backtop


Print 41 comment(s) - last by mrdelldude.. on Aug 27 at 5:07 PM

AnandTech puts "Penryn" to the test

Intel’s quad-core Penryn processors are just a quarter away from launching, but the company already has plenty of samples floating around. Hardware analysis site AnandTech managed to snag a dual-core Wolfdale based Penryn processor clocked at 2.33 GHz to run a slew of tests. AnandTech puts the 2.33 GHz Wolfdale up against the Core 2 Duo E6550, similar to the previously released Wolfdale benchmarks by HKEPC.

DailyTech previously reported Intel’s Penryn receives upgrades to its L2 cache. In addition to the larger cache size, Penryn processors have a 24-way associative. Conroe and Kentsfield processors have a 16-way associative L2 cache. The cache enhancements reduce the overall latency of the L2 cache, taking off a single clock cycle, when compared to Conroe.

AnandTech computes an average performance increase of close to 5%, when compared to Conroe. The most notable performance boost is with DivX 6.6 using Xmpeg. With an unoptimized build of DivX 6.6, AnandTech observed a 10.5 percent performance boost. Wolfdale also shows notable performance improvements in 3D rendering applications such as Cinebench and Lightwave – 9.1 percent and 6.7 percent, respectively.

Gaming improvements are all over the place, ranging from 0.4 percent to 8.5 percent, in Oblivion and Lost Planet, respectively. However, none of the benchmarks ran were optimized for SSE4, which showed a 116 percent performance lead in DivX 6.6.

Intel slightly reduced power consumption with its 45nm fab process. Although idle power consumption between Wolfdale and Conroe are quite similar, Wolfdale sips 20 watts less power than Conroe under load.

With a new 45nm process, Wolfdale overclocks fairly well. AnandTech managed to raise the front-side bus to 460 MHz, bringing the clock speed up to 3.22 GHz.

Expect dual-core Wolfdale to show up in 2008, in Xeon and Core 2 Duo variants.


Comments     Threshold


This article is over a month old, voting and posting comments is disabled

Barcelona Where is it?
By jdun on 8/22/2007 2:48:16 PM , Rating: 1
Where is AMD's Barcelona numbers because Intel will roll over them two months from now.




RE: Barcelona Where is it?
By Martimus on 8/22/2007 3:05:19 PM , Rating: 4
The benchmarks show that Penryn isn't much better than Conroe. (On average ~5%) So if Penryn would roll over Barcelona, so would Conroe. After seeing these benchmarks, I am pretty confident that Barcelona will be competitive to Penryn. I am not really worried anymore that Intel will have a vastly superior chip to Barcelona by the time it is released.


RE: Barcelona Where is it?
By Martimus on 8/22/07, Rating: 0
RE: Barcelona Where is it?
By johnsonx on 8/22/2007 10:03:30 PM , Rating: 1
I've seen that a couple of times too... The moment I click 'Post', it shows up as a 1 rather than a 2. No one could possibly have read it that fast and clicked 'Not Worth Reading'. Besides, ALL of my posts are Worth Reading!


RE: Barcelona Where is it?
By TomZ on 8/22/2007 10:28:15 PM , Rating: 1
I've noticed that once before, too.


RE: Barcelona Where is it?
By johnsonx on 8/23/2007 7:47:00 PM , Rating: 1
You've only noticed ALL my posts are worth reading once? You should notice it EACH AND EVERY TIME!


RE: Barcelona Where is it?
By wordsworm on 8/23/2007 8:06:17 AM , Rating: 2
Yeah, I've had that happen to me.


RE: Barcelona Where is it?
By wordsworm on 8/23/2007 8:09:48 AM , Rating: 2
I tried to rate you up Martimus... but it didn't get recorded for some reason. I hit 'worth reading', but your 0 stayed at 0. I think there's a bug at DailyTech. I guess the rating system isn't reliable. They should look into it.


RE: Barcelona Where is it?
By Martimus on 8/23/2007 10:35:15 AM , Rating: 2
It's not really a problem for me. I just thought it was strange, and felt like mentioning it. Thank you for trying to help though.


RE: Barcelona Where is it?
By MrBungle123 on 8/22/2007 3:24:57 PM , Rating: 2
Except Intel already has a vastly superior chip, and given AMD's recent history of price cut after price cut after loudmouth claim after delay after paper launch after more loudmouth clams, you should be worried.

It's time for AMD to put up or shut up. Their chip is supposed to drop in a couple weeks and we still haven't seen any concrete numbers (the dailytech cinebench test as the only exception) and Intel is already letting people run tests on their chips that won't be out for a couple months. I'm starting to think that barcelona will be another HD 2900, yeah it will be quick, but it won't be earth shatteringly fast. I'm guessing it's something that would have been competitive a year ago when conroe launched, I'm thinking it will end up being vanilla a bunch of hype around something we've seen before... except in this case about a year ago.


RE: Barcelona Where is it?
By Martimus on 8/22/2007 3:38:05 PM , Rating: 2
I'm also guessing that it will be about equal to Conroe, at similar speeds. But that should make it competitive with Penryn at similar speeds as well, if these benchmarks are any indication.

Either way, I am not upgrading my computer until next year when I get my taxes back, so I will be able to tell which platform to get by then. I have always liked getting AMD processors in the past, because I could get so much more for the money, but now Intel processors are also very cheap. We'll see which is the best buy in February when I have the cash to spend though.


RE: Barcelona Where is it?
By aguilpa1 on 8/23/2007 4:50:16 PM , Rating: 3
Barcelona has to be more than "competitive" to have a chance. Consider this, a Q6600 priced at $270 will easily overclock to 3.2Ghz and with some good cooling and a little more voltage boost will get it to 3.6Ghz and higher. The barcelona has to be able to "compete" with that to win any footing in the enthusiast section. I know this because my Q6600 is running smoothly at 3.5Ghz with only minor voltage bump. Benchmarks are through the roof and with two 8800GTX easily hits in the mid 17000's for 3Dmark06 scores.


RE: Barcelona Where is it?
By mrdelldude on 8/22/2007 3:29:00 PM , Rating: 1
The benchmarks will also change once BIOS have been tuned for this chip.

Also, Barcelona when released will "only" be around 2GHz. The mobile Penryn's will be launched at similar speeds.

This 2.33 sample is one of the slower versions of the Wolfdales. The dual-core Xeons will be up to 3.33 and the quads are upto 3.1GHz. The reviewer was able to take this early sample to 3.22GHz.

I don't think that Penryn will stomp all over Barcelona once AMD ramps up the speed. But by the time that happens we'll be seeing Dailytech's previews of Nehalem.


RE: Barcelona Where is it?
By MrBungle123 on 8/22/2007 3:37:34 PM , Rating: 2
quote:
Also, Barcelona when released will "only" be around 2GHz. The mobile Penryn's will be launched at similar speeds.


except Barcelona is supposed to be AMD's "high end" server chip, moble Penryn and Barcelona aren't aimed at the same market.


RE: Barcelona Where is it?
By omnicronx on 8/22/2007 4:08:07 PM , Rating: 2
not only that barcelona is not the mainstream desktop variant, its the replacement for the opteron. When the agena core comes out expect max clocks of 2.6-2.8 ghz.


RE: Barcelona Where is it?
By Vanilla Thunder on 8/22/2007 5:07:17 PM , Rating: 4
Thank you. I get so tired of hearing the Barcelona vs C2D comparisons. Granted, the Agena is the desktop variant of the Barcelona, they are not the same chips. I just don't think it's fair to put the 2 up against each other(C2D vs. Barcelona) when they're aimed at such different markets. Also, it's all hypothetical until the benchmarks are released. So let's be patient, and see if AMD can't pull one from their arses. I for one, am hoping they do. It will benefit us, the end user, more than them falling on their faces.

Vanilla


RE: Barcelona Where is it?
By mrdelldude on 8/22/2007 5:18:29 PM , Rating: 2
My intended point was that server chips usually - exception being low voltage versions - are clocked higher than mobile chips. Obviously as they deal with higher loads - database, virtualization, etc - they are higher powered. The systems they are in are also typically are better ventilated than a laptop system - many times those system have quite a few redundant fans.

The launch of Barcelona - AMD's new flagship and halo chip - will be released at similar clock speeds as Intel's mobile chip.


RE: Barcelona Where is it?
By Brockway on 8/22/2007 5:38:21 PM , Rating: 2
Uhh, no. Server chips ship at lower clock speeds than desktop variants. Stability is way more important on a server. The ghz wars are over anyways, its all about performance per watt. Lets see those mobile chips run in a multi socket environment.


RE: Barcelona Where is it?
By mrdelldude on 8/22/2007 5:56:13 PM , Rating: 2
A) I was comparing clock-rates of the soon to be released Intel mobile vs Barcelona. I was not refering to desktop chips.

B) If I go to dell.com for example:

Laptop - fastest is a dual-core, which usually have a faster clock speeds than a quad core, at 2.33GHz for Intel and 2.0GHz for AMD.

Server - Quad Core from Intel 2.66GHz, dual core from Intel 3.0GHz, and dual core from AMD at 3.0 GHz.


RE: Barcelona Where is it?
By TomZ on 8/22/2007 7:17:08 PM , Rating: 2
Small correction, you can get 2.4GHz dual-core in Dell laptops, e.g., Inspiron 1720.


RE: Barcelona Where is it?
By Brockway on 8/22/2007 11:35:56 PM , Rating: 3
Its a pointless comparison. A 3ghz opteron isn't comparable to the 3ghz xeon, and a 2ghz barcelona is certainly not comparable to a 2.4ghz mobile cpu. Equivalent clock speeds do not mean equivalent performance anymore. Why would you even compare a dual core laptop to a quad core server chip?


RE: Barcelona Where is it?
By mrdelldude on 8/23/2007 10:36:47 AM , Rating: 2
The original point was that AMD is taking a long time to get Barcelona to ramp up it's speed. When they are releasing mid 2s and 3GHz versions, which they said they will be doing, it's quite likely that Intel will have released engineering samples of the Nehalem family.

Had AMD given out engineering samples of chips running at the mid 2s for people to test and preview, then the Barcelona would really give Intel a run for the money right away. Quite a few people - posters in forums and professional reviewers - have said that this indicates that AMD is having a hard time getting their chips up to speed.

Intel has been on a roll - continually releasing updates to existing chips, delivering quad-core to the market, kicking butt on bechmarks. What has AMD done lately? Released speed updates to existing chips.

AMD has really gotten itself in spot with Barcelona. Had they given out chips to DailyTech et al which ramped up like this Penryn did, don't you think it would have caused a lot of positive industry buzz? Don't you think that they lost a lot of the positive buzz when they announced Barcelona only would be launched at speeds up to 2.0 GHz? The chip is many months late, and this is the best the can do at this point?

Look at what started this pointless debate, Intel released (officially or unofficially - Intel had a hand in getting it out) an engineering sample to testers. This for a chip that will launch 2 months after Barcelona. For Barcelona, we have all kinds of information about the chip and what it potentially will do. Sure they "demonstrated" a 3GHz system, but it was strickly handsoff for the reviewers - with 6 120 fans. For all we know that could have been only one of a few chips that could ramp up to those speeds without crashing.

Unless I am mistaken, which certainly is likely, I don't know of any hands on real-world testing of Barcelona systems, like this one for a Penryn chip.

This engineering sample again is able to be overclocked - without increasing voltage - to speeds faster than the Barcelona intial reports. The Intel chip may not as efficient at the same clock-speed, but it certainly has shown that it will be faster.

Certainly a clock-for-clock comparison isn't always the best for comparison. But comparing these two families of chips at similar speeds is a lot more relevant than a x86 vs Itanic or T1/2 chip.

To further my point (source: http://blogs.zdnet.com/Ou/?p=567): At a simulated 2.6GHz (at that chip is not a hard release) on a SPECint_rate2006 test, the Barcelona is barely faster than the fastest existing (and soon to be previous generation) Xeon chip.

So once again, if all AMD can do for a while is 2GHz, most of the benchmarks will show that the current generation - not the next ones - Intel chips can outperform Barcelona.


RE: Barcelona Where is it?
By Brockway on 8/26/2007 6:21:17 PM , Rating: 2
You are still comparing the mobile chip to a server chip. Amd's ramping problem with Barcelona was due to having to work in a multi-socket server environment. The Agena desktop line will be shipping at 2.6ghz in Q4 (barring AMD lying of course) which is faster than the mobile chips you keep wanting to compare them to.

Really though, you gotta wonder why AMD didn't release a quad core Amd64. Even if they just glued 2 together like the Core 2 quads, Hyper transport would have given it an advantage over the fsb design on the Intel's. Would probably just run the same as their FX74 models (meaning not as good as the core 2 quads) but at least they would have released some kind of improvement in the last year.


RE: Barcelona Where is it?
By mrdelldude on 8/27/2007 3:04:57 PM , Rating: 2
quote:
You are still comparing the mobile chip to a server chip. Amd's ramping problem with Barcelona was due to having to work in a multi-socket server environment. The Agena desktop line will be shipping at 2.6ghz in Q4 (barring AMD lying of course) which is faster than the mobile chips you keep wanting to compare them to.


In my most recent post, where did I say mobile anywhere in there?

The entire post was about AMD appearantly not being able speed up their Barcelonas. Whereas Intel's engineering samples being able to do so without problem, 2core going from "stock" 2.33GHz up to 3.22GHz.

In my post I even talked about how AMD's 2.6GHz Barcelona is about 1% faster on SPECint_rate2006 vs current generation 2.66GHz Xeon - and the 3.0GHz 5365 is actually faster than the Barcelona.

Once again, the main point is that AMD is not doing alot in the public to help their own cause.


RE: Barcelona Where is it?
By mrdelldude on 8/27/2007 5:07:13 PM , Rating: 2
quote:
Amd's ramping problem with Barcelona was due to having to work in a multi-socket server environment.


I'd be interested in reading about this. I've only seem ambigious statements like this:

http://news.com.com/8301-13579_3-9762375-37.html
"In an interview published in Sunday's Mercury News (the excerpts don't seem to be online yet), Ruiz said "every time we ran into a gotcha (or a technical glitch), it created a six-week-or-so hole in the schedule as we went back and fixed it. We hoped we wouldn't get many of those, but in the Barcelona case, we got more than we thought. By the time we got through fixing them all, we were six months-plus later from where we originally wanted to be."

http://www.firingsquad.com/news/newsarticle.asp?se...
AMD has been struggling of late to catch up with Intel and their plans for a true quad-core server processor, code named Barcelona, will finally be realized when it is scheduled to launch on Sept. 10. However a new Q&A at SiliconValley.com with AMD CEO Hector Ruiz reveals some interesting behind-the-scenes issues with the Barcelona chip:
It's awfully complicated. We're doing something that nobody has ever done. As strong as our competitor is, they have not done a quad core on a chip (four chip cores, or brains, laid out on a single piece of silicon instead of on two pieces attached together). . . . Every time we ran into a gotcha (or technical glitch), it created a six-week or so hole in the schedule as we went back and fixed it. We hoped we wouldn't get many of those, but in the Barcelona case, we got more than we thought. By the time we got through fixing them all, we were six months-plus late from where we originally wanted to be.


RE: Barcelona Where is it?
By deeznuts on 8/22/2007 5:39:47 PM , Rating: 2
quote:
The benchmarks show that Penryn isn't much better than Conroe. (On average ~5%) So if Penryn would roll over Barcelona, so would Conroe. After seeing these benchmarks, I am pretty confident that Barcelona will be competitive to Penryn.


One thing Penryn will have is clockspeed. Barcelona/Agena will be competitive to Penryn I believe. But it needs to be more than competitive for two reasons. First, more people are inclined to buy Intel when performance is quite similar or even slightly in favor of AMD. Second, and this relates to the first, AMD needs to charge a premium for their new chips, they need the money.

AMD has to compete on performance (when the netbursts were getting their ass handed to them by AMD chips, and when X2's were expensive, remember that?) or price. They can't battle on price on their new architecture. They desperately need the money.


Good For Power Users, Good for Competition
By JasonMick (blog) on 8/22/2007 3:01:01 PM , Rating: 1
Looks to be a fairly impressive processor depending on the clock speeds, though a bit pricey.

Really the performance gains are mild, which is to be expected, but relatively nice.

The only thing that scares me is the average guy walking into Best Buy and thinking that this is going to be a radically different computing experience over a Conroe/Kentsfield. Maybe to an overclocker this would be a radically different experience, but not to that guy. But he may end up paying the extra cash because of the advertising by Intel, which is typically very assertive and snazzy, and will invariably dub this release in a rather deceptive manner of a "computing revolution" or something of that nature.

Now I think all processor manufacturers are equally guilty of doing this with die shrinks, so I'm not just picking on Intel. Its the marketing that bothers me. Also I think die shrinks are a good thing for practical reasons like heat, but the desktop markets advertising aura bothers me.

The fact that the dual core mainstream version is not coming out till Q1 2008, has got to be making AMD's day. They can finally hope to promote their Barcelona as a level competitor before Intel one ups them again with a new architecture.

All in all, I have mixed feelings about this release as I am happy that the timing is very good for promoting competition. Both Intel and AMD have grand plans for their next generation architecture, a level playing field should push them to do their best and hopefully realize them. I am also happy to see modest performance gains for the high end consumer.

However, it still bothers me that this processor will invariably be overmarketed to the average consumer as a radical new computing experience. Maybe I am wrong, but I would be shocked if Intel does not do this!




RE: Good For Power Users, Good for Competition
By fk49 on 8/22/2007 3:11:44 PM , Rating: 5
As these chips will still be branded under Core 2 Duo, I doubt most consumers will even realize there's been an upgrade. The sticker on the box may say "Now with even faster performance!" but the average user can't tell you the difference between Conroe and Corsair so I doubt Penryn will make a huge difference.

What will be advertised is the Quad-cores, as Intel is trying to shift people's attention from the old metric of MHz count to the core count. I have some friends that still clamor for Pentium 4's, even over Athlon X2's and Core 2 Duo's just because the clock speed is advertised as so much higher!


RE: Good For Power Users, Good for Competition
By Rike on 8/22/2007 5:21:03 PM , Rating: 3
quote:
I have some friends that still clamor for Pentium 4's, even over Athlon X2's and Core 2 Duo's just because the clock speed is advertised as so much higher!


That's just so sad! :(


By Belard on 8/23/2007 3:55:57 AM , Rating: 2
Not very smart either...

These are the same "gamers" who blow $1000 on those P4-EE chips and boost how great they are - when the $250 AMD 3200 was about as fast... never mind the 3500 or 3800 which were still much cheaper.... of course this was 2 years ago.


RE: Good For Power Users, Good for Competition
By TomZ on 8/22/2007 3:20:41 PM , Rating: 2
quote:
...level playing field should push them to do their best and hopefully realize them

That's an optimistic sentiment, except that it really isn't much of a level playing field. For one thing, you're comparing AMDs Barcelona "vaporware" to Intel's silicon in the field. With AMDs past delays, how do you know they'll hold to their current version of ship dates?

Second, Intel is very profitable and has lots of cash on hand, while AMD is runing losses and is juggling its various debts. That just has to have an impact on next-gen architecture R&D.

Third, there is nothing close to a level playing field in terms of either marketing or manufacturing, where Intel completely dominates. And I don't see that changing any time soon.


By JasonMick (blog) on 8/22/2007 3:26:03 PM , Rating: 2
Obviously Intel is dominating the current generation.

AMD very well may be nothing but vaporware, but if they fall, I sure hope some other manufacturer will step into the void.

I like Intel a lot, and don't care much for AMD's current offerings, but it is rather scary to think of Intel as being the sole manufacturer of desktop PC chips.

I know they currently control the market, but having some competition versus no competition is essentially the difference between a successful corporation of the capitalist economy (very efficient) and the nonexistant competition (at a high cost) of a purely communist/communist dictatorship run economy (very inefficient).


By TomZ on 8/22/2007 3:36:20 PM , Rating: 3
I agree, and I sure hope that AMD can fulfill their potential. If Intel no longer has competition from another big player, that won't be good for anybody.


RE: Good For Power Users, Good for Competition
By MrBungle123 on 8/22/2007 3:33:06 PM , Rating: 2
quote:
The only thing that scares me is the average guy walking into Best Buy and thinking that this is going to be a radically different computing experience over a Conroe/Kentsfield. Maybe to an overclocker this would be a radically different experience, but not to that guy.


why would that scare you? If someone is dumb enough to spend $1000 or more on something they did no research on before making the purchase they either have enough money that it doesn't matter or are too stupid to know the difference.


By omnicronx on 8/22/2007 4:05:55 PM , Rating: 2
Most people just don't know what to look for, its not that they did not do the research, its that they don't understand what they are looking at when they do their research. Whenever i help a friend or a family member out, i have to dub it down to a level they can understand. And i don't blame them a bit, with the move from mhz to more efficient, to more cores, then back to mhz, its no wonder people don't understand.

Most people just dont get that spending 1000$ instead of 500$ is usually not going to be 2x the performance. In the computer world you always pay to be on the bleeding edge, and it the difference in price does not always convert to difference in speed.


By grenableu on 8/22/2007 5:10:14 PM , Rating: 5
quote:
Intel, which is typically very assertive and snazzy, and will invariably dub this release in a rather deceptive manner of a "computing revolution" or something of that nature.
Don't you think you should actually wait till Intel does something wrong before you convict them? I don't think they've even come CLOSE to calling Penryn a "computing revolution".


E
By conrad13a on 8/22/07, Rating: 0
RE: E
By Anonymous Freak on 8/22/2007 3:02:15 PM , Rating: 5
What do you mean "at the cost of the OC"?

This is a first-generation silicon chip, a individual that was binned for 2.33 GHz, that OCed to 3.22 GHz. At stock voltage. This generation of silicon (A0) won't even be released, it's a pure test batch. A1 should overclock better, since they will have more experience manufacturing these, and since we will have 3.33 GHz chips right off the bat, I can't imagine they will overclock WORSE than the existing Conroes.


RE: E
By jeromekwok on 8/22/2007 9:22:04 PM , Rating: 2
I think Intel intentionally gives out the lowest multiplier Penryn for partners or reviewers. It is not Intel can't make a 3GHz+. But Intel does not want AMD to know the headroom of Penryn right now. No matter how hard we overclock, the sample will be limited by the mobo FSB.


Tom's Hardware?
By JonnyDough on 8/22/2007 8:17:46 PM , Rating: 3
Tom's had posted the same exact benchmarks yesterday. I've asked Anandtech if they were in any way associated with Tom's over email, and they said they weren't. Yet here are both companies with the same benchmarks and the same data, both claiming to have a sample chip. Furthermore, they both use DailyTech news. So who can be trusted? Do they feed us whatever they want to keep us buying?




RE: Tom's Hardware?
By johnsonx on 8/22/2007 10:10:14 PM , Rating: 2
What makes you think Tom's uses DT? I've certainly never heard of them using the DT news feed, and I don't find it anywhere on their site. They have a news section, but it doesn't look like DT to me.

There's no connection at all between DT and TH; more of a gentle hate.


"Google fired a shot heard 'round the world, and now a second American company has answered the call to defend the rights of the Chinese people." -- Rep. Christopher H. Smith (R-N.J.)

Related Articles
Intel Sets "Penryn" Launch Date
August 14, 2007, 6:18 PM
"Penryn" Benchmarks Hit The Web
August 7, 2007, 4:04 PM













botimage
Copyright 2014 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki