Official Desktop Penryn Discussion Thread

**red** · 08-05-2007, 01:57 PM

Originally Posted by informal

It won't be enough clock/clock against K10 derivatives.Especially for quad core ones.

lol

**Shintai** · 08-05-2007, 01:57 PM

Originally Posted by informal

As firts results for Penryn show up,http://forum.coolaler.com/showthread...=158813&page=4,
we can put to an end the talk and FUD about Penryn pefromance.
As for gaming benches OCed Penryn and Kentsfield score almost the same in CPU test in mark06
http://img.coolaler.com.tw/images/wd...jylnvnqtzn.png
http://img.coolaler.com.tw/images/ih...j3inlymtug.png

Difference at OCed speed is mere 2%.(where Kentsfield nullifies the FSB advantage of Yorkfield since it is reaching high FSB)
So the scores are practicly the same.
SPI benches show also a small advantage for Penryn(1.5s faster) that equally clocked Conroe at 2.33Ghz.
Where is the fast radix and other improvements intel was bombing the press?The only thing responsible for this small improvements is 6MB of L2 and higher FSB of Penryn.

Next look at Tmpengc perfomance in that topic.NO IMPROVEMENT AT ALL,and that test uses 128bit SEE! Where is so much desired improved super duper shuffle engine?Funny stuff indeed.

So Penryn is almost the same as Conroe,only featuring larger L2 and higehr FSB(later gets absolete since most of overclockers reach the same FSB with their Core2Duo & Quad systems).
Advantages are the SSE4 which COULD bring improvements once there is SSE4 aware software,possible higher OC(this hasn't been confirmed yet) and lower wattage at same clock as Conroe.Everything else is just a PR.Generally it is 5-6% faster or the same as Conroe and Core2Quad.

A CPU score of 5000 for a 3.2Ghz kentsfield? A 2.93Ghz Quad scores 4000.

And a 3.43Ghz is 4500.
http://www.computerpoweruser.com/edi...09%2F09c09.asp

Did you just make an ups informal?

**dippyskoodlez** · 08-05-2007, 01:59 PM

LOL.

post 1: Penryn shows no improvement!

post 2: Penryn shows MASSIVE 10%+++++ gaines!@!!!11111oneoneone

So, which benchmarker cheated?

**informal** · 08-05-2007, 02:03 PM

Yeah massive 10% on average is really amazing,i am amazed by Penryn,zomG.

On average it's ~9.2% faster clock/clock(with that 115% test) and ~7% without it.

BTW Wolfdale@2.33Ghz scores 2118 in CPU mark test in mark06,while Conroe e6600(2.4Ghz) scores 2111(techreport). It's ~3 faster ,zomG.

**Shintai** · 08-05-2007, 02:06 PM

I think informal is on thin ice

**informal** · 08-05-2007, 02:10 PM

Originally Posted by Shintai

I think informal is on thin ice

Sure i am.You have been walking on it for days in K10 topics,must be a cold feeling

**Shintai** · 08-05-2007, 02:14 PM

Originally Posted by informal

Sure i am.You have been walking on it for days in K10 topics,must be a cold feeling

Why dont you start with finding a 5000+ 3Dmark06 CPU scoring kentsfield.

**safan80** · 08-05-2007, 02:17 PM

I can get a 3dmark06 cpu score of 5036 with my qx6700 at 3.6Ghz so I hoping that a quad core penryn can do much better at 3.3Ghz

**Shintai** · 08-05-2007, 02:19 PM

Originally Posted by safan80

I can get a 3dmark06 cpu score of 5036 with my qx6700 at 3.6Ghz so I hoping that a quad core penryn can do much better at 3.3Ghz

See this is the fun part. You need 3.6Ghz to do what the other need 3.2...

**informal** · 08-05-2007, 02:21 PM

I just did,didn't I?I posted that SS,you could believe it or not.So does the other SS (yorkfileds) mean that it too was a fake?Or it is not real since it must be way faster than C2Q,and in reality it is not?
That OCed C2Q had 1800 MHZ FSB(much higher than Kentsfield in your picture) so it had much better scaling from 2 to 4 cores(81% versus 57% ).Also it was running uber fast RAM @570Mhz(efectively 1040Mhz).And plus it was benched on a probably highly tweaked system(also goes for Yorkfield).
So this is not logical to you?Hmm

**Shintai** · 08-05-2007, 02:23 PM

Originally Posted by informal

I just did,didn't I?I posted that SS,you could believe it or not.So does the other SS (yorkfileds) mean that it too was a fake?Or it is not real since it must be way faster than C2Q,and in reality it is not?
That OCed C2Q had 1800 MHZ FSB(much higher than Kentsfield in your picture) so it had much better scaling from 2 to 4 cores(81% versus 57% ).Also it was running uber fast RAM @570Mhz(efectively 1040Mhz).And plus it was benched on a probably highly tweaked system(also goes for Yorkfield).
So this is not logical to you?Hmm

http://www.hkepc.com/bbs/hwdb.php?ti...iew&rid=837360

I know this is a shocker to you. But they do match this!
http://www.anandtech.com/cpuchipsets...spx?i=2972&p=3

Obviously we'll reserve final judgments on Penryn for our official review of the CPU, but these initial results look very promising. We would expect to see clock for clock Penryn vs. Conroe improvements to be in the 5 - 10% range at minimum depending on the application. Factor in higher clock speeds and you can expect our CPU performance charts to shift up by about 20% by the end of this year.

OMG, that was what we all said!!!

So the only one that expected something else must have been you.

**informal** · 08-05-2007, 02:31 PM

What is a shocker?I got the average of 9%(including that large improvement in one test-sse4) looking at that exact review,~7% without that one.
The improvement in other tests(non sse4) is solely thanks to larger L2 and fsb.
No shocker for me,maybe for you since it's a minor improvement over existing(very good) Conroe.

**Shintai** · 08-05-2007, 02:34 PM

Originally Posted by informal

What is a shocker?I got the average of 9%(including that large improvement in one test-sse4) looking at that exact review,~7% without that one.
The improvement in other tests(non sse4) is solely thanks to larger L2 and fsb.
No shocker for me,maybe for you since it's a minor improvement over existing(very good) Conroe.

FSB? They use the same FSB little one

And try remove all the synthetics now you dont like the DiVX encoding

Oh, and in the one that most of us use..games..it does look pretty well!

**informal** · 08-05-2007, 02:42 PM

Little one??Tin ice indeed.

I said larger L2 and FSB in general,not in that hkepc test(meaning Conroe vs Penryn)
On the serious note ,big guy,the average in games is still around the number which corresponds to larger 6MB of L2.It is good there is an improvement with Penryn,not big in most of the cases,but a solid 5-10%.

**red** · 08-05-2007, 02:44 PM

I got the average of 9%(including that large improvement in one test-sse4) looking at that exact review,~7% without that one.

If you use that logic, including Minesweeper, Calculator, Wordpad, Firefox for "overall" performance, K8 would be equivalent to P4. http://www.anandtech.com/cpuchipsets...spx?i=3038&p=8 Or that "on average", A 3GHz dualie is just 3.7% worse than a 2.4GHz quad. Good thing for weighted averages, ignoring of pointless benches, individual workloads

The "minor" improvement of 100%+ in encoding and 10-30% in gaming is going to be quite appreciated I thunk.

http://pc.watch.impress.co.jp/docs/2.../comp08_10.jpg There is still gains to be had from X38 and from absolute clock gain (10%+, 3.33 vs 3).

**Shintai** · 08-05-2007, 02:44 PM

Originally Posted by informal

Little one??Tin ice indeed.

I said larger L2 and FSB in general,not in that hkepc test(meaning Conroe vs Penryn)
On the serious note ,big guy,the average in games is still around the number which corresponds to larger 6MB of L2.It is good there is an improvement with Penryn,not big in most of the cases,but a solid 5-10%.

http://www.bit-tech.net/hardware/200...o_processors/5
http://www.anandtech.com/cpuchipsets...spx?i=2795&p=4

And 4 to 6 would be even less.

**hollo** · 08-05-2007, 04:54 PM

Originally Posted by informal

http://img.coolaler.com.tw/images/wd...jylnvnqtzn.png
http://img.coolaler.com.tw/images/ih...j3inlymtug.png

not identical setups, i think he was just having a play around with the new toy (see memory speed (5:6 vs 4:5) & timings & windows style).

any clock for clock improvement is a bonus tbh, i wouldn't mind if penryn was 10% slower clock for clock. those temps posted by red are banana-worthy

10 degree temp drop with 50% extra cache
_b@intel

**Donnie27** · 08-05-2007, 07:47 PM

Originally Posted by red

I got the average of 9%(including that large improvement in one test-sse4) looking at that exact review,~7% without that one.

If you use that logic, including Minesweeper, Calculator, Wordpad, Firefox for "overall" performance, K8 would be equivalent to P4. http://www.anandtech.com/cpuchipsets...spx?i=3038&p=8 Or that "on average", A 3GHz dualie is just 3.7% worse than a 2.4GHz quad. Good thing for weighted averages, ignoring of pointless benches, individual workloads

The "minor" improvement of 100%+ in encoding and 10-30% in gaming is going to be quite appreciated I thunk.

http://pc.watch.impress.co.jp/docs/2.../comp08_10.jpg There is still gains to be had from X38 and from absolute clock gain (10%+, 3.33 vs 3).

Yea, IMHO 3D Marks anything is useless for cross platform comparisons since it doesn't compare directly to any Game Engine. Real world commonly used apps matter a lot more to me.

**Donnie27** · 08-05-2007, 08:00 PM

Benchmark Yorkfield Performance Advantage
3DMark '06 V1.1.0 Pro CPU.... (score) ..: 21.8%
3DMark '06 V1.1.0 Pro Overall.. (score) ..: 7.6%
Mainconcept H.264 Encoder (seconds) .: 18.0%
Cinebench R9.5 (CPU test).................: 24.9%
Cinebench R10 Beta (CPU test)...........: 25.5%
HL2 Lost Coast Build 2707 (fps) ...........: 37.3%
DivX 6.6 Alpha w/ VirtualDub 1.7.1 (seconds) 111%

If you remove the top, remove the bottom. 18 + 24.9 + 25.5 + 37.3 Divided by 4 for an average of 26.4% average of those tests. I don't know where the 5 to 10% keeps coming from

**Epsilon84** · 08-05-2007, 11:11 PM

Umm... why is Informal throwing a hissy fit over the performance enhancements in Penryn?! Most people here are impressed with the gains, considering it's mainly a die shrink, but he comes along with childish remarks of "5 -10% zOMG I'm amazed" WTF is wrong with you? It's one thing to be a fanboy, but please, don't act like a tard.

Anyway, a 9% increase on average is quite impressive for a die shrink with minor architectural improvements. Trying to say all that improvement comes from cache is ridiculous, considering 4MB C2Ds are only ~3.5% faster than 2MB C2Ds on average, and that is a 100% increase in cache size. 6MB is only 50% more than 4MB, and considering the law of diminishing returns, I'd be surprised if that extra 2MB accounted for anything more than a 1 - 2% increase in overall performance. Obviously the 'super shuffle' engine and radix 16 helps in certain situations, probably more so than the larger cache size.

**gallag** · 08-06-2007, 02:01 AM

All I want to see is Intel45nm vs AMD K10 so I know what to buy, And non of this clock for clock or performance per watt

, I just wanna know which is faster

**onethreehill** · 08-06-2007, 02:07 AM

Intel Sheds Light on "Penryn" Enhancements

On the surface, Penryn looks like die shrink of last year’s Conroe micro architecture, but Intel sought additional tweaks to the micro architecture to achieve greater performance at the same clock speeds as Conroe processors.

http://www.dailytech.com/Intel+Sheds...rticle8313.htm