backtop

Hardware AMD 65nm Product Plans Unveiled
Anh Tuan Huynh (Blog) - November 14, 2006 3:17 PM
Print E-mail del.icio.us 75 comment(s) - last by sdsdv10.. on Nov 17 at 11:18 PM
AMD unleashes 65nm beginning in December

AMD’s long awaited 65nm Brisbane core products are just around the corner. DailyTech previously reported that Brisbane is expected to launch in December. AMD’s latest roadmap shows Brisbane 65nm products will arrive as scheduled. Brisbane will be AMD’s first 65nm core and is expected to launch with four parts. The four 65nm Brisbane core based products include the Athlon 64 X2 5000+, 4800+, 4400+ and 3800+. All four models will have 2x512KB of L2 cache and a 65W TDP rating.

AMD Athlon 64 X2 Brisbane 
Model
 Core
Frequency 		L2 Cache
 TDP Expected
5400+ 2.8 GHz 2x512KB 76W Q2'07
5200+ 2.7 GHz 2x512KB 65W Q2'07
5000+ 2.6 GHz 2x512KB 65W
December
4800+ 2.5 GHz 2x512KB 65W December
4400+ 2.3 GHz 2x512KB 65W December
4000+ 2.1 GHz 2x512KB 65W
December

Two additional products will switch over to the Brisbane 65nm core in Q2’2007. This includes the Athlon 64 X2 5200+ and 5400+. The Athlon 64 X2 5600+ will be clocked at 2.8 GHz with a 2x512KB L2 cache configuration. Unlike the lower Brisbane products it will have a 76W TDP rating. AMD’s Brisbane based Athlon 64 X2 5200+ will differ from the current Windsor offering. Instead of the 2.6 GHz clock speed and 2x1MB L2 cache configuration of the Windsor based Athlon 64 X2 5200+, the Brisbane based Athlon 64 X2 5200+ will have a 2x512KB L2 cache configuration and 2.7 GHz clock speed. The Brisbane Athlon 64 X2 5200+ also has a 65W TDP like other Brisbane products.

AMD Athlon 64 X2 Brisbane Energy Efficient
Model
 Core
Frequency 		L2 Cache
 TDP Expected
4400+ 2.3 GHz 2x512KB 35W Q3'07
4200+ 2.2 GHz 2x512KB 35W
Q2'07
4000+ 2.1 GHz 2x512KB 35W Q2'07
3800+ 2.0 GHz 2x512KB 35W
Q2'07

Also arriving in Q2’2007 is energy efficient Brisbane based products. These Energy Efficient processors will have a 35W TDP and be available in Athlon 64 X2 4200+, 4000+ and new 3800+ models. Joining the energy efficient product lineup in Q3’2007 will be the Athlon 64 X2 4400+. The energy efficient Athlon 64 X2 4400+ carries the same 35W TDP as the other low power Brisbane products.

AMD Athlon 64 Lima
Model
 Core
Frequency 		L2 Cache
 TDP Expected
4000+ 2.6 GHz 512KB 45W Q2'07
3800+ 2.4 GHz 512KB 45W
Q1'07
3500+ 2.2 GHz 512KB 45W Q1'07

AMD’s dual-core product lineup isn’t the only lineup switching over to 65nm. Single-core Athlon 64 products based on the 65nm Lima core will arrive in January 2007. Two initial single-core Lima products will be introduced. These products are the Athlon 64 3800+ and Athlon 64 3500+. Joining the 65nm single-core party in Q2’2007 is the Athlon 64 4000+. All single-core Lima based products carry a 45W TDP rating and have 512KB of L2 cache.

AMD Sempron Sparta
Model
 Core
Frequency 		L2 Cache
 TDP Expected
3800+ 2.2 GHz 256KB 35W Q2'07
3600+ 2.0 GHz 256KB 35W
Q2'07
3500+ 2.0 GHz 128KB 35W Q2'07
3400+ 1.8 GHz 256KB 35W
Q2'07

Not to be left out of the 65nm transitions is the Sempron product lineup. In Q2’2007 AMD will release four 65nm Sparta based Sempron products. The Sempron Sparta based lineup includes the 3800+, 3600+, 3500+ and 3400+ clocked at 2.2 GHz, 2.0 GHz, and 1.8 GHz respectively. All four Sparta based products carry a 35W TDP. Three of the Sparta based products have 256KB of L2 cache while the Sempron 3500+ only has 128KB of L2 cache.

Comments     Threshold


This article is over a month old, voting and posting comments is disabled
Too little too late?
By wingless on 11/14/2006 3:31:12 PM , Rating: 2
Im an AMD fanboy but I get the feeling this is too little too late. Its a necessary step for AMD but we all wish it came 6 months ago. Im holding out to see what the K8L will bring to the table and if AMD will remain competitive with Intel's Core 2's before I upgrade from my 2500mhz X2 4200+. 65nm should be badass for overclocking though so yippee!




RE: Too little too late?
By wingless on 11/14/2006 3:34:08 PM , Rating: 1
....maybe I should go to a 35watt X2 just to save on that light bill. Its hurting our bottom line and If I get an 8800GTX then Ill need a 35watt proc just to be able to keep my 500watt power supply going.

BTW, do any of you know the typical power usage of a harddrive? I have 5 now and really need to consolidate into a couple of 500+gb drive lol. Anybody else a harddrive collector like me?


RE: Too little too late?
By Vertigo101 on 11/14/2006 4:37:25 PM , Rating: 1
I also have 5 Harddrives in my case, plus an external drive.

As far as I know, modern drives consume about 10 watts driving activity, and less than 1 watt while asleep.


RE: Too little too late?
By lucyfek on 11/14/2006 5:12:31 PM , Rating: 1
i've got six. the number has built up over time and the oldest 2 x 80 ata133 raid0 will have to go (sooner or later) as they lag behind at boot (sometimes the controller has hard time to detect them) and nvidia seems to forget about updating parallel raid drivers (vista had never seen them as raid array (actually this probably saved the data on these drives ;)), linux ("stright out of the box" also does not see partitions, not only because ntfs).
i could go for two more (silicon controller on my asus x? premium mb) to hit ten (including cd/dvd drives) but this would be an overkill and i'd have to sacrifice fdd (not an issue) and fan controller (pain).
plenty of hdd comes handy when trying different os - no need for boot loader to be aware of (and screw) the other os, as the bios f8 option gives enough flexibility/convenience (vm is other nice option but it has its own limits). for some os (vista) separate hdd does not seem enough to keep away from other os (or be exterminated by them) taking with itself a bite of your data at random partition it could see, which is one more reason (for me) not to go any further with hdd/os number - too much headache.
since i said goodbye to vista, i've had plenty spare capacity for backups of my data (that survived vista/xp animosity).


RE: Too little too late?
By BrassMonkey on 11/14/2006 6:30:38 PM , Rating: 2
watts
18 ---two hard drives
10 ---dvd drive
65 ---E6400 Core2
135 ---8800GTX
9.9 --DDR2 2 dimms
30 ---4 fans & cpu fan
? -----mobo?

around 268 watts for a high end PC.


RE: Too little too late?
By GaryJohnson on 11/15/06, Rating: 0
RE: Too little too late?
By blwest on 11/15/2006 11:01:45 AM , Rating: 2
If you have trouble paying the light bill... spend less on computer equipment.


RE: Too little too late?
By othercents on 11/14/2006 3:36:02 PM , Rating: 2
No I don't think it is too late. They have to ramp up the production on their new plant and get the process running. I'm really just waiting to see what the new Quad Core processors will look like next year.

Hopefully with their new process they will be able to produce them for less and increase yield.

Other


RE: Too little too late?
By ADDAvenger on 11/14/2006 9:51:58 PM , Rating: 2
Exactly what he said.

Intel transitioned Netburst to 65nm before introducing Core, AMD is doing the same kind of thing. They're going to practice 65nm on the K8 so they'll have the experience needed to produce the K8L with good yeilds and good bins.


RE: Too little too late?
By JackPack on 11/14/2006 3:56:04 PM , Rating: 1
It is pretty late.

These numbers are 65nm for K8. By the time they transition K8L to 65nm, Intel will be well ahead in using 45nm. Even assuming K8L is on par with Core 2 in IPC, Intel will still have the 45nm advantage for clock and power.


RE: Too little too late?
By drebo on 11/14/2006 4:29:03 PM , Rating: 2
You have to remember that K8L is an evolutionary platform targeted toward Socket F and the introduction of quad-core processors. K8L is not a revolutionary step, like Intel's move from Netburst to Core (2).

K8L is more of a stepping stone for AMD to put all of their next-gen features into a current-gen processor. This has been stated frequently, and I don't know why your fallacy still persists.

Do not expect the kind of performance gains from K8L that you saw with Core 2. They simply won't be there. But, then, they weren't ever intended to be either.


RE: Too little too late?
By Slaimus on 11/14/2006 5:41:14 PM , Rating: 2
Actually, K8L will have many of the features that made Conroe so fast:

- Double number of FP units
- Double width of SSE units
- Improved Branch Prediction and Prefetch


RE: Too little too late?
By OneEng on 11/14/2006 11:07:40 PM , Rating: 2
[quote]Do not expect the kind of performance gains from K8L that you saw with Core 2. They simply won't be there. But, then, they weren't ever intended to be either.[/quote]
Actually, there is ample reason to believe that K8L will gain more than Core 2 did over Pentium M (~25%).

The big question for AMD will be clock scaling. Once IPC is evened up (or surpassed) with K8L, AMD will still have to deal with Intel's ability to scale their product.

On the current 65nm, it appears that Intel could move to around 3.2-3.4Ghz in a released product with some small mask changes.

AMD on the other hand is releasing its new 65nm parts at a max of only 2.8Ghz by Q2 2007 if recent information is to be believed.

Intel is also beating on the doors of a Q4 release of its 45nm die shrink. This is really bad news for AMD.

While I can buy that AMD may equal or suprass Intel's current 65nm core 2 product line with K8L, they most certianly don't seem primed to surpass a potential 4.0Ghz Core 2!


RE: Too little too late?
By Targon on 11/15/2006 12:47:47 AM , Rating: 2
AMD is able to make changes to their process technology without a complete upgrade of their fabs. As a result, the 65nm we are starting to see and will see in December in the channel will be only the first rev of the AMD 65nm process. By the time K8L comes around with the architectural improvements, we should be seeing 3.2GHz dual core products.

Keep in mind that due to the increased number of transistors, it's difficult to know just how high AMD will be able to get the K8L design.


RE: Too little too late?
By Tanclearas on 11/15/2006 11:26:10 AM , Rating: 2
quote:
Do not expect the kind of performance gains from K8L that you saw with Core 2. They simply won't be there.


Actually, it is entirely possible that those types of gains are exactly what we'll see. I'm not saying it's likely, but that it's possible.

Your comparison of Netburst to Core 2 vs K8 to K8L is flawed. The performance improvements that should be examined are not those made from Netburst to Core 2, but rather from Core to Core 2, and there were significant performance improvements made. The changes AMD is planning for K8L aren't all that different from Intel's changes to Core. So, once again, it is entirely possible K8L will see significant performance improvements. Only time will tell.


RE: Too little too late?
By saratoga on 11/15/2006 5:31:03 PM , Rating: 2
quote:
You have to remember that K8L is an evolutionary platform targeted toward Socket F and the introduction of quad-core processors. K8L is not a revolutionary step, like Intel's move from Netburst to Core (2).


Conroe is an evolutionary reworking of Yonah, which is itself a revision of the P6. Its an extremely conservative design, and theres nothing revolutionary about it. They basically just took what made the P3 so great, and added more. More OOO ops, more flexibility in reordering, more ALUs, more vector units, more cache, wider alus and vector units, etc.

The P4 was the only revolutionary chip to come out of either Intel or AMD since the P6, and look how that turned out. They needed your "revolution" to undo it!


RE: Too little too late?
By saratoga on 11/15/2006 5:31:03 PM , Rating: 2
quote:
You have to remember that K8L is an evolutionary platform targeted toward Socket F and the introduction of quad-core processors. K8L is not a revolutionary step, like Intel's move from Netburst to Core (2).


Conroe is an evolutionary reworking of Yonah, which is itself a revision of the P6. Its an extremely conservative design, and theres nothing revolutionary about it. They basically just took what made the P3 so great, and added more. More OOO ops, more flexibility in reordering, more ALUs, more vector units, more cache, wider alus and vector units, etc.

The P4 was the only revolutionary chip to come out of either Intel or AMD since the P6, and look how that turned out. They needed your "revolution" to undo it!


RE: Too little too late?
By cgrecu77 on 11/14/2006 5:28:16 PM , Rating: 2
if I remember correctly amd will almost catch intel for 45nm as they are planning a very quick transition from 65 to 45. AMD is doing more than great right now, they don't have enough chips to satisfy demand, once they launch the new generation next year (which will probably overtake intel for the performance lead) combined with the switch to 45nm they should be able to post record financial numbers.


RE: Too little too late?
By JackPack on 11/14/2006 5:38:38 PM , Rating: 2
Lack of chips to satisfy demand is bad thing.

The main reason why AMD can't produce enough chips to satisfy demand is due to the Intel-led shift to dual-core combined with 90nm.

Giving chips to Dell for free (or very cheaply) doesn't help either.

Either way, ASPs are held low.


RE: Too little too late?
By drebo on 11/14/2006 6:24:09 PM , Rating: 2
quote:
the Intel-led shift to dual-core


Correct me if I'm wrong, but didn't AMD release their X2 before Intel released the Pentium D?


RE: Too little too late?
By Russell on 11/14/2006 6:42:21 PM , Rating: 2
Actually no. Intel beat them out the door by a day (an entire day!) with the hack-job of a "dual-core" chip that the first Pentium D's were. I think AMD announced their X2 chips first though, and they were in development a lot earlier (like earlier than 1999 when the K8's were first demoed).

But what the poster means by "led" is that it was Intel who created a market for insanely cheap dual-core systems. Look at how high AMD dual-core prices were before the post-Conroe price cuts. AMD was perfectly happy to leave dual-core chips as high-end until they reached 65nm (or later). It was Intel that forced their hand, mading dual-core mainstream and subsequently forcing AMD to constrain their capacity by mass-producing dual-core CPU's.


RE: Too little too late?
By ShapeGSX on 11/14/2006 6:44:16 PM , Rating: 2
They both released dual core chips within days of each other.

However, Intel has definitely led the shift to dual core. Intel lowered dual core prices to levels that were on par with single cores. And they started producing mostly dual core chips sooner than AMD (if AMD has actually shifted in that direction yet).

And all of this was made possible by Intel leading the industry in 65nm CMOS technology.

AMD has been following Intel reluctantly on price and dual core because they MUST if they want to keep up. And they are attempting to catch up with their process (lagging by a full year right now). But it has to be hurting their bottom line big time.


RE: Too little too late?
By drebo on 11/14/2006 8:15:02 PM , Rating: 2
quote:
Intel lowered dual core prices to levels that were on par with single cores.


Intel's chips didn't perform well enough to demand the premiums that AMD's dual cores could. When your "extreme edition" processor is outperformed by the lowest-end of the competition's processor(955 performed worse than X2 3800+), there's no question about the fact that you have to massacre your prices.

I would hardly call that "leading" anything.

From experience, I've sold many more dual core Athlon64s than I have dual core Pentiums, even when the Pentiums were much cheaper. Yes, the PD915 is at a great price point right now, but in a majority of mainstream applications (as in, uses/implementations), our Athlon64 3500+s still perform just as well.

quote:
AMD has been following Intel reluctantly on price and dual core because they MUST if they want to keep up.


AMD's X2 processors demanded huge premiums because they were great processors and far better than anything the competition had to offer. THAT is why they were so high priced for so long. Now that Intel has processors that are competetive with the X2s, AMD has lowered their prices in such a way that they are now competetive on a dollar/performance basis.

Sure, it might be hurting their gross margin a bit, but they're still pushing a profit. If they weren't, they wouldn't slash their prices. Unlike Intel, AMD isn't slashing prices to vacate an excess inventory of worthless parts.


RE: Too little too late?
By ShapeGSX on 11/14/2006 10:08:21 PM , Rating: 1
quote:
Unlike Intel, AMD isn't slashing prices to vacate an excess inventory of worthless parts.


Of course not. That would imply that AMD could make enough processors to actually have an excess inventory. ;)

Regardless of what caused what, Intel slashed the dual core prices first. As well, even when Intel launched the new Core 2 Duo and had a very good lead over the competition, they didn't raise prices. This has furthered the dual core adoption, and it is what ultimately forced AMD to drop their own prices.

Furthermore, Intel also introduced the first (and excellent) laptop dual core processor back in January. Oddly, the dual core laptops have probably done more to open the market's eyes to the benefits of dual core than any desktop processor has. You can't possibly say that Intel isn't leading the laptop dual core market. I don't think the Turion X2 even competes performance-wise with the Core Duo or Core 2 Duo mobile processors.


RE: Too little too late?
By JackPack on 11/14/2006 11:22:20 PM , Rating: 2
As mentioned, desktop chips are only half the story. Since 2005, notebooks have surpassed destkop sales in the U.S. retail market. The Centrino Duo campaign speaks for itself.

Intel has definitely led the shift to dual-core. Even if the Pentium D didn't perform as expected, the pricing more than made up for it. Look at how Dell continued to use Smithfield/Presler in the face of Toledo/Manchester. Why is Dell using AMD now, when Conroe is clearly the leader? Once again, price.


RE: Too little too late?
By kmmatney on 11/14/2006 4:18:08 PM , Rating: 2
This is still a good thing - AMD should have smaller die and be able to crank out more cpus per wafer. This should help out with supply issues. I'm sure that there are some revisions in the shrink-down which will also help out in performance and reduce the gap from Conroe. Progress is always good, even if its a bit late.


RE: Too little too late?
By Samus on 11/14/2006 11:59:52 PM , Rating: 2
An x2 running at 2.8GHz is quite competitive price/peformance-wise with a E6600 Core Duo. 65nm should enable AMD to make chips for roughly 1/3rd the cost of current 90nm parts, not to mention, higher yields and no more shortages.


RE: Too little too late?
By Oregonian2 on 11/15/2006 3:05:33 PM , Rating: 2
You're comparing an overclocked AMD X2 with a non-overclocked E6600? A fair comparison? Reports on the Conroe line's overclockability puts even the very overclockable X2's to shame. There's been reports of even a lowly E6300 running at 4Ghz, more than twice the rated speed. I'm a VERY conservative fellow and my E6600 is running at 2.9Ghz, about a 20% overclock that in at least video applications that I upgraded for is equivalent to about an X2 7500+ AMD processor.

And until a couple months ago I was an AMD fanboy. My previous last Intel processor was a 486-50.


RE: Too little too late?
By Oregonian2 on 11/15/2006 5:02:15 PM , Rating: 2
Note that I was speaking about "now". For the future X2 5400+ in Q2 2007, it should nearly equal the E6600 for non-multimedia use (takes about X2 6300+ to equal a 2.4Ghz E6600 for things like video rendering in my testing which is something that takes a long time and needs the cpu power in my setup). However by that time Intel should be through at least one or two "regularly scheduled" price reductions, which means AMD will have to be very aggressive with their pricing as they will be competing with an "old" "mature" Intel part by then (being nearly a year behind), and Intel may have come out with something newer by then too (which normally pushes down the prices of the older products). Sorry about my overclocking rant (although all true for "now"), it was just before my lunch and I had a sugar deficiency at the time. :-)


RE: Too little too late?
By StevoLincolnite on 11/15/2006 8:54:39 AM , Rating: 3
I honestly have no use for these new fangled processors.
I'm still using my trusty old
Athlon XP 3000+
Gigabyte Nforce 2 mobo
A gig of ram.
And a Geforce 7 7800GS AGP.
And a 200gb of DDR.
You see I only have a 17Inch monitor and the sweet spot for me is 1024x768. I only just recently upgraded from a Radeon 9700pro. I could still play Oblivion on it, the system ran fine, Now all my games work flawlessly with everything on high @ 1024x768.
And a Athlon 3000+ is plenty for web surfing, MP3's, Watching movies and what not. Untill I dunno, It becomes a minimum system requirement for windows, I dont think my processor will change, I did try overclocking it once, from 2.16Ghz to 2.47ghz Yet I didnt need the extra power, so it sits on stock settings.
And the motherboard I have has all the features I can ever need, and the bonus of it is that I found it down at the dump when I was dropping rubbish off. And she works like a dream! (Its a Gigabyte 7n400pro2 Revision 2). Seriously, as the question yourself do you really need the latest Athlon 64 X2 60000000000+ Just to web surf or play a few games?


2.7ghz? I dont get it
By OcHungry on 11/14/2006 4:01:43 PM , Rating: 2
What is the base FSB for these 2.7, 2.3ghz?
is it odd multiplier or has the fsb changes to 2xx?
please elaborate on this?
thanks




RE: 2.7ghz? I dont get it
By Furen on 11/14/2006 4:19:55 PM , Rating: 2
The CPUs will support half-multipliers, which is why we'll see lots of speed grades.


RE: 2.7ghz? I dont get it
By superunknown98 on 11/14/2006 4:19:58 PM , Rating: 2
13.5 x 200=2700


RE: 2.7ghz? I dont get it
By OcHungry on 11/14/2006 7:20:48 PM , Rating: 2
I don’t think this is the case because, as I understand it, AMD's 1/2 multi will make system unstable and this is the first time AMD is using a “.5” multi (if any). I still don’t quite get the 2.3, 2.5, and 2.7ghz.
Those who wonder about HT vs. FSB, AMD uses a base speed of 200mhz and commonly been referred to as FSB (mistakenly or not). the term FSB is less confusing in the overclocker's arena than using "HT" that confuses w/ Hypertransport link which is a completely different beast.


RE: 2.7ghz? I dont get it
By johnsonx on 11/15/2006 11:46:32 AM , Rating: 2
The .5 multiplier that has been implemented so far on K8 CPU's has been a bios/motherboard hack, not sanctioned by AMD. The earlier attempts were nothing more than complete BS, as all the BIOS would do was raise or lower the FSB/HTT clock to simulate a half multiplier. I think some later attempts took advantage of partial support implemented in the CPU, but such support was not complete and not intended to be used.

There is nothing inherent about a .5 multiplier that makes a system unstable if it's implemented correctly. All of AMD's previous CPU's (K7, K6, K5) used half-step multipliers from the beginning.

By the way, the simple method to implement a .5 multiplier is to do the .5 first, then multiply by 2x the desired multiplier. In other words, if HTT is 200Mhz, and desired multiplier is 13.5 (to yield 2.7Ghz), then first divide by 2 (a clock divider is a very simple circuit, and there's at least one already in the K8 memory controller, and no doubt several more), then multiply by 27. Easy.


RE: 2.7ghz? I dont get it
By archcommus on 11/14/2006 4:21:27 PM , Rating: 2
Uhh I'm pretty sure Athlon 64's haven't had an FSB for a long time.


RE: 2.7ghz? I dont get it
By archcommus on 11/14/2006 4:22:51 PM , Rating: 2
To clarify, I thought Athlon 64's didn't use the whole FSB/multiplier system. Isn't it just an HT speed or something? I forget I haven't checked out my BIOS in a long time.


RE: 2.7ghz? I dont get it
By superunknown98 on 11/14/2006 4:26:29 PM , Rating: 2
They don't really, but from how I understand it, there is a clock from which it's final speed is multiplied. I could be wrong but I think it's 200mhz.


RE: 2.7ghz? I dont get it
By JumpingJack on 11/14/2006 9:13:46 PM , Rating: 2
quote:
Uhh I'm pretty sure Athlon 64's haven't had an FSB for a long time.


He mis-spoke, he should have asked what is the base system clock, at it is 200 MHz, AMD will be using fractional 1/2 multipliers to hit these clocks.

AMD does indeed have a FSB, it is serial and is called HT and DirectConnect. Each are clocked based on the system clock just like Intel platforms.


RE: 2.7ghz? I dont get it
By xFlankerx on 11/14/2006 9:45:13 PM , Rating: 2
There's even more confusion when you hear that the base system clock's real name is the HTT, lol. Its so weird that no one even knows that HTT stands for (it has nothing to do with HT bus, afaik).


RE: 2.7ghz? I dont get it
By Yawgm0th on 11/15/2006 6:01:24 AM , Rating: 2
HyperTransport is by nature not a front side bus. It is a bus, but it is not the same thing. Hypertransport is used to directly connect the CPU to the memory controller, northbridge, and other chipsets, each with their own direct connection. A FSB is parallel by nature, as opposed to an HT connection, which is serial.

In concept, HT is simply a replacement for a FSB and is functionally the same, but it is not a form of a FSB.


Nothing interesting
By obeseotron on 11/14/06, Rating: 0
RE: Nothing interesting
By mforce on 11/14/2006 5:02:57 PM , Rating: 2
Yes , ufortunately AMD is once again going to be the choice for budget like it once was. But things are a bit different now because you cand find AMD CPUs in Dells and other important brands which you couldn't back in the K6-2 or even K7 days .
Let's also not forget Intell used to sell quite a bit of stuff even though thier P IV cookers were hardly competitive. If you judge their performance I don't think AMD is doing worse than Intel was with the P IV. AMD has a good price , good power consumption and decent performance.
I would buy a Core 2 Duo but the MBs ( 965 P ) are pretty expensive and if the new AMDs will OC asa well as the Core 2 Duo I will get those instead.


RE: Nothing interesting
By saratoga on 11/14/2006 9:57:47 PM , Rating: 2
quote:
These chips are still going to be much slower than the Core2's, and use more power while doing it.


Power-wise they should be about the same as Core 2 at the same clock speed, or slightly less. Clock for clock the 90nm Turions were comparable to the 90nm Dothan in terms of actual dissipation (but not TDP of course), and power use per clock actually went up slightly when Intel jumped from Yonah > Merom. Since this is a straight die shrink, the 65nm Turion should have slightly less power consumption thent he 65nm then Merom, or perhaps the same if AMD doesn't do as well as Intel at the shrink.

Performance/watt will lag a little though, particularly in laptops and servers with less then 4 cores.


RE: Nothing interesting
By raven3x7 on 11/15/2006 2:06:25 AM , Rating: 2
No that's wrong. Intel's TDP is 75% of maximum heat dissipation while AMD's is 100%. Therefore AMD 65nm parts are quite a bit more energy efficient than Intel parts. In fact Core2 max TDP is 87 Watts.


RE: Nothing interesting
By Accord99 on 11/15/06, Rating: 0
RE: Nothing interesting
By Diasper on 11/15/2006 5:38:22 AM , Rating: 3
*Impartial non-fanboy post*

Ok there is some selective quoting and linking here. That link there is not comparing with the AMD Energy Efficient models.

A better and more comprehensive set of links:
http://www.xbitlabs.com/articles/cpu/display/amd-e...
http://www.xbitlabs.com/articles/cpu/display/amd-e...
http://www.xbitlabs.com/articles/cpu/display/amd-e...

If you are talking about average use for a home user which means the CPU is generally idle than even the 85W AMDs are more energy efficient than Intel's most energy efficient the C2D E6300.

If you talk about light usuage, then the E6300 is much better than AMD's 85W chips. However, the E6300 is still about matched by AMD 65W variants. The AMD's 35W chips are alot better.

If you are talking full load, the the E6300 improves once again but still AMD's 35W CPUs beat Intel's most efficient efficient. However, this is a scenario the average user will rarely find themselves.

Now to be clear this is all while AMD is still at 90nm. This is not at 65nm which should be even more energy efficient. Now for the future it makes sense for AMD to continue to offer 35W (or even 65W) variants since they've already established the market/branding and customers are used to it. Also, it should provide some headroom to manoeuvre against Intel when needed.

So really, if you are talking about the average user who wants a cheap, quiet PC, AMD is a better choice. If you are talking about office PCs ATI's chipsets are quite nice and provide better graphics performance than Intel's integrated options. I don't know about now but some of the S939s used to include DVI output.

Of course, if you need the extra power Intel's C2Duo's provide and use it constantly, the E6300 is the better choice as it does have a better performance/watt ratio. Then when it can be overclocked as well as it can it's a no-brainer for any enthusiast/Anandtecher.

But for average Joe user it is a more complicated question: performance/watt/cost/platform/the maximum performance needed.


RE: Nothing interesting
By Accord99 on 11/15/2006 3:59:13 PM , Rating: 2
The EEs are rare. AMD has to use 1.075v to beat the power consumption of a 1.325v Core 2 Duo. The XBitlabs's E6300 doesn't appear to even have working SpeedStep, considering how the idle system power consumption is higher with it on then with it disabled.

And then there's Techreport's measurements:

http://www.techreport.com/reviews/2006q4/core2-qx6...

The E6400 uses less power at idle and load than the EE 3800+.

Besides, I was responding to the mistaken belief that Intel's TDP is under-rated, which is clearly not the case.


RE: Nothing interesting
By saratoga on 11/15/2006 5:36:16 PM , Rating: 3
quote:
Intel's TDP is 75% of maximum heat dissipation while AMD's is 100%.


TDP is not proportional to maximum dissipation. Also, on Intel systems, maximum dissipation is typically higher then TDP (but not always!).

You can't call this from spec sheets, you have to measure it.


Already in the Channel
By Goty on 11/14/2006 3:39:36 PM , Rating: 2
I don't remember where I read it, but certain retailers have reportedly already been selling 65nm chips labeled as revision F2.




RE: Already in the Channel
By JackPack on 11/14/2006 3:45:54 PM , Rating: 2
It was another wrong article from The Inq... now retracted.
http://www.theinquirer.net/default.aspx?article=35...


RE: Already in the Channel
By Russell on 11/14/2006 6:38:40 PM , Rating: 2
Which, if you look at the article, wasn't the Inq's fault at all. The stores were mistakenly advertising the 65nm chips for sale.


RE: Already in the Channel
By Tsuwamono on 11/14/2006 8:19:17 PM , Rating: 2
the Inq. is always wrong anyway though. If they were intelligent they would have reported on how companies were mislabeling CPUs


RE: Already in the Channel
By coldpower27 on 11/16/2006 2:08:55 AM , Rating: 2
Not exactly, no, you have to be able to correlate what they said with another source and then your fine.

The issues is htat they mix in truths with lies so you have to have a good eye for picking out what is right.


RE: Already in the Channel
By johnsonx on 11/15/2006 12:17:22 PM , Rating: 3
How come The Inq has never posted a retraction like that for the Rydermark fiasco?

A headline like "Inq hack has wool pulled over his eyes, then participates in Rydermark fraud" would be most welcome.


is tdp really the same?
By Pirks on 11/14/2006 4:21:00 PM , Rating: 2
what bothers me here is that it's always different clocked chips with exactly the same tdp for all of them. like I gonna believe the 2.2 GHz is going to dissipate same 35 Watts of heat as the 1.8 GHz one.




RE: is tdp really the same?
By raven3x7 on 11/14/2006 4:39:41 PM , Rating: 2
You have to remember that it is max tdp not average. Additionally the chips voltages probably differ therefore you can't really compare them that way.


RE: is tdp really the same?
By acejj26 on 11/14/2006 4:42:01 PM , Rating: 2
the TDP that AMD quotes is the MAXIMUM thermal dissipation from that particular processor. apparently you have never read up on the TDP's that AMD publishes for its processors. never do they say that the 1.8 GHz model will put out exactly 35 W and the 2.2 GHz model will put out exactly 35 W. however, all the chips rated for 35 W will put out 35 W OR LESS.


RE: is tdp really the same?
By Pirks on 11/14/2006 5:44:24 PM , Rating: 2
woot! so I can still buy athlon X2 clocked at 1.8 GHz, put silent hsf on it and forget about BOTH noise and heat, while getting pretty decent speed & multitasking. thanks for clarification.


RE: is tdp really the same?
By Furen on 11/14/2006 5:07:11 PM , Rating: 3
AMD is just cutting itself some slack. The 2.3GHz parts will likely get close to the 35W power draw but the 2GHz parts could, for example, use 25W. The nice thing about rating all of these for 35W is that manufacturers can design systems for 35W parts and use the whole range of parts, and any future 35W AMD releases. Then there's also the fact that some chips may actually draw more power than normal and get close to 35W even at 2.0GHz, and if these don't fit the low-power TDP they'd have to be released as 65W parts.


RE: is tdp really the same?
By smilingcrow on 11/15/2006 9:51:28 AM , Rating: 2
"AMD is just cutting itself some slack. The 2.3GHz parts will likely get close to the 35W power draw but the 2GHz parts could, for example, use 25W."

What you suggest is the typical way that things work, although 25W at 2GHz seems a bit optimistic; 30W seems more likely to me.
It also gives them headroom if they struggle to produce many 35W chips at 2.3GHz. In this case the chips below 2.3GHz could actually produce 35W, rather than it being a max figure for the range of chips which is normal practice.
I’m not saying this to criticise AMD, but because they seemed to have struggled in producing many EE chips at 90nm, especially the 35W version. This isn’t surprising and AMD resorted to factory under-volting the 90nm X2 EE chips; the 35W EE SFF was unsurprisingly even more heavily undervolted than the 65W EE chips.
The very poor availability of the 35W EE SFF chip makes me think that they were cherry picked and that they had trouble in finding many that could match the specs.
At 65nm they are obviously more confident as they are offering higher speeds and a wider range of chips. That good news provided they can deliver this time and the pricing isn’t as exorbitant as well.

Incidentally, you can undervolt almost all current AMD chips using software and achieve results similar or even better than that obtainable with the EE chips and save money at the same time. Just don’t expect to match the 35W of the EE SFF chip though.


939 or AM2?
By Exodus220 on 11/14/2006 8:17:38 PM , Rating: 2
This might be a dumb question, but are these for socket939 or AM2? I am running socket 939 and am wanting to upgrade my 3200+ eventually, but I was wondering if AMD is leaving that socket behind. Is it EOL for 939? Thanks.




RE: 939 or AM2?
By sdsdv10 on 11/14/2006 11:03:01 PM , Rating: 3
All AM2. See the graphic in this other news item.

http://www.dailytech.com/article.aspx?newsid=4943

The 939 EOL was posted here before. See this link

http://www.dailytech.com/article.aspx?newsid=3235


RE: 939 or AM2?
By Iger on 11/15/2006 6:10:18 AM , Rating: 2
No love for my poor soc. 754... Where is my 65nm dual-core? :)


Price points?
By sdsdv10 on 11/14/2006 10:59:15 PM , Rating: 2
Has DT or anyone published any price points for these 65nm units?

It seems that with the x2 3800+ running ~$150(OEM)-$175(retail) that doesn't leave a lot of open space for the single cores and Semprons to play in, that is before you hit zero! Which I assume AMD would want to avoid.




RE: Price points?
By coldpower27 on 11/16/2006 2:15:27 AM , Rating: 2
They are probably identical for existing SKU's.

4000+ will probably be 165US?
4400+ will probably be 210US?
4800+ will probably be 270US?

I find this many SKU's a bit much to be told.

4000+ will probably slot in at 112US, knocking the 3800+ downward.

We will also have to see what AMD does with pricing in the Q1 and Q2.


RE: Price points?
By sdsdv10 on 11/17/2006 11:18:41 PM , Rating: 2
Thanks for the information, but what I was really curious about is the single core Lima's and the Sempron's.

With the lowest dual core x2 core for ~$150 give or take a few dollars, what would be the pricing structure for the lower units. There is seven SKU's which presumably will cost less than the dual cores? If you assume the cheapest would be ~$50, that leaves just a $100 spread for seven processors to fill, or about a $10-15 difference for each one. Seems a little tight to me. Again, not important just wondering.


Doesn't make sense...
By Phynaz on 11/14/2006 4:50:02 PM , Rating: 2
That there's a 100Mhz clock difference between the 4000+ and the 4400+.




RE: Doesn't make sense...
By JackPack on 11/14/2006 5:27:44 PM , Rating: 2
It's a DailyTech Typo (TM).

4400+ is 2.3 GHz.


Pretty cool for the budget end of things
By Zurtex on 11/14/2006 4:50:05 PM , Rating: 2
When helping people make a budget PC these days I always recommended AMD. Cheep and cheerful, and for home users I can build very quite systems, without them running very slowly compared to the days of the Pentium and the Celeron, which is always appreciated.




By Oregonian2 on 11/15/2006 5:13:58 PM , Rating: 2
You might look at the E6300. They've been really cheap w/motherboard in Fry's ads, and their throughput should be very good. If the stock heatsink that came with my retail E6600 is the same one on the E6300, it's a very nice fan/heatsink, a LOT nicer and quieter than any recent AMD processor's stock fan that I've gotten (and I've been all AMD inbetween the 486-50 and the E6600).



can we even get any?
By kenyee on 11/14/2006 8:44:22 PM , Rating: 2
It's harder to find any of the other AM2 chips now supposedly because Dell is using most of them for their systems...

I really want to know when we can actually get these parts in December... :-)




Minor typos
By OCedHrt on 11/15/2006 1:16:46 PM , Rating: 2
So many responses and no one noticed...

"This includes the Athlon 64 X2 5000+ and 5600+. The Athlon 64 X2 5600+ will..."

I think this should be:
"This includes the Athlon 64 X2 5200+ and 5400+. The Athlon 64 X2 5400+ will..."




By Regs on 11/15/2006 9:42:39 AM , Rating: 1
We knew 65nm was coming and coming late. We knew they are just die shrinked 90nm K8s. Just like how it was with the 130nm CPUs. We know AMD has no real plans but to reconstruct and plot their strategy for the next 6-8 quarters.

What would be news is if AMD actually made a 65nm CPU instead of announcing one.




AMD is so screwed
By alstonfds on 11/15/2006 2:40:25 PM , Rating: 1
The future sure don't look bright for AMD. Don't get me wrong, until the Core2 family arrived i was still swearing by AMD but how many people actually look at PPW. As long as it can be air cooled without sticking a Nvidia Fx Dustbuster (pardon the pun) in your rig you have the performance to call the shots in the industry. AMD better have something in those K8L cores that we haven't anticipated.




"I f***ing cannot play Halo 2 multiplayer. I cannot do it." -- Bungie Technical Lead Chris Butcher











botimage
Copyright 2010 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki