Intel’s Six Core, 12-Thread, 32nm Core i7 980X; Obsecenly fast in threaded Apps

It was not that long ago that everyone was eagerly awaiting the first dual core CPUs. I can still remember the discussions [and arguments] about this new technology and the ever present law of diminishing returns. But the AMD’s and later Intel’s dual core CPUs came and we had new things to argue about. Things like, yeah but is it any good for gaming? These same arguments played out again when the first quad cores hit the market. At the time Intel was first to bat and for some reason those first Kentsfield CPUs seemed able to bypass the law of diminishing returns and just flat out perform. But as with everything these same CPUs that performed so well reached a saturation point. So Intel had to come up with something new.

This was Nehalem, it brought in many new [to Intel desktop CPUs] features and a couple of old ones that made sense with the new architecture. Again these new CPUs just seemed to fly, they flat out beat everything else on the market even faster clocked CPUs from Intel’s older architecture. Still, no technology company can sit still; it is a world of evolve or die. Well ok, evolve or get passed up by your competition. So the Nehalem has evolved; Intel has reduced the process size to 32nm, thrown in some extra goodies into the feature set, and added two more cores. This when rolled up into a neat package arrives on the market as the Gulftown CPU, known to you and I as the Core i7 980x CPU.  Let’s see how well this 6-core, 12-thread, 32nm monster performs when you drop it into your favorite X58 motherboard.

What’s new?
Well one of the simplest things that are new with the Core i7 980X is Intel’s new CPU cooler. This falls under the name of the DBX-B and is one serious cooler. Although we did not use this cooler for our testing we do plan to take a look at it and see how well it can keep up compared to some of the third party coolers out there.

But on more of the CPU side we have some nice ‘new’ features that will only add to the performance you should expect from a 12 thread capable CPU. As we mentioned above the 980X is being manufactured on a 32nm process. This means that despite the increase in cores and cache the Gulftown CPU is actually smaller than the Core i7 975 and its 45nm process. Where the Core i7 975 Bloomfield measures 263mm2 the Gulftown is a much smaller 248mm2 – all that while having 429 million more transistors on the CPU die [1.17 Billion vs. 731 Million]. This translates into more ‘power’ in a smaller package while using less power from the wall and generating less heat.

But there is, of course, more to it than that [even if not much more]. Intel has placed the new AES-NI [Advanced Encryption Standard – New Instructions] into the CPU. This should give a handy little boost to any encryption/decryption you have going on [like hiding all that porn you have amassed]. The smart cache also gets a boost up from 8MB to 12MB. This is to account for the extra threads flying around inside the 980X.

The rest of the features on Gulftown are what you have come to expect from the new Core i7 CPUs such as Turbo Boost, High-K metal gate [no lead in the CPU], Triple Channel memory support for up to 1066MHz and so on. So while we do get some great new stuff; we are not seeing an amazing revolution in terms of CPU design. But then again as Gulftown is the ‘Tick’ part of Intel’s ‘Tick-Tock’ roadmap it is not really meant as one. That will be the next in line when Sandy Bridge comes around.

Overclocking and 6 cores
So we know that the new Core i7 980X has 6 cores, 12 threads and a 32nm Process. What we do not know yet is if this will affect overclocking one way or the other. We have seen in the past that the fewer cores a CPU has the better you can usually overclock it and maintain stability. If anyone remembers the AMD’s Athlon 64 FX-62 then you probably also remember having issues where one CPU core would be fine at a certain clock speed but the other would fail miserably. This issue has only been complicated as we see more and more cores in the CPU.  On the other hand, usually when a new CPU comes out at a new process size you can get a few extra MHz out of it without too big of a problem. What we wanted to find out was if this was true with the Core i7 980X.

Our testing yielded some puzzling results. On the one hand we were able to hit 4.47GHz [172×26] and boot into Windows. However, almost all of the tests we tried to run would fail. These were not catastrophic failures [like a blue screen of death] but minor application crashes. In the end to get everything working we had to back off to 4.2GHz [168×25] to get everything nice and stable. This is about 300MHz faster than my best table clock on the i7 975 [on air]. The voltage used here was 1.375v. Even with the voltage up that high we did not see any indications of overheating, in fact even under the load of 12, 32M SuperPi runs we never peaked over 69c on any single core. 

As always, our overclocking results may vary from yours. This is due to differences in hardware [including our CPU being an engineering sample] and cooling methods. Our final OC speed is the speed that we can run all of our tests back-to-back without any crashes or application failures. The performance numbers from our overclocking are listed with the stock scores below.

Test System and Comments

Processor: Intel Core i7 908X
Mainboard Asus P6X58D Premium [Supplied by Asus]
Memory: 6GB Kingston KHX12800D3T1K3/6GX [Supplied by Kingston]
Hard Disk: Kingston SSD Now M [Intel X25-M 80GB SSD] [Supplied by Kingston]
Graphics Card: AMD Radeon HD 5870 [Supplied by AMD]
Graphics Card: Zotac GTX 285 AMP! Edition 1GB [flashed to stock BIOS] [Supplied by Zotac]
Cooling: Cooler Master Hyper 212 [with an extra fan] [Supplied by Cooler Master]
Operating System: Microsoft Windows 7 Ultimate x64
Drivers: Intel INF 9.1.0.1007, ForceWare 196.21, Catalyst 10.2

Comments
We are making some changes to the way we test here at BSN*. One item that came as quite a surprise was that my old beloved classic cameras in LightWave were not as efficient or effective as I had thought. After a few issues with render times that were literally all over the place. I started up a conversation with both Intel and Newtek. I will cover the details of this in another article but it is noteworthy as it has changed the entire method for how we test using LightWave and this will be the first time we use it in a full-fledged review. We also have changed our GPU of choice; as such we wanted to ensure that there were no additional GPU related performance aspects. We know that Photoshop CS4 has limited CUDA support but we did not know exactly how much that would impact performance. Because of this we have included Photoshop times for both a GTX 285 and the HD 5870. Other new items are the first time inclusion of a DirectX 11 First person Shooter [Battlefield Bad Company 2] and the use of the new Cinebench R11.5. For the most part though, our testing is going to be identical to what you have become used to.

Synthetics
For our performance testing we like to have a combination of synthetic and real-world testing.  For Synthetics we have the usual bank of tests from FutureMark, Maxon, Sisoft, and HyperPi. These tests cover raw performance and potential; this side of testing brings out the basics of the CPU on its own while our real-world testing covers actual usage. So for the numbers crowd; let’s get started with two of the industry standards for benchmarking system and gaming performance.

FutureMark
PCMark Vantage
PCMark Vantage is a suite of tests designed to give an easily reproducible result. It is also often used for bragging rights. The suite combines some of the most common PC tasks into an easy to use format.

Here we see Gulftown give us some great numbers at stock speeds and even better when we push the CPU but overall the increase is less than a 10% difference compared to the stock Core i7 975. We have to wonder if PCMark Vantage is not able to scale well over 8 threads.

3DMark Vantage CPU Score
3DMark Vantage is the second in the twin tests from FutureMark.  This one turns to gaming. It is one of the first tests to use DX10 for the rendering loops. We also see the first full-scale usage of PhysX in a gaming test.  As we have moved to using the HD 5870 in our testing the scores shown are all without GPU accelerated PhysX.

Here things are looking a little better as we note a 32% increase between the two top Intel CPUs. Considering that these are the RAW CPU numbers with no GPU acceleration it is impressive indeed.

Sisoft Sandra
Next up we have Sisoft’s Sandra up on the board. Here we see the numbers that represent the raw potential of our CPUs.

In this tangle of confusing lines we note on very important thing. The 980X manages to pull off an incredible 245.77 Mega Pixels/second in the multi-media testing at stock speeds.  This is a 25% increase over the next fastest performer the i7 975. In terms of raw CPU power the 980X is 32% faster in our Whetstone test and 45% faster in Dhrystone testing.

Cinebench R10 x64
This test from Maxon [the makers of Cinema 4D) covers CPU based rendering. The test renders a single file using a single CPU and then again using all the available CPUs/ Cores. As this test is no longer supported by Maxon we will be phasing it out and moving to the new Cinebench R11.5. This new test improves threading efficiency [up to 64 threads] and also includes some advanced rendering techniques to keep the CPU working hard. CB R11.5 also includes a revamped OpenGL test for GPU performance.

This will be the last review where we use Cinebench R10.  Still even with the older less threaded engine we see some impressive improvements to our scores. We did see a slight slowdown in single core rendering but once we put all cores and threads in play the 980X simply shined.

For our Cinebench R11.5 testing we see some interesting numbers. In terms of CPU performance the 980X falls in line right behind the big dual 1366 Nehalem rigs. This is exactly where it should be, so while this is not a major ‘win’ it is good performance anyway.

HyperPi 0.99
HyperPi is a great test to find out how well a CPU can handle large amounts of math based information. With HyperPi you have a single instance of SuperPi Mod 1.3 running on each logical and physical core. This puts a great deal of stress on the CPU and also on the internal memory controller.

Our HyperPi results again show us that trying to stuff two instances of SuperPi into a single CPU core is going to bog your system down. Still if you think about it considering the work load it is an impressive performance.

Photoshop CS4
Photoshop is an industry standard image manipulation application. It is great for not only editing photographs but also for creating original content. For out testing here we used the Driverheavn Photoshop script. This is a custom built action that puts a stock image through a series of filters and allows you to record the time each takes to run.

As we mentioned above, there was some concern with us moving from a Cuda accelerated GPU to one that is not. Adobe Photoshop CS4 has limited GPU acceleration and we wanted to make sure there would be no noticeable performance hits from moving to the AMD GPU. As you can see there were not major performance differences when using the Driver Heaven bench.  However, we were surprised to see that the Core i7 980X was only slightly faster overall and was actually slower in many of the individual filters. This could be an indication that Photoshop might not be as threaded as Adobe likes to claim.

Real World
LightWave 3D 9.6 x64
Our first real world test is LightWave 3D. This is an industry standard 3D Animation and rendering software from Newtek.  Our rendering tests with LightWave 9.6 have changed. I have always been a fan of the Classic Camera and the multi-pass PLD anti-aliasing that it offered. However, what I did not know was that this type of camera model was only capable of about 75-80% CPU utilization. This makes it very inefficient [as I am sure you guessed] so we made the shift to the newer perspective camera and are using its much better [100% CPU utilization] threaded engine as well as a newer and more efficient form of anti-aliasing. Along with this new camera model we get ray tracing and some other nice features. We have also leapt from the 1080p resolutions we used before and are now setup with a 35mm 4k resolution of 4096 x 3072; this should give the CPU a nice workout. To show off the vast difference in performance between the two we ran both and show you the render times here. This also shows what you can do when your application is truly written for a multi-threaded CPU.

Perspective Camera Stock Performance

It looks like the perspective camera does have a massive impact on render times and with it more efficient threading you can knock off almost half the time needed for the same quality frame. This is great performance and I am sure will make the 980X of great interest to digital animators and other content creators. We will be coming out with a more complete explanation of the differences and the performance delta between the two camera types in the near future.

AutoGK 2.55
AutoGK is a compilation of transcoding applications wrapped up into a very nice installer and front end application. It is a great ‘one-stop’ for transcoding or even re-encoding files. For our testing we transcode a 2-hour movie at 100% quality and record the time it takes to complete.

This is another time where I want to say Wow and leave it at that.  The performance here comes close to some GPU accelerated transcoding applications. Very impressive indeed.

Gaming
Gaming is a very real-world test. We do not use benchmarking scripts but actually play the games though a pre-planned level and record the frame rates using FRAPS. This allows up to see exactly how the CPU benefits [or hinders] performance. We have moved to a new format and will now be bringing you a game of each of the three common Direct X Levels. This should give a broader idea of CPU performance across multiple gaming APIs.

Modern Warfare 2 DX9
Call of Duty Modern Warfare 2 is a fast paced first person shooter style game. It covers the gamut of modern ‘low intensity’ and covert style combat that is actually going on in today’s world. Yes the plot line is farfetched but some of the actual types of missions are not far from the mark. As it is a console port it is limited to DX9 for its engine. However due to the massive ‘bar fight’ AI it can be a good test for a CPU. Settings are shown below.

The performance here is great. However, the vast majority of it comes from the HD 5870 and not the CPU.

FarCry 2 DX10
FarcCry 2 is a large “sandbox” style game that does not have any real levels. It is all mission based but allows for a great deal of free movement in the environment. You take the role of a mercenary sent to kill “The Jackal” a dangerous gun runner. Unfortunately you are overcome by your malaria and end up serving as an errand boy for a local thug. Settings are show below.

Again great performance. In this case the CPU has a little bit more of a hand in it, but still most comes from the GPU.

Battlefield Bad Company 2 DX11
This title from EA also is our first DX11 First Person Shooter. It is a game that is heavily multi-player oriented, but also has a short single player game. In the beginning of that game you take the part of a World War II commando infiltrating a small island to recover a Japanese defector with a secret weapon which you spend the rest of the game searching for. Settings are show below.

With the extra rendering load from the advanced DX 11 features we are not surprised to see the frame rates we did. But as the AI was not that advanced and there was nothing to indicate that Bad Company 2 was using more than two CPU cores we still have to give credit to the GPU on this one.

Gaming recap
I have said this for a long time now; gaming at high resolutions is GPU limited. Yes you will get a more fluid feel to a game and in many real time strategy games you can see more rendered enemies on the field but for the most part gaming is about the GPU. This might not always be the case though, as we have heard that there are a few games in the works that will be able to handle more than 2 threads. This is great news and something that has been promised to us gamers since the first quad core CPU this the market. But, these games are coming at a time when the two major GPU makers are pushing to get more on the GPU and as more developers are coding for the console first them porting to the PC. Only time will tell if these optimizations will make a global difference in gaming.

Value
Intel has made something of a statement with the new Core i7 980X. They are pushing it right into the same top pricing segment as the i7 975. This means that from Intel, in 1000 part lots you can get the i7 980X for $999. This does not mean that is what it will cost you from e-tailers like Newegg and Frys. The out of pocket cost there will probably be more in the range of $1200-$1400. This healthy little markup is quite annoying and makes me very frustrated with the guys at the end of the supply chain. Speaking with a few they claim that due to the high cost, and limited market scope they need this mark up to recoup the costs of stocking these in quantity. Personally I do not buy that one. Still for what it is worth you are getting a single CPU that is ‘drop-in’ compatible with existing X58 chipsets [with only a BIOS flash] that can run close to and outperform many dual CPU systems. When you think about it like that, it is not too terribly bad. However, if you want to buy one of these you will have a little bit of a wait. The Core i7 980X will not be in the channel for a few more weeks yet. Still, once they do hit I can imagine that they will find a home in some of the top end systems out there.

Conclusion
There is nothing quite like the feeling of having a six core, 12-thread CPU running in your system. I think that I was more impressed with this single CPU that I was with the V8 monster that Intel put out a few years ago. The 32nm process really shows off with not only the fairly cool operation of the CPU but also with its impressive overclocking headroom.  Performance was up and down, but did serve to illustrate something.  The current tests for multi-core/multi-threaded CPUs are not up to scratch. Many of these simply fail to utilize all the available threads on the core i7 980X. However when an application does use them all you can see an almost obscene level of performance over CPUs with less threads to work with. This puts the Core i7 980X and future many-cored CPUs into an interesting position; they are faster and more capable by design, but are limited by some fairly short-sighted software development.

We have talked about this in the past especially with gaming, too many companies are going ‘cheap’ in development and are really failing to take advantage of the hardware that is out. However, all of that is for another article. There is good news though; this is in the professional and Prosumer level. Here companies seem to be working overtime to get software out that can use the full potential of the newer and more massively threaded CPUs out. These are companies that have always had to deal with SMP and SMT so they have the experience to do this. The big lack is in the consumer market; here the dual and quad core CPU is becoming the ‘defacto’ standard even for laptops. Yet, many software developers are still coding for single core.

Back to the CPU at hand, the Intel Core i7 980X is impressive. There is simply no two ways about it. As a workstation CPU it runs rings around many of today’s dual CPU [non hyper threaded] systems and is capable of doing much more at the same price, energy use level, and while fitting into your existing X58 board.  As time and necessity forces software to deal with more than two cores we will only see this CPU’s performance increase. For now, we are happy to have one of these to tinker with and also to award it with our Editor’s Choice for Enthusiast/Prosumer Performance.

Original Author: Sean Kalinich

Webmaster’s note: This news article is part of our Archive, if you are looking for up-to date articles we would recommend a visit to our technology news section on the frontpage. Additionally, we take great pride in our Home Office section, as well as our VPN Reviews, so be sure to check them out as well.