Barcelona debut in June, 2.1 - 2.3 GHz ?

[NoobyDoo]

Link

Original Source

They say the chips will come out in June in three main flavors clocked at 2.1GHz, 2.2GHz, and 2.3GHz. Each flavor will available in multi-socket (82xx) and dual-socket (22xx) variants, and all models will have 512KB of L2 cache per core, 2MB of L3 cache per chip, DDR2-667 memory support, and a 95W thermal envelope. HKEPC goes on to say that AMD won't introduce faster quad-core Opterons until the third quarter, and that those chips will run at 2.4GHz and 2.5GHz with 120W thermal envelopes. The fastest quad-core Opteron on AMD's roadmap is a 2.6GHz model scheduled for the second quarter of next year, the site claims.

 

[Brunnis]

The low frequencies indicate that these new AMD CPUs are faster than Core 2 clock-for-clock. It could of course be that they can't run them on higher frequencies, but I kind of doubt that. It would be pretty cool if these CPUs are some 15% faster on average, compared to C2D. That's one hell of an IPC.

 

[f4phantom2500]

I'm hoping they run slower, somewhere in the mid-high 1GHz range (assuming that this would be sufficient to make them competitive with Conroe/Penryn), more ocing headroom

 

[dmens]

Originally posted by: Brunnis
It could of course be that they can't run them on higher frequencies, but I kind of doubt that.

from experience, that's the usual reason. scratch that... always the reason.

 

[Furen]

Remember that these are quad-core, so they only need to compete against the 2.66GHz Kentsfield. The TDPs are quite nice (95W) and the clock speed MAY be enough to match the Kentsfield but I doubt they'll outperform it to a significant degree (with the possible exception being FP performance). The higher-clocked dual-core parts may be more interesting in an absolute performance race. Hopefully these will actually overclock nicely enough to match Intel when overclocked, too.

 

[Lonyo]

Originally posted by: Furen
Remember that these are quad-core, so they only need to compete against the 2.66GHz Kentsfield. The TDPs are quite nice (95W) and the clock speed MAY be enough to match the Kentsfield but I doubt they'll outperform it to a significant degree (with the possible exception being FP performance). The higher-clocked dual-core parts may be more interesting in an absolute performance race. Hopefully these will actually overclock nicely enough to match Intel when overclocked, too.

Intel are planning 3GHz quad core and a low voltage 50W TDP quad core by mid 2007
http://www.xbitlabs.com/news/cpu/display/20070221162530.html

Still, the company has a plenty of headroom with its 65nm chips and has plans to introduce 3.0GHz Intel Xeon ?Clovertown? chips going forward.

the company plans to introduce low-power quad-core chip with 50W thermal design power in early March

 

[Viditor]

Originally posted by: Furen
Remember that these are quad-core, so they only need to compete against the 2.66GHz Kentsfield. The TDPs are quite nice (95W) and the clock speed MAY be enough to match the Kentsfield but I doubt they'll outperform it to a significant degree (with the possible exception being FP performance). The higher-clocked dual-core parts may be more interesting in an absolute performance race. Hopefully these will actually overclock nicely enough to match Intel when overclocked, too.

Slight correction there...they are matched against Cloverton (Kentsfield is desktop). In the 2P they should still surpass the 2.66 Cloverton (because of the superior scaling), and they might break even in the 1P configuration.

Edit: It's worth noting that the 3rd quarter is only a few weeks later...

Edit 2: Also don't forget that the 95w TDP is even less accurate on these...Barcelona can adjust power on each core individually to minimize usage.

Edit 3: I just read the stuff, and I am quite dubious about the HKEPC report...they said "The fastest quad-core Opteron on AMD's roadmap is a 2.6GHz model scheduled for the second quarter of next year, the site claims"

I'm fairly sure that this is incorrect...

 

[NoobyDoo]

Edited : Sorry, my bad.

 

[A554SS1N]

Originally posted by: dmens

Originally posted by: Brunnis
It could of course be that they can't run them on higher frequencies, but I kind of doubt that.

from experience, that's the usual reason. scratch that... always the reason.

Agreed.

 

[sdsdv10]

Originally posted by: Viditor
Edit: It's worth noting that the 3rd quarter is only a few weeks later...

I guess I'll pick some nits, but by my calendar 3Q starts July 1st which as of today (2/23) is at least 18 weeks away. In my opinon, that is a little more than "a few weeks". Not trying to be jerk, as I am waiting as anxiously as everyone else here but it will come out when it comes out. That doesn't stop me from checking here and at xtremesystems every day for leaked benchmarks.

 

[Furen]

Originally posted by: sdsdv10

Originally posted by: Viditor
Edit: It's worth noting that the 3rd quarter is only a few weeks later...

I guess I'll pick some nits, but by my calendar 3Q starts July 1st which as of today (2/23) is at least 18 weeks away. In my opinon, that is a little more than "a few weeks". Not trying to be jerk, as I am waiting as anxiously as everyone else here but it will come out when it comes out. That doesn't stop me from checking here and at xtremesystems every day for leaked benchmarks.

He's saying that Q3 starts a few weeks after the June launch for the 2.1-2.3GHz parts...

 

[jpeyton]

I'm glad they're launching their low-end parts first. I hate how Intel always intros the most expensive "Extreme" versions of new technology first, while budget quads are nowhere to be found.

 

[SickBeast]

Originally posted by: dmens

Originally posted by: Brunnis
It could of course be that they can't run them on higher frequencies, but I kind of doubt that.

from experience, that's the usual reason. scratch that... always the reason.

So then why are the C2D chips still being sold at ~1ghz slower than they are capable of?

 

[dmens]

Originally posted by: SickBeast

Originally posted by: dmens

Originally posted by: Brunnis
It could of course be that they can't run them on higher frequencies, but I kind of doubt that.

from experience, that's the usual reason. scratch that... always the reason.

So then why are the C2D chips still being sold at ~1ghz slower than they are capable of?

because the tester doesn't strap a fancy heatsink onto the part and runs at a lower voltage?

besides, i doubt those overclocked parts will last the 7 years promised by intel... of course, the warranty is voided as soon as it goes above the rated speed. tough beans.

 

[Lonyo]

Originally posted by: SickBeast

Originally posted by: dmens

Originally posted by: Brunnis
It could of course be that they can't run them on higher frequencies, but I kind of doubt that.

from experience, that's the usual reason. scratch that... always the reason.

So then why are the C2D chips still being sold at ~1ghz slower than they are capable of?

Because they can be.
If Intel offered higher speed-bin parts, what would they do? Cut the rest of their prices and hurt their real margins? Add the new SKU with an astronomically high price?
There's no NEED for Intel to release anything higher than 3GHz at the moment, they have the fastest chip.

 

[sdsdv10]

Originally posted by: Furen

Originally posted by: sdsdv10

Originally posted by: Viditor
Edit: It's worth noting that the 3rd quarter is only a few weeks later...

I guess I'll pick some nits, but by my calendar 3Q starts July 1st which as of today (2/23) is at least 18 weeks away. In my opinon, that is a little more than "a few weeks". Not trying to be jerk, as I am waiting as anxiously as everyone else here but it will come out when it comes out. That doesn't stop me from checking here and at xtremesystems every day for leaked benchmarks.

He's saying that Q3 starts a few weeks after the June launch for the 2.1-2.3GHz parts...

Oops, sorry Viditor I misread your post. My apologies, please ignore previous stupidity on my part. I should read a little closer. Thanks for the correction Furen...

 

[HurleyBird]

Also remember that bulk silicon generally needs greater headroom than SOI because you can't test each individual chip to the point of failure.