AMD Athlon 64 X2 6400+

[mcrooster]

I found this article the other day, its seem's nobody's posting or maybe just nobody's interested, either way i decided to find out what you're all thinking.

Sneaking around the leaks
By INQUIRER newsdesk: Tuesday 24 July 2007, 15:31

IT SEEMS THAT the era of the Athlon 64 X2 will end with a product with numbered 6400+.
The company is obviously battling out leakage issues with K10 processors, as Charlie reports.
So it makes perfect sense to tweak up the existing line-up as high as possible.
A guinea-pig processor to try to sort out the problems is sampling right now to various partners, preparing for a launch designed ot coincide with the end of the summer holidays.
The highest-numbered model you can buy today is the 6000+. The 200MHz clock bump will yield a 400 PR mark increase, in line with AMD's policy of, um, exaggerated calibration.
This processor is expected to go up against Intel's Core 2 Duo E6850, which is nothing but a first Core 2 Extreme with a faster FSB - 1333MHz instead of the 1066MHz of yesteryear.
Preliminary benchmarks put the 3.2GHz Athlon above the 2.66GHz C2D E6700, but it comes a bit late, and performance is likely to be lower than that of the E6850. µ (I have sinse read it will be 10 - 15 % slower than the 6850).

After a bit of reading the only advantage i can see with this chip is the 200 mhz increase and that it runs at ddr2-800 unlike the 6000+ which runs at ddr2-750, im not sure what this means for performance but im sure we will find out on August 20th when its released in japan and we see some benchies.
Also i wonder how well it will overclock, 3.6Ghz on air??? lol i can but hope.

 

[darkfalz]

More embarrassment for AMD. 125 watts - might as well be a heater. Don't expect to overclock that piece of junk.

 

[hans007]

Originally posted by: darkfalz
More embarrassment for AMD. 125 watts - might as well be a heater. Don't expect to overclock that piece of junk.

thats really not THAT bad since the way they calculated heat output is different than intel and it has a memory controller on board .


for 90nm its pretty impressive , a 2.8 ghz dual pentium D 820 put out 120 or so as well and wasnt nearly as fast.

 

[Lonyo]

The 200MHz clock bump will yield a 400 PR mark increase, in line with AMD's policy of, um, exaggerated calibration.

I'm not sure I get that. For 1MB cache chips:
2GHz = PR or 4000+
2.2 = 4400+
2.4 = 4800+
...
3.2 = 6400+

PR = 2x clock speed. And they aren't even being compared to anything really anymore. Seems simple enough to have the PR at 2x clock speed.

 

[bradley]

Originally posted by: hans007

Originally posted by: darkfalz
More embarrassment for AMD. 125 watts - might as well be a heater. Don't expect to overclock that piece of junk.

thats really not THAT bad since the way they calculated heat output is different than intel and it has a memory controller on board .


for 90nm its pretty impressive , a 2.8 ghz dual pentium D 820 put out 120 or so as well and wasnt nearly as fast.

Yep. And Pentium D is still being sold till this December if I'm not mistaken. AMD is actually quite liberal when it comes to their energy estimates. I expect actually real-world usage to fall quite under that number.

 

[nyker96]

about time they killed off the X2 line, been using it like on and on along with this 90nm junk.

 

[meson2000]

Actually, untill recently the 3ghz 6000+ and the 3ghz Opteron were 120watt parts on 90nm. However, the latest core revision dropped these down to 95watt parts. This allowed AMD to make one more release of a 3.2ghz processor at the 120watt level. Its 200mhz faster, but it won't put out any more heat than the older 6000+ processors did.

 

[SlowSpyder]

I wouldn't exactly call it junk... it may not be as fast as Intel's current offerings, but the A64's have had a pretty good, long, and useful life. They can still do just about anything asked of them today.

 

[firewolfsm]

It took them a year, but they tweaked out an extra 600MHz from the chip.

It's priced right too, below the E6850.

 

[RussianSensation]

It's interesting how quickly people forget that A64 outperformed P4 by approximately 50% per clock cycle. Today, C2D is approximately 25-30% faster on average per clock cycle than A64, which is no small achievement. However, it isn't anything special either considering A64 architecture was released in September 2003 - almost 4 years ago! It took intel until June/July of 2006 to outperform A64, which is almost 3 years. Now considering that Barcelona was in development for many years now, what is the probability that it will outperform the old A64 architecture by 30%? I cannot believe how many people are so pessimistic about Phenom. AMD's problem with Phenom won't be its ability to outperform C2D per clock cycle - that's 99% given. Its main problem will be is scaling of the chip in mhz.

It wouldn't be unreasonable for AMD to increase efficiency by 50% over A64. But unfortunately a 2.4ghz Phenom even at 50% efficiency increase of A64 won't be enough to match a Penryn at 3.66ghz. The story would be completely difference if Phenom started out at 3.0ghz.

 

[BitByBit]

Personally, I don't think Barcelona will hit the clock speeds necessary to compete with Penryn. I expect its IPC to be roughly equivalent to Penryn's in integer code, and slightly above in FP, but certainly nothing that will compensate for Penryn's clock advantage. Shanghai will perform even better per clock than Barcelona, but more importantly should scale sufficient high to give Penryn a real run for its money. Hopefully, Shanghai will put Intel and AMD neck and neck, and we all know what that would mean

 

[darkfalz]

Intel caught AMD snoozing after they outperformed the netburst era, and now they have no answers for over a year and still sandbagging like crazy.

 

[BitByBit]

I don't subscribe to the idea that AMD got complacent with K8, for the following reasons.

1. New architectures take years to develop. When Netburst first started to ship, AMD would have been already working on K9.

2. We know AMD cancelled K9 for various reasons, in favour of what would become K10. Turning the ship around is probably the primary reason for AMD's apparent lack of response to Core 2.

3. Given the scaling and leakage issues Intel faced with Netburst, is it reasonable to assume that AMD didn't think Intel would have something else in the works? I'd certainly be willing to bet that AMD knew about Conroe before we did.

One thing is certain - the future will see Intel and AMD performing far more closely. AMD, like Intel, is promising a new architecture every two years. Having said that, one thing I feel AMD should have done is update K8. A few core tweaks resulting in maybe a 5-10% IPC increase could have softened the blow dealt by Core.

 

[sdsdv10]

Originally posted by: BitByBit
Shanghai will perform even better per clock than Barcelona, but more importantly should scale sufficient high to give Penryn a real run for its money. Hopefully, Shanghai will put Intel and AMD neck and neck, and we all know what that would mean

By the time Shanghai comes out in 2008 ( http://www.dailytech.com/AMD+4...n+2008/article7132.htm ), it will be competing with Nehalem ( http://www.dailytech.com/article.aspx?newsid=8147 ) not Penryn.

 

[Zstream]

Originally posted by: sdsdv10

Originally posted by: BitByBit
Shanghai will perform even better per clock than Barcelona, but more importantly should scale sufficient high to give Penryn a real run for its money. Hopefully, Shanghai will put Intel and AMD neck and neck, and we all know what that would mean

By the time Shanghai comes out in 2008 ( http://www.dailytech.com/AMD+4...n+2008/article7132.htm ), it will be competing with Nehalem ( http://www.dailytech.com/article.aspx?newsid=8147 ) not Penryn.

Well considering at stock the CPU's are closer then what people here make them out to be. Granted yes they do perform less but is it really that big of a difference?

I think something else caused the AMD decline such as Dell, ATI transaction, Hector and FUD.

AMD will get back on track but when is the question everyone is asking...

 

[bryanW1995]

I don't think that amd can hold out until shanghai unless barcelona/phenom produce the goods.

 

[darkfalz]

AMD are only going to "catch up" if Intel stays still. We are still waiting. Meanwhile Penryn is around the corner and Nehalem coming soon after that. Things look very, very bad for AMD right now.

 

[Gikaseixas]

Just a thought
if barcelona chips have only 100 mhz difference beteewn models then their IPC must be insanely good. People lets wait 2 more weeks and then make our assumptions. I say Barcelona will be a good cpu, hope so...

 

[Idontcare]

Originally posted by: Gikaseixas
Just a thought
if barcelona chips have only 100 mhz difference beteewn models then their IPC must be insanely good. People lets wait 2 more weeks and then make our assumptions. I say Barcelona will be a good cpu, hope so...

Even still you can't escape the mathematical certainty that if 2 chips differ by merely 5% in clockspeed (1.9 vs. 2.0 GHz) then the maximum performance difference between the two chips is likewise merely 5%...regardless of IPC.

/remembers the days where 486 chips doubled and tripled their performance (33 to 66 to 100 MHz) from chip-to-chip

 

[Idontcare]

Originally posted by: darkfalz
AMD are only going to "catch up" if Intel stays still. We are still waiting. Meanwhile Penryn is around the corner and Nehalem coming soon after that. Things look very, very bad for AMD right now.

Only one thing is certain, always has been and always will be, and that is Intel will continue to make barrels of cash regardless where the chips may fall performance wise.

Even in their darkest hour with the P4 cluster-fvck that was Prescott they still made some $6B in profit.