backtop

Hardware Intel 45nm "Penryn" Tape-Out Runs Windows
Sven Olsen (Blog) - January 10, 2007 2:13 AM
Print E-mail del.icio.us 83 comment(s) - last by SacredFist.. on Jan 30 at 10:07 PM

Intel engineers celebrate after "Penryn" boots Windows
Intel's 45nm CPUs might get here sooner than later

Sources inside Intel have confirmed the company recently received the taped-out 45nm Penryn processor.  Reports from the ensuing champagne toast claim the first Penryn boot was capable of booting Windows.

Even though the processor is labeled as "A0" silicon, or first revision, the successful boot is a huge milestone and bonus for Intel.  Reports of operational first-run tape-outs are few and far in-between, especially in the CPU industry.

Penryn, the 45nm optical-shrink of the Core architecture, was prepped for tape-out in late November, and returned to the Intel development team just a few weeks ago. 

The early tape-out may even have beneficial consequences further down the line.  Any early progress will pull the launch date in several weeks, though Intel has not publicaly stated a target date for launch.  The latest Intel roadmaps claim Penryn is simply a "late 2007" processor.

Comments     Threshold


This article is over a month old, voting and posting comments is disabled
feeling the squeeze
By Spoelie on 1/10/2007 4:42:27 AM , Rating: 5
AMD should be feeling pretty squeezed at the moment. I have every confidence in their ability to compete and outdo Intel in chip design, but if they, from this point on, will always be a process node behind, it's still almost impossible to compete. The reason why the 90nm AM2 processors didn't extend the performance of the cores, is because between the dual cores, the new memory controllers, the new virtualization etc. features, they are on too tight a transistor budget. That's why they so desperately need the die shrinks to happen.

Intel had that transistor budget from the beginning of last year (launch of Yonah), while AMD will only ship some proper 65nm in the middle of this year, one and a half years behind. And by that time, Intel will be on the verge of increasing their transistor budget another time, if their schedule isn't overly optimistic.

I'll always have a special place for AMD, but they better do not screw up on their execution the coming year.




RE: feeling the squeeze
By Brand0 on 1/10/2007 5:26:35 AM , Rating: 2
I agree. AMD was dominating the gaming processor market but is now playing catch-up. They need to pull it together.


RE: feeling the squeeze
By BladeVenom on 1/10/2007 7:05:34 AM , Rating: 5
You need to give AMD some time. Look how long it took Intel to come out with something better than AMD's Athlon 64s.


RE: feeling the squeeze
By nah on 1/10/2007 8:09:10 AM , Rating: 1
AMD has been the leader since Athlon's debut in 1999--test after test showed that it beat the heck out of the orig P IIIs--even with the Coppermines it had a 33 % advantage in FP ops--if it wasn't SSE optimised ( the old 3Ds Max 6)--they repeated this in 2003 with the Athlon 64--Intel responded finally in 2006(sigh)--can AMD repeat this trend over the next year with K8L---lets see


RE: feeling the squeeze
By Master Kenobi (blog) on 1/10/2007 8:30:34 AM , Rating: 1
quote:
AMD has been the leader since Athlon's debut in 1999

Leader of what?


RE: feeling the squeeze
By Samus on 1/10/2007 8:43:16 AM , Rating: 3
Well...performance. duh.


RE: feeling the squeeze
By Master Kenobi (blog) on 1/10/2007 8:46:51 AM , Rating: 3
quote:
Well...performance. duh.

The word "performance" makes a huge difference in the validity of his statement. When your leading in a race that is measured by many factors, its a good idea to state exactly what your leading in.


RE: feeling the squeeze
By ScythedBlade on 1/10/2007 5:25:46 PM , Rating: 3
Performance ...

Well, actually, Intel creamed AMD almost entirely with Northwood. They actually took back the performance crown until A64 came out. Of course, even when A64s came out, they were pretty much tied. Intel was better for comiling stuff, while AMD was better for gaming, in that area. For a while, there wasn't too much of a clear winner until the first tier dual cores.

However, the Core 2 Duos were a definitely a different story. They outright won like crazy.


RE: feeling the squeeze
By ScythedBlade on 1/10/2007 9:38:38 PM , Rating: 2
... I got voted down for stating the truth? Sucks ...
But technically, its true. Pentium 4's owned in media creationg (all except LAME Mp3, which was single threaded). Anyway, I would post links, but that guy already did.


RE: feeling the squeeze
By Pirks on 1/11/07, Rating: 0
RE: feeling the squeeze
By mindless1 on 1/11/2007 7:17:23 PM , Rating: 2
No, technically it's misleading. Joe Benchmarker likes to use new premium software to show off the latest CPU but nevermind if that is what the end users were running. Further, I would speculate that most people are not spending the bulk of their time content-creating. It's somewhat surprising that not so many people even consider this issue, as if we are compelled to change everything to revolve around the CPU we bought.


RE: feeling the squeeze
By ScythedBlade on 1/30/2007 10:06:31 PM , Rating: 2
XD, the key word here is "technically"


RE: feeling the squeeze
By InsaneScientist on 1/27/2007 2:54:49 PM , Rating: 2
Well, sorta....
The Northwood P4s as a whole did cream AMD in encoding (of any sort), but they didn't really take back the all around performance crown until the Revision where P4s with 800MHz FSBs started showing up... (Revision C, I think?)
IMHO, that was when they really took the crown back. They were close before, but it wasn't really until then that they actually surpassed the Athlon XPs.

With the majority of the Athlon 64s, especially the more recent revisions, the lead that the P4s/PDs had over the A64s/X2s in encoding wasn't as significant as the A64/X2's lead over the P4/D in just about everything else... (With the early Athlon 64s, the P4 did have a huge lead in encoding) so they weren't quite equal, but I don't think that Intel really ever allowed AMD to relax.

And then Core2 came out and massacred everything. ;)


RE: feeling the squeeze
By SacredFist on 1/30/2007 10:07:51 PM , Rating: 2
of gaming you mean


RE: feeling the squeeze
By StevoLincolnite on 1/10/2007 9:25:25 AM , Rating: 4
http://www.aceshardware.com/read.jsp?id=55000281
http://www.anandtech.com/showdoc.aspx?i=1834&p=8
http://www.anandtech.com/showdoc.aspx?i=1595&p=6
http://www.anandtech.com/showdoc.aspx?i=1615&p=13
http://www.aceshardware.com/read.jsp?id=55000281
http://www.tomshardware.com/2003/06/23/bidding_adi...

Pentium 4 3.2ghz vs Athlon XP 3200. (AThlon 64 wasnt released.
And actually your wrong about the Pentium 3, The Pentium 3 coppermine managed to beat the Athlon in most tests, That is until AMD moved the cache on die.
The Pentium 3 vs Athlon was never clear cut, Alot of technology's were brought in.

So no, AMD hasn't been the leader for 7 - 8 years.


RE: feeling the squeeze
By Master Kenobi (blog) on 1/10/2007 10:11:33 AM , Rating: 2
I believe its more in line with a 4-5 year lead.


RE: feeling the squeeze
By gramboh on 1/10/2007 1:01:37 PM , Rating: 2
Nope, 4 years ago the Northwood P4 "C" was the best performing CPU. AMD had the lead for roughly 2.5-3 years with A64 until Core 2 Duo came out.


RE: feeling the squeeze
By Master Kenobi (blog) on 1/10/2007 4:46:40 PM , Rating: 2
I stand corrected then.


RE: feeling the squeeze
By Tsuwamono on 1/12/2007 12:02:04 AM , Rating: 2
no, you were correct initially. You just failed to state which field they had the lead in. Gaming performance AMD has been better since Athlon XP. I would almost say since Athlon with the 900mhz core. I had one of those with a 9600 and it beat my buddys P3 and then when he upgraded to P4(i think it was one of the originals) it still ended up pretty close but my Athlon 900mhz still came out on top in gaming.

You could argue its the mobo or ram or whatever but all i know is my Athlon 900Mhz with the same video card as him was producing a better FPS then his Early P4 and late P3


RE: feeling the squeeze
By carl0ski on 1/10/2007 8:24:12 PM , Rating: 2
Athlon Thunderbird outperformed Pentium 3 Coppermine
mainly due to higher clock speeds
1.4 vs 1.13

Tualatin was very promising it reached 1.4ghz and lower voltages however was abrutly discontinued to cover up the fact it out perform new flag ship product Pentium 4 willement 1.6Ghz


I dont care what the word on the street is
Pentium M = Upgraded Pentium 3 Tualatin


Athlon XP was way out of it's depth the moment Pentium 4 hit 3.06Ghz
it just could scale past 2.2Ghz very well.

Enter Athlon64


Depends what you mean by leader
AMD didn't see considerable increases in sales until 2001.


RE: feeling the squeeze
By StevoLincolnite on 1/11/2007 2:03:07 AM , Rating: 2
Considering the Pentium 3 Coppermine 1.13Ghz was recalled...
Like I said the Pentium 3 VS Athlon was never clear cut.


The Pentium M was a hybrid of technologies from the Pentium 3 Tualatin and Pentium 4. Its still based upon the P6 Architecture found in the Origional Pentium Pro.
I'll make it easier, The Pentium M core is the Pentium 3, Intel added the 400/533/800 etc Frontside buses to the processor, Upgraded the cache from 256k to 1Mb/2Mb/4Mb etc.
Then They borrowed the Pentium 4 Branch Tree Predictor and added it on, Parts of the Pentium M cache were altered so that when that part of the cache is not in use, it shuts it off completly, Giving it lower power consumption.
The Barton (Athlon XP 3200) Was just a stop gap measure to steal Some of intls thunder. In general there has NEVER been an entire series of processors that have been on top for there entire life time.

And untill AMD releases there counter to the core series, Intel will remain on top.
And your wrong About the Athlon XP scaling past 2.2ghz, I managed to overclock mine to a respectfull 2.4ghz.
And that was on the crap tastic stock air cooling.


RE: feeling the squeeze
By stromgald on 1/10/2007 11:34:08 AM , Rating: 3
quote:
You need to give AMD some time. Look how long it took Intel to come out with something better than AMD's Athlon 64s.


AMD has a lot less room and time to work with compared to Intel. Even though AMD was in the lead from at least 2003 - mid-2006, Intel was still the dominant chip maker during that time since Dell and other companies basically committed to selling Intel.

Although that stranglehold by Intel has disappeared, AMD was dependent on holding that performance crown and won't survive for as long as Intel without it. IMHO, Dell and HP are now selling AMD because they just want to keep two chip manufacturers around. This is very similar to how even though Lockheed won the F-35 contract, Boeing is still building like 40% of the aircraft and getting paid for it.

If the K8L isn't competitive with the new quad cores coming from Intel this year, AMD might be basically on life support.


RE: feeling the squeeze
By carl0ski on 1/10/2007 8:40:32 PM , Rating: 2
AMD has made a long stretched business of being second fiddle

Being best got their name in the spot light and their brand name known

You don't need to be the biggest and best to have a healthy business.

If need be they can just fall back on their brand name to sell lower grade (cheaper to manufacture) chips.
and still make a healthy turn over


RE: feeling the squeeze
By wien on 1/10/2007 7:58:26 AM , Rating: 1
Would it be possible for AMD to skip an entire process node in order to "catch up" to Intel? (Skip 45nm for instance.) Or is the knowledge gained by shrinking your process imperative to making the next shrink work?


RE: feeling the squeeze
By Master Kenobi (blog) on 1/10/2007 8:07:07 AM , Rating: 3
Skipping a process is generally a no-go idea. Since your engineers have to redesign a bit with each shrink. Shrinking such a large amount would send them back to the drawing board for years trying to figure out how to build something on a much much smaller die. Die shrinks show you potential problems with the next shrink, and give you ample time to find solutions to those problems. Technologies like Strained Silicon, etc.... are the result of people seeing an issue with the current process and realizing that shrinking it again would make this issue a huge problem, and thus must be planned for.

In short, no, you don't skip a process node.


RE: feeling the squeeze
By thereaderrabbit on 1/10/2007 9:14:01 AM , Rating: 2
So just what is IBM doing?


RE: feeling the squeeze
By Master Kenobi (blog) on 1/10/2007 10:00:36 AM , Rating: 2
IBM is also on the 65nm node, (See Xbox 360 Revision 2). IBM is also currently working on a 54nm process. Intel's however is already in production, so Intel likely has a 6-8 month lead.


RE: feeling the squeeze
By Master Kenobi (blog) on 1/10/2007 10:12:11 AM , Rating: 4
That "54" should be "45". Blasted Typos.


RE: feeling the squeeze
By thereaderrabbit on 1/11/2007 11:33:18 AM , Rating: 2
IBM contracts out the 65nm node to Charter semi, while they focus their fab on 45nm and beyond for in-house. They seem to be planning to skip 65nm in-house.


RE: feeling the squeeze
By therealnickdanger on 1/10/2007 9:24:46 AM , Rating: 5
Yeah dude, haven't you ever played an RTS? You can't skip technology leaps... :D


RE: feeling the squeeze
By therealnickdanger on 1/10/2007 2:33:50 PM , Rating: 2
"an RTS" --> "a RTS"

I am teh suk @ gramer.


RE: feeling the squeeze
By Smoza on 1/10/2007 6:24:52 PM , Rating: 2
quote:
I am teh suk @ gramer.


and spelling it would seem...


RE: feeling the squeeze
By danrien on 1/10/2007 8:04:19 AM , Rating: 2
In general, AMD has always been behind Intel on their process nodes. In fact, I can't recall when they've been ahead.


RE: feeling the squeeze
By Master Kenobi (blog) on 1/10/2007 8:10:09 AM , Rating: 2
quote:
In general, AMD has always been behind Intel on their process nodes. In fact, I can't recall when they've been ahead.


Correct, AMD has never been ahead of Intel on the process front. They were for a few years better architecturally. This has in recent times slipped(See Core2 Microarchitecture http://en.wikipedia.org/wiki/Intel_Core_microarchi...


RE: feeling the squeeze
By Spoelie on 1/10/2007 9:09:41 PM , Rating: 2
Yes they were always behind, but as far as I can remember, a few months (6?) at most. This is one year and a half, and at a point where their architecture isn't 'compensating' for the process gap...


RE: feeling the squeeze
By Viditor on 1/12/2007 2:39:54 AM , Rating: 2
quote:
This is one year and a half


Intel is shipping their 45nm at the end of 07...this means they will start manufacture in Q3 07.

AMD is shipping 45nm in mid 08...this means they will be starting manufacture at the beginning of 08.

AMD is still only 6 months behind...


RE: feeling the squeeze
By sprockkets on 1/10/2007 11:24:54 AM , Rating: 2
Agree. Though aside from AMD and now IBM, I have not heard of anyone else doing 65nm yet. Intel really is steam rolling not just AMD, but everyone.

The weird part, is like, aside from adding more cores, does anyone need them to be operating at such a ridiculous pace? Oh, wait, we don't care, we benefit from them spending billions on 45nm fabs!


RE: feeling the squeeze
By Hoser McMoose on 1/10/2007 3:23:15 PM , Rating: 2
TSMC and Chartered Semi are also doing 65nm production now. I don't know about UMC, but they might be as well. However all of them were well behind Intel in first rolling out their 65nm products and will likely to continue to trail Intel in the move to 45nm.

As for the reason for moving to 45nm fabs, if all else is equal chips made on these new fabs should eventually be cheaper. Yeah, there is a large up-front capital cost involved with the change, but the per-chip costs drops due to the smaller die size. Certainly for some chips and some manufacturers that is the key reason for the switchover (case-in-point: Intel's Celeron chips, no performance reason for them to be made on 65nm fabs, but it's cheaper than making them on a 90nm fab). The new chips might also be able to clock higher and/or consume less power as well, which is a good thing. And worst comes to worst, you can always just throw more cache at the thing.


AMD needs to pull their head out...
By aguilpa1 on 1/10/2007 8:58:46 AM , Rating: 1
after years of performance leading architecture, instead of working on a replacement for what was an already aging chip architecture, they kept milking the cow. I guess AMD figured Intel would take another 2 years to recover from the netburst fiasco. Now what do they have 4x4? whatever! They need a whole redesign of that tired K series and a rename, not just a die shrink.

A die shrink is a luxury only Intel can afford to play with right now. So they are.




By Master Kenobi (blog) on 1/10/2007 9:08:24 AM , Rating: 2
Well, to be fair AMD will need atleast 12-18 months to mature their 65nm process, which by the way is not quite there yet, it still has some minor adjustments that need to be made. Once they square it all away then they can think about moving to 65nm. Remember Intel has been on 65nm for over a year now. AMD just got there. Intel also has far more money, R&D, and Fab capacity to play around with. They are already toying around with 32nm and they havent even launched 45nm into the market yet. Intel just has the power to finance and maintain multiple large R&D projects at once. No surprise the 45nm process is here so quickly, I would expect Intel to be this far with their 32nm process in Q4'08.


RE: AMD needs to pull their head out...
By Goty on 1/10/2007 9:26:45 AM , Rating: 1
K8L is not just a die shrink. Learn the facts.


RE: AMD needs to pull their head out...
By Master Kenobi (blog) on 1/10/2007 10:10:25 AM , Rating: 2
quote:
K8L is not just a die shrink. Learn the facts.


It is also not a large departure from the current K8 architecture. It will be a K8 with several technologies updated, and possibly larger cache and slightly higher clock speeds. It is mostly a K8 revision designed and built on a 65nm process, instead of the K8 which was built and designed for the 90nm process.

K8 and K8l would be like building a similar car, but one generation newer, using updated technologies of what was found in the older one. Very little in the way of new features will be added.


By nurbsenvi on 1/10/2007 11:46:22 AM , Rating: 2
Would K8L be able to execute 4 instructions per clock cycle like the Core 2s?

I heard that K7/8 was able to handle 3 instructions Per clock cycle where else P4 was able to handle only 2 or 2.5 instructions at a time. So highly clocked P4 struggled to beat K7 just like K8 struggling with Core2.




RE: AMD needs to pull their head out...
By johnsonx on 1/10/2007 3:37:25 PM , Rating: 2
quote:
K8 which was built and designed for the 90nm process.


No, K8 was designed for a 130nm process. It was nearly 2 years after K8 first launched that they transistioned to a 90nm process.


RE: AMD needs to pull their head out...
By Master Kenobi (blog) on 1/10/2007 4:50:22 PM , Rating: 2
I was referring to the Athlon64 series, not the AtlonXP chips.


By Smoza on 1/10/2007 6:19:43 PM , Rating: 2
quote:
I was referring to the Athlon64 series, not the AtlonXP chips.


A64 was released on a 130nm process, see clawhammer and newcastle http://en.wikipedia.org/wiki/Athlon_64#Clawhammer_...

It wasn't until Winchester was released that A64 was based on 90nm process.


RE: AMD needs to pull their head out...
By Viditor on 1/11/2007 1:38:29 PM , Rating: 2
1. AMD's 65nm is quite mature and was at launch. AMD tweaks their process all the time (they are still tweaking 90nm as well), as does Intel. They will be all 65nm by mid 2007 (talk about a fast ramp!)...

2. K8L is a name given by the Inquirer, not AMD. The chip being released in Q2 is the K10 according to AMD, and it's as similar to the K8 as Core2 is to the Piii.
New features include:
32B Fetch
Enhanced Branch prediction
Out-of-order loads
Up to 4 DP FLOPS/cycle
Dual 128-bit SSE dataflow
Dual 128-bit loads/cycle
Additional HT links (HT-3)
Quad Crossbar
Enhanced power management

3. Being able to retire 4 ops/clock sounds really cool, but I strongly doubt that anyone here is running software that will let them actually do that (most all max out at 2.5-3 clocks/cycle).


By Tsuwamono on 1/12/2007 12:11:56 AM , Rating: 2
Thank you for that. Its nice to see someone post some actual facts for a change.

(google it if you dont believe him)


Does it really make a huge difference today?
By Vesuvius on 1/10/2007 12:44:09 PM , Rating: 2
I remember when CPU speed and a very critical thing for consumers and game enthusiasts. But today I didn't think it really made a huge difference. Am I wrong?

I built a AM2 4200+, got the CPU for about 130 on a frys deal. From everything I looked at. A $300+ C2duo (supposedly the reigning champ) would get me 2-3 frames in any game. Split seconds it application boot up. And maybe a few min in any encoding.

At this point both companies have extremely competitive products that are at times so far ahead of the game it really doesn't make a difference in the real world. So Intel is ahead of the game, thats great. But I wouldn't say that AMD is lagging behind by any means.

Just my opinion maybe I am wrong. But in my eyes these two companies are well matched with no absolute champ of the CPU.





RE: Does it really make a huge difference today?
By Smurfer2 on 1/10/2007 12:50:41 PM , Rating: 2
Uh, yes... As I rebrowse through the Core 2 Duo article, a $300 E6600 increases FPS over your 4200 X2 by at least 25fps... That is not a small margin.

http://anandtech.com/cpuchipsets/showdoc.aspx?i=27...

Actually, reread that whole article. The E6600 is significantly faster.

Intel is well in the lead at the current time.


RE: Does it really make a huge difference today?
By rhog on 1/10/07, Rating: -1
RE: Does it really make a huge difference today?
By Thorburn on 1/11/2007 9:23:25 AM , Rating: 2
quote:
In terms of heat the Conroe processor is a Hot Processor, much more so than the AMDs and this is most likely due to the large caches.


The Core 2 Extreme X6800 has a TDP of 80w with real world measurements showing peak draw to be closer to 70w.
The Athlon FX-62/70/72/74 all have a 120-130w TDP....

The only Core 2 over 100w is the QX6700 (and perhaps the Q6600, although the 2.33GHz quad core Xeon is 80w)

Cache doesn't consume a great deal of power and can easily be switched to lower power states when the CPU is idle. The Core 2 architecture is even able to flush the contents of L2 into memory and power off the cache completely at very low CPU loads.


RE: Does it really make a huge difference today?
By rhog on 1/11/2007 1:33:03 PM , Rating: 2
If you read my response you would know that what I am taling about is "Real World" heast values. Yes you can power down the cache when it is not in use but when you are using the procesor you need the cache! I just assumed the cache was the source of the heat as I could not think of a better reason. I am not an EE but many EEs I speak to indicate that the cache and its associated high transistor count are what cause a lot of the heat. I don't really care what the specs says I actually have the processors in front of me! I am using the processors in the real world and performing measurements with digital thermometers (I know they are not that accurate but the give me a general trend) the Conroe processors are hotter! I Don't really know why, I don't care what intel or some other sites say. I have many of the procesors myself as due my employees and the Conroe is a hot processor. Just a fact. I have compared the E6600 against a 2.6ghz FX60 as they are similar in clock speed and I can tell you that the E6600 is hotter all the time. Not really that good a deal when you look at the fact that the AMD is a 90nm and the Intel is a 65nm. I also hear the new AMD 65nm are even cooler and I will be testing this fact as well. When you are in an office with 6 or more workstations it really makes a difference and a 10-20 percentage points and a few seconds quicker compile times make very little difference.

In regards to the Quad processors I will wait a bit until I see what AMD has in response (not the FX7x series) and I will wait for the price of the Intels come down a lot. Intel is known for Gluing multiple chips in a single package and calling it a dual (P4) or in this case a Quad (Core 2 Quad). The AMD design "should" be better, and at the current prices of the Intel Quad cores I can wait awhile to see it. Don't get me wrong Intel has done a great job in creating a fast Processor but a gain of 10-20% over the AMD makes it evolutionary and not revolutionary and needing the 65nm or the 45nm process to keep it cool does not give me a lot of confidence. Just my opinion nothing more.


By Vidmar on 1/11/2007 4:01:46 PM , Rating: 2
How are you ignoring the multitude of data that refutes your heat claim?? Having “gut feelings” and about how much heat your processor put out will get you no where with this discussion.

I bet you also believe in 911 conspiracy theories too… Don’t ignore the facts.


RE: Does it really make a huge difference today?
By Einy0 on 1/11/2007 9:14:20 PM , Rating: 2
Intel and AMD measure TDP values differently. Comparing the two TDP values is apples to oranges. AMD calculates based on maximum possible. Intel calculates at average usage.


By Thorburn on 1/13/2007 6:41:15 AM , Rating: 2
This statement is thrown around all over the place and could perhaps hold true for the later Netburst cores, but look at Lost Circuits power measurements (measuring the entire VRM draw, not just the CPU's no less) and you'd see the peak of the X6800 being around 70w, 10w under its TDP....


Intel......
By Master Kenobi (blog) on 1/10/2007 8:02:22 AM , Rating: 2
Well, this goes to show you the Intel process technologies have not lost steam over the years and continue to lead the industry. Now that they are back on track with a solid architecture we should see Intel increase the distance between them and AMD quickly. K8L might be the supposed holy grail with the AMD guys right now, but I have to wonder.... would a 65nm K8L be able to compete with a 45nm Core2 architecture. Somehow I think not.




RE: Intel......
By Goty on 1/10/07, Rating: 0
RE: Intel......
By RamarC on 1/10/2007 10:44:34 AM , Rating: 2
but a lower mfg process usually translates into higher yields and speeds (once the kinks get smoothed out). that means more pricing flexibility and higher performance. k8l will have an uphill climb to keep up.


RE: Intel......
By Phynaz on 1/10/2007 12:02:10 PM , Rating: 2
quote:
since Penryn is otherwise identical to Conroe


Penryn is a little more than a Conroe shrink.


RE: Intel......
By Thorburn on 1/10/2007 1:26:30 PM , Rating: 2
I believe Penryn infact has the L2 cache increased from 4MiB to 6MiB and features SSE4 instructions, so its not 'just' a shrink.

Also if AMD match IPC with there coming architectures that doesn't help them if they have a clockspeed deficit, AMD roadmaps suggested clockspeeds up to 2.5GHz for there next generation parts, last I heard Penryn was shooting for '3.46-3.73GHz'.


RE: Intel......
By VooDooAddict on 1/10/2007 3:24:53 PM , Rating: 2
Depending on Intel's thermal and power flexibility ... it looks like they could already hit those speeds with Conroe.

This is why we need AMD to come out with K8L ... push Intel into a clockspeed push that would hopefully lower the price of the perfectly good 1.8 and 2.1Ghz Conroe and Alendale parts.


RE: Intel......
By Viditor on 1/12/2007 3:08:30 AM , Rating: 2
quote:
we should see Intel increase the distance between them and AMD quickly


It is most likely that when Penryn is released, Intel will need it to be catching up to AMD again.

1. K10 (called by some the K8L) should have a roughly equal core (equal in integer, superior in FP) to Core2, but a quicker platform (due to HT).
2. The boost that the larger cache gives Penryn is to make the latency equivalent to AMD's on-die mem controller.

quote:
would a 65nm K8L be able to compete with a 45nm Core2 architecture


I would think so...

But you are asking the wrong question...a shrink to 45nm and addition of cache won't garner a huge amount for Intel. What the key questions are:

1. Will Penryn have the high-k metal gates as originally planned, or will they be delayed until Nehalem? That process would give Intel some significant headroom...

2. Will DSL be successful on K10? If it is, then AMD will also have some serious headroom...

3. Will CSI be ready by the end of 2008 (even it's cut-down form)?

4. Will AMD be able to utilize Z-Ram for their cache? And when?


But...!
By keitaro on 1/10/2007 3:59:26 AM , Rating: 1
Can it run OS X, Linux, FreeBSD, NetBSD, OpenBSD, or Solaris?! :)

I'm sure they went ahead with Windows since it's the first thing they'd want to test. But the least thing they could've done is help pimp the other operating systems too! *grin*




RE: But...!
By Calin on 1/10/2007 6:16:34 AM , Rating: 5
I'm sure they will try all those operating systems, and maybe others (BeOS) in a "compatibility testing". However, Windows is pretty hard on the processor, and is THE operating system. Also, maybe as of now the new Penryn runs OpenBSD. Why would any PR person would put OpenBSD in a press conference, when they could put Windows?


yikes!
By casket on 1/10/2007 9:02:30 PM , Rating: 2
AMD is already selling 65nm in limited volume. However, this 65nm transition for AMD seems like a crappy job.

In the past... AMD has caught up with Intel on manufacturing, which gave them a 3 to 6 month lead on Intel. AMD is scheduled for 45 nm in 2008.

This is terrible news for AMD. From the advent of Core 2 until 2008, AMD will have a clearly inferior product.
(Core 1 is debateable)

A manufacturing advantage like this could potentially lead to the eventual death of AMD.




RE: yikes!
By Viditor on 1/12/2007 3:42:25 AM , Rating: 2
quote:
this 65nm transition for AMD seems like a crappy job


Why? Keep in mind that AMD is starting production now on the new K10 chips, so much of that 65nm volume will be going for that.
Also, remember that AMD is changing over 2 different things...65nm AND 300mm for all of their lines.

quote:
In the past... AMD has caught up with Intel on manufacturing, which gave them a 3 to 6 month lead on Intel. AMD is scheduled for 45 nm in 2008


Ummm...AMD has never released a die shrink before Intel AFAIK. The 2008 release of 45nm puts AMD roughly 6 months later than Intel (which is better than they have been on this last node).

quote:
From the advent of Core 2 until 2008, AMD will have a clearly inferior product


Is this what your Magic Eight Ball tells you? :)


RE: yikes!
By JumpingJack on 1/16/2007 10:35:24 PM , Rating: 2
You don't have the first clue what you are saying in most posts.... you are funny a guy.


Time to buy Intel stock?
By JNo on 1/10/2007 11:06:23 AM , Rating: 3
Hmmm... looks like it's a good time to buy some intel stock... especially while my £ goes further to the $...

(Can't wait for Viditor's knee-jerk response about how AMD's new technologies and roadmaps mean they're going to kick butt :) And just to clarify, I have never directly owned intel stock but happily made profits owning AMD over a year ago)




RE: Time to buy Intel stock?
By Viditor on 1/12/2007 3:19:40 AM , Rating: 2
quote:
looks like it's a good time to buy some intel stock


Yes, it is...but not because of Penryn (that would be silly!). I happen to have 10k share of Intel myself...
I also have more than 50k shares of AMD, but for a different reason.

Intel shares should do very well for the next 2 quarters, and today's price is a bargain.
AMD's shares are also a bargain right now, but don't expect a big jump for another year or so...

quote:
Can't wait for Viditor's knee-jerk response about how AMD's new technologies and roadmaps mean they're going to kick butt


My knee is jerking, and they are...but not for another 9 months to a year (at least as far as the stock price). :)


K8L really K9
By jay2o01 on 1/10/2007 1:10:12 PM , Rating: 2
K8L will be an updated microarchitecture. Essentially AMD wanted to avoid calling any of their chips K9, so they came up with k8l. In short, myself and the entire industry are hoping for much more than just a simple die shrink (or native quad core for that matter) out of K8L.




RE: K8L really K9
By cheburashka on 1/10/2007 3:32:39 PM , Rating: 1
No, the name K8L is the moniker Intel gave to the AMD CPU in question. AMD has no internal project named K8L (I'm not sure of it's true code name).


RE: K8L really K9
By Viditor on 1/12/2007 3:22:20 AM , Rating: 2
quote:
AMD has no internal project named K8L (I'm not sure of it's true code name)


It's the K10...


Tape-out not production
By HaryHr on 1/10/2007 5:54:25 PM , Rating: 1
Just wanted to point out that this is just 1 processor in "expiremental" phase.
Intel is ahead of AMD in manufacturing process, no doubt there, but AMD usually makes faster swich to new process once it starts transfer.

So what's the real situation in availability? There are still quite a few Intel processors being made in 90nm (most of them I think, maybe someone has exact numbers) and are swithing to 65nm.
Where is AMD? Just started switching to 65nm from 90nm, so they are probably working on their 45nm shrink, just don't have anything yet.

Bottom line is: nothing much changed (other than manufacturing process) ;))




RE: Tape-out not production
By afkrotch on 1/10/2007 11:01:21 PM , Rating: 2
AMD has to make the switch to a different process faster. It's not by choice, it's by need. They don't have the fabs to be able to make multiple different processes, which is quite the opposite from Intel.

Intel can easily make 130nm, 90nm, 65nm, and 45nm procs and be able to meet customer demands. AMD can't even meet demands for 90nm procs.

According to Intel half of their current procs being manufactured are 65nm. Which to me, isn't all too surprising. Intel likes to have a long overlap period. Intel has 3 65nm fabs. I think AMD only has 3 fabs.


RE: Tape-out not production
By Viditor on 1/12/2007 3:35:44 AM , Rating: 2
quote:
They don't have the fabs to be able to make multiple different processes, which is quite the opposite from Intel


You are forgetting that AMD can ramp and switch processes faster than Intel because of their APM system...
AMD is scheduled to be completely 65nm in 6 months, while Intel takes about 15 months.

quote:
AMD can't even meet demands for 90nm procs


Sure they can...what they screwed up in Q3 was the TYPE of 90nm procs. They had a majorly unexpected increase in mobile demand (50% increase), and they weren't ready for it.

quote:
Intel has 3 65nm fabs. I think AMD only has 3 fabs


One of those Intel Fabs (D1D) has started to back off of 65nm in order to retool for 45nm already (so it's actually 2.5). Also, AMD's Fabs have a higher capacity than many of Intel's Fabs...


hmmm
By Comdrpopnfresh on 1/10/2007 1:25:08 PM , Rating: 2
Will this die-shrink bring better performance? In the case with amd's current architecture going to 65nm, they experienced lower power consumption, and therefore less heat, but so far their L2 cache is slower to allow for the die shrink- and the performance generally lags behind the 90nm counterparts. Is C2D more scalable than A64? Unless someone has a clear-cut answer I suppose time will tell.




RE: hmmm
By carl0ski on 1/10/2007 8:43:52 PM , Rating: 2
the cache didnt slow because of the die shrink

AMD chose the increase the delay on the cache (dont ask why/possible larger cache)

Intel will not have a slower cache on the 45nm unless they decide like AMD to support much larger cache


So much discussion so little prospective.
By ChipDude on 1/10/2007 8:43:09 PM , Rating: 2
What does this mean? Confirms earlier annoucment from INTEL that 45nm was on track is indeed on track. There were few details on INTEL 45nm. INTEL was curiously absent from this years IEDM. This is usually the year ( even year ) that you heard about their new logic and saw yield, Ion/Ioff and all that good stuff; 1998 180nm, 2000 130nm, 2002 90nm, 2004 65nm, 2006 nothing. The fact that A0 booted says something about the health of the design as well as the process. Looks like Penrym 45nm products will be there in 2nd half 2007 is what this means. Figure a few design tweeks and steppings and production in 2007.

Where is IBM/AMD? They annouced sparse details about their 45nm showing nothing revoluationary. Lots of noise about strain and lower K BE all incremental things done on 90nm and 65nm by many others. Lots of noise about wet lithography but again it not matters how you get there but once you get there what you have. THe performance stuff I saw wan't marginally any better then INTEL 90nm. Thus AMD design may have transistor density parity 18 months behind INTEL but that is about it. AMD still lags a great deal in performance.

Lets not forget that the ONLY reason AMD had any performance was due to INTELs very ill advised adventure into deep pipeling for MegHz at any cost. If they had instead permitted Coppermine extensions 6 years ago we'd be on Core4 by now and AMD would be at 10% MS or less.

K8L/Barcelona ( why not call it K9 ?? ) has already been reviewed. I see it has opportunity to match Core2 @ 65nm and maybe even be a little faster at platform level. Actually there is nothing special here. AMD is a year later then Core2 it better be faster. If it isn't like Prescott then its nothing but a disaster for the company. From what I see there is little there that says it'll beat Core2 in 2nd half 07 by more then low single digits. There is no second rabbit. Don't get fooled by the preception of a rabbit last time around. That was simplly INTEL screwing up vs AMD brilliance.

For consumer/mobile space I see INTEL dominating all benchmarks. With 65nm maturity they will continue to hold the upper hand. They will dictate the stack, the pricing and control the amount of pain AMD sees.

FOr > 4 ways servers AMD will continue to lead but momentum is ALL INTEL. The latest sever platforms hold well and with 45nm release will make this arena in early 2008 look like end of 2006 for AMD on the desktop front.

Finally the the average Joe really doesn't care if your quadcore is 4 CPUs on a single package, 4 CPU/packages on a board. All the matters is performance and power. Here again INTEL's superior engineering resource and manufacturing capacity will give them the upper hand and with their marketing muscle manage the preception and how far down the stack the quadcore goes.

someone asked does it matter? TO the consumer its the best situation now as each company provide more and more at lower cost. Why can't car companies do this too? But long-term the situation isn't looking good for AMD. No product leadership leads to losing pricing leadership. not competitive on technology means cost handicap. Not having growing MS means loss of volume/cost advantage. Lets face it AMD isn't a Apple. AMD makes interchangeble widgets on a circuit board. If it aint faster or cheaper they don't get designed in. I see red on the balance sheet all of 2007.




By Commandodan2001 on 1/16/2007 1:24:39 AM , Rating: 2
This post is the most insightful of all of them, save the original article. The comparison to cars is useful here. Buying and owning certain cars elicits an emotional response and indentity in MUCH more of the population than which widget runs their computer. Rote speed/cost is all that is important with processors, not the amount of cache, clocks, or cores used to get there. When driving 300KPH on the autobahn, only the drivers care WHAT car they are driving, but the radar and the police radio do not.

AMD's code names or process technology be damned, if they can give up better performance for the cost with an older process, then they will sell. Conroe's selling point is not the mere fact that it is a 65nm chip and Penryn's selling point will not be the mere fact that it is on 45nm process. Their selling points will be the sheer speed advantage over the rest of Intel's and AMD's lines at the point of release. If Penryn doesn't perform (unlikely, but what if) then it will not make Intel any money. Endeavors that don't make money go away for something that does make money.

If Intel's next part is better than AMD's, they will enjoy higher margins, and vice-versa. I think AMD shot their own feet when they priced their top-end processors higher than Intel's in certain cases. That decision stifled their growth as much as their inability to match what volume-selling segments demanded. AMD's 3 fabs can't make enough chips to satisfy world demand for everything. Intel's 15 fabs make it the de facto volume leader. They can afford to shut one down for upgrade where AMD cannot. Processor speed and fan-boy allegiance doesn't drive companies, profit and survivablity do.

We have seen Intel, with inferior chips (for the most part P4/D vs A64/X2) weather that storm in the black and release a leapfrog architecture and process. AMD simply must release a faster processor with lower cost, because they will not be able to stay afloat as a company as long as Intel can. Innovation of the business model and improvement of the products(or reasonable facsimiles) are every company's bread and butter.


say
By nah on 1/11/2007 10:47:10 AM , Rating: 2
All I was trying to say was that GENERALLY--on apps that weren't SSE/2/3 optimized--AMD stayed slightly ahead of intel in performance--from 1999-2003, until the Athlon64 when the gap widened by a not inconsiderable margin--until Core2Duo showed up--
talking about process nodes--anyone remember when copper was a reality for AMD and intel were struggling to keep up--or the time when thunderbirds were 'go' in reality when intel was just struggling to make 1 Ghzs--

for a fascinating guide as to which processors were actually at top and when from someone who was closely related to the industry try this---
http://www.redhill.net.au/iu.html




RE: say
By mindless1 on 1/11/2007 7:22:22 PM , Rating: 2
You're correct, I have to assume most of the people were either pretending their P4 was faster or were stealing all their software as there's no way they can make a comparison between spending $200 for some AMD CPU or multiple times as much for P4 plus software.

People like colored graphs and can understand numbers though, if you get much more involved with the data then there is a constant conflict between those who considered the rest of the details and those that just keep repeating the colors and numbers on the graphs as if merely repeating them often enough was a strength in redundancy.


Native quad cores?
By hstewarth on 1/10/2007 11:33:16 AM , Rating: 2
I believe this version of cpu has in AMD words - Native quad core support - it would be funny if this chip is release before AMD releases there quad cores.

What is the possible of Intel doing Woodcrest/Clovertown on this series and release an 8 core cpu late 2007. My guess is this chip will be release this summer.

Are these drop in chips for 775/771 systems. that would be really cool.




"I want people to see my movies in the best formats possible. For [Paramount] to deny people who have Blu-ray sucks!" -- Movie Director Michael Bay











botimage
Copyright 2009 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki