-e lanes then Intels ones
Yes, but it hasn't done them a lot of good. Its only a matter of time before Intel equals or exceed them on the inter-chip links. And really, the number of PCIe lanes on the Intel chipsets is sufficient for most purposes so the extras are just wasteful in those cases. For the edge cases Intel will have a solution eventually (if not already), but they hardly matter for the bulk of the sales.
Yes, it would be increadible... to the point that it stretches credibility.
I have no doubt that in some test case it will out perform Clovertown by 40%, but that case might be so obscure or narrow or contrived as to be irrelevant. And about the time it arrives AMD will have to compete with Penryn which ought to outperform Clovertown by a similar margin, at least. Intel has a substantial lead in term of process, and core design. And I don't say this because I'm an Intel or AMD fanboy, I say it because they do and they do because they have lots of money.
I'm no fanboy either. If AMD is able to produce such a fine chip I would be in favor of Apple using it in the Mac Pro. Looks like an ideal match, if it lives up to AMD's promise. This little chip war between AMD and Intel is really going to benefit users.
Have been reading up on Barcelona, AMDs Quad core chip due out this summer. AMD is really pumoing it up with claims of a 40% faster performance over Clovertown. That would be pretty incredible. Wonder if this will prove to be true.
AMD has a history of talking but not doing the walking.
The problem is that Barcelona is set to be released at a maximum clock speed of 2.30Ghz.
Current Core 2 Duos easily overclock to nearly 4.0Ghz and Penryn is conservatively clocked in that respect.
An Intel 3.0GHz Penryn will walk all over the AMD 2.30Ghz Barcelona.
Doesn't Santa Rosa include GMA X3000 and Robson, so you can just say Leopard + Santa Rosa + Penryn.
Penryn will probably be released on the server first at speeds higher than the current Woodcrests and Clovertowns. That's where Intel is most exposed wrt AMD's Barcelona. There is no need to rush Penryn for the mobile and desktop segments.
I rarely see new procs hitting servers before Desktop/Workstations. Servers always require a tad bit more care. If your Workstation crashes you reboot and hope you haven't lost data. If your Server crashes services don't get completed and many people can be affected. Santa Rosa will contain two IGP the X3000 and the GMA3000 which is the sucky one. We cannot assume that x3000 is with every Crestline chipset
AMD still has the better cpu to cpu link, cpu to chipset links and the chip sets have alot more pci-e lanes then Intels ones
So what? Core2 cache coherency isn't bad and the dynamic partitioning of shared cache will be back with 6MB of L2 in Penryn. Right now Intel is cleaning AMD's clock. In the end overall compute speed rules the day and AMD is getting a bit old. Barcelona better get a speed boost because they will not catch Intel and if Nehalem ships on time it could be game over.
Doesn't Santa Rosa include GMA X3000 and Robson, so you can just say Leopard + Santa Rosa + Penryn.
Penryn will probably be released on the server first at speeds higher than the current Woodcrests and Clovertowns. That's where Intel is most exposed wrt AMD's Barcelona. There is no need to rush Penryn for the mobile and desktop segments.
I thought santa rosa and penryn were different platforms. How can they come at the same time
I thought santa rosa and penryn were different platforms. How can they come at the same time
Santa Rosa is the chipset/platform, Penryn is the CPU.