Sign in with
Sign up | Sign in
Your question

45nm Penryn CPU's - 1st Qtr 2007?

Last response: in CPUs
Share
November 29, 2006 1:25:26 PM

Reported in X-bit labs: (I assume you guys have seen this ... no fanboy wars please - just information :) )
Quote:
Intel Corp. on Monday said that the first prototypes of microprocessors using 45nm process technology have been produced. The announcement once again illustrates Intel’s very aggressive process technology transition roadmap and proves that Intel Corp. has delivered on its promise to tape out its first 45nm chip this quarter....
.... Intel’s sampling of the code-named Penryn processor at 45nm fabrication process occurs amid the company’s arch-rival Advanced Micro Devices still have not announced a single processor made using 65nm fabrication process, which is used to manufacture the majority of Intel’s central processing units (CPUs) these days. According to the director of Intel, the company is on-track to produce Penryn in volumes and ship them to customers in the second half of 2007.

Intel Samples 45nm “Penryn” Microprocessors
Copyright (c) 1999-2006 X-bit labs
November 29, 2006 1:28:31 PM

We've already discussed this quite a bit in another thread. AMD is either really quiet because they're going to take the world by storm, or really quiet because they really don't have anything. There's the 4x4 vs. QX6700 test that was just done, and it didn't seem promising to say the least...

Then again, AMD has been on top for a long time now, which means not only is Intel pushing really hard to get back on track to be the top again, but AMD might have gotten a little lazy and slacked off...

We'll see what the future holds.
November 29, 2006 1:37:29 PM

Quote:
We've already discussed this quite a bit in another thread. AMD is either really quiet because they're going to take the world by storm, or really quiet because they really don't have anything. There's the 4x4 vs. QX6700 test that was just done, and it didn't seem promising to say the least...

Then again, AMD has been on top for a long time now, which means not only is Intel pushing really hard to get back on track to be the top again, but AMD might have gotten a little lazy and slacked off...

We'll see what the future holds.

It is what I was just thinking; AMD has definitely been caught by the Intel syndrome; They did nothing special after the first K8's were launched except for the straightforward moves like X2 and dual channel controller,...ah yes, they also switched two sockets :lol: 
Related resources
November 29, 2006 3:18:20 PM

Quote:
Reported in X-bit labs: (I assume you guys have seen this ... no fanboy wars please - just information :) )
Intel Corp. on Monday said that the first prototypes of microprocessors using 45nm process technology have been produced. The announcement once again illustrates Intel’s very aggressive process technology transition roadmap and proves that Intel Corp. has delivered on its promise to tape out its first 45nm chip this quarter....
.... Intel’s sampling of the code-named Penryn processor at 45nm fabrication process occurs amid the company’s arch-rival Advanced Micro Devices still have not announced a single processor made using 65nm fabrication process, which is used to manufacture the majority of Intel’s central processing units (CPUs) these days. According to the director of Intel, the company is on-track to produce Penryn in volumes and ship them to customers in the second half of 2007.

Intel Samples 45nm “Penryn” Microprocessors
Copyright (c) 1999-2006 X-bit labs

Expected to roll out in Q3 07.
a c 471 à CPUs
a c 115 å Intel
November 29, 2006 3:21:00 PM

Yeah, it's actually quite refreshing that despite own admitance (back in July 2006) that mid 2007 was a very aggressive target date, it seems that they are capable of meeting that goal. At least for the time being.
November 29, 2006 3:41:00 PM

I'm hating Intel for this simple reason ... I was going to build a e6600 (C2D) system after the holidays (hoping for DDR2 price drop, and mobo's etc) - so now, I see, I need to wait on the Penryn chips as these 45 nm C2D chips will, clearly require a different socket ... right?

Or maybe the 65 nm stuff will drop enough in price that it's a much better deal to build with the current socket chips and just OC and pocket the difference. Does this sound right? :?
November 29, 2006 3:46:43 PM

Not necessarily. What might get you is the different VRM. Intel went from 90nm to 65nm all on LGA775. But from Prescott to Conroe, the voltages needed changed. Thats why 915P chipsets don't work with Core 2 Duo's.
November 29, 2006 3:56:56 PM

Quote:
Not necessarily. What might get you is the different VRM. Intel went from 90nm to 65nm all on LGA775. But from Prescott to Conroe, the voltages needed changed. Thats why 915P chipsets don't work with Core 2 Duo's.


915 chipsets don't work with C2Ds becuase they do not support dual-core, but not the VRM problem.
November 29, 2006 3:58:29 PM

You're telling me the 915 VRM can support the Core 2?
November 29, 2006 4:01:22 PM

Quote:
I'm hating Intel for this simple reason ... I was going to build a e6600 (C2D) system after the holidays (hoping for DDR2 price drop, and mobo's etc) - so now, I see, I need to wait on the Penryn chips as these 45 nm C2D chips will, clearly require a different socket ... right?

Or maybe the 65 nm stuff will drop enough in price that it's a much better deal to build with the current socket chips and just OC and pocket the difference. Does this sound right? :?


Yes that sounds right. The low end of core2 are cheap anyway. They offer the best for your money. You won't see these Penryn's around for quite a while. The only other option i can think of is play the wait game, where you wait forever for the new thing and never buy anything :p 

No point hating them though, they could put it off for another year, but that would just mean technology would advance slower. That's not really feasible either.
November 29, 2006 4:03:11 PM

Quote:
You're telling me the 915 VRM can support the Core 2?


Grantsdale (915 series) and Alderwood (925X) do not support dual-core due to the silicon design.

The reason for that is that Asus tried to use 875P to power up dual Xeon processors. Then Intel designed Grantsdale and Alderwood so that they cannot support multi-core designs.
November 29, 2006 4:07:22 PM

Penryn wouldn't necessarily require a new socket but that doesn't mean a new one might not come out for it and that would eventually support Intel's native quad core down the road. Right now what to watch out for is Intel's new chipsets that are coming as you might want to wait for them. Not that big of a deal but upgrade-ability could definitely come into play.
November 29, 2006 4:08:31 PM

You keep missing my point. I'm saying the 915 VRM cannot support Core 2. You keep coming at me with the fact that I already know, that it cannot support Dual Core. Do you understand me yet? I'm talking about the VRM. Nothing else.
November 29, 2006 4:15:13 PM

Quote:
You keep missing my point. I'm saying the 915 VRM cannot support Core 2. You keep coming at me with the fact that I already know, that it cannot support Dual Core. Do you understand me yet? I'm talking about the VRM. Nothing else.


If you know that a Athlon64 cannot be plugged in a Socket 478 motherboard, will the motherboard be prepared (or to be prepared) for the voltages (1.1V / 1.3-1.35V) for Athlon64s?
November 29, 2006 4:19:06 PM

I give up. My point was pretty simple yet you complicate it so much. The analogy is quite simple. 915P is a 775 chipset as is 975. I was using the different VRM's in each chipset in this case to illustrate the difference and the possible problems that would occur when moving from 65nm to 45nm. Seriously. It wasn't supposed to take up that many posts. *sneeze*
November 29, 2006 4:20:29 PM

Quote:
I give up. My point was pretty simple yet you complicate it so much. The analogy is quite simple. 915P is a 775 chipset as is 975. I was using the different VRM's in each chipset in this case to illustrate the difference and the possible problems that would occur when moving from 65nm to 45nm. Seriously. It wasn't supposed to take up that many posts. *sneeze*


I don't think there are VRM-11 powered 915s.

Also 915 chipset will be discontinued very soon.
November 29, 2006 4:21:22 PM

Quote:
You keep missing my point. I'm saying the 915 VRM cannot support Core 2. You keep coming at me with the fact that I already know, that it cannot support Dual Core. Do you understand me yet? I'm talking about the VRM. Nothing else.


If you know that a Athlon64 cannot be plugged in a Socket 478 motherboard, will the motherboard be prepared (or to be prepared) for the voltages (1.1V / 1.3-1.35V) for Athlon64s?

Where the bloody hell did you hear that from?

They haven't had the same socket since the days of the K6-2!
November 29, 2006 4:23:07 PM

*Sniff* Do you get my point though? I don't give a flying boat about if there is a VRM-11 for 915P. I was trying to show the different VRM's for different process sizes even among the same chipset. Ah what the hell. I'm gonna go find that flu medicine.
November 29, 2006 4:23:18 PM

Quote:
You keep missing my point. I'm saying the 915 VRM cannot support Core 2. You keep coming at me with the fact that I already know, that it cannot support Dual Core. Do you understand me yet? I'm talking about the VRM. Nothing else.


If you know that a Athlon64 cannot be plugged in a Socket 478 motherboard, will the motherboard be prepared (or to be prepared) for the voltages (1.1V / 1.3-1.35V) for Athlon64s?

Where the bloody hell did you hear that from?

They haven't had the same socket since the days of the K6-2!

I just use an example to illustrate that motherboard manufacturers will not update the VRM of 915-based motherboards to "fit" newer processors.
November 29, 2006 4:24:18 PM

Quote:
*Sniff* Do you get my point though? I don't give a flying boat about if there is a VRM-11 for 915P. I was trying to show the different VRM's for different process sizes even among the same chipset.


Sorry.... I still can't get your point :(  :(  :(  .
November 29, 2006 4:24:56 PM

Quote:
You keep missing my point. I'm saying the 915 VRM cannot support Core 2. You keep coming at me with the fact that I already know, that it cannot support Dual Core. Do you understand me yet? I'm talking about the VRM. Nothing else.


If you know that a Athlon64 cannot be plugged in a Socket 478 motherboard, will the motherboard be prepared (or to be prepared) for the voltages (1.1V / 1.3-1.35V) for Athlon64s?

Where the bloody hell did you hear that from?

They haven't had the same socket since the days of the K6-2!

I just use an example to illustrate that motherboard manufacturers will not update the VRM of 915-based motherboards to "fit" newer processors.


Ahh that makes sense...although the AMD/Intel crossover idea wasn't too good!
November 29, 2006 4:25:40 PM

I don't think I could break it down any more without patronizing you.
November 29, 2006 4:26:39 PM

Bit o' good ole patronizing aint that bad every now and then!
November 29, 2006 4:28:34 PM

Quote:
I don't think I could break it down any more without patronizing you.


I just wonder why AMD boards can go smoothly over 130nm / 90nm / 65nm CPUs without changing the powering module. My guess is from the Cool'n'Quiet idea. But it seems that the support of new 65nm CPUs will "require" a BIOS update.
November 29, 2006 4:33:51 PM

It was still the same m/arch. Core 2 and Netburst are polar opposites. Not to mention different m/archs.
November 29, 2006 4:34:25 PM

Wow looks like Intel wants to compete with itself. Bad for us consumers, since our cpu will now gets outdated really fast. 8O
November 29, 2006 4:34:45 PM

Quote:
I don't think I could break it down any more without patronizing you.


I just wonder why AMD boards can go smoothly over 130nm / 90nm / 65nm CPUs without changing the powering module. My guess is from the Cool'n'Quiet idea. But it seems that the support of new 65nm CPUs will "require" a BIOS update.

Didn't AMD just switch to AM2 anyway.
They did 130/90 with 939, 90/65 with AM2.

At the end of the day all that matters is that the end user has to buy a new mobo.
November 29, 2006 4:35:09 PM

Quote:
It was still the same m/arch. Core 2 and Netburst are polar opposites. Not to mention different m/archs.


To be honest Core 2 has a lot more in common with Pentium 3 m/arch than Netburst, in my opinion.
November 29, 2006 4:35:22 PM

Quote:
It was still the same m/arch. Core 2 and Netburst are polar opposites. Not to mention different m/archs.


Same architecture......
Northwood => Prescott
Prescott => Cedar Mill

The transitions required some modifications in the power modules. :( 
November 29, 2006 4:38:31 PM

Quote:
It was still the same m/arch. Core 2 and Netburst are polar opposites. Not to mention different m/archs.


To be honest Core 2 has a lot more in common with Pentium 3 m/arch than Netburst, in my opinion.

Without Netburst, Core 2 will not be as successful as now.
There are some small, but important features of Core arch. from Netburst:
1. Extremely good pre-fetching logics
2. Quad-pumped FSB
3. Streaming SIMD extensions.
......
November 29, 2006 4:42:54 PM

Quad pumped FSB isn't really a great thing as such, HTT would be much better.
November 29, 2006 4:53:12 PM

Well, let's just cross our fingers and hope that intel/nvidia anticipated this. Maybe the new digital VRM's rolling out for 680i and R600 will be able to cope. If the architecture isn't changing, only the process size, then the only impact should be reduced voltage requirements, right? Or are there subtle architecture changes as well?
a c 471 à CPUs
a c 115 å Intel
November 29, 2006 6:05:01 PM

Quote:
I'm hating Intel for this simple reason ... I was going to build a e6600 (C2D) system after the holidays (hoping for DDR2 price drop, and mobo's etc) - so now, I see, I need to wait on the Penryn chips as these 45 nm C2D chips will, clearly require a different socket ... right?

Or maybe the 65 nm stuff will drop enough in price that it's a much better deal to build with the current socket chips and just OC and pocket the difference. Does this sound right? :?


I'm no tech expert, but I expect Penryn will be compatible with LGA775. I could possibly see a new chipset for it, but I think the current generation of P965 and 975X chipsets should be fine. I do hope that Penryn will not require mobos manufacturers to update the VRM on current mobos; I would like to drop in a Penryn into my MSI 975X Platinum when I update my HTPC to Core 2 Duo (from an Athlon XP-M 2600+).

I'm sure the 65nm Conroes will drop in price by the time Penryn is avaiable. I would like to get a Penryn for increased performance and lower power consumption. I like to give a helping to F@H as long as the hit on my electricity bill isn't too much.
November 29, 2006 7:36:05 PM

Quote:
Wow looks like Intel wants to compete with itself. Bad for us consumers, since our cpu will now gets outdated really fast. 8O

It's not the first time they do so; back to the days of the pentium pro, they boasted the first release, and then, soon after they sent out the MMX enhanced version.
November 29, 2006 7:44:16 PM

well right now, unfortunately AMD's not doing much in the way of competition.
November 29, 2006 7:46:05 PM

Quote:
It was still the same m/arch. Core 2 and Netburst are polar opposites. Not to mention different m/archs.


To be honest Core 2 has a lot more in common with Pentium 3 m/arch than Netburst, in my opinion.

Without Netburst, Core 2 will not be as successful as now.
There are some small, but important features of Core arch. from Netburst:
1. Extremely good pre-fetching logics
2. Quad-pumped FSB
3. Streaming SIMD extensions.
......
You forgot HT, however, they were imposed evolutionary steps and as we see, both AMD and Intel have found a way to adjust efficiency at this point, the only thing to meke a difference remains raw power and IPC of the processing units themselves.
It may be as you say but it might even have been better if they were forced earlier to push on th P3 ->Core ->Core2 path.
November 29, 2006 7:50:24 PM

Quote:
well right now, unfortunately AMD's not doing much in the way of competition.

They were sitting, like intel did for years and inertia is still holding them back, I am curious to see the performance of the 65nm athlons (rumoured to be out Decembet the 5-th). By no way will they be the C2D killers but at least are supposed to get nearer in performance and offer some price competition too,... hopefully!
November 29, 2006 7:55:57 PM

Quote:
I'm hating Intel for this simple reason ... I was going to build a e6600 (C2D) system after the holidays (hoping for DDR2 price drop, and mobo's etc) - so now, I see, I need to wait on the Penryn chips as these 45 nm C2D chips will, clearly require a different socket ... right?

Or maybe the 65 nm stuff will drop enough in price that it's a much better deal to build with the current socket chips and just OC and pocket the difference. Does this sound right? :?


I'm pretty sure there's a socket change on Penryn, and why wait? it's never going to get any better as far as that stuff goes with computers.
December 1, 2006 1:01:47 AM

There will be no performance difference between AMDs 65nm chips vs their 90nm chips when clocked at the same speed. Its a dumb shrink w/o any other architecture changes. The only improvement would be power consumption and usually it resulted in the ability to increase clockspeed but it doesn't seem like it will give AMD that much more headroom that one would have expected.
December 1, 2006 1:20:38 AM

Quote:
There will be no performance difference between AMDs 65nm chips vs their 90nm chips when clocked at the same speed. Its a dumb shrink w/o any other architecture changes. The only improvement would be power consumption and usually it resulted in the ability to increase clockspeed but it doesn't seem like it will give AMD that much more headroom that one would have expected.


So I see you've gotten a 65 nm cpu, can you post your benchmarks for us?
December 1, 2006 1:39:10 AM

I see that you just don't understand what's going on here, this is simply a "dumb" die shrink...which means absolutely no changes in the architecture, which means there is no increase in performance, therefore the only improved result is a lower power consumption. usually though with a die shrink the real advantage is the increase ability to run at higher speeds due to the lower power. (Just like GDDR4, and such.)
December 1, 2006 1:45:08 AM

Quote:
Quad pumped FSB isn't really a great thing as such, HTT would be much better.


Why? As we have seen, the FSB currently causes no bottlenecks.

This of course, is not to say CSI will not be a relief.
December 1, 2006 2:23:23 AM

isnt the bearlake mobo's made for the 45nm c2d? 1333fsb


the only things sad is that Bearlake X og Bearlake G+ (IGP) wont support crossfire :\ tho i would probably not buy 2 gpu's anyway but the opportunity would still be good
December 1, 2006 2:37:33 AM

Quote:
Wow looks like Intel wants to compete with itself. Bad for us consumers, since our cpu will now gets outdated really fast. 8O


Last thing i remember in computing was HYPER THREADING was the new "it" and only a few years from the 90Nm Prescott's we're already on the Dual/Quad-core 45nm.
By the time AMD decides to shift gear into 45nm (18 months from now) Intel might already be sampling 32nm-16WMax chips, or maybe even mass producing them?? :? <
December 1, 2006 2:37:47 AM

Quote:
I'm hating Intel for this simple reason ... I was going to build a e6600 (C2D) system after the holidays (hoping for DDR2 price drop, and mobo's etc) - so now, I see, I need to wait on the Penryn chips as these 45 nm C2D chips will, clearly require a different socket ... right?

Or maybe the 65 nm stuff will drop enough in price that it's a much better deal to build with the current socket chips and just OC and pocket the difference. Does this sound right? :?


I wouldn't worry about it too much Jake. Intel is sampling 45nm, which bodes very well for their process, as they are right on the schedule on their older roadmaps, but I seriously doubt anyone will be able to buy a retail 45nm chip until at least July 2007.

We will start to see some leaked engineering samples in the spring, which I'm very interested to see the improvements in multimedia workloads with the 40-50 new SSE4 instructions, but yeah you'll probably need a bearlake mobo at least for penryn.
December 1, 2006 2:44:58 AM

Quote:
I see that you just don't understand what's going on here, this is simply a "dumb" die shrink...which means absolutely no changes in the architecture, which means there is no increase in performance, therefore the only improved result is a lower power consumption. usually though with a die shrink the real advantage is the increase ability to run at higher speeds due to the lower power. (Just like GDDR4, and such.)

I understand fully what's going on here, what you're talking about is well illustrated in articles such as this...
http://www.tomshardware.com/2005/10/07/a_sneak_peak_at_...
But I'm just saying that you haven't seen a Brisbane 65 nm yet so why speak of it when there is a chance that you're wrong? When I'm most positive some minor changes have been made to the core?
December 1, 2006 3:32:20 AM

on the bright side all Dual Core Mobos Support Quad............
December 1, 2006 3:40:30 AM

Quote:
40-50 new SSE4 instructions

I'm curious to find if you're referring to:
"Implementation and possibly adding extensions of SSSE3 (which was called 'SSE4' prior to its official name announced) and/or SSE4 , which AMD codenamed SSE4a."
Or a new instruction set? Just wondering, because if you're referring to SSSE3 then that'll be on the K8l architecture.
December 1, 2006 4:23:05 AM

Well, some AMD fanboy might have started a thread about this already but at least AMD has sampled 65nm native Quad opty's.

LINK
December 1, 2006 7:50:02 AM

Quote:
I'm hating Intel for this simple reason ... I was going to build a e6600 (C2D) system after the holidays (hoping for DDR2 price drop, and mobo's etc) - so now, I see, I need to wait on the Penryn chips as these 45 nm C2D chips will, clearly require a different socket ... right?

Or maybe the 65 nm stuff will drop enough in price that it's a much better deal to build with the current socket chips and just OC and pocket the difference. Does this sound right? :?


I'm pretty sure there's a socket change on Penryn, and why wait? it's never going to get any better as far as that stuff goes with computers.

Mmmmm, Penyrn is the mobile update of the Merom processor. The Wolfsdale is the desktop update of the Conroe. There is no socket change needed for Wolfdale. On the other hand Intel is planning to come out with X33, X35 cancelled and X38 chipsets. These are to support the new and improved 1333 FSB along with half multipliers. If they keep the same VRM11 then as long as the bios has the correct uCode in it and the 965 or 975 chipset can reach the 1333 FSB speed then the our current Intel chipset based platforms can run the 45nm Wolfsdale.
!