Techgage logo

Intel Core 2 Extreme QX9650 – The 45nm Era Begins

Date: October 29, 2007
Author(s): Rob Williams

Intel will be launching their first 45nm desktop processor in a few weeks, but we have an in-depth look for you here today. Penryn brings improved efficiency, new high-k metal gate transistors, additional cache and something that will make multimedia buffs rejoice: SSE4. Welcome to the 45nm era!



Introduction

The past two years have been exciting for fans of the processor world, with Intel doing their job well to keep the momentum going. Since Conroe’s launch last July, Intel has successfully kept their foot atop AMD by offering products that not only outperform theirs by a significant margin, but are also offered at such low prices that only hardcore AMD fans would ponder purchasing a non-C2D product.

Everyone had hoped that AMD’s Barcelona architecture would bring them back to the competitive table, but not unlike the current-gen ATI vs. NVIDIA battle, there wasn’t a mark to be left. While AMD undoubtedly needs to release a product that prove to people that they haven’t completely lost their grip on things, consumers have been benefiting either way.

Today, it would be a challenge to purchase a processor that didn’t perform exceptionally… at any price range. Quad-Cores can be had for under $300 and both Intel and AMD have superb dual-core processors that retail for under $200. Without question, it is a great time to build a new machine or for being in the mood to upgrade.

True to the tick tock model, Intel today is unveiling their first 45nm processor. For those unaware, the tick tock model is Intel’s term for describing their product cycle, with tick referring to the initial launch of a new microarchitecture and tock denoting it’s successor. In today’s case, tick is Penryn, while tock is Nehalem, which we’ve taken a look at earlier this year and also learned more about at last months IDF.

Also at last month’s IDF, we were told the official launch date for Penryn: November 12. This date still holds true, with the first and only model being offered on that day being the QX9650. Intel hasn’t stated pricing yet, but previous leaks have stated the usual $999 in shipments of 1,000. Once the QX9650 hits your favorite retailer or e-tailer, prices you pay will be greater than this, around $1,300 at a guess.


Top: Q6600, Bottom Left: QX6850, Bottom Right: QX9650

Before we delve into different processor models, we will first see what Intel’s 45nm architecture is bringing to the table. Penryn and it’s derivatives are much more than a simple speed boost, and we will tackle all of that here today.

Big Things In Small Packages

The overall theme of 45nm is efficiency. Thanks to the smaller process, benefits should include better thermal properties, better power efficiency, faster performance at equal frequencies and for the enthusiasts, potentially better overclocking. Add to the fact that 45nm has gone lead-free, the new transistors seem like a win / win.

But just how small is 45nm? Go grab a metre stick and divide it into one billion equal sections. One of those pieces is a nanometre. If that feels like too much work, just consider the fact that you could fit 2,000 45nm transistors across the span of a single human hair. Or for an even more mind-boggling comparison… 30 million on the head of a pin. Dual-Core Penryn chips will include 410 million of these transistors, while Yorkfield will be double this number.

There’s more to these transistors than just being small though. Intel has been talking about High-k Metal Gate transistors all year, and according to co-founder Gordon Moore, this new implementation “marks the biggest change in transistor technology since the introduction of poly-silicon gate MOS transistors in the late 1960s.”

The above diagram shows the differences between the previous generation of transistors and the new High-k offering. Because transistors are constantly shrinking, the risk for current leakage will increase, hence the need for a new insulator. Because a high-k dielectric insulator is more efficient than silicon dioxide in smaller transistors, Intel (and others) felt it absolutely necessary to include them with 45nm. Interestingly however, information has recently leaked that AMD has produced initial 45nm processors using current materials. How efficient they will prove to be will likely remain unknown for some time to come… if they choose to stick along that route with the final product.

Metal gates in the MOS stack have been used in the past, but were since removed due to inefficiencies. However, with 45nm, its become necessary to bring them in again to replace the common poly-silicon in order to increase efficiency with the high-k insulator. Together, they should prove far more efficient than previous transistors with lower current leakage and increased reliability.


Cache, SSE4, Future Processors

Penryn brings more than just smaller transistors and better efficiency, however, with the introduction of additional L2 cache, a beefier FSB and the SSE4 instruction set.

Previous generation Core 2 processors featured 2MB per core, while Penryn ups that to 3MB per core, or 12MB L2 Cache for a Yorkfield chip and 6MB for the dual-core Penryn. With a larger cache, the issue of latency arises, however Intel believes there will be no such issue here. With larger caches, there is more of a chance that each core can access the data it needs quickly. Smart Cache is also introduced to improve the way that split loads are accessed and stored, by predictive means.

Though already seen in recent launches, the initial Penryn line-up will feature a 1333MHz front-side-bus. Because of this, processor multipliers will vary differently than what we are used to. Future Penryn-based Wolfdale dual-cores will feature x.5 multipliers because of this, in order to keep frequencies in order. Because of the 1333MHz FSB, DDR3-1333 would prove the most popular option to throw in a system, however that adoption is still slow due to high memory prices.

2004 saw the launch of SSE3, and now Penryn will see the launch of SSE4. Intel well knows what users are looking for and have picked up on the fact that media-centric lifestyles are the new black, because not one of the new 54 instructions are designed for anything but multimedia-type processes. To the end-user, this means that videos should, and will, encode faster the the encoder utilizes SSE4.

To increase the benefits even further, new to the feature set is the Super Shuffle Engine, which allows Penryn to perform full-width shuffles on units 128 bits wide in a single cycle. This effectively doubles the speed for more byte, word and dword SSE data, which results in reduced latencies and throughput for all SSE instructions that have shuffle-like operations. You can expect popular encoders to utilize the new instructions to their fullest in the months to come.

I’ve touched on all of the basic new features with Penryn, but there are others as well. Most of those have already been covered in the past. If you are interested in recent information regarding Intel’s goals and products beyond this launch, you can refer to our IDF coverage from last month.

Future Processors

Although this is a fresh launch, all of the product packaging and logos will remain untouched. This could be due to the fact that the products share the same name, or the fact that new packaging for a single product launch would not have made much sense. The reason is probably both.

Intel has not disclosed pricing at this time either, but it’s safe to assume it will not be far different than what we’ve seen from XE processors in the past. Current rumors are $999 in quantities of 1,000, but again, that’s just a rumor and we might see something different on Nov. 12 when this is launched alongside its Xeon counterparts.

Model
Clock Speed
FSB
L2-Cache
TDP
Pricing
Release Date
Intel Core 2 Extreme QX9650
3.0GHz
1333MHz
2x 6MB
130W
$999
Nov. 12, 2007
Intel Core 2 Extreme QX9770
3.2GHz
1600MHz
2x 6MB
136W
$1,399
Early 2008
Intel Core 2 Extreme QX9775
3.2GHz
1600MHz
2x 6MB
150W
$1,499
Early 2008
Intel Core 2 Quad Q9300
2.5GHz
1333MHz
2x 3MB
95W
$266
Jan 2008
Intel Core 2 Quad Q9450
2.66GHz
1333MHz
2x 6MB
95W
$316
Jan 2008
Intel Core 2 Quad Q9550
2.86GHz
1333MHz
2x 6MB
95W
$530
Jan 2008
Intel Core 2 Duo E8190
2.66GHz
1333MHz
6MB
65W
$163
Jan 2008
Intel Core 2 Duo E8300
2.83GHz
1333MHz
6MB
65W
$173
Jan 2008
Intel Core 2 Duo E8400
3.0GHz
1333MHz
6MB
65W
$183
Jan 2008
Intel Core 2 Duo E8500
3.16GHz
1333MHz
6MB
65W
$266
Jan 2008

According to roadmap leaks, processor pricing for Penryn is going to be right in line with what we’ve seen recently, with the lower-end offerings beginning out at $163. Early 2008 will also see new XE Quad-Cores with a 1600MHz FSB, with the QX9775 being designed for the Skulltrail platform, we assume.

Another interesting listing is the Q9300, which features 2x 3MB of L2 Cache. It’s likely that there will be a dual-core counter-part that’s currently unknown, since no other dual-core Penryn CPUs feature a 3MB cache. If all of this pricing remains true, then enthusiasts are going to be treated well in the new year.

Without further ado, let’s jump right into our exhaustive testing of the QX9650, beginning with a quick overview of our testing methodology.


Testing Methodology

Regardless of the OS we are running or product being reviewed, there are a few conditions that need to be met to assure accurate, repeatable results.

Here is the machine used for testing, followed by our operating systems configurations.

Operating Systems

For our CPU reviews, we use two different versions of Windows and one version of Linux. Even though Vista has been out for almost a year, we focus on XP because it has a much wider user base, and is preferred for the best performance and compatibility. Vista is used only for our SYSmark 2007 Preview and PCMark Vantage suites.

Game Benchmarking

By now, we are well aware that Intel’s 45nm processors are set to make huge improvements in multi-media work, but what can it do for our games? Since IDF, Intel has been touting both Half-Life 2 and FEAR as two games that see double-digit increases. Today, we will see if these claims are true, while throwing seven other games into the pile to see if the increases are commonplace.

Our selection of games includes some older titles as well as current hot sellers so that we can throw our Yorkfield through a variety of different genres and game engines to see where improvements will arise. Eight of our nine games are played through manually, with FRAPS 2.9.2 acting as the FPS capture tool. Our lone automated benchmark is Lost Planet. Since this is a CPU review, we wanted to choose at least one automated benchmark that was more CPU bound than GPU bound. To our benefit, Lost Planet has a built-in benchmarking option.

All manual playthroughs last between 4 – 6 minutes. Because they are played manually, the average FPS will vary between tests due to changing circumstances in the games. Each setting is played through twice, and the result is the average. We tested each game at both 1280×1024 and 2560×1600 for the sake of seeing if the benefits were greater with either.

Below, you can view all of the games we will be using, as well as the settings used.

Specification
1280×1024
2560×1600
BioShock Vertical Sync Off
All other options On
Vertical Sync Off
All other options On
Call of Juarez High Quality setting
0x Anti-Aliasing
High Quality setting
4x Anti-Aliasing
Call of Duty 2 Anisotropic Filtering
0x Anti-Aliasing
High Dynamic Lights
Soften Smoke Edges: Everything
Corpses: Medium
Anisotropic Filtering
4x Anti-Aliasing
High Dynamic Lights
Soften Smoke Edges: Everything
Corpses: Insane
FEAR All settings default All settings default
Half-Life 2: Episode Two 0x Anti-Aliasing
8x Anisotropic Filtering
All other options High
8x MS Anti-Aliasing
8x Anisotropic Filtering
All other options High
Lost Planet All settings default
0x Anti-Aliasing
All settings default
4x Anti-Aliasing
NFS: Carbon High Detail Maximum Detail
STALKER Full dynamic lighting
Maximum quality

Full dynamic lighting
Maximum quality
Unreal Tournament 3 (Demo) All settings maxed All settings maxed

All other non-game benchmarks will be explained along the way.


System: SYSmark 2007, PCMark Vantage

There is no better way to evaluate a system and its components than to run a suite of real-world benchmarks. To begin our testing, we will use two popular benchmarking suites that emulate real-world scenarios and stress the machine the way it should be… by emulating tasks that people actually perform on a day to day basis.

Both SYSmark and PCMark are hands-free, using scripts to execute all of the real-world scenarios, such as video editing and image manipulation. Each one of these suites output easy-to-understand scores once the tests are completed, giving us a no-nonsense measure of seeing which areas our computer excels in.

SYSmark 2007 Preview

SYSmark, from Bapco, is a comprehensive benchmarking application that emulates real-world scenarios by installing popular applications that many people use every day, such as Microsoft Office, Adobe Photoshop, Sony Vegas and many others.

SYSmark grades the overall performance of your system based off of different criteria, but mostly it will depend on how fast it could complete certain tasks. Once the suite is completed, five scores are delivered, one as an overall average and the others for each of the four categories.

The QX9650 stormed past the QX6850 throughout all of SYSmark’s tests… a good sign.

A six-point gain for two processors of the same clock speed, not bad at all. The extra cache could have been the factor to help it pull in front.

PCMark Vantage

The most recent recruit to our testing suite is PCMark Vantage, an application that proves to be far more than a simple upgrade from a previous version. Vantage is a completely overhauled application, and this was evidenced by the fact that it took more than two full years to produce. Rather than having a PCMark that could complete in 15 minutes, Vantage’s entire run will take around 90 minutes, testing seven primary areas, such as high-definition video, image manipulation, music conversion, et cetera.

Like SYSmark, PCMark delivers simple scores once completed, one for each of the seven main categories and an overall “PCMark Suite” score, which is what most folks will use for comparisons. Because Futuremark’s products are notorious for fluctuating scores, the suite is run twice on each processor to make sure results are accurate.

PCMarks results are somewhat difficult to go by as they can flucatate a lot, but overall, neither the QX6850 nor the QX9650 have a real advantage here.


Multi-Media: DivX 6.7, TMPGEnc 4.4

It’s time to get our hands dirty with manual benchmarks… true real-world tests! Yorkfield brings many interesting things to the table, so we are using a variety of new applications to benchmark in our review today to better test the processor. To start off our battery of multi-media tests, we will be taking a look at two video encoding applications that both benefit from SSE4, DivX and TMPGEnc.

DivX 6.7

The DivX encoder has supported SSE instruction sets to some degree in the past, but beginning with 6.6.0, SSE4 support has been added in the form of “Experimental Full Search”. Although I have been unable to see quality differences between the original encoding method and this new algorithm, I am told that overall quality should prove better with the newer method. The downside is that the newer process takes longer than a standard process, but that’s where the benefits of SSE4 come into play.

When using DivX 6.6.0+, you will notice the Experimental Full Search option available under the Codec tab, which is left at Disabled by default. This, as we found out, is a good thing since it does indeed take longer overall. If you are a media enthusiast who cares a lot about quality and doesn’t mind the extra wait, then this might be the route to take. The end result may vary depending on certain factors, such as original video codec, original video quality and video length.

Using the default options, the differences are small. Our Yorkfield did prove faster overall, but it was just over 4%. We decided to stick with our Experimental Full Search testing here, since we have a Nero Recode test coming up with takes care of a similar “stock” option.

For our testing, we are using a 0.99GB high-quality DivX .AVI of Half-Life 2: Episode Two gameplay. The video is just under 4 minutes in length and is in 720p resolution, which equates to a video bit rate of ~45Mbps, not dissimilar to standard 720p movies. We converted the video two different ways.

First, we encoded the video at the same resolution but a lower quality, so as to achieve a far more acceptable file size (~150MB). The second method is encoding of the same video, but to a 478×256 resolution, similar to what some mobile devices use. This last method is not entirely realistic as it’s unlikely the exported video would work right off the bat on such a device, but the test is to see the benefits of SSE4 in general. Though DivX is the only popular codec to utilize SSE4 at this time, many more should offer support in the new year, making these results relevant.

It’s hard not to be impressed with the results seen here. Although the differences are somewhat small with our Mobile Video conversion, the larger 720p was 41% faster thanks to SSE4 enhancements. We believe the mobile results are less than stellar due to the fact we rendered it in an unconstrained DivX profile. However, now we have a general idea of what to expect from non-official DivX profiles.

That all said, we didn’t quite hit the same performance as we saw at last months IDF, but 40% is nothing short of incredible.

TMPGEnc 4.4 Express

TMPGEnc XPress is a great application for being able to take a variety of video types and convert them to numerous different file formats for use elsewhere, such as DivX, H.264, QuickTime and HDV. Like DivX above, TMPGEnc 4.4 and onward supports SSE4. That’s the good news. The bad news is that this version is currently unavailable to consumers, but will be beginning in the new year. The developers consider it to be “very experimental” at this point in time.

Since we’ve tackled DivX AVI already, for this test we will be taking a 1:16 video which weighs in at 1GB and export it to an MPEG file at 1440×1080 resolution, 29.97fps.

It’s safe to say that 40% faster rendering with SSE4-encoders is a safe average to go by, as both our DivX and TMPGEnc runs delivered the same performance gains. What Core 2 Duo originally did for desktop performance, SSE4 does for multimedia enthusiasts. The addition of these instructions is a huge benefit.


Multi-Media: Nero Recode, Adobe After Effects CS3

Nero Recode

Where video conversion is concerned, one of the applications I’ve grown to enjoy over the years is Nero Recode. Though it’s export options are extremely limited, they offer high image quality and decent file weight. Nero 8 was released earlier this month, but lacks support for SSE4. In a recent meeting, we questioned whether or not we would see SSE4 support in a future update, but we were told that there is no immediate plans to implement it, although the “guys in the lab” are taking a look at it. Nero exhibits confidence that their application is optimized enough as is, and SSE4 is not needed.

For this test, we’ve first ripped our copy of our concert DVD, Killadelphia, by Lamb of God. The original DVD rip weighs in at 7.7GB, but we are using Nero to reconvert it to 4.5GB so that it will fit on a normal-sized DVD to use as a backup.

Absolutely no performance gains were seen in our standard DVD re-encoding, but taking the main movie and converting it to a mobile file proved 8% faster on our QX6950. The extra L2-Cache might have played a role in that instance.

Adobe After Effects CS3

For our last video-specific benchmark, we will be using one of the leading video creation tools available, Adobe After Effects CS3. The test does not benefit from SSE4, however it will take advantage of a multi-core processor in general. Our test will include a workload that applies numerous filters to a variety of file types, ultimately exporting it as an AVI movie.

After Effects saw 6% gains with our new processor. Though not entirely impressive, clock for clock, we are pleased with anything over a 2% increase.


Multi-Media: Adobe Lightroom, 3DS Max 9

Adobe Lightroom 1.2

Years ago, you’d have to fork over many bills in order to get a piece of great technology, but that’s not the case anymore. For a modest fee, you can set yourself up with some absolutely killer hardware. Luckily, one area where that’s definitely the case is with digital cameras. It’s cheaper than ever to own a Digital-SLR, which is the reason why they are growing in popularity so quickly. As a result, RAW photo editing is also becoming more popular, hence the topic of our next benchmark.

Adobe Lightroom is an excellent RAW photo editor/organizer that’s easy to use and looks fantastic. For our test, we take 100 RAW files (Nikon .NEF) which are 10 Megapixel in resolution and then export them as JPEGs in 1000×669 resolution… a result that could be easily passed around online or saved elsewhere on your machine as a low-resolution backup.

The benefits continue to roll in, as Lightroom completed our project 8% faster using the new processor over our QX6850.

3DS Max 9

As an industry-leading 3D graphics application, 3DS Max is one of our more important benchmarks. If there are people who will benefit from faster CPUs with lots of cores, it’s designers of 3D models and environments and animators. Some of these projects are so comprehensive that they can take days to render. At this time, the application does not support SSE4 and will likely not in the future due to irrelevant instructions, but 45nm has been promising us a lot, so should we not expect to see some nice improvements here as well?

For our test, we are taking a dragon model which is included with the application, Dragon_Character_Rig.max and rendering it to 1080p resolution (1920×1080). We’ve used a helicopter model in previous processor reviews, but this Intel-recommended scene proved better as it takes longer to render and is included free with the application.

Even our rendering is being treated to faster performance, with 5.5% and 7% gains, respectively. Small projects like these are not too impressive, but render a comprehensive project and the benefits will be even more appreciated.


Multi-Media: Cinebench 10, POV-Ray 3.7, 7-Zip Archiving

Cinebench R10

Like 3DS Max, Cinema 4D is another popular cross-platform 3D graphics application that’s used by new users and experts alike. Its creators, Maxon, are well aware that their users are interested in huge computers to speed up rendering times, which is one reason why they released Cinebench to the public. Cinebench R10 is based on the Cinema 4D engine the test consists of rendering a high-resolution model of a motorcycle and gives a score at the end. Like most other 3D applications on the market, Cinebench will take advantage of as many cores as you can throw at it.

Intel promised us double-digit increases with Cinebench, and while I am unsure whether they meant double-digit percentages, we did hit 7.7%. Since Cinebench is based off of Cinema 4D, performance increases will be seen there as well.

POV-Ray 3.7

POV-Ray is a ray tracing application which can render scenes with outstanding results. There are three major benefits of this application. First and foremost… it’s free. Second, it has a built-in benchmark that’s capable of stressing all of your processor’s cores and also scales well with frequency. Third, it’s available on a variety of platforms. There’s little doubt it’s an application that deserves inclusion in our processor reviews.

As we can now see, not everything will benefit from our newer processor. POV-Ray is heavily CPU-intensive, but doesn’t seem to derive much benefit from the larger cache.

7-Zip Archiving

If you are a power-user and love free software (such as myself), then you no doubt have heard of 7-Zip. Although the application is similar to other compression applications on the market, such as WinZip and WinRAR, 7-Zip is completely free and lacks nothing that you need to effectively archive your documents.

For our test, we take a 4GB folder that’s complete with pictures, music, documents and other random files and compress it to a .7z file using the Bzip2 algorithm. The reason we choose to use Bzip2 over LZMA is because the former is entirely multi-threaded and proves a much faster result. It’s around 1% less effective than LZMA, but it takes 40% as long to compress anything, in my prior experiences.

Our QX9650 doesn’t show huge increases all over the board, but it is indeed showing some increases in everyday tasks.


Linux: GCC Compiler, ImageMagick

GCC Compiler

When thinking about faster processors or processors with more cores, multi-media projects immediately come to mind as being the prime targets for having the greatest benefit. However, anyone who regularly uses Linux knows that a faster processor can greatly improve application compiling with GCC. Programmers themselves would see the greatest benefit here, but end-users who find themselves compiling large applications often would also reap the rewards.

Even if you don’t use Linux, the results found here can benefit all programmers, as long as you are using a multi-threaded compiler. GCC is completely multi-core friendly, so the results found here should represent the average increase you would see with similar scenarios.

For testing, we used SabayonLinux Professional 1.1 x86. We normally use Fedora, but no version that we tried would function properly, and due to a tight deadline, we chose SL instead. Our problems with Fedora are not due to the processor, but rather either the motherboard chipset or our 8800GTX card.

Our target is a copy of Wine 0.9.47. The distro is based on the 2.6.23 Linux kernel and we are using GCC 4.1.2 as our compiler. For single core testing, “time make” was used while dual and quad core compilations used “time make -j 3” and “time make -j 5”, respectively.

We didn’t expect much of an increase here, and in single-thread mode, there wasn’t. Using all four cores though, however, we saw a 3.6% increase over the dual-core E6750.

ImageMagick

Even though multi-core processors are not new, it’s tricky finding a photo application that handles them properly. Lightroom was one, Photoshop is another. In light of the fact that it’s difficult to write scripts for more popular image manipulation applications, we are going to test the single core benefit of ImageMagick, a popular command line image application for Linux, Windows and Mac OS. It’s most commonly used on web servers, but can be used locally as well.

The benchmark consists of taking the 100 output JPGs from our Adobe Lightroom tests and watermarking them. Then, our script creates 500×335 thumbnails for each of those. This benchmark accurately represents the exact same process that our own review images go through.

We even saw a minor increase with our ImageMagick run, but minor is the key word here. Next up, we will be taking the QX9650 through nine different games, to see if 45nm benefits can be seen there.


Gaming: BioShock, Call of Juarez, Call of Duty 2

In most of our processor reviews, we focus on just a few popular titles to see if improvements can be seen. Because the QX9650 brings numerous benefits to the table, moreso than a just a simple speed bump, we are using a total of nine games in our review.

At this point, you may choose to skim our testing methodology page which lists full system specs and complete settings for each game used. Each game was run using 1280×1024 resolution, in addition to 2560×1600. It’s our goal to see if greater benefits will be seen at either the low-end or high-end scheme of things. Once again, all runs have been completed twice and the average FPS was recorded using FRAPS 2.9.2.

BioShock

The result of the collaboration between 2K Boston and 2K Australia is BioShock, a game with outstanding visuals, creepy monsters, incredible sound and gameplay that can scare you out of your wits. The game takes place in an underwater city called Rapture, built in the mid-1940s, headed by Andrew Ryan. You know someone with two first names has got to be a little messed up, and that proves true quickly.

Built on the UE3 engine, the graphics are top-rate, making this a perfect game for stressing your system regardless of how high-end you believe it is. The level used is the first in the game. Our save begins us right outside the capsule and the gameplay leads us to a large room with a huge balcony (at this point you receive a second radio transmission from a friend).

Call of Juarez

No gaming platform on the market sees that many western-inspired titles, so when Call of Juarez came out and proved to be a solid game, people rejoiced. The game features two main characters, the young Billy who has been accused of murder and also a religious Reverend Ray (perhaps the coolest western name ever), an ex-outlaw who doesn’t find it anti-religious to kill whomever gets in his way.

The level chosen here is Episode 3, which begins Billy out in a field along a path that leads to a farmers ranch. The goal is to progress through to the ranch, killing wolves that chase you down along the way, ultimately making it inside without being spotted in order to take off with the farmers horse in order to jump over a small chasm.

Call of Duty 2

It’s hard to believe that we continue to use a game in our tests that has had two sequels released since. When Call of Duty 2 was released two years ago, though, it offered exceptional gameplay and crisp graphics. It was a total package, and I clearly remember dropping my jaw at various points throughout the game.

Demolition is our preferred level, which takes place immediately after the training mission. You start out in a small room with other soldiers and make the trek through to another building and then yet another building that resides beside a large field of trenches. This is where we end the level, which equates to around 3 – 4 minutes worth of gameplay.

Not one of these games exhibited real increases in performance, but next up are two games that Intel claims will prove faster on a Yorkfield, Half-Life 2 and FEAR.


Gaming: FEAR, HL2: Episode Two, Lost Planet

F.E.A.R.

Like Call of Duty 2, FEAR first hit our PCs in fall of 2005. When it did, it proved to almost everyone just how badly our computers needed upgrading. It was one of the first games to truly benefit from having 2GB of RAM installed, but of course also a massive graphics card. Even today, running a high-resolution FEAR is a visual treat.

The third level is our destination today, which begins us out beside two friends who send me off through various buildings, kicking some ass en route. I am unsure where the final destination is, as I’ve never explored that far, but throughout our five-minute gameplay we encounter four enemies, outdoor and indoor areas and even have a strange horror sequence occur.

Intel’s claims of double-digit increases are spot-on, with a 12FPS increase at 1280×1024 and 13FPS at 2560×1600. This would be more impressive if the overall frame rates were not already in the hundreds, however.

Half-Life 2: Episode Two

Yet another game that needs no introduction, Half-Life 2: Episode Two was a proper sequel to Episode One, although the duration in which people had to wait between the two was a little questionable. Luckily for fans though, Episode Two proved to be more of what we love. It was a win/win. Introduced with this version were achievements as well, which let you know how much of a fan you really are.

We are using the Silo level for our testing, which is a level most people who haven’t even played the game know about, thanks to Valves inclusion of it in their Episode Two trailers over the past year. During our gameplay, we shoot down a total of three striders (their locations are identical with each run, since we are running a saved game file) and a barn is blown to smithereens.

At our max resolution, neither processor had a real benefit, although our dual-core came out on top. Using a lower resolution acheived the double-digit gains Intel expected to see.

Lost Planet

If you are not a fan of Capcom, you might not be a fan of gaming. They’ve produced some true classics over the years, but sadly, Lost Planet is not one of them. You take the role of Wayne Holden who is on the snowy planet E.D.N. III and attempts to recover memory of his past while doing battle with an insane amount of aliens who want him dead.

The game is available with DX10 enhancements, but since all of our game testing is being done on Windows XP, we are overlooking those. Lost Planet is the only game in our testing today that is run using the built-in benchmarking tool. This test drags the user through various parts of a level and reports an average FPS at the end. That’s the number we’ve noted here.

Despite being a CPU intensive benchmark, we saw no advantages between either of our processors. Even our dual-core performed on part with the Quads.


Gaming: NFS: Carbon, STALKER, UT III Demo

Need for Speed: Carbon

Need for Speed has been a leading racing series for years, and for good reason. Each game delivers what gamers want… the ability to drive very fast cars that they are unlikely to afford. Carbon brought a few new things to the series, such as the ability to recruit help, Canyon Races and more decals and car modifications than most people need.

Because EA is to slow to catch on to what gamers want, Carbon does not support wide-screen resolutions on the PC. This… is absolutely ridiculous, but I have hopes that the upcoming Pro Street will prove to me that EA gets a clue. Because of this, we are not using 2560×1600 here, but rather 1280×1024 and 1600×1200. The race we are using is the first available under the Quick Race section, using the Koenigsegg CCX.

S.T.A.L.K.E.R.

Did we need another “post-apocalyptic” FPS game? Of course we did, or at least THQ thought so. STALKER proved to be a good game though overall, and very different from your standard FPS, which is why it’s done well in the marketplace. The game has great visuals, with a realistic world that isn’t out to give feelings of joy. Rather, it’s dark and gritty, appropriate for what you’d expect after a power plant explosion.

Our mission is one that geeks all over can relate to… to deliver a thumb drive to it’s rightful owner. Seems odd given the situation everyone is in, but that’s not what is important! What is important is that we find out which processor handles this quest the best.

Unreal Tournament III Demo

It’s been quite some time since we’ve seen a UT release, but fans are soon to be treated with the third installment, appropriately called Unreal Tournament III. This game as you would expect has superb graphics and surprisingly, it should run well on most current “decent” gaming machines. Anti-Aliasing is disabled in the game, helping to speed up performance all around. Our single 8800GTX card proved more than capable to run the game with maxed out settings at 2560×1600.

Shangri-La is our level of choice, a standard deathmatch affair. Our run consists of going through as much level as we can, while making fifteen kills total.

With these last three games, differences are seen, but they are minor. The only games to have actually exhibited real increases out of the nine we tested were Half-Life and FEAR… both games that offer high frame rates to begin with.


Final Thoughts

Although it seems like 45nm has been a long time coming, in reality, it hasn’t even been that long since 65nm made its own debut. If one thing is certain, this industry doesn’t slow down for anything, even technical hurdles such as the one the High-k dieletric corrects.

We’ve known for a while that Penryn was much more than a process shrink, but some of the performance increases have exceeded our expectations. The performance advantage of Penryn depends upon what the processor will be used for, however. As far as video conversion is concerned, SSE4 is going to prove a godsend for those who deal with these scenarios everyday. No matter how you look at it, a 40% increase clock for clock is incredible.

The problem with SSE4 at this point is the support. DivX does offer consumers a codec with support for the new instructions, but others do not. TMPGEnc, for example, does not believe their 4.4 XPress will be available until early 2008. Intel notes that other SSE4-capable applications will be available in the new year as well, including Sony Vegas, Pinnacle Studio Plus and MainConcept H.264 Encoder.

Given the performance increases exhibited with encoders that utilize SSE4 however, it’s unlikely that developers will ignore this and dilly-dally to add support. As far as non-SSE4 video encoders are concerned, we saw increases there as well, although at closer to 6%. Even still, clock for clock, any increase is welcomed, especially by those who regularly render very large video projects. These increases likely have the extra L2-Cache to thank.

One area where Penryn will not cause stark increases in performance is gaming, however. Seven out of the nine games we benchmarked saw virtually no increase, while the other two, FEAR and Half-Life 2, saw double-digit gains. But as mentioned on those pages, the benefits of these gains are minor, as those games already have an average FPS well into the hundreds. Both of those titles are known for being extremely CPU-intensive, however, so future games that prove the same should also see similar increases in performance.

Of course one of the largest benefits of a smaller process is overall improved efficiency. We were pleased to see that when comparing the QX9650 to the QX6850, both power consumption and temperature were improved. With the QX6850 installed, our system wattage hit 303W at max CPU load, while our QX9650 decreased that to 273W, despite both processors having the same TDP rating of 130W.

Once again, pricing has not been disclosed and at this point it’s difficult to make a guess. Previous XE processors have been released for $999 in quantities of 1,000, but because this is a new microarchitecture and product release, there might be a premium. Right now, no one knows except Intel. Once these hit e-tailers, you could expect the price to be in the range of $1,100 – $1,300.

For those who choose to wait until the rest of the 45nm fleet arrives, you should be able to pick from a variety of different models in January, including non-XE Quad-Cores and Dual-Core models. From what we’ve seen, pricing will prove lower than what we saw with the initial Core 2 Duo launch. If reports are true, a 3.0GHz E8400 with 6MB of Cache can be had for $183 in quanties of 1,000, whereas last summer, a 2.4GHz E6600 with 4MB of Cache was offered for $316.

Should you purchase now or wait until January? It’s all a matter of wants and needs. If you are in need of a new machine now, then you have little choice. If you have patience and can hold out for two more months, you will be well rewarded.

Compared to current offerings, Penryn and Yorkfield processors offer faster performance overall, additional L2-Cache, the SSE4 instruction, better power efficiency and operates with cooler temperatures. The bottom line? Penryn is something worthy of being excited over.

Discuss in our forums!

If you have a comment you wish to make on this review, feel free to head on into our forums! There is no need to register in order to reply to such threads.

Copyright © 2005-2021 Techgage Networks Inc. - All Rights Reserved.