DailyTech - AMD PDF Leaks Surprise Athlon X2 5000+ Info

Submit News

Hardware AMD PDF Leaks Surprise Athlon X2 5000+ Info
Kristopher Kubicki (Blog) - March 23, 2006 2:58 AM

32 comment(s) - last by glennpratt.. on Mar 29 at 11:47 PM

Aside from the discrepancy about cache sizes, the AMD thermal whitepaper reveals much about AMD's newest processor

AMD surprises all with a 2.6GHz Socket 939 processor that wasn't on the roadmap

Matthew Werner emails to tell us that the upcoming AMD Athlon 64 X2 5000+ processor information was accidentally leaked by AMD earlier this week -- and is still visible in the corporate "tech doc" whitepaper (PDF) (Google HTML cache).

The Athlon 64 X2 will continue along AMD's A5BV Manchester 2x512KB L2 core, but a side note on the document claims the 5000+ will have 2MB of L2 cache. This is almost certainly a typo because while there are 2x1MB Toledo Athlon 64 cores with some cache disabled, additional cache cannot just be added to a core design.

The 2.6GHz 5000+ comes as a bit of a surprise to many as AMD had originally no plans for new Socket 939 processors on the most recent roadmaps. Like the FX-60, the X2 5000+ will have a TDP envelope of 110W.

AMD frequently updates its whitepapers to give system builders and integrators a heads up on exact specifications for certain processors. Typically, these updates either go unnoticed or do not end up on the public AMD site until after the actual product launch.

Comments

Threshold

Username
Password
remember me

This article is over a month old, voting and posting comments is disabled

By boinkle on 3/23/2006 6:31:27 AM , Rating: 2

110w? Give me a break... AMD's power requirements are getting just as obscene as Intel's.

RE: Ha

By weskurtz on 3/23/2006 6:40:19 AM , Rating: 2

Sure the 90 nano process is reached it's end, but with the new tech's and 65 nano, it will be a whole different story.

Parent

RE: Ha

By Viditor on 3/23/2006 6:46:30 AM , Rating: 2

quote:
110w? Give me a break... AMD's power requirements are getting just as obscene as Intel's

That's TDP, not the power requirements...AMD and Intel give values for different things on their TDPs.

Parent

RE: Ha

By Phynaz on 3/23/2006 12:45:59 PM , Rating: 3

This has been proven to be a myth.

Parent

RE: Ha

By Viditor on 3/23/2006 6:59:41 PM , Rating: 3

quote:
This has been proven to be a myth

Ummm...no. It's in the spec sheets, and you may read for yourself.
http://www.silentpcreview.com/arti cle169-page3.htm...

That's just an article listing and explaining the specs, which you may acquire at both Intel and AMD.

Parent

RE: Ha

By Phynaz on 3/24/2006 10:29:03 AM , Rating: 2

Nice link there (broken).

You should read the specs yourself for an education. I'll quote from Intel for you:

"Thermal Design Power (TDP) is a power dissipation target based on worse-case applications."

From further into the same document:

"The TDP and Maximum Case Temperature are defined as the maximum values"

So quit spreading the myth Intel under reports TDP. They don't.

Parent

RE: Ha

By Viditor on 3/25/2006 9:04:48 PM , Rating: 2

Try this link...
http://tinyurl.com/zt768

quote:
You should read the specs yourself for an education

They're sitting in front of me, mate!

“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.”

“Analysis indicates that real applications are unlikely to cause the processor to consume maximum power dissipation for sustained periods of time. Intel recommends that complete thermal solution designs target the Thermal Design Power (TDP) indicated in Table 26 instead of the maximum processor power consumption. The Thermal Monitor feature is intended to help protect the processor in the unlikely event that an application exceeds the TDP recommendation for a sustained period of time.”

AMD's definition:

“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.”

The Intel data sheet is available here at http://tinyurl.com/mnglz
You want Page 80, paragraph 2...

Parent

RE: Ha

By Viditor on 3/25/2006 9:39:05 PM , Rating: 2

quote:
"Thermal Design Power (TDP) is a power dissipation target based on worse-case applications."

That's right...Intel's TDP is based on their idea of the worse-case application, NOT the maximum output (as AMD does).

quote:
"The TDP and Maximum Case Temperature are defined as the maximum values"

You need to learn to read what they are saying...
What they mean is that for the purposes of Thermal Design, you should consider TDP to be the maximum power, and NOT the ACTUAL maximum power...

BTW, as you can now see, they actual myth is that Intel and AMD report TDP the same...

Parent

RE: Ha

By Targon on 3/23/2006 8:53:54 AM , Rating: 2

AMD gives a max power requirement for the socket, not the power demand for a given chip. A CPU might only use 55 watts, but the TDP that AMD lists might be 89 watts. A part of this is for those who overclock and to give an idea where the chip won't be able to handle as well(even if you use liquid nitrogen and manage to get and keep the temperatures low).

Parent

RE: Ha

By Slaimus on 3/28/2006 4:34:53 PM , Rating: 2

It is not per socket as per CPU class. All 90nm non-low power A64 X2 have the same TDP designation. This is to make things easier for system builders when choosing cooling options. As a result, this ends up being the heat production of the highest planned processor in this class.

Parent

Who in their right mind pay $1200 for this when Conroe creams it in 3 months?

By DallasTexas on 3/23/2006 8:16:38 AM , Rating: 1

$1200 is a chuck of change for 3 months. By then Conroe will run circles around it and at HALF the power consumption. Dunno about this.

If you insist on AMD, which is understandable for the David & Goliath crowd, you might as well wait for AMD's 65nm offering later this year (presumably).

RE: Who in their right mind pay $1200 for this when Conroe creams it in 3 months?

By SexyK on 3/23/2006 9:29:21 AM , Rating: 2

Seems that it'll be a longer wait than that for AMD 65nm chips. From Anandtech's 965 review:

quote:
From Intel's perspective, AMD won't be producing 65nm in volume until 2007, the same year when Intel will start playing with 45nm chips.

Parent

RE: Who in their right mind pay $1200 for this when Conroe creams it in 3 months?

By Stele on 3/23/2006 10:21:51 AM , Rating: 2

Well 65nm would be useful in lowering power consumption/dissipation and helping frequency scaling, but it's not a do-or-die (pun intended) thing.

As the whole Athlon vs. Pentium saga has often illustrated, Intel likes emphasizing numbers (MHz, nm, etc.) - to be fair, so does everyone but Intel puts rather more weight than many into marketing them - while AMD demonstrates that numbers aren't necessarily everything. After all, the first Athlon was on 0.25 micron vs. Pentium III's 0.18 micron, and we all know how much that helped Intel when the Athlon first hit. ;)

Parent

RE: Who in their right mind pay $1200 for this when Conroe creams it in 3 months?

By defter on 3/23/2006 3:12:37 PM , Rating: 2

quote:
Intel likes emphasizing numbers (MHz, nm, etc.)

AMD also liked to emphasized numbers when they were in lead, e.g. they were first to hit 1GHz mark.

quote:
After all, the first Athlon was on 0.25 micron vs. Pentium III's 0.18 micron

When Athlon was launched, only 0.25 micron Pentium III existed. 0.18 micron Athlon was launched about two months later than 0.18 micron Pentium III:

Parent

RE: Who in their right mind pay $1200 for this when Conroe creams it in 3 months?

By killerroach on 3/23/2006 12:30:33 PM , Rating: 2

Also, where's this $1200 number coming from, other than one of the earlier posts? C'mon, lay off of the hallucinogens, people. If AMD's flagship CPU (the FX-60) is $1000, then I doubt this will end up costing even more than that. More likely than anything, this chip will be priced closer to where the x2 4800+ is right now, around $600.

Parent

RE: Who in their right mind pay $1200 for this when Conroe creams it in 3 months?

By Viditor on 3/24/2006 2:33:57 AM , Rating: 2

quote:
Seems that it'll be a longer wait than that for AMD 65nm chips

Even Anand gets it wrong occasionally...Hector Ruiz has stated a couple of times now that 65nm begins volume production in Q3, with shipping at the end of this year.
Of course this is not all models, but all production will be 100% 65nm by next June (2007)

Parent

RE: Who in their right mind pay $1200 for this when Conroe creams it in 3 months?

By Viditor on 3/23/2006 7:05:52 PM , Rating: 2

quote:
By then Conroe will run circles around it and at HALF the power consumption. Dunno about this

Mate, you must have received glasses during Valentines Day as they appear to be rose coloured...

1. You are making up the "$1200"
2. Where on Earth did you get "HALF the power consumption" from?

Parent

Woohooo!!!

By astrodemoniac on 3/23/2006 4:30:16 AM , Rating: 2

An extra 2 FPS for a mere $1200 give or take... it's a win win situation <0.0>

RE: Woohooo!!!

By timmiser on 3/23/2006 12:02:51 PM , Rating: 2

You are insinuating you already have a 4800+ and therefore the improvement is minimal. How is that different from any other speed upgrade in the history of processors or any other computer component that regularly comes out with faster products?

$1200? More probably about $800. Your used 4800+ is currently worth about $600 so it would only be about $200 out of your pocket.

Parent

RE: Woohooo!!!

By TomZ on 3/23/2006 2:56:45 PM , Rating: 2

Used 4800+ for $600... Good luck!

Parent

RE: Woohooo!!!

By timmiser on 3/23/2006 6:44:04 PM , Rating: 2

Well then, feel free to correct my number with your number, subtract the total from 800 to get your revised subtotal that you agree with. Insert your revised subtotal over my subtotal to come up with your net total. After this, re-read the entire message with your numbers, rinse, and repeat.

Parent

RE: Woohooo!!!

By glennpratt on 3/29/2006 11:47:12 PM , Rating: 2

Someones about to get owned!

http://cgi.ebay.com/AMD-ATHLON-64- X2-DUAL-CORE-480...

Parent

Confused?

By timmiser on 3/23/2006 11:55:51 AM , Rating: 3

A 2.6 GHz chip with 2x1MB cache should be a 5200+.

A 2.6 GHz chip with 2x512MB should be a 5000+.

If it's the latter, I would think the 5200+ should not be far away.

RE: Confused?

By AkumaX on 3/23/2006 2:59:01 PM , Rating: 3

AMD confused a lot of us when they decided not to give a codename for Toledos with 1/2 the cache disabled.

For instance:

Barton with 1/2 cache = Thorton
Tbred with 1/2 cache = Applebred

BUT

Toledo with 1/2 cache = ??? - I decided to call it Toledo512
It's not a Manchester, because Manchester's don't have any disabled cache.

Parent

fill in gap

By fliguy84 on 3/23/2006 4:45:51 AM , Rating: 2

this new cpu will nicely fit between 4800+ and fx-60 if priced around ~$800

RE: fill in gap

By vtohthree on 3/23/2006 6:18:59 AM , Rating: 2

LONG LIVE S939!!!

Parent

So this is what I've had all along.

By bunnyfubbles on 3/23/2006 10:55:55 AM , Rating: 2

Yes my Opteron 165 gives me the performance of an FX60 at stock, but its got the locked upper multis so its more in line with the 5000+ ;)

Granted I'm sure the FX60/5000+ have a far greater potential to reach higher OCs, I'm maxed out around 2.75GHz with stability at relatively low volts (1.4v) @ 2.6GHz, I hit a pretty big wall around 2.7 where more volts doesn't do much except produce more heat (and thus noise).

RE: So this is what I've had all along.

By Ringold on 3/23/2006 11:37:18 PM , Rating: 2

I've got 2.6 on my 3800+ myself easy, 2.7 if I use ACTIVE watercooling (pointing a fan at my Zalman Reserator, lol), and your experience with the OC and mine are why I don't let the pro-intel anti-amd doom-and-gloom rhetoric get to me.

With chips that aren't cutting edge, we get OC's that Intel benchmarked with on an implication it was the best AMD could do down the road. We really think that even the 90nm process can't be pushed higher than 2.6? We really think that 2.6, a speed original X2s that came out quite a little while now achieve, is the best AMD can deliver another six months or so down the line?

I think Intel is finally putting up a fight instead of simply using its size to avoid confrontation, so AMD is and will be under pressure, but no worries. 2.6 today, no reason why not 2.8 or 3ghz couldn't be common OC's on chips purchased later this year.

Okay. back to enjoying the power of a cheap, easy OC that Intel apparently tried to make look like was break-neck hard-core bleeding edge difficult that I've been enjoying for about six months.

Parent

So what's the difference between this and an FX-60?

By Heatlesssun on 3/23/2006 12:28:53 PM , Rating: 2

I am a little confused. This looks to me to be an FX-60 rebranded. Maybe no unlocked multiplier would be the biggest difference I could think of.

RE: So what's the difference between this and an FX-60?

By jkresh on 3/23/2006 3:55:40 PM , Rating: 2

This has 2 * 512kb cache vs. 2 * 1mg cache, a 5200 if it existed would be an fx 60 without the unlocked upper multiplier (all athlon64's can lower the multiplier only fx's are completely unlocked).

Parent

Can you say....

By keitaro on 3/23/2006 4:32:52 AM , Rating: 2

Oops?

If... keyword = IF ... they decide to release this, I hope for AMD's sake that they do it with a limited run/supply.

It's funny. thing changes

By clnee55 on 3/29/2006 3:34:53 PM , Rating: 2

It is funny to see AMD fans now have to stand up to defend AMD power consumption. The same thing Intel fans did 2 years ago. This is just life, the two companies leap frog and grow. Every customer enjoys the competition. Don't get hung up on the TDP definition. Get a life instead.