Penryn only 0-10% over Conroe?

[Polish3d]

http://www.anandtech.com/cpuch...howdoc.aspx?i=3069&p=3



Bummer! I was looking for 10%-20%+

 

[frythecpuofbender]

With a Q6600 OCed to 3Ghz+...I'm probably be able to skip Penryn completely and wait for Nehalem.
When 3.33Ghz Quad 45nm will be affordable (300$) it will probably be around Christmas 08/09. So it really was the right time now to upgrade from that stupid P4.

 

[Lonyo]

Seems odd they suggest the dual core will only clock to 3GHz and quad will hit 3.33GHz. I would expect parity between the two at least, if not dual core still clocking higher.

 

[HopJokey]

Originally posted by: Lonyo
Seems odd they suggest the dual core will only clock to 3GHz and quad will hit 3.33GHz. I would expect parity between the two at least, if not dual core still clocking higher.

Yes definitely the dual core chips have more headroom in the frequency department. However, what the article is referring to is what actual clockspeeds Intel will release the chips as, not their actual capabilities in terms of frequency.

 

[zach0624]

I am happy to see even a 5% speed increase from conroe with about 5%decrease in power consumption + (from the es chips that intel has supplied) better overclocking.

 

[idiotekniQues]

bring on nehalem!

will p35 support it? or at this point may as well wait for x68?

 

[SerpentRoyal]

The key selling point of 45nm is lower heat, which should result in a higher core speed. C2D is still the KING of value and performance (E2160/E4300). Wait another year before you upgrade to Peryn (45nm quads). 65 nm quads put out too much heat. I personally would not upgrade the CPU unless I could pickup another 15 to 20% from the new architechture.

 

[Nemesis 1]

Interesting results but a little more info would have been nice.

Like what M/B and ram was used.

Since X38 chipset is close to release . With its memory tweaks made for penryn I think those numbers between the performance delta's will change.

I found O/C really interesting. 1000 ghz. increase at same V. Is great. Put some more V. to wolfdale and show us what it has.

Can't wait for the Yorky update. It will be interesting to see how fast programmers optimize for SSE4 instructions. Which is were Penryn shines.

This was an interesting article. But for enthusist we need to see more . Such as.

Test done with X38 chipset .
O/C with higher V.
Testing done with DDR3 low latency done at 1600+ fsb . And of course the follow up test of Yorky.

On the Wolfdale/ Yorky stock clocks . It would seem that intel is binning the best chips for Yorky. Woldale is the rest of the bins. Which makes perfect Sense considering Intel has only one fab putting out these chips at the moment.

New steppings as 08 comes around is probably going to be in Intels plan . So I found this to be very good article all and all. I liked the O/C at stock Volts . Very impressive. But would have been nice to crank up the V. a bit to see how far these chips could really go.


On the cpu-z O/C charts what with the V. on the second chart?

 

[zsdersw]

Originally posted by: idiotekniQues
bring on nehalem!

will p35 support it? or at this point may as well wait for x68?

No. Nehalem will be Socket H (LGA715) and Socket B (LGA1366).

 

[Skott]

Not hugely surprised its only a max of 10% increase. More would have been better but we have yet to see full testing in OCing. I'm still thinking waiting till 2nd half of 2008 is better as far as upgrading to quadcore goes.

 

[Acanthus]

Dont forget that:

1. the overclocking was done at stock voltage only.

2. the benchmarks did not include SSE4, which is an enormous performance gain in certain applications.

 

[idiotekniQues]

Originally posted by: zsdersw

Originally posted by: idiotekniQues
bring on nehalem!

will p35 support it? or at this point may as well wait for x68?

No. Nehalem will be Socket H (LGA715) and Socket B (LGA1366).

ahh thanks.

good reason to stick with my setup until then.

 

[Idontcare]

I am not surprised at all regarding the measured performance delta between Penryn and Conroe, neither should anyone who has had a pulse and the ability to read any of the web articles from Intel and others as to what Penyrn was supposed to accomplish.

What I am surprised about is the meager power reduction as measured (albit sample size = 1 here) by Anandtech.

65nm to 45nm shrink AND poly-silicon gate to HiK/MG implementation and all that affords is a measly 10% lower power usage?

Granted there are an extra 2MB worth of L2$ xtors operating inside the Penryn chip, and it is 10% system power reduction so we can't say for certain what the total chip power usage reduction has been.

Hopefully that is what A1 (and beyond, just like B3 to G0 for 65nm quads) silicon is intended to bring, drastically reduced power usage.

 

[Accord99]

10% reduction in system power probably translates to the 2.33GHz Wolfdale idling at 1-3W and full load power of less than 35W.

 

[Rogue 2]

Maybe Intel is getting lazy again now that they are on top...
Or maybe people just expect too much.
Either way, the bar's set.. time for AMD to answer.

 

[Idontcare]

Originally posted by: Accord99
10% reduction in system power probably translates to the 2.33GHz Wolfdale idling at 1-3W and full load power of less than 35W.

Oh sure, bust out some actual numbers and stuff :laugh:

OK, I guess I should not have been so lazy. So from Anand's article, the conroe chip used 36.9W more at full-load than when idle (189.1-152.2 = 36.9W).

The Penryn system used 22.6W more at full-load than when idle (170.3-147.7 = 22.6W).

So we can see that the reduction in power savings at the CPU is perhaps in-line with expectations then. Penryn's active leakage appears to be roughly 61% that of Conroe (22.6/36.9 = 61%), or to say it differently Penryn is 39% less leaky than Conroe ((36.9-22.6)/36.9 = 39%).

Once you account for the xtor increase in going from Conroe to Penryn, you probably get pretty close to the expected 50% reduction in active leakage for the node shrink.

So I stand corrected, looks like a good job was done by the process guys after all. (not that they need me to tell them this)

 

[Zenoth]

I can't wait for Penryn, it'll be my next upgrade from my current system. I've never owned a C2D, completely passed over them since I thought my X2 was doing a fine job still. Now however I'm starting to see advantages in that architecture and two extra Cores, not to mention the potential in OC'ing. And it'll run on a X38 Mobo as well. I guess Penryn is supposed to be released in November/December right ?

 

[SickBeast]

According to Fudzilla, Barcelona will have a golden opportunity seeing as Penryn is slower than expected.

I really do expect Barcelona to be much better than people are saying.

 

[Dainas]

Much better as in competing on overall performance at stock speeds, sure why not. Still even after all this time Barcelona reeks of delay; Enough that I would not be surprised if AMD does not get a worth buying line until the eve of Penryn, by which time Intel will have yields atleast over-clocking sky high.

 

[bryanW1995]

Originally posted by: Zenoth
I can't wait for Penryn, it'll be my next upgrade from my current system. I've never owned a C2D, completely passed over them since I thought my X2 was doing a fine job still. Now however I'm starting to see advantages in that architecture and two extra Cores, not to mention the potential in OC'ing. And it'll run on a X38 Mobo as well. I guess Penryn is supposed to be released in November/December right ?

QX6950 will be a 3.33 ghz penryn quad core in Q4 07. The other penryns will be released in Q1 08. They will range from 2.33 to 3.16 with 2 not yet named speeds (unsure if they will be higher or lower).

 

[bryanW1995]

Originally posted by: SickBeast
According to Fudzilla, Barcelona will have a golden opportunity seeing as Penryn is slower than expected.

I really do expect Barcelona to be much better than people are saying.

I still don't expect barcelona to be much better than people are saying, but at least amd has a chance to be competitive before nehalem comes out now. 4.7 % is not much improvement at all and will probably get me to at least wait for phenom before building my next rig. I've noticed a "few" heat issues with running an e6750 at 3.6 with an opteron 180 in a small office together. These issues mainly involve some combination of: 1.) pregnant wife and 2.) south texas summer. I'll be interested to see if phenom keeps the heat in check better at a similar performance level to penryn. Don't get me wrong, if I expect to get 3.8 out of a penryn quad core and only 2.6 out of a phenom after oc, then we all know where I'm headed. But if phenom is a little bit better clock for clock (say 10%) and I can reasonably expect to get 3.3+ vs the 3.8 penryn after oc, plus it keeps heat under control much better than penryn, then I would probably get the phenom. Then, I would get pissed that my wife had a better computer than me and still buy the penryn. Maybe I could set up a dual-unit watercooling system and see which system dissipated more heat individually...hmmm...maybe AT will sponsor such a test...(mental gears grinding into action)...

 

[SickBeast]

Originally posted by: bryanW1995

Originally posted by: SickBeast
According to Fudzilla, Barcelona will have a golden opportunity seeing as Penryn is slower than expected.

I really do expect Barcelona to be much better than people are saying.

I still don't expect barcelona to be much better than people are saying, but at least amd has a chance to be competitive before nehalem comes out now. 4.7 % is not much improvement at all and will probably get me to at least wait for phenom before building my next rig. I've noticed a "few" heat issues with running an e6750 at 3.6 with an opteron 180 in a small office together. These issues mainly involve some combination of: 1.) pregnant wife and 2.) south texas summer. I'll be interested to see if phenom keeps the heat in check better at a similar performance level to penryn. Don't get me wrong, if I expect to get 3.8 out of a penryn quad core and only 2.6 out of a phenom after oc, then we all know where I'm headed. But if phenom is a little bit better clock for clock (say 10%) and I can reasonably expect to get 3.3+ vs the 3.8 penryn after oc, plus it keeps heat under control much better than penryn, then I would probably get the phenom. Then, I would get pissed that my wife had a better computer than me and still buy the penryn. Maybe I could set up a dual-unit watercooling system and see which system dissipated more heat individually...hmmm...maybe AT will sponsor such a test...(mental gears grinding into action)...

Do they really generate that much heat while pregnant? :Q

I've never heard of a wife affecting an overclock before. :light:

Barcelona should be cooler and more efficient than Penryn, but who knows until it's out.

I'm sure the Texas heat lowers your overclocks by a good 10%.

 

[bryanW1995]

maybe I should ship her up north to relatives in minneapolis or ohio until her bun is out of the oven. I can't let something stupid like this compromise my oc'ing!

 

[Acanthus]

Originally posted by: SickBeast
Barcelona should be cooler and more efficient than Penryn, but who knows until it's out.

Err... 45nm vs 65nm?

 

[Pabster]

Originally posted by: SickBeast
Barcelona should be cooler and more efficient than Penryn, but who knows until it's out.

Yeah, because 65nm will surely run cooler than 45nm! :laugh:

IMHO Penryn isn't going to be huge. It'll be like Yonah -> Merom. Nehalem looks tasty though.