Penryn 5 - 10% clock-for-clock improvement over Conroe

[continuation]

http://www.hkepc.com/bbs/hwdb.php?tid=837360&tp=Intel-Penryn-Preview&rid =837360

Some sample clock-for-clock improvements:

FarCry: 10.48%
Half-Life 2: 31.12%
Adobe Photo Elements 5.0.2 SmartFix: 7.82%
Divx 6.6 Alpha w/SSE4 1080p Mpeg2 to Mpeg4: 115.63% (SSE4)
PowerPoint - Print to XPS: 8%

These are all on top of any frequency improvement.

Also showed significant drop in power consumptions and temperature.

Overall quite impressive.

 

[r0ck]

oooh

 

[YO_KID37]

Does it make you wonder that the Average 9.8% Overall performance boost may have come because of a 4mb to 6mb boost too?

Because average boost in performance from a 2Mb vs 4Mb Core 2 Duo is about 10% too

Did you Wonder that???

 

[BaronMatrix]

The best thing about Penryn will be lower power. It will still need DUAL\QUAD FSB to compete on the 4P side. With Barcelona due this month, it may not be enough.

 

[NMDante]

The best thing about Penryn will be lower power. It will still need DUAL\QUAD FSB to compete on the 4P side. With Barcelona due this month, it may not be enough.

Based on what? Slides?

Also, with Barcelona due this month, as you stated, is it not odd that they also introduced a higher clocked Opteron. Why? Would that just eat into Barcelona sales?

 

[turboflame]

I think Barton *could* be right, barcelona supposedly will have a massive hypertransport bus along with an L3 cache to decrease latency. If it turns out that the fsb is a detriment to Intel's design, it's gonna show.

and yea, I doubt any one here is gonna touch a barcy when they actually "launch", AMD will need some time to get them pumping out plus the first chips will be going right into the server industry.

 

[bydesign]

Does it make you wonder that the Average 9.8% Overall performance boost may have come because of a 4mb to 6mb boost too?

Because average boost in performance from a 2Mb vs 4Mb Core 2 Duo is about 10% too

Did you Wonder that???

That is kind of funny E6400 when clocked to E6600 speeds don't show a 10% defict in performance... That 2 vs. 4MB...

Penryn is just a minor change wait until Nehalem in 08' AMD currently has nothing in the near future to compete with. Concidering that AMD is already on the clock for clock song, translation we can't compete for top performer, they're in deep trouble.

 

[San Pedro]

31% increase in Half Life 2 is pretty damn impressive!

 

[htoonthura]

http://www.hkepc.com/bbs/hwdb.php?tid=837360&tp=In tel-Penryn-Preview&rid =837360

Some sample clock-for-clock improvements:

FarCry: 10.48%
Half-Life 2: 31.12%
Adobe Photo Elements 5.0.2 SmartFix: 7.82%
Divx 6.6 Alpha w/SSE4 1080p Mpeg2 to Mpeg4: 115.63% (SSE4)
PowerPoint - Print to XPS: 8%

These are all on top of any frequency improvement.

Also showed significant drop in power consumptions and temperature.

Overall quite impressive.

I skip the whole C2d line due to my motherboard and budget limitations. I think it is worthwhile to wait finally.

 

[pausert20]

I'm looking at using the new Wolfdale in a HTPC. Going get a Satellite card for it.

 

[BaronMatrix]

Based on what? Slides?

Also, with Barcelona due this month, as you stated, is it not odd that they also introduced a higher clocked Opteron. Why? Would that just eat into Barcelona sales?

AMD id do a closed box demo of Barcelona at Analyst Day. http://www.vnunet.com/tv/?channel=vnunet.com+TV&clipid=1386_vnunet_0096

The SPECfp score was 78 for Barcelona at 2GHz and 60 for Cloverton at 2.33GHz. Take it how you like but that's right around the 40% they touted for fp operations. Come August 12th the Phenom is to be demoed at 3GHz in Japan so some numbers should come out of that, preliminary or not.

I still don't understand why everyone is so certain AMD can't improve Opteron enough to make up for K8s deficiencies.

 

[BaronMatrix]

I think Barton *could* be right, barcelona supposedly will have a massive hypertransport bus along with an L3 cache to decrease latency. If it turns out that the fsb is a detriment to Intel's design, it's gonna show.

and yea, I doubt any one here is gonna touch a barcy when they actually "launch", AMD will need some time to get them pumping out plus the first chips will be going right into the server industry.

Actually, the first Barcelona's wil run on the current nVidia 3600 chipset with HT1.1. Phenom will ship with HT3 and I'm assuming 1P Barcelona will also. Anad's test of the new 3.2GHz Opteron show that as clockspeed increases, the available bandwidth increases so at 2GHz they won't need HT3 yet.

 

[Mephistopheles]

These numbers look really good!!

The whole Barcelona thing is kind of unknown though. We have no idea of the real improvements in barcelona. What would be nice to know was the relative improvement in IPC going from current opterons to Barcelona.

Intel's offerings look more tasty to us simply because we have more information. Now with these Conroe to Penryn improvements, and a whole lot of more tests (a full suite), we can tell that Intel strengthened their offer.

All we have for Barcelona are a few rather vague spec benches... something more representative would help AMD's image a lot right now.

 

[NMDante]

AMD id do a closed box demo of Barcelona at Analyst Day. http://www.vnunet.com/tv/?channel=vnunet.com+TV&clipid=1386_vnunet_0096

Don't care to watch something I already read about in December 2006 Analyst Day. Same stuff, different year. All talk about great things, nothing to really show.

The SPECfp score was 78 for Barcelona at 2GHz and 60 for Cloverton at 2.33GHz

. Take it how you like but that's right around the 40% they touted for fp operations. Come August 12th the Phenom is to be demoed at 3GHz in Japan so some numbers should come out of that, preliminary or not.

Really?

Where was this done? Got a link? It wasn't another "estimated" slide presentation, was it?
The 3GHz demo is going to be another hands off presentation, I assume. No real testing applications running, or anyone allowed to even run through the systems specs, I am assuming (like at Analyst Day).

I still don't understand why everyone is so certain AMD can't improve Opteron enough to make up for K8s deficiencies.

Because it's prepping to launch a product that is suppose to replace the Opteron? So, why keep improving on a product that you want to replace?