Whatever happened to Moore's law, amd??

[Knowname]

I have a Phenom II from some 18 months ago.... I don't think there's anything faster (from amd) than the first batch of Phnom II's! am I right about this or not? It doesn't seem like amd's done anything except for backpedal with Athlon II's. I haven't really kept up, but am I correct or what?

 

[Hyperlite]

404 error: teh 18 months not found.

http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=3492

You're wrong about the time frame...but you may be right in the end. I don't think AMD will have a qualifying part out by july.

 

[cbn]

Next month RD890/SB850 chipsets (with SATA III ports) will be released alongside Phenom II x6.

The Phenom II x6 will be a 45nm Hex core/6 MB L3 cache and will include a "Turbo mode".

 

[Rhoxed]

AMD Released a new C3 revision that got them 3.4 in the 125W envelope and soon releasing a 975 @ 3.6ghz (higher clockspeeds at 45nm then intel had, yet the clock vs clock is still a bit behind)

AMD of course is advancing, but they need money to do that, covering the extreme low-end to midrange with a plethora of bang-for-buck chips assures them a couple more years to get bulldozer out. If bulldozer flops.... then we will see sad days i believe.

 

[EarthwormJim]

I have a Phenom II from some 18 months ago.... I don't think there's anything faster (from amd) than the first batch of Phnom II's! am I right about this or not? It doesn't seem like amd's done anything except for backpedal with Athlon II's. I haven't really kept up, but am I correct or what?

Phenom II launced in December of 2008, it hasn't been 18 months yet bud.

Moore's Law pertains to transistor density, not speed.

 

[Hulk]

Ever watch MMA?

The Core2Duo was like Intel giving AMD a flying knee.

The ref never pulled Intel off and they are still pounding AMD.

This is all meant in jest boys as I realize that AMD does have some very good buys for the money these days. Intel is dominating the high end though.

 

[cbn]

Ever watch MMA?

The Core2Duo was like Intel giving AMD a flying knee.

The ref never pulled Intel off and they are still pounding AMD.

This is all meant in jest boys as I realize that AMD does have some very good buys for the money these days. Intel is dominating the high end though.

Have you read this thread yet?

I am a little worried AMD will not be able to respond in the same way Intel did back in 2006.

In fact, we might very well see a slow down in IPC improvements.

 

[sandorski]

FYI-Moore's Law isn't really a "Law". Eventually it will be broken.

 

[edplayer]

Moore worked for Intel

AMD's law is "Hey! Our stuff is cheaper."

 

[LoneNinja]

Phenom II launched Jan 2008, only 13 months ago, not 18. LOL

Thuban will be out soon, but I doubt that'll help performance for anything but the limited number of applications that can use 6 threads. Server side will also have a 12 core chip soon too, but as far as a new architecture or major IPC improvement, that'll be in 2011. It is a shame AMD isn't able to keep up with Intel's tick tock cycle, 45nm K10 just isn't cutting it against Nehalem, and by the time Bulldozer arrives Intel will have Sandy Bridge.

 

[nyker96]

I have a Phenom II from some 18 months ago.... I don't think there's anything faster (from amd) than the first batch of Phnom II's! am I right about this or not? It doesn't seem like amd's done anything except for backpedal with Athlon II's. I haven't really kept up, but am I correct or what?

when you think that you have to wait another 18 months to get BD, that's like 36 months or 3 years for something new from amd. I'm personally putting Intel at about 18 months ahead of amd in process technology as well as core design. It's really sad to see this lack of competition in cpu market. just look at intel pricing above $200. they haven't lowered prices since these cpu's intro days. why should they? zero competition.

 

[deimos3428]

I have a Phenom II from some 18 months ago.... I don't think there's anything faster (from amd) than the first batch of Phnom II's! am I right about this or not? It doesn't seem like amd's done anything except for backpedal with Athlon II's. I haven't really kept up, but am I correct or what?

I'm pretty sure Moore's law was intended as an upper bound. There's no compelling reason technology must advance that quickly -- it merely can advance that quickly.

 

[Kraeoss]

Moore worked for Intel

AMD's law is "Hey! Our stuff is cheaper."

hahah rofl i like this law

 

[dguy6789]

I have a Phenom II from some 18 months ago.... I don't think there's anything faster (from amd) than the first batch of Phnom II's! am I right about this or not? It doesn't seem like amd's done anything except for backpedal with Athlon II's. I haven't really kept up, but am I correct or what?

AMD has 6 and 12 core processors around the corner. Bulldozer is most likely coming 18 months after that.

I'm not finding any kind of violation of Moore's Law?

 

[Kraeoss]

hrmmm the 6 and 12 core chips are gonna be server chips or might they have desktop variants ?

 

[aigomorla]

AMD has 6 and 12 core processors around the corner.

gets raped by gulftown.

 

[alyarb]

more cores, integrated radeons, super-wide vectors are exciting, but these things can/will be achieved using a core very similar to K10.5 and the complexity per area doesn't really increase. i don't know how they will get their general purpose performance per thread back in line with intel's.

 

[dguy6789]

gets raped by gulftown.

Magny Cours will destroy Gulftown in any application that can keep twelve threads fully fed. Applications that take advantage of 6 or fewer threads will perform better on Gulftown.

I don't know what this has to do with the claim that AMD isn't following Moore's Law.

 

[aigomorla]

Magny Cours will destroy Gulftown in any application that can keep twelve threads fully fed. Applications that take advantage of 6 or fewer threads will perform better on Gulftown.

I don't know what this has to do with the claim that AMD isn't following Moore's Law.

^_^

I'll take 6 x32nm i7 cores + HT = 12

over 12 real AMD cores, any day.. ()

So.. once again.. sorry to say..
AMD gets RAPED by Gulftown.

To me moores law is my worst enemy.. lol..
Means i need to upgrade...

This is how i will answer moores law in a few months..

24 working threads makes me a obsessive pig...

 

[Cogman]

^_^

I'll take 6 x32nm i7 cores + HT = 12

over 12 real AMD cores, any day.. ()


So.. once again.. sorry to say..
AMD gets RAPED by Gulftown.

To me moores law is my worst enemy.. lol..
Means i need to upgrade...

This is how i will answer moores law in a few months..

24 working threads makes me a obsessive pig...

In highly threaded applications, it is almost definite that 12x real cores will beat 6 real cores + 6 logical cores. Hyperthreading is cool and all, however, unless amd really screws up things, I don't see a 6 core i7 beating a 12 core amd (Of course this is all specuation)

That being said, I would pick the i7 over the 12 core. Barring a miracle leap in performance, single, and lightly threaded applications are most likely to perform better on the i7.

 

[aigomorla]

In highly threaded applications, it is almost definite that 12x real cores will beat 6 real cores + 6 logical cores. Hyperthreading is cool and all, however, unless amd really screws up things, I don't see a 6 core i7 beating a 12 core amd (Of course this is all specuation)

...

a single i7 cpu has more production then dual optys.

This has been talked about over and over again.

And the OC advantage intel has over AMD, its a landslide victory.

And i already beat a 12c/24t Gulftown @ 2.4ghz... why cant i beat an AMD clocked even higher?

 

[Hyperlite]

...

a single i7 cpu has more production then dual optys.

This has been talked about over and over again.

And the OC advantage intel has over AMD, its a landslide victory.

And i already beat a 12c/24t Gulftown @ 2.4ghz... why cant i beat an AMD clocked even higher?

Link? and you cant add OCing to the argument when we're talking about server processors.

And what is the source of this delusion that a logical core is as good as a physical core?

 

[veri745]

...

a single i7 cpu has more production then dual optys.

This has been talked about over and over again.

And the OC advantage intel has over AMD, its a landslide victory.

And i already beat a 12c/24t Gulftown @ 2.4ghz... why cant i beat an AMD clocked even higher?

And you're overclocking a server why?

*edit* aww, hyperlite beat me to it...

 

[IntelUser2000]

Magny Cours will destroy Gulftown in any application that can keep twelve threads fully fed. Applications that take advantage of 6 or fewer threads will perform better on Gulftown.

Actually, Magny Cours is said to be clocked around 25% lower than Istanbul, so you can't look at it like that. Even in highly threaded apps, Gulftown will be competitive with Magny Cours due to its high clock. At worst, its probably only 10-15% slower than Magny Cours.

(In server apps)
-6 core Istanbul performs 10-15% less than 4 core Bloomfield
-Gulftown with 6 cores will clock slightly higher than 4 core Bloomfield
-12 core Magny Cours clock 20-25% lower than 6 core Istanbul

2x more cores with 25% lower clock vs. 50% more cores and 10% higher clock, and Bloomfield is already a bit better than current gen.

Destroy? I think not.

 

[Hyperlite]

Actually, Magny Cours is said to be clocked around 25% lower than Istanbul, so you can't look at it like that. Even in highly threaded apps, Gulftown will be competitive with Magny Cours due to its high clock. At worst, its probably only 10-15% slower than Magny Cours.

(In server apps)
-6 core Istanbul performs 10-15% less than 4 core Bloomfield
-Gulftown with 6 cores will clock slightly higher than 4 core Bloomfield
-12 core Magny Cours clock 20-25% lower than 6 core Istanbul

2x more cores with 25% lower clock vs. 50% more cores and 10% higher clock, and Bloomfield is already a bit better than current gen.

Destroy? I think not.

OK OK OK so since we're into a full blown server discussion now, this "speed" argument is completely moot until we see power consumption figures. And if Gulftown requires a 25% clockspeed increase + HT to meet Magny Cours, well, you can fill in the blanks. All this fanboy-ism with utter disregard for all the variables bugs me. With my budget this time around and needing 4 cores, i was only allowed an AMD solution. I'm gladly going to play both sides, but the fact of the matter is, we cannot declare a winner in this argument yet.