Sign in with
Sign up | Sign in
Your question

AMD modifies 65nm Athlon 64 X2 roadmap

Last response: in CPUs
Share
September 27, 2006 7:47:38 AM

Looks like AMD is finally going to push 3 GHz, eh? Sounds fun. Looks like a November release is a pretty good time. Make some holiday sales and get the enthusiast gamers something new to buy/overclock (those that don't have Core 2 Duos, anyways).

As long as they're priced nicely, AMD will probably do fine in the upcoming financial quarters. Of course, Intel is making good headway on it's new models so only time will tell if AMD and Intel's next-year models will wow us.

Nice to see them making good progress :) 
September 27, 2006 9:49:50 AM

Jack, you seem like a smart fellow so please enlighten me on why you think AMD is having problems with their 65nm ramp.
Related resources
September 27, 2006 10:24:36 AM

Quote:
Looks like AMD is finally going to push 3 GHz, eh? Sounds fun. Looks like a November release is a pretty good time. Make some holiday sales and get the enthusiast gamers something new to buy/overclock (those that don't have Core 2 Duos, anyways).

As long as they're priced nicely, AMD will probably do fine in the upcoming financial quarters. Of course, Intel is making good headway on it's new models so only time will tell if AMD and Intel's next-year models will wow us.

Nice to see them making good progress :) 
The X2 6000+ is probably going to be the worst overclocking chip in history. It's pretty much tapped-out already. :wink:
September 27, 2006 11:45:33 AM

Can I ask something, why is that as soon as the new Intel came out that everybody dislikes AMD.As I recall some of the ppl in the this form use to have an AMD. It’s like as soon as something better comes alone you all stand by it and degrade something that maybe be better or hate the other team. If you have a computer or like technology I think that you should have respect for it..
a b à CPUs
September 27, 2006 11:52:24 AM

Quote:
Amd making good progress on 65nm.

http://www.digitimes.com/mobos/a20060926PR215.html


Not really.. it just looks like AMD are free to play with there naming scheme. There 65nm products do not performs better then there 90nm products. I mean look at the new Athlon64 X2 6000+, it's only 200MHz faster then the AMD Athlon64 5600+ (2.8GHz vs 3GHz) but AMD gave it a rating of 6000+??

Hmm... looks like mis-leading marketing here.
September 27, 2006 11:53:05 AM

Quote:
Can I ask something, why is that as soon as the new Intel came out that everybody dislikes AMD.As I recall some of the ppl in the this form use to have an AMD. It’s like as soon as something better comes alone you all stand by it and degrade something that maybe be better or hate the other team. If you have a computer or like technology I think that you should have respect for it..
It's called...being a fanboy of technology...not just one company. It's not a bad thing to change brands(like cheering for a different hockey team gets you called...a bandwagon jumper). Sticking around, and defending the older/slower tech and ignoring the newer/faster tech will get you labelled a fanboy though.
September 27, 2006 11:57:06 AM

I'm not saying its a bad thing but it does make ppl look stupid. Offcourse anyone can change but to take it too far if fucken stupid also. I think ppl who make one liner comments to a subject don't know nothing and you make your self look like a fool. plus the form convo don't go anywhere.
a b à CPUs
September 27, 2006 11:57:22 AM

Quote:
Can I ask something, why is that as soon as the new Intel came out that everybody dislikes AMD.As I recall some of the ppl in the this form use to have an AMD. It’s like as soon as something better comes alone you all stand by it and degrade something that maybe be better or hate the other team. If you have a computer or like technology I think that you should have respect for it..


We don't hate AMD, we route for the better product. We also dislike mis-leading efforts made by companies who's products are trailing the competition.

We dislike Intel's stance with Netburst during the Pentium 500, 600, D etc days and we dislike this new AMD naming scheme where they, out of nowhere, decide to increase the performance ratings of there products substantially.

Honestly.. they should lower the performance rating, AS AMD'S PROCESSORS perform slower clock for clock then Core 2 Duo. So what we currently know as an AMD Athlon64 4600+ should be renamed to 1900+ (as it performs about on par with a Core 2 Duo E6300 clocked at 1.86GHz).

But AMD claims there performance rating is when comparing to there earlier Athlons.. well what has happened.. have the K7s slowed down or something?
September 27, 2006 12:02:23 PM

Yea i guess your right. It is pretty lame.
a b à CPUs
September 27, 2006 12:08:50 PM

Quote:
And till this day I still don't get where AMD got the standard for their QuantiArseSpeed rating. :lol: 
It's not exactly close to PR rating either....


Yeah.. AMD is doing like Intel used too it seems.

Where as Intel believed MHz was everything and marketed there CPU's that way, AMD is going towards PR Rating is everything and is continuously increasing there PR rating to ridiculously high numbers without actually making any substantial changes to there Core Processing features.

And as you stated.. it's much easier to understand Intel's performance ratings as it's like BMW. They compare performance between there own current products.

I just wish AMD would explain how they came up with there PR Ratings.
September 27, 2006 12:42:36 PM

Yea but most ppl still don't understand the difference between the new intel and the old one , some ppl come into me work and buy the D's not the C2D because of the mhz even if you explain to them what’s the difference. I would be a bit confusing if u only knew that the high the Mhz the faster it would be then you here that the newer intel is better but the mhz shows its slower.
September 27, 2006 12:54:28 PM

Quote:
I just wish AMD would explain how they came up with there PR Ratings.


From what I remember it was to counteract Intels Mhz campaign. Basically they wanted to make everyone aware that a chip running at the same Mhz from AMD could do more than a intel chip. i.e. a 1.83ghz chip from AMD running side by side with an Intel 1.83ghz would kick butt.

The numbering system represents what mhz level an intel chip would have to run at to compete with AMD. So for instance an athlon 2000+ means an intel chip would have to run at 2000mhz to be comparable to the AMD chip. Of course the numbering doesn't mean much anymore especially with dual cores.

I wish amd would get rid of the numbering. It was confusing even when it came out.
a b à CPUs
September 27, 2006 12:58:40 PM

Quote:
I just wish AMD would explain how they came up with there PR Ratings.


From what I remember it was to counteract Intels Mhz campaign. Basically they wanted to make everyone aware that a chip running at the same Mhz from AMD could do more than a intel chip. i.e. a 1.83ghz chip from AMD running side by side with an Intel 1.83ghz would kick butt.

The numbering system represents what mhz level an intel chip would have to run at to compete with AMD. So for instance an athlon 2000+ means an intel chip would have to run at 2000mhz to be comparable to the AMD chip. Of course the numbering doesn't mean much anymore especially with dual cores.

I wish amd would get rid of the numbering. It was confusing even when it came out.

You see that is the logical explanation but AMD have stated that it is not a number used to compare there products with Intel's but rather to compare there newer products with there original K7 Athlon.

Which is why I ask if the Athlon (original) has slowed down or something because AMD is increasing the PR rating without increasing the performance.
September 27, 2006 12:58:53 PM

Quote:
Quote:
And till this day I still don't get where AMD is going towards PR Rating is everything and is continuously increasing there PR rating to ridiculously high numbers without actually making any substantial changes to there Core Processing features.

an a64 2ghz was rated 3200+...it would be logical that a 2ghz a64x2 would be rated 6400+...instead, amd refreined from such easy tricks..2.6 ghz x2 are currently rated 5000+...so it s pure moronic thoughts that claiming that amd is exagerating its current prs...yet another intel fanboy with no clues about the subject...
a b à CPUs
September 27, 2006 1:12:36 PM

Quote:
Quote:
Quote:
And till this day I still don't get where AMD is going towards PR Rating is everything and is continuously increasing there PR rating to ridiculously high numbers without actually making any substantial changes to there Core Processing features.

an a64 2ghz was rated 3200+...it would be logical that a 2ghz a64x2 would be rated 6400+...instead, amd refreined from such easy tricks..2.6 ghz x2 are currently rated 5000+...so it s pure moronic thoughts that claiming that amd is exagerating its current prs...yet another intel fanboy with no clues about the subject...


Hey kiddo, I'm no Intel fanboi and my knowledge on IT technology would make yours look like it came from BestBuy.

You don't seem to understand where the PR rating came from. As such you should refrain from posting.

Quote:
AMD doesn't arbitrarily select model numbers according to how they feel performance stacks up to the competition, rather they use a collection of 14 widely used independent benchmarking applications and games (including SYSMark 2001, but more on that later) and determine the rating based on overall performance in those benchmarks.

In order to explain in simple but catchy terms why the Athlon XP is able to operate at a lower clock frequency yet obtain these high model numbers, AMD has coined the term "QuantiSpeed Architecture." In reality there is nothing new about the Athlon XP's "QuantiSpeed Architecture"; it's merely a way of saying that the Athlon XP can do more work in a single clock (higher IPC) than the Pentium 4.


So are you saying MR wiseguy that AMD is comparing there numbering scheme on there Dual Core X2's to single core Pentium4's still? AMD have tried to state that there numbering scheme is not comparing AthlonXP and higher to Intel products but rather there own original AMD Athlon Classic processor.

Either way.. it's BLOATED and you're acting like an idiot.

Quote:
In reaction to the consumers' misconception, AMD reinstalled the PR rating to compare their Athlon XP microprocessors. AMD made sure to advertise the PR number of its microprocessors rather than their raw clock speeds believing that consumers would compare the PR of AMD's processors to the clock speed of Intel's processors. The PR number was originally believed to show the clock speed (in megahertz) of an equivalent Pentium 4 processor, but this was never confirmed by AMD. As part of its marketing, AMD even made sure that motherboard manufacturers conspicuously showed the PR number of the microprocessor in the motherboards' POST and not include the processors' clock speeds anywhere except within the BIOS.

The use of the convention with these processors (which are rated against AMD's earlier Athlon Thunderbird CPU core) is less criticized, as the Athlon XP is a capable performer in both integer and FPU operations, and manages to out-perform an Intel Pentium 4 at a PR rating equalling the P4's MHz.

Wikipedia


Now look at the picture bellow.. why is it that with the AMD Athlon64 X2 6000+ an increase of 200MHz is worth 400+ points where as in the past it has been worth 200+? What has changed? NOTHING! AMD's K8 architecture has not increase in performance per clock therefore it's bullox.
September 27, 2006 1:20:55 PM

Sir, your logic is flawed. Think about it ... they are attempting to asses performance with thier naming scheme, right? Ok, lets deal with your example, a A64 3200+ @ 2 Ghz. If we went to a 2x 2Ghz you think 2x3200 = 6400, ah but you see you are missing a very vital point, cores do not scale perfectly by any stretch of the imagination. I believe the advantage over the single core initially was 15-20% (ballpark figure ONLY) therefore to keep the performance numbering accurate we should do 3200 x 120% (easier math) and we get 3840 and we all know the Manchester X2 @ 2 Ghz is ... drum roll ... 3800+. So next time you have a logical thought, let it go. You clearly fail to include context and reality into your logic which by definition negates any semblence of logic you once had.
September 27, 2006 1:27:05 PM

The X2 6000+ is a renamed locked multiplier of the FX-64. Does this mean a X2 with FX-64 performance at a good price which will work with 4X4?
September 27, 2006 1:27:47 PM

well, despite amd s spinner s claims, the pr rating was a comparison to intel s netburst craps...it was very conservative at the start, with 1.8ghz athlon xp (2200+) matching early p4/512k/400fsb at 2.4ghz...the scheme became less convincing when intel raised the bus to 533 with the p4 2.53...since, the axp were overated in most benches...yet, amd used again this pr rating with the a64 s, comparing them to advanced p4 with 1mega cache and 800fsb.....as intels backed from frequency courses and returned to ipc efficiency, the model numbering of amd is now in deep trouble...surely, the time when intel was the reference is now gone, and these pr ratings are no more needed....
September 27, 2006 1:29:20 PM

You make a very good point. Once, back in the original introduction days I think AMD tried to be reasonable with thier PR naming scheme, but as of late it has gotten totaly out of hand. This 5600+ and 6000+ crap is a bunch of marketing BS that has no bearing on performance and only leads to confusion among poorly informed individuals which leads the more imformed people into migranes having to explain it constantly.

Intel has at least adopted an internally comparartive naming scheme. ie E6600 < E6700, where the 6 is the class of processor and the second 6/7 denote clock speed and cache in a general sense, although maybe not explicity. Intel has gotten away from trying to name things to appear better than AMD and it was worked well in general. I still find their naming scheme confusing sometimes but with a click of google I can find everything I need to compare procs inside Intel. AMD requires effort... unacceptable lol.
September 27, 2006 1:31:08 PM

Wow. I don't think I've ever seen you smack someone down, Super. Awesome. As to the holes in AMD's numbering scheme, perhaps they're leaving room for the 2x1MB processors still?

-J
September 27, 2006 1:36:47 PM

You know everyone is an expert, and everyones rig is better than the other guys! It's easy to get cuaght up in the BS if you start buying into it. Just lok at the number of post some of these ppl have. Do you get an ideal of who you are talking too? Your talking to ppl who do this for a living, they eat sleeep piss & $hit with one hand on a keyboard. No I'm not a fanboy or a wanbabe, but some ppl in these forms have nother better to do but post bla bla bla....

Quote:
We don't hate AMD, we route for the better product. We also dislike mis-leading efforts made by companies who's products are trailing the competition.


There will always be products are trailing the competition.
There will always be mis-leading efforts made by companies.
And there will always be new and better technology.

Just use it for the reasons you need it and don't get caught up in The Hype!
September 27, 2006 1:38:29 PM

I really try to be respectful towards people, but when someone starts using the term logicly you really raise my red flag and I start tearing them apart. I am not the most logical person in the world, but after doing logical proofs (no not logic puzzles, actual proofs that are 100+ lines) and having my reasoning compared to spock (ok, not really but it paints a picture non the less :wink: ). One of my most cynical qualities is deductive reasoning and logic and I use it. I just get mad when people invoke "logic" and aren't trully logical.

Back on topic: Yeah, I don't know what AMD is doing these days with thier PR dept. I can only hope they get it fixed. They started to do that with the Turion series with the ML and MT categories, I just wish it would spill over into the desktop arena. I do have to give Intel credit for making a sweeping nomenclature change unlike AMD which has just changed a line... way to confuse us more lol. :roll:
September 27, 2006 1:38:34 PM

hope that 65nm will allow for greater speed than 3Ghz, cause otherwise AMD will be on the "Dark" side of sales.

Those C2D really rock with that 10 stage integer pipeline (K8 needs 12 clocks if I remember correctly).

Maybe K8L will be more efficient.
September 27, 2006 1:39:03 PM

very few softs were efficiently conceived with multitreads capabilities....so, it wasn t possible for amd to simply double the pr along with the core doubling...those who were using optimised softwares saw that the performance boost was far more than your 20%...so, you missed the point, blaming the processor instead of innacurately designed softs....
September 27, 2006 1:48:49 PM

Quote:
very few softs were efficiently conceived with multitreads capabilities....so, it wasn t possible for amd to simply double the pr along with the core doubling...those who were using optimised softwares saw that the performance boost was far more than your 20%...so, you missed the point, blaming the processor instead of innacurately designed softs....


Really, I missed the point? Aren't we trying to assess processor performance? What do you use to test processor performance? Software, right? Ok, so now 18 months after dual core intro (give or take) we have some well coded apps for dual core (not perfect, but code never is) and yes the performance difference is more than 20%, I agree, but take the market a whole. Take every software application in existence, take every task a computer could perform, on average they run 15-20% faster on a dual core. Certain programs run better on dual core, some won't run at all (much older ones but still valid). You want to blame software for AMD not being able to double thier PR rating... why? Because programmers aren't perfect? Quad core shows noticable improvement over dual core but is it double? Nope.

If you have twice as many engines, do you go twice as fast? No, because they don't scale perfetly just like code and CPU cores. There are enhancements you can make to gain more efficiency out of each core, but they will never scale perfectly. The problem rests in the principles of engineering (which I am definately not qualified to speak on). I am sure someone around here knows more than I do and could explain it better.

Quote:
BTW, I'm going back to study now. If anyone sees JumpingJack be sure to tell him to read my first post in this thread in response to him. :wink:


I should be studying because I have an audit test in 30 min... talk about going to kick my ass. High of 83 last year, avg of 64. Yup, its going to hurt!
September 27, 2006 2:01:03 PM

Jack- there is a site that compares Intels TDP to AMDs I just can't find the link. They measure them slightly differently but its valid point. I'll try and find the site, it was awhile ago and I will post it if I can find it.

Sorry, in a rush its test time baby!
a b à CPUs
September 27, 2006 2:07:25 PM

Quote:
very few softs were efficiently conceived with multitreads capabilities....so, it wasn t possible for amd to simply double the pr along with the core doubling...those who were using optimised softwares saw that the performance boost was far more than your 20%...so, you missed the point, blaming the processor instead of innacurately designed softs....


And what has changed? How many Applications in the Consumer Arena make use of multiple threads? How much of those same applications give anything more then a +/- 20% increase?

You'll be left with VERY few applications. Surely not enough to warrant the naming scheme being used by AMD as of late.

Here's a list..

AMD Athlon64 X2 4200+ 2x512KB = 2.2GHz
AMD Athlon64 X2 4400+ 2x1MB = 2.2GHz
AMD Athlon64 X2 4600+ 2x512KB = 2.4GHz = Intel Core 2 Duo E6300 @ 1.86GHz
AMD Athlon64 X2 4800+ 2x1MB = 2.4GHz
AMD Athlon64 X2 5000+ 2x512KB = 2.6GHz = Intel Core 2 Duo E6400 @ 2.13GHz
AMD Athlon64 X2 5200+ 2x1MB = 2.6GHz
AMD Athlon64 X2 5400+ 2x512KB = 2.8GHz
AMD Athlon64 X2 5600+ 2x1MB = 2.8GHz = Intel Core 2 Duo E6600 @ 2.4GHz
AMD Athlon64 X2 6000+ 2x1MB = 3.0GHz

To me, AMD opted to change how they rate there processors. I'd like them to explain it now. Since they're not comparing to Pentium4's anymore, certainly not to Core 2's and definitly not Athlon Thunderbirds.
September 27, 2006 2:12:46 PM

i agree with you in most cases...as you noticed it, dualcores are 18 months old,at least in the consumer s perspective...and the softs we re using are all 10 years old, with some minor improvement.
September 27, 2006 2:20:39 PM

your comparison list is flawed...a 1.86 g c2d is not as performant as an a64 2.4 g...if i had sometimes, i would explain that c2d are performant only when using heavily sse optimisation to replace the x87 fp unit....most of the benches are now sse optimised...but when x87 fpu unit is used, as in sciencemark, the c2d collapse....
a b à CPUs
September 27, 2006 2:38:34 PM

Quote:
your comparison list is flawed...a 1.86 g c2d is not as performant as an a64 2.4 g...if i had sometimes, i would explain that c2d are performant only when using heavily sse optimisation to replace the x87 fp unit....most of the benches are now sse optimised...but when x87 fpu unit is used, as in sciencemark, the c2d collapse....


And Sciencemark is a game, application that is VERY important to the consumer market why?

Honestly you're accusing me of fanboism when I'm just telling you like it is. The reason you take offense is because you, yourself, are a fan of AMD's. Pretty self apparent.

My comparison list is not flawed.. it is based on a HUGE number of consumer oriented applications ranging from games, to video/audio editing, synthetic benchmarks to Office applications.

Oh and here is the comprehensive list of applications tested that I've used to come to my conclusion. I've also used information from other sources such as Anandtech and THG to formulate my opinion. Now This article from xbitlabs ALSO took sciencemark into question to formulate there opinion and there overall score. You can view the article here.














Last image here shows the OVERALL avg. As you can see teh Core 2 Duo E6300 falls in between the 4200+ and the 4600+ OVERALL. Now if you compare the scores you'll see that the E6300 is closer to a 4600+'s performance then it is a 4200+'s performance. Thus it is more comparable to a 4600+.


As for Price/Performance.. that's right the E6300 has the best Price/Performance Ratio.


And you've got sciencemark... :p  :p  :p 
September 27, 2006 3:03:21 PM

thanks for all the insights....it s clear that the ax2 is no match for the c2d in matter of ipc..most of intel s dualcore performance has the 128 bit executions units as the root....while the ax2 got only 64 bit units, so it need 2 passes to execute a 128 bit sse precision calculus...all the secret of the c2d is here....once amd release its k8l upgrade, with 128 bit precision sse units, we will see the same picture as the athlon/p3 duel....but that s another story, still in the starting blocks....
a b à CPUs
September 27, 2006 3:07:58 PM

Quote:
thanks for all the insights....it s clear that the ax2 is no match for the c2d in matter of ipc..most of intel s dualcore performance has the 128 bit executions units as the root....while the ax2 got only 64 bit units, so it need 2 passes to execute a 128 bit sse precision calculus...all the secret of the c2d is here....once amd release its k8l upgrade, with 128 bit precision sse units, we will see the same picture as the athlon/p3 duel....but that s another story, still in the starting blocks....


Yes.. that's true.

And the leap frog effect will being. With one company jumping over the other with each successive Core revision.

Things will be competitive again. Only problem is that K8L doesn't arrive until 1H08 (1st Half of 2008). So there's still ALOT of time left.
September 27, 2006 3:19:45 PM

Interesting stuff. Bottom line AMDs numbering scheme is now crap. As was pointed out earlier in this thread they were conservative with the numbers when it started. But now AMD has even been flippant and denies what the numbers meant in the first place.

Get with the program and come up with a numbering and naming scheme everyday consumers can follow. Just my 2 cents.
September 27, 2006 3:35:42 PM

Yeah, that is what really sinks my boat so to speak. 1H 08? Isn't Intel supposed to be pushing native quad core out in 1Q 07 err maybe 1H07, then switching over to 45nm Q4 07 or 1Q08 and pushing out octcore in 08, which leaves AMD in a pickle. My understanding is K8L will be a native quad core, which is what Intel is putting out in 07... I forget the code names, I just took an audit test so my brain is all fudged up.

Point being, K8L should be nice but my question is will it be timely. It could be a great core by today's standards but by the time it comes out it could end up a mainstream release instead of the high-end core revision they are looking for. I am skeptical but I am just floating my 2 cents.
September 27, 2006 3:48:24 PM

i think,but i may be wrong, that k8l revision core with improved ipc will be released with the 4core variant af the athlon, in early h2 2007.....until this outcome amd will battle as the same way intel did with the netburst, as a poster accurately said it, by using the 65 nm process to step up the frequencies, as the ax2 is less power hungry that a c2d at this technological step....i must say that i can t advice one to go c2d, because intel is a specialist to get rid of previous sockets...am2 seems more upgradable for the times coming...i just remember when i bought a duron 700 and my friend a p3 733...2 years later, i was still with the same pc, upgraded with a 1.6 g axp, while my buddy upgraded to the axp!!..those who bought the first socket 423 p4 were ever more frihtened by the non upgradibility when the p4 s478 came to light....as a chipset producer, intel prefer to change sockets to sell both new chipsets and new processors...on the other side, amd rely on third parties for the chipsets, so it can see the upgrade market as interessant, cause pople are not very willing to change systematically their mobos, and often memories....
September 27, 2006 4:04:01 PM

Quote:
The X2 6000+ is a renamed locked multiplier of the FX-64. Does this mean a X2 with FX-64 performance at a good price which will work with 4X4?

CPUID microcode, that's hardburned into every CPU nowdays although they may be the same core with same potential.

BTW, I'm going back to study now. If anyone sees JumpingJack be sure to tell him to read my first post in this thread in response to him. :wink:
This could atleast come close the the Q6600 on the 4X4 if AMD manages a better % increase with the 4X4 than Intel is getting with the quad's. This will mean little as Intel is also lunching a Q6700 which I dont see 2 2X6000+ getting anywhere close. This may move AMD a little up the ladder but we'll have to see.
September 27, 2006 4:11:58 PM

I wish they'd quit d*cking with their roadmap and model names and get on with the good chips.
September 27, 2006 4:19:32 PM

Question for you all: The 65 watt processors on Newegg right now are listed as "Brisbane" cores. I've never seen the same codename used for different runs of processors? So, what is going on there?
September 27, 2006 4:30:29 PM

yeah...a64 is now 3 years old...let s just hope that it won t take them 3 years ,like intel, to re enter the good rails...surely that they are going to maximise the a64 profits to death before switching to new products...it must be said that actually,we, common pc users,are short of new usages for our machines,excepts perhaps the games afficionados, we have now enough ,if not to much,power for our daily work..i still use an old 2001 pc for my work, and it s far enough to send some buy/sell orders to the french stock market......
September 27, 2006 4:54:07 PM

Quote:
the power/transistor is much higher.


Jack, AMD and Intel have different definitons about TDP.

AMD's definition for TDP is the scenario when maximum power is used:
Quote:
Thermal Design Power (TDP) is measured under the conditions of Tcase Max, IDD Max, and VDD=VID_VDD,
and include all power dissipated on-die from VDD, VDDIO, VLDT, VTT, and VDDA.

URL: http://search.amd.com/cs.html?url=http%3A//www.amd.com/...||+TDP,+|+language%3Aen&col=&n=2
Intel's definition for TDP is the average used power:
Quote:
Thermal Design Power (TDP) should be used for processor thermal solution design targets. The TDP is not the maximum power that the processor can dissipate.

http://www.intel.com/cd/channel/reseller/asmo-na/eng/products/box_processors/desktop/proc_dsk_p4/technical_reference/183499.htm


Also,
Quote:
Jack, you seem like a smart fellow so please enlighten me on why you think AMD is having problems with their 65nm ramp.


I will elaborate more later --- too much new stuff, add in SOI makes all other processing much more complicated.

Jack
waiting for your reply.
September 27, 2006 5:06:04 PM

Quote:

AMD Athlon64 X2 4200+ 2x512KB = 2.2GHz
AMD Athlon64 X2 4400+ 2x1MB = 2.2GHz
AMD Athlon64 X2 4600+ 2x512KB = 2.4GHz = Intel Core 2 Duo E6300 @ 1.86GHz
AMD Athlon64 X2 4800+ 2x1MB = 2.4GHz
AMD Athlon64 X2 5000+ 2x512KB = 2.6GHz = Intel Core 2 Duo E6400 @ 2.13GHz
AMD Athlon64 X2 5200+ 2x1MB = 2.6GHz
AMD Athlon64 X2 5400+ 2x512KB = 2.8GHz
AMD Athlon64 X2 5600+ 2x1MB = 2.8GHz = Intel Core 2 Duo E6600 @ 2.4GHz
AMD Athlon64 X2 6000+ 2x1MB = 3.0GHz

To me, AMD opted to change how they rate there processors. I'd like them to explain it now. Since they're not comparing to Pentium4's anymore, certainly not to Core 2's and definitly not Athlon Thunderbirds.


The X2 numbering system hasn't changed. The 200 jumps occur with a change from 512KB cache to 1MB or from 1MB cache to 512KB cahce +200mhz increase of speed. You get a 400 jump when you go up 200mhz with the same cache. So when you look at the numbering of the new chips and their specs it fits in with the exisiting numbering scheme. The 5000+ is a 2X512KB cache chip at 2.6GHz. The 6000+ is a 2X1MB cache chip at 3.0 ghz to make this jump we increased 400mhz, which under the numbering scheme would be an increase of 800 , and jumped from 512KB cache to 1MB cache which gives us a 200 increase. Last time I checked 5000+800+200=6000.
The numbering scheme is a PR to the A64 PR and not to anything from intel.

You know the average Joe will walk into a store and see an X2 6000+ system for $2500 and a C2D 6300 system for $1000 and get the C2D because the number is higher so therefore it is better and it's cheaper to boot. I haven't seen GHz advertised in a while in the local ads just the model numbers, and higher is better.
September 27, 2006 5:22:40 PM

We don't know if AMD is less power hungry, we haven't seen a working sample out to reviews (that I am aware of). As for complaining about sockets, that is a function of architectural changes. Somtimes it is necessary to make a new socket. Most recently Intel gave drop in support for 975 chipsets with C2D because it was techinicly possible because there weren't enough changes in the way the processor relates to the chipset to demand a new socket.

AMD switched from 939 to AM2 due to DDR2, Intel went from socket 423 to 478 for DDR dual channel support as opposed to RDRAM they intially used with Williamette. Before you fly off the handle, think about what is going on at the forest level, not the tree level. If you focus on the fact that they change sockets you will fail to see why they have to. Intel will have to switch sockets when they shift to DDR3 towards the end of 2007.

AMD just doesn't shift standards as fast as Intel does. Intel starts early gets penetration and revises the chipsets as it goes. AMD waits more for maturity in products, such a DDR 2, then jumps in so they look like they don't shift sockets as much. Calm down and think of the big picture. Its all good, in the end we all win.
a b à CPUs
September 27, 2006 5:29:13 PM

Quote:
We don't know if AMD is less power hungry, we haven't seen a working sample out to reviews (that I am aware of). As for complaining about sockets, that is a function of architectural changes. Somtimes it is necessary to make a new socket. Most recently Intel gave drop in support for 975 chipsets with C2D because it was techinicly possible because there weren't enough changes in the way the processor relates to the chipset to demand a new socket.

AMD switched from 939 to AM2 due to DDR2, Intel went from socket 423 to 478 for DDR dual channel support as opposed to RDRAM they intially used with Williamette. Before you fly off the handle, think about what is going on at the forest level, not the tree level. If you focus on the fact that they change sockets you will fail to see why they have to. Intel will have to switch sockets when they shift to DDR3 towards the end of 2007.

AMD just doesn't shift standards as fast as Intel does. Intel starts early gets penetration and revises the chipsets as it goes. AMD waits more for maturity in products, such a DDR 2, then jumps in so they look like they don't shift sockets as much. Calm down and think of the big picture. Its all good, in the end we all win.


Plus I think he forgot about the last 3 years. Intel went from Socket 423 to 478 to 775 for it's Netburst architecture in the span of 5-6years. AMD went from 754 & 940 (Athlon64 FX CPU's) to 939 to AM2. 4 Sockets in 3 years!!

Sure Socket A lasted long, but AMD certainly isn't doing the same anymore.
September 27, 2006 5:54:58 PM

Quote:

+ They pulled in from Dec to Nov. This puts them at 13 months behind Intel instead of 14 months.

Jack


This is very good news for AMD, as Intel are trying to be at LEAST 2 years ahead of everyone else, according to tomshardware.com article anyway.

I wonder if the 65nm jump will bring prices down, or if it will push them up? Who knows.
a c 83 à CPUs
a b À AMD
September 27, 2006 6:49:44 PM

Seeing as you bestbuy educated people cant' seem to use the search feature, I suggest you "refrain from posting". (see, rudeness always comes back to bite you eventually...)

For those to lazy to search AMD.com.
http://www.amd.com/us-en/Processors/TechnicalResources/...

Quote:
Q: What do AMD Athlon 64 processor models represent?
A: Since frequency alone is not an accurate measure of performance, AMD model numbers serve as a simple, accurate representation of relative AMD performance within a processor series. The higher the model number, the better the overall software performance running on the processor. The AMD Athlon 64 FX-60 dual-core processor is the first member of the dual-core FX processor series from AMD.


Hopefully we all now understand what these numbers mean. Higher numbers = better performance.

(Ok, I'm done stepping on the toes, don't forget to bash me for being an AMD fanboy.)
a b à CPUs
September 27, 2006 6:57:47 PM

Quote:
Seeing as you bestbuy educated people cant' seem to use the search feature, I suggest you "refrain from posting". (see, rudeness always comes back to bite you eventually...)

For those to lazy to search AMD.com.
http://www.amd.com/us-en/Processors/TechnicalResources/...

Q: What do AMD Athlon 64 processor models represent?
A: Since frequency alone is not an accurate measure of performance, AMD model numbers serve as a simple, accurate representation of relative AMD performance within a processor series. The higher the model number, the better the overall software performance running on the processor. The AMD Athlon 64 FX-60 dual-core processor is the first member of the dual-core FX processor series from AMD.


Hopefully we all now understand what these numbers mean. Higher numbers = better performance.

(Ok, I'm done stepping on the toes, don't forget to bash me for being an AMD fanboy.)

Not really... this means AMD changed the way they name processors and no longer base it on an equivalent P4 or Athlon T-Bird CPU as it used to be.

See this is what they say about the Athlon XP

Quote:
AMD Athlon™ XP Product Information

AMD model numbers, based on industry-standard benchmarks on a wide range of popular software, are a simple, accurate representation of relative AMD processor performance.

The model number methodology is designed to help end users simplify their PC purchase decision. The higher the model number, the better the overall software performance on the processor. The "+" at the end of each model number indicates the added performance benefits delivered by AMD's innovative processor designs.


So they've changed there philosophy without informing anyone... it's VERY mis-leading.

In other words.. we're correct to find it mis-leading and confusing. No doubt AMD had to change the way they did there Performance Rating due to Intel's use of model numbers themselves rather then MHz.
a c 83 à CPUs
a b À AMD
September 27, 2006 7:05:58 PM

Sorry for being an @$$ today.

I still don't see where this is supposed to be a comparison to Intel or thunderbird CPUs. You can easily find this claim in so many review websites it makes me sick. Back when everyone was concerned about MHz/GHz, I think AMD let people think thats what the numbers ment. I for one don't believe doubling the L2 cache means your CPU would now run (equivalent) 200MHz faster. These numbers are simply numbers. (if they were frequncies, then as has already been pointed out, they lost meaning a long time ago.)
a b à CPUs
September 27, 2006 7:10:57 PM

Quote:
Sorry for being an @$$ today.

I still don't see where this is supposed to be a comparison to Intel or thunderbird CPUs. You can easily find this claim in so many review websites it makes me sick. Back when everyone was concerned about MHz/GHz, I think AMD let people think thats what the numbers ment. I for one don't believe doubling the L2 cache means your CPU would now run (equivalent) 200MHz faster. These numbers are simply numbers. (if they were frequncies, then as has already been pointed out, they lost meaning a long time ago.)


Well the line "AMD model numbers, based on industry-standard benchmarks on a wide range of popular software, are a simple, accurate representation of relative AMD processor performance. "

So what AMD is saying is that they based there model numbers on the performance there Athlon's had relative to other industry offerings (Pentium4 and T-Bird come to mind). Although no one knows which one AMD were basing there Performance Rating on.

Who knows really. But there's quite a change that has occured with how AMD use the PR numbers going from the AthlonXP to the Athlon64 as you highlighted.
!