So, how do the latest processors stack up in Windows Vista? Will a sub-$200 CPU suffice for your needs? Have price cuts allowed the Athlon 64 to catch up to the Core 2 Duo in terms of price-performance? What about power consumption and energy efficiency? Can any of these processors stand up under the weight of killer new games like Supreme Commander? Can I possibly squeeze any more questions into one paragraph? Keep reading for answers to all of these questions and more.
The setup for this one is fairly simple. We’re directly comparing processors from Intel at AMD at a range of price points. Intel has had a lock on the overall performance lead since the Core 2 Duo first hit the scene, but AMD has made clear its intention to maintain a competitive price-performance ratio. To do so, AMD will have to meet or beat each of the processors in Intel’s current desktop lineup, which looks like so:
|Model||Clock speed||Cores||L2 cache (total)||Fab process||TDP||Price|
|Core 2 Duo E6300||1.83GHz||2||2MB||65nm||65W||$183|
|Core 2 Duo E6400||2.13GHz||2||2MB||65nm||65W||$224|
|Core 2 Duo E6600||2.4GHz||2||4MB||65nm||65W||$316|
|Core 2 Duo E6700||2.66GHz||2||4MB||65nm||65W||$530|
|Core 2 Extreme X6800||2.93GHz||2||4MB||65nm||75W||$999|
|Core 2 Quad Q6600||2.4GHz||4||8MB||65nm||105W||$851|
|Core 2 Extreme QX6700||2.66GHz||4||8MB||65nm||130W||$999|
Intel does offer lower cost options like the Core 2 Duo E4300 and its Celeron value-oriented processors, but our list includes the meat of the lineup.
AMD, on the other hand, offers a dizzying array of Athlon 64 X2 models, from 3600+ to 6000+, generally in increments of 200 (or is it 200+?). Not only that, but AMD often sells multiple products under the same performance-related model number, just to keep smug members of the general public from becoming overconfident. For instance, the Athlon 64 X2 4400+ comes in a 90nm “Toledo” flavor that runs at 2.2GHz, has 1MB of L2 cache per core, and is intended for Socket 939 motherboards. The X2 4400+ also comes in the form of a 65nm chip code-named “Brisbane” that runs at 2.3GHz, has 512K of L2 per core, and slips into Socket AM2 mobos. Several of these featuresfab process, clock frequency, cache size, and socket/memory typemay vary within the same model number.
With that said, we’ve chosen the following members of the Athlon 64 lineup as the most direct competitors to their Core 2 counterparts. Because we live in the now, all of these are newer-style Socket AM2 processors:
|Model||Clock speed||Cores||L2 cache (total)||Fab process||TDP||Price|
|Athlon 64 X2 4400+||2.3GHz||2||1MB||65nm||65W||$170|
|Athlon 64 X2 5000+||2.6GHz||2||1MB||65nm||65W||$222|
|Athlon 64 X2 5600+||2.8GHz||2||2MB||90nm||89W||$326|
|Athlon 64 X2 6000+||3.0GHz||2||2MB||90nm||125W||$459|
|Athlon 64 FX-70||2.6GHz||4||4MB||90nm||125W x 2||$599|
|Athlon 64 FX-72||2.8GHz||4||4MB||90nm||125W x 2||$799|
|Athlon 64 FX-74||3.0GHz||4||4MB||90nm||125W x 2||$999|
As you can see, AMD has a fairly direct answer for most members of the Core 2 range. Things start to get shaky at the high end, where the Athlon 64’s lower performance takes its toll. The Athlon 64 X2 6000+ sells at a discount versus the Core 2 Duo E6700, and AMD has no answer to the Core 2 Extreme X6800, Intel’s fastest dual-core processor. If you match up the two product lines against one another, the results look something like this:
|Core 2 Duo E6300||$183||Athlon 64 X2 4400+||$170|
|Core 2 Duo E6400||$224||Athlon 64 X2 5000+||$222|
|Core 2 Duo E6600||$316||Athlon 64 X2 5600+||$326|
|Core 2 Duo E6700||$530||Athlon 64 X2 6000+||$459|
|Core 2 Quad Q6600||$851||Athlon 64 FX-72||$799|
|Core 2 Extreme QX6700||$999||Athlon 64 FX-74||$999|
So the comparisons are remarkably direct, by and large.
These things are never entirely simple, though, so we should roll out some caveats. One of the big ones involves those FX-series processors. You’ll need two of them in order to populate a Quad FX motherboard, so they’re priced (and listed above) in pairs. However, there’s currently only one Quad FX motherboard available, and it costs about $350, which throws the value equation out of whack.
The value equation sometimes goes off-kilter the other way when AMD employs guerrilla price-war tactics like selling the Athlon 64 X2 4600+ for $125.99 on Newegg, well below the slower 4400+. AMD has several of these “Crazy Hector” deals going at Newegg right now, and none of them seem to involve the Athlon 64 models we’ve identified as direct competitors to specific Core 2 Duo models. That’s probably an intentional facet of AMD’s strategy. This practice throws a wrench in our nice, neat comparsion, but there’s little we can do other than tell you about it.
One other thing we should tell you about is why we’ve included two versions of the Athlon 64 X2 5000+ in our testing. Regular readers may recall that we’ve already tested the 65nm version of the 5000+ against its 90nm predecessor and found that the 65nm one had lower power consumption. But the 65nm version also has a slower L2 cache, so we’ve tested the 65nm and 90nm chips head to head to see how the slower cache affects performance.
Our testing methods
As ever, we did our best to deliver clean benchmark numbers. Tests were run at least three times, and the results were averaged.
In some cases, getting the results meant simulating a slower chip with a faster one. For instance, our Core 2 Duo E6600 and E6700 processors are actually a Core 2 Extreme X6800 processor clocked down to the appropriate speeds. Their performance should be identical to that of the real thing. Similarly, our Athlon 64 FX-72 results come from an underclocked pair of Athlon 64 FX-74s, our Athlon 64 X2 4400+ is an underclocked X2 5000+ (both 65nm), and our Athlon 64 X2 5600+ is an underclocked Athlon 64 X2 6000+.
Our test systems were configured like so:
|Processor|| Core 2 Duo E6300 1.83GHz
Core 2 Duo E6400 2.13GHz
Core 2 Duo E6600 2.4GHz
Core 2 Duo E6700 2.66GHz
Core 2 Extreme X6800 2.93GHz
Core 2 Quad Q6600 2.4GHz
Core 2 Extreme QX6700 2.66GHz
|Athlon 64 X2 4400+ 2.3GHz (65nm)
Athlon 64 X2 5000+ 2.6GHz (65nm)
Athlon 64 X2 5000+ 2.6GHz (90nm)
Athlon 64 X2 5600+ 2.8GHz (90nm)
Athlon 64 X2 6000+ 3.0GHz (90nm)
|Athlon 64 FX-70 2.6GHz
Athlon 64 FX-72 2.8GHz
Athlon 64 FX-74 3.0GHz
|System bus||1066MHz (266MHz quad-pumped)||1GHz HyperTransport||1GHz HyperTransport|
|Motherboard||Intel D975XBX2||Asus M2N32-SLI Deluxe||Asus L1N64-SLI WS|
|North bridge||975X MCH||nForce 590 SLI SPP||nForce 680a SLI|
|South bridge||ICH7R||nForce 590 SLI MCP||nForce 680a SLI|
|Chipset drivers||INF Update 22.214.171.1240
Intel Matrix Storage Manager 6.21
|ForceWare 15.00||ForceWare 15.00|
|Memory size||2GB (2 DIMMs)||2GB (2 DIMMs)||2GB (4 DIMMs)|
|Memory type||Corsair TWIN2X2048-6400C4
DDR2 SDRAM at 800MHz
DDR2 SDRAM at 800MHz
|Crucial Ballistix PC6400
DDR2 SDRAM at 800MHz
|CAS latency (CL)||4||4||4|
|RAS to CAS delay (tRCD)||4||4||4|
|RAS precharge (tRP)||4||4||4|
|Cycle time (tRAS)||12||12||12|
|Audio||Integrated ICH7R/STAC9274D5 with
Sigmatel 126.96.36.19974 drivers
|Integrated nForce 590 MCP/AD1988B with
Soundmax 188.8.131.5200 drivers
|Integrated nForce 680a SLI/AD1988B with
Soundmax 184.108.40.20600 drivers
|Hard drive||Maxtor DiamondMax 10 250GB SATA 150|
|Graphics||GeForce 7900 GTX 512MB PCIe with ForceWare 100.64 drivers|
|OS||Windows Vista Ultimate x64 Edition|
Our Core 2 Duo E6400 processor came to us courtesy of the fine folks up north at NCIX. Those of you who are in Canada will definitely want to check them out as potential source of PC hardware and related goodies.
Thanks to Corsair for providing us with memory for our testing. Their products and support are far and away superior to generic, no-name memory.
Also, all of our test systems were powered by OCZ GameXStream 700W power supply units. Thanks to OCZ for providing these units for our use in testing.
The test systems’ Windows desktops were set at 1280×1024 in 32-bit color at an 85Hz screen refresh rate. Vertical refresh sync (vsync) was disabled.
We used the following versions of our test applications:
- SiSoft Sandra XI 2007.2.11.17 64-bit
- CPU-Z 1.39
- POV-Ray for Windows 3.7 beta 19a 64-bit
- Cinebench 9.5 64-bit Edition
- Windows Media Encoder 9 x64 Edition
- picCOLOR 4.0 build 598 64-bit
- 3DMark06 1.0.2
- notfred’s Folding benchmark CD 10/31/06 revision
- The Panorama Factory 4.4 x64 Edition
- CASE Lab Euler3d CFD benchmark 2.2
- MyriMatch proteomics benchmark
- Valve Source Engine particle simulation benchmark
- Valve VRAD map build benchmark
- LAME MT 3.97a 64-bit
- 3DMark06 1.10
- The Elder Scrolls IV: Oblivion 1.1
- Rainbow Six: Vegas 1.02
- Supreme Commander 3220
The tests and methods we employ are generally publicly available and reproducible. If you have questions about our methods, hit our forums to talk with us about them.
The Elder Scrolls IV: Oblivion
We tested Oblivion by manually playing through a specific point in the game five times while recording frame rates using the FRAPS utility. Each gameplay sequence lasted 60 seconds. This method has the advantage of simulating real gameplay quite closely, but it comes at the expense of precise repeatability. We believe five sample sessions are sufficient to get reasonably consistent results. In addition to average frame rates, we’ve included the low frame rates, because those tend to reflect the user experience in performance-critical situations. In order to diminish the effect of outliers, we’ve reported the median of the five low frame rates we encountered.
For this test, we set Oblivion‘s graphical quality to “Medium” but with HDR lighting enabled and vsync disabled, at 800×600 resolution. We’ve chosen this relatively low display resolution in order to prevent the graphics card from becoming a bottleneck, so differences between the CPUs can shine through.
Notice the little green plot with four lines above the benchmark results. That’s a snapshot of the CPU utilization indicator in Windows Task Manager, which helps illustrate how much the application takes advantage of up to four CPU cores, when they’re available. I’ve included these Task Manager graphics whenever possible throughout our results. In this case, Oblivion really only takes full advantage of a single CPU core, although Nvidia’s graphics drivers use multithreading to offload some vertex processing chores.
Obviously, any and all of these CPUs will run Oblivion at acceptable performance levels. If you’re keeping score, though, note that Core 2 processors take the top five spots, generally outpacing their like-priced competition and even jumping ahead of their more expensive Athlon 64 rivals.
Interestingly, the dual-socket Athlon 64 FX systems trail their dual-core siblings, probably because there’s a slight performance penalty to be paid for the Quad FX platform’s non-uniform memory architecture. Even though Windows Vista is supposed to be better at handling NUMA-type architectures, these systems have a small yet measurable disadvantage in this application and, as you’ll see, in some of our other tests. Since this game doesn’t make use of four cores, the FX processors can’t make up for it here. The Quad FX platform’s four cores and plentiful memory bandwidth can produce good results in other applications, and we’ll see that, too.
Rainbow Six: Vegas
Rainbow Six: Vegas is based on Unreal Engine 3 and is a port from the Xbox 360. For both of these reasons, it’s one of the first PC games that’s multithreaded, and it ought to provide an illuminating look at CPU gaming performance.
For this test, we set the game to run at 800×600 resolution with high dynamic range lighting disabled. “Hardware skinning” (via the GPU) was disabled, leaving that burden to fall on the CPU. Shadow quality was set to very low, and motion blur was enabled at medium quality. I played through a 90-second sequence of the game’s Terrorist Hunt mode on the “Dante’s” level five times, capturing frame rates with FRAPS, as we did with Oblivion.
Once again, the Core 2 processors lead their Athlon 64 competition, but this is a much tighter contest. Note to self: Don’t expect a game designed for the weakling Xbox 360 CPU to stress any modern PC processor.
We’ve been hearing about this game for a while now, including the fact that it’s multithreaded and thus a good candidate to take advantage of dual-core (or better) CPUs. Now we get to test it out. We ran into some snags when we first tried to test this game with FRAPS. Getting consistent results proved difficult, and the sound didn’t want to work on our Intel D975XBX2 motherboard, whose Vista x64 audio drivers may not yet be up to snuff. I was also developing the first signs of extreme RTS addictiona grave condition indeed. I found myself analyzing unit types and lusting after level-two engineer bots. Fortunately, we were able to overcome these problems by using Supreme Commander‘s very nice built-in benchmark, which plays back a test game and reports detailed performance results afterward. We launched the benchmark by running the game with the “/map perftest /nosound” options. (Normally, we prefer to test games with audio enabled, but we made an exception here.) We tested at 1024×768 resolution with the game’s default quality settings.
Supreme Commander’s built-in benchmark breaks down its results into several major categories: running the game’s simulation, rendering the game’s graphics, and a composite score that’s simply comprised of the other two. The performance test also reports good ol’ frame rates, so we’ve included those, as well.
The most striking thing about these results is, relatively speaking, how little separation there is between the fastest and slowest systems. One of the reasons for that is clear. The Task Manager plot tells us (and the results confirm) that, despite rumors the game would put more than two CPU cores to good use, Supreme Commander doesn’t use four cores very effectively.
Another reason may be that the game’s simulation component is limited by memory performance, at least on the Athlon 64 systems. For instance, the Athlon 64 X2 5600+ turns out to be faster than the 6000+, which seems counterintuitive. However, some speed grades of the A64 don’t run their memory at its full potential clock speed, and that’s the case with the X2 6000+. The 6000+ clocks its DDR2 RAM at 375MHz (or 750MHz effective), while the 5600+ runs its memory at 400MHz (or 800MHz effective). The Quad FX platform’s memory access latency penalty seems to come into play here, as well, since the FX systems occupy the bottom rungs of the simulation results.
That said, the Intel processors clearly dominate, with the Core 2 Duo E6300 outrunning the Athlon 64 X2 6000+.
The AMD chips get their groove back in the rendering portion of the test, where they prove to be very much competitive with the Core 2 processors.
Add up the two scores, though, and the Core 2 CPUs come out on top overall. Again, the total separation in scores between the slowest and fastest processors is relatively small.
We can perhaps put these results into better context by reporting a familiar, real-world result like frames per second. Once again, we have an average frame rate and a median low frame rate. This time, I’ve chosen to sort the results by the median low number, since I think that’s easily the most important performance factor in this game.
Now we can see why the overall scores in the game’s benchmark are so close together. The average frame rates don’t vary much from the fastest to slowest CPUs. The minimum frame rates range from about six FPS at bottom to nine FPS at most. That’s either a whole lot, if you think in terms of percentage gains, or not very much, if you consider that both six and nine FPS are, er, really slow.
Our GeForce 7900 GTX graphics card may be limiting our performance somewhat. I ran this same test on a Core 2 Quad QX6700 with the game’s graphics quality settings turned down, and it improved scores in the render portion of the test by about 1000 points, boosting the minimum frame rate to 13 FPS. Simulation scores, however, were unchanged.
Valve Source engine particle simulation
Next up are a couple of tests we picked up during a visit to Valve Software, the developers of the Half-Life games. They’ve been working to incorporate support for multi-core processors into their Source game engine, and they’ve cooked up a couple of benchmarks to demonstrate the benefits of multithreading.
The first of those tests runs a particle simulation inside of the Source engine. Most games today use particle systems to create effects like smoke, steam, and fire, but the realism and interactivity of those effects is limited by the available computing horsepower. Valve’s particle system distributes the load across multiple CPU cores.
The quad-core systems stake out the top positions here, finally getting to show their stuff. Overall, the Core 2 processors are clearly faster.
Valve VRAD map compilation
This next test processes a map from Half-Life 2 using Valve’s VRAD lighting tool. Valve uses VRAD to precompute lighting that goes into its games. This isn’t a real-time process, and it doesn’t reflect the performance one would experience while playing a game. It does, however, show how multiple CPU cores can speed up game development.
This is a very different test from the last one, but the finishing order is quite similar, with the quad cores and Core 2s doing best. Apparently, the word “core” has magical powers.
3DMark06 combines the results from its graphics and CPU tests in order to reach an overall score. Here’s how the processors did overall and in each of those tests.
3DMark’s graphics tests are very much GPU limited, as is appropriate, but its CPU tests scale well with faster processors. The CPU test contributes to the overall 3DMark score, dictating the separation between our contestants. The Athlon 64s are more competitive here, especially in the lower speed grades, where they’re outright faster than the similarly priced Core 2 Duos.
The Panorama Factory
The Panorama Factory handles an increasingly popular image processing task: joining together multiple images to create a wide-aspect panorama. This task can require lots of memory and can be computationally intensive, so The Panorama Factory comes in a 64-bit version that’s multithreaded. I asked it to join four pictures, each eight megapixels, into a glorious panorama of the interior of Damage Labs. The program’s timer function captures the amount of time needed to perform each stage of the panorama creation process. I’ve also added up the total operation time to give us an overall measure of performance.
Quad-core processors have tremendous potential for image processing. Nearly every operation in this application uses four or more threads, and the quad-core processors take advantageespecially the Intel “Kentsfield” processors. Intel’s quad-core CPUs take the top two spots, but the mid-range Athlon 64s come out ahead of the competing Core 2 Duos.
picCOLOR was created by Dr. Reinert H. G. Müller of the FIBUS Institute. This isn’t Photoshop; picCOLOR’s image analysis capabilities can be used for scientific applications like particle flow analysis. Dr. Müller has supplied us with new revisions of his program for some time now, all the while optimizing picCOLOR for new advances in CPU technology, including MMX, SSE2, and Hyper-Threading. Naturally, he’s ported picCOLOR to 64 bits, so we can test performance with the x86-64 ISA. Eight of the 12 functions in the test are multithreaded, and in this latest revision, five of those eight functions use four threads.
Scores in picCOLOR, by the way, are indexed against a single-processor Pentium III 1 GHz system, so that a score of 4.14 works out to 4.14 times the performance of the reference machine.
Yep, the Core 2 processors wind up on top again, this time thanks in no small part to strong performances in the rotation test that uses floating-point interpolation. The fastest CPU of the bunch, by the way, is over 13 times the speed of a 1GHz Pentium III, despite having a clock frequency of “only” 2.66GHz.
Windows Media Encoder x64 Edition
Windows Media Encoder is one of the few popular video encoding tools that uses four threads to take advantage of quad-core systems, and it comes in a 64-bit version. For this test, I asked Windows Media Encoder to transcode a 153MB 1080-line widescreen video into a 720-line WMV using its built-in DVD/Hardware profile. Because the default “High definition quality audio” codec threw some errors in Windows Vista, I instead used the “Multichannel audio” codec. Both audio codecs have a variable bitrate peak of 192Kbps.
The fastest CPU finishes encoding this clip in less than half the time it takes the slowest one to finish. Broken record here, but look, the Core 2s are faster than the Athlon 64s!
LAME MP3 encoding
LAME MT is a multithreaded version of the LAME MP3 encoder. LAME MT was created as a demonstration of the benefits of multithreading specifically on a Hyper-Threaded CPU like the Pentium 4. Of course, multithreading works even better on multi-core processors. You can download a paper (in Word format) describing the programming effort.
Rather than run multiple parallel threads, LAME MT runs the MP3 encoder’s psycho-acoustic analysis function on a separate thread from the rest of the encoder using simple linear pipelining. That is, the psycho-acoustic analysis happens one frame ahead of everything else, and its results are buffered for later use by the second thread. That means this test won’t really use more than two CPU cores.
We have results for two different 64-bit versions of LAME MT from different compilers, one from Microsoft and one from Intel, doing two different types of encoding, variable bit rate and constant bit rate. We are encoding a massive 10-minute, 6-second 101MB WAV file here, as we have done in many of our previous CPU reviews.
At best, you can expect to shave about 10 seconds per song (and this is a long song) off of LAME encoding jobs by choosing a Core 2 Extreme X6800 over a Core 2 Duo E6300. Going quad-core won’t get you anything here (except perhaps the ability to run two or three encoding tasks simultaneously). The Core 2 Extreme X6800 takes the top spot throughout our LAME tests, but the competitive landscape shifts quite a bit depending on the encode type and compiler used.
Graphics is a classic example of a computing problem that’s easily parallelizable, so it’s no surprise that we can exploit a multi-core processor with a 3D rendering app. Cinebench is the first of those we’ll try, a benchmark based on Maxon’s Cinema 4D rendering engine. It’s multithreaded and comes with a 64-bit executable. This test runs with just a single thread and then with as many threads as CPU cores are available.
The tables turn dramatically, with Athlon 64 FXs occupying the top ranks of the results and the X2s clearly leading their Core 2 counterparts. Will this pattern hold in another rendering app?
We’ve finally caved in and moved to the beta version of POV-Ray 3.7 that includes native multithreading. The latest beta 64-bit executable is still quite a bit slower than the 3.6 release, but it should give us a decent look at comparative performance, regardless.
Yes, the pattern holds in POV-Ray, with the FX processors finishing 1-2-3 and the Athlon 64 X2 4400+ edging out the much pricier Core 2 Duo E6600.
POV-Ray’s officially sanctioned benchmark scene takes advantage of features that our old Chess2 scene doesn’t, and some of those features aren’t multithreaded. That’s why the dual-core processors gain back some ground here, with the Core 2 Extreme X6800 completing the render before the Core 2 Quad Q6600. Even so, the Athlon 64s retain their edge.
Our benchmarks sometimes come from unexpected places, and such is the case with this one. David Tabb is a friend of mine from high school and a long-time TR reader. He recently offered to provide us with an intriguing new benchmark based on an application he’s developed for use in his research work. The application is called MyriMatch, and it’s intended for use in proteomics, or the large-scale study of protein. I’ll stop right here and let him explain what MyriMatch does:
In shotgun proteomics, researchers digest complex mixtures of proteins into peptides, separate them by liquid chromatography, and analyze them by tandem mass spectrometers. This creates data sets containing tens of thousands of spectra that can be identified to peptide sequences drawn from the known genomes for most lab organisms. The first software for this purpose was Sequest, created by John Yates and Jimmy Eng at the University of Washington. Recently, David Tabb and Matthew Chambers at Vanderbilt University developed MyriMatch, an algorithm that can exploit multiple cores and multiple computers for this matching. Source code and binaries of MyriMatch are publicly available.
In this test, 5555 tandem mass spectra from a Thermo LTQ mass spectrometer are identified to peptides generated from the 6714 proteins of S. cerevisiae (baker’s yeast). The data set was provided by Andy Link at Vanderbilt University. The FASTA protein sequence database was provided by the Saccharomyces Genome Database.
MyriMatch uses threading to accelerate the handling of protein sequences. The database (read into memory) is separated into a number of jobs, typically the number of threads multiplied by 10. If four threads are used in the above database, for example, each job consists of 168 protein sequences (1/40th of the database). When a thread finishes handling all proteins in the current job, it accepts another job from the queue. This technique is intended to minimize synchronization overhead between threads and minimize CPU idle time.
The most important news for us is that MyriMatch is a widely multithreaded real-world application that we can use with a relevant data set. MyriMatch also offers control over the number of threads used, so we’ve tested with one to four threads. Also, this is a newer version of the MyriMatch code than we’ve used in the past, with a larger spectral collection, so these results aren’t comparable to those in some of our past articles.
Perhaps the most striking thing about these results is how typical they are of what we’ve seen elsewhere. The Athlon 64 X2 6000+ falls behind not only the Core 2 Duo E6700 but the E6600, as well, and the Core 2 Extreme QX6700 is all alone at the topagain.
STARS Euler3d computational fluid dynamics
Our next benchmark is also a relatively new one for us. Charles O’Neill works in the Computational Aeroservoelasticity Laboratory at Oklahoma State University, and he contacted us recently to suggest we try the computational fluid dynamics (CFD) benchmark based on the STARS Euler3D structural analysis routines developed at CASELab. This benchmark has been available to the public for some time in single-threaded form, but Charles was kind enough to put together a multithreaded version of the benchmark for us with a larger data set. He has also put a web page online with a downloadable version of the multithreaded benchmark, a description, and some results here. (I believe the score you see there at almost 3Hz comes from our eight-core Clovertown test system.)
In this test, the application is basically doing analysis of airflow over an aircraft wing. I will step out of the way and let Charles explain the rest:
The benchmark testcase is the AGARD 445.6 aeroelastic test wing. The wing uses a NACA 65A004 airfoil section and has a panel aspect ratio of 1.65, taper ratio of 0.66, and a quarter-chord sweep angle of 45º. This AGARD wing was tested at the NASA Langley Research Center in the 16-foot Transonic Dynamics Tunnel and is a standard aeroelastic test case used for validation of unsteady, compressible CFD codes.
The CFD grid contains 1.23 million tetrahedral elements and 223 thousand nodes . . . . The benchmark executable advances the Mach 0.50 AGARD flow solution. A benchmark score is reported as a CFD cycle frequency in Hertz.
So the higher the score, the faster the computer. I understand the STARS Euler3D routines are both very floating-point intensive and oftentimes limited by memory bandwidth. Charles has updated the benchmark for us to enable control over the number of threads used. Here’s how our contenders handled the test with different thread counts.
Chalk up another one for the Core 2, especially the quad-core variants. The Quad FX systems stumble, probably due to latency issues related to their NUMA memory subsystems.
Next, we have another relatively new addition to our benchmark suite: a slick little Folding@Home benchmark CD created by notfred, one of the members of Team TR, our excellent Folding team. For the unfamiliar, Folding@Home is a distributed computing project created by folks at Stanford University that investigates how proteins work in the human body, in an attempt to better understand diseases like Parkinson’s, Alzheimer’s, and cystic fibrosis. It’s a great way to use your PC’s spare CPU cycles to help advance medical research. I’d encourage you to visit our distributed computing forum and consider joining our team if you haven’t already joined one.
The Folding@Home project uses a number of highly optimized routines to process different types of work units from Stanford’s research projects. The Gromacs core, for instance, uses SSE on Intel processors, 3DNow! on AMD processors, and Altivec on PowerPCs. Overall, Folding@Home should be a great example of real-world scientific computing.
notfred’s Folding Benchmark CD tests the most common work unit types and estimates performance in terms of the points per day that a CPU could earn for a Folding team member. The CD itself is a bootable ISO. The CD boots into Linux, detects the system’s processors and Ethernet adapters, picks up an IP address, and downloads the latest versions of the Folding execution cores from Stanford. It then processes a sample work unit of each type.
On a system with two CPU cores, for instance, the CD spins off a Tinker WU on core 1 and an Amber WU on core 2. When either of those WUs are finished, the benchmark moves on to additional WU types, always keeping both cores occupied with some sort of calculation. Should the benchmark run out of new WUs to test, it simply processes another WU in order to prevent any of the cores from going idle as the others finish. Once all four of the WU types have been tested, the benchmark averages the points per day among them. That points-per-day average is then multiplied by the number of cores on the CPU in order to estimate the total number of points per day that CPU might achieve.
This may be a somewhat quirky method of estimating overall performance, but my sense is that it generally ought to work. We’ve discussed some potential reservations about how it works here, for those who are interested. I have included results for each of the individual WU types below, so you can see how the different CPUs perform on each.
This one’s clear-cut. The Athlon 64 processors dominate the Tinker and Amber work unit types, and the Core 2 processors thrive with the two types of Gromacs WUs. We’ve seen this pattern before, but having so many different CPU speed grades shows how very pronounced it can be. For example, the Athlon 64 X2 4400+ comes out ahead of the Core 2 Extreme X6800 in the Tinker WU test, amazingly enough.
Once we average the different WU types and account for multiple processor cores, though, things snap into perspective. The quad-core systems take a pronounced lead, and the slower dual-core models have under half of their point-production potential.
SiSoft Sandra Mandelbrot
Next up is SiSoft’s Sandra system diagnosis program, which includes a number of different benchmarks. The one of interest to us is the “multimedia” benchmark, intended to show off the benefits of “multimedia” extensions like MMX, SSE, and SSE2. According to SiSoft’s FAQ, the benchmark actually does a fractal computation:
This benchmark generates a picture (640×480) of the well-known Mandelbrot fractal, using 255 iterations for each data pixel, in 32 colours. It is a real-life benchmark rather than a synthetic benchmark, designed to show the improvements MMX/Enhanced, 3DNow!/Enhanced, SSE(2) bring to such an algorithm.
The benchmark is multi-threaded for up to 64 CPUs maximum on SMP systems. This works by interlacing, i.e. each thread computes the next column not being worked on by other threads. Sandra creates as many threads as there are CPUs in the system and assignes [sic] each thread to a different CPU.
We’re using the 64-bit version of Sandra. The “Integer x16” version of this test uses integer numbers to simulate floating-point math. The floating-point version of the benchmark takes advantage of SSE2 to process up to eight Mandelbrot iterations in parallel.
This test underscores the vector math potential of the Core microarchitecture, which can execute a 128-bit SSE2 operation in a single cycle. The Athlon 64 is completely overmatched here, but as I’ve said before, the tables may turn when AMD’s new microarchitecture arrives later this year with similar vector math capabilities.
Power consumption and efficiency
We’re trying something a little different with power consumption. Our Extech 380803 power meter has the ability to log data, so we can capture power use over a span of time. The meter reads power use at the wall socket, so it incorporates power use from the entire systemthe CPU, motherboard, memory, video card, hard drives, and anything else plugged into the power supply unit. (We plugged the computer monitor and speakers into a separate outlet, though.) We measured how each of our test systems used power during a roughly one-minute period, during which time we executed Cinebench’s multithreaded rendering test. All of the systems had their power management features (such as SpeedStep and Cool’n’Quiet) enabled during these tests.
You’ll notice that I’ve not included some members of the Athlon 64 family here. That’s because our “simulated” Athlon 64 CPUs are underclocked versions of faster processors, and we’ve not been able to get Cool’n’Quiet power-saving tech to work when CPU multiplier control is in use. I expect to have in-the-flesh examples of the Athlon 64 X2 4400+ and 5600+ soon, and I’ll provide an update once I’ve tested their power consumption.
I have included our simulated Core 2 Duo E6600 and E6700, because SpeedStep works fine on the D975XBX2 motherboard alongside underclocking. The simulated processors’ voltage may not be exactly the same as what you’d find on many retail E6600s and E6700s. However, voltage and power use can vary from one chip to the next, since Intel sets voltage individually on each chip at the factory.
The differences between the CPUs are immediately obvious by looking at these plots of the raw data. We can slice up the data in various ways in order to better understand them, though. We’ll start with a look at idle power, taken from the trailing edge of our test period, after all CPUs have completed the render.
Among dual-core setups, the Core 2s have a slight advantage in idle power consumption, but not more than 10 watts or so. There’s a much larger gap between the quad-core rigs, since the Quad FX platform is a dual-socket solution with dual core-logic chipsets. Meanwhile, the Intel quad-core systems draw only about 10-15 watts more than their dual-core brethren.
Next, we can look at peak power draw by taking an average from the five-second span from 10 to 15 seconds into our test period, during which the processors were rendering.
The peak power consumption numbers tell us very good things about the Core 2 processors. The system based on Intel’s power-hungriest dual-core processor, the Core 2 Extreme X6800, draws 15W less under load than the one based on the lowest power AMD processor here, the 65nm version of Athlon 64 X2 5000+. AMD’s 65nm chip does shave off some power drawroughly 12Wversus its 90nm variant. The Athlon 64 FX-74, meanwhile, is just plain silly.
Another way to gauge power efficiency is to look at total energy use over our time span. This method takes into account power use both during the render and during the idle time. We can express the result in terms of watt-seconds, also known as joules.
The Core 2-based systems led in terms of both peak and idle power draw, so it’s unsurprising they do so well over the total duration of our test period.
However, we can quantify efficiency even better by considering the amount of energy used to render the scene. Since the different systems completed the render at different speeds, we’ve isolated the render period for each system. We’ve chosen to identify the end of the render as the point where power use begins to drop from its steady peak. There seems to be some disk paging going on after that, but we don’t want to include that more variable activity in our render period.
We’ve computed the amount of energy used by each system to render the scene. This method should account for both power use and, to some degree, performance, because shorter render times may lead to less energy consumption.
These results tell us several things. First, multi-core processors can be very energy efficient when handling parallel tasks like rendering. The quad-core Q6600 and QX6700 run away with the top two positions, and the FX-70 turns out to be the most efficient Athlon 64, even with its considerable built-in platform power handicap. Second, Intel’s Core 2 processors are much more efficient overall than the current Athlon 64s. It’s really no contest. And finally, the Athlon 64 does gain some efficiency when moving to AMD’s 65nm process, but not yet enough to put it back into contention against the Core 2.
The fact that Intel retains the overall performance crown comes as no surprise. As we said at the outset, AMD has no real answer to the Core 2 Extreme X6800 among its dual-core processors. Also, Intel’s quad-core CPUs tend to scale better than AMD’s Quad FX platform, especially for typical desktop-class applications. Our move to Windows Vista x64 has done little to alter this dynamic. At the same time, Core 2 processors tend to draw less power and to be more energy efficientsometimes markedly sothan Athlon 64s. Right now, Intel has the magic combination of a superior processor microarchitecture and a more mature, fully realized 65nm manufacturing capability working together on its side.
This one-two punch has allowed Intel to maintain a performance edge at most price points, despite standing pat through AMD’s aggressive pricing moves and new model introductions. AMD’s current weaknesses manifest themselves most fully in its high-end models, like the Athlon 64 X2 6000+, which draws more power at peak than the Core 2 Extreme QX6700 yet is often outperformed by the less expensive Core 2 Duo E6600. The Athlon 64 looks more competitive in its lower-end incarnations like the X2 5000+ and 4400+, which match up better on both performance and power characteristics against the Core 2 Duo E6300 and E6400. These processors have the benefit of being available in 65nm form, and I’d say the minor performance penalty one pays in performance at 65nm (due to the slower L2 cache) is worth it for the reduced power draw.
This low-to-mid-range territory, incidentally, is where I’d be looking to buy. Many of our tests have shown the benefits of quad-core processors, but honestly, finding applications that will make good use of four cores is not easyand the list of games that really use four cores is approximately zero. I’d probably grab a Core 2 Duo E6400 and overclock it until it started to glow, if I were putting together a system right now. I must admit, though, that I have an almost irrational fondness for the Core 2 Quad Q6600, probably because it’s the most energy efficient processor in our Cinebench power test. The thing is by no means a great dealtwo E6600s will set you back over $200 less than a single Q6600but it’s easy to imagine a near-silent multitasking monster built around one.
AMD would do well to expand its 65nm offerings into higher clock frequencies as soon as it can reasonably do so. That may take a while yet, given the limited overclocking headroom we’ve seen from early 65nm Athlon 64 X2s. Meanwhile, Intel isn’t likely to sit still for much longer. Rumors of an April price cut abound, and in light of the Core 2’s ample frequency headroom, higher speed grades are a definite possibility, as well. For AMD, its next-generation microarchitecture can’t come a moment too soon.