Evaluating the latest processors based on a price-performance analysis is turning into a yearly tradition here at TR. Our first stab at the concept dates back to June 2007, and we tried again with a new batch of CPUs last May. Setting the stage for these articles is a bit like waiting for the stars to align, because a number of criteria must be met: we need performance numbers from a broad enough cross-section of current processors, we need to skirt new processor launches, and we need to wait for prices to be reasonably stable.
The stars have now fallen into place again (or so it would seem), and we’ve therefore taken another gander at CPU price-performance relationships using fresh numbers from our Socket AM3 Phenom II review. Once again, we threw our test results and official pricing information into a big spreadsheet, laid out the data into a veritable smorgasbord of graphs, and compiled everything neatly here for your reading pleasure.
Before inviting you to bask in the glow of our many charts and scatter plots, we should clarify this article’s purpose. This isn’t an exhaustive value assessment of all current desktop processors, nor should it be your one-stop guide to picking a new CPU. Rather, this is an attempt to determine how our collection of test processorsalmost exclusively enthusiast items priced above $100compare when we study both pricing and performance simultaneously.
Is a Core i7-920 worth the extra cost over a Phenom II X4 940 for video encoding buffs? Are dual-core CPUs like the Core 2 Duo E8400 still compelling choices compared to low-end triple- and quad-core offerings? Those are some questions this article should help answer.
The test subjects
With that in mind, let’s take a look at which CPUs we’ll be comparing today. Here’s the list on the Intel side of the playground:
|Model||Clock speed||Cores/threads||L2 cache/L3 cache||Fab process||TDP||Price|
|Core 2 Quad Q9550||2.83GHz||4/4||12MB||45nm||95W||$266|
|Core 2 Quad Q9400||2.66GHz||4/4||6MB||45nm||95W||$213|
|Core 2 Quad Q9300||2.5GHz||4/4||6MB||45nm||95W||$266|
|Core 2 Quad Q6600||2.4GHz||4/4||8MB||65nm||95W||$183|
|Core 2 Quad Q8200||2.33GHz||4/4||4MB||45nm||95W||$163|
|Core 2 Duo E8600||3.33GHz||2/2||6MB||45nm||65W||$266|
|Core 2 Duo E8400||3GHz||2/2||6MB||45nm||65W||$163|
…and on the AMD side:
|Model||Clock speed||Cores/threads||L2 cache/L3 cache||Fab process||TDP||Price|
|Phenom II X4 940||3GHz||4/4||2MB/6MB||45nm||125W||$225|
|Phenom II X4 920||2.8GHz||4/4||2MB/6MB||45nm||125W||$195|
|Phenom II X4 810||2.6GHz||4/4||2MB/4MB||45nm||95W||$175|
|Phenom X4 9950||2.6GHz||4/4||2MB/2MB||65nm||140W||$173|
|Phenom II X3 720||2.8GHz||3/3||1.5MB/6MB||45nm||95W||$145|
|Phenom X3 8750||2.4GHz||3/3||1.5MB/2MB||65nm||95W||$122|
|Athlon X2 6400+||3.2GHz||2/2||2MB||90nm||125W||~$90|
Since retail and e-tail prices oscillate a little too much for our liking, we took our prices straight out of Intel’s official list and AMD’s processor pricing page. These are figures for bulk orders, but they should only be within a few dollars of retail prices. The only exception is the Athlon X2 6400+, which doesn’t appear on the AMD pagewe got that CPU’s $90 price tag from Newegg.
Why include a discontinued CPU to begin with? Though it’s growing long in the tooth, the Athlon X2 6400+ should be roughly representative of the type of performance you can expect from some of today’s faster sub-$100 dual-core processors. It should thus serve as a useful baseline against which to compare newer and dearer offerings.
Low-end processors aside, you may see some other missing links in the lists above. There, too, time constraints forced us to make some compromises and exclude CPUs like the Core 2 Duo E7400 or Core 2 Quad Q8300. Thanks to our relatively broad cross-section of data, however, figuring out where those chips would be situated shouldn’t be too hard. (For instance, the Core 2 Quad Q8300 should perform somewhere between the Q8200 and Q9300 with a price tag of around $185.)
Finally, sharp-eyed readers might notice we didn’t factor platform or power costs into in our processor prices. That’s partially true. We’d rather keep things simple for now, but we’ll have a look at platform and power costs a little later.
Laying out our data
So, how does one represent value in graph form? We’ve re-enlisted our two trusty friends from previous value articles: performance-per-dollar bar charts and performance-versus-price scatter plots. The former should be self-explanatorythink “score points per dollar” or “frames per second per dollar.” In cases where performance is measured as a time in seconds (and the shortest time is best), we’ll use “rate” as our metric. We’ll usually define rate in kilohertz or megahertz, which we work out with a formula like “1/seconds × 1000” or “1/seconds × 1000000.”
You can use the perf-per-dollar charts to get a more precise look at which CPUs offer more bang for your buck, but be careful: getting the most thingamajigs per dollar isn’t the whole story. You’ve gotta look at the whole picture. Processor prices don’t rise linearly with performance, so faster offerings will almost always seem like poorer deals than low-end dual-core chips. That doesn’t mean extra performance (and the resulting time saved) isn’t worth it.
Our scatter plots look like so, mapping performance to the Y axis and price to the X axis:
As you’d expect, the best possible processor would sit at the top left of the plot, offering the highest performance at no cost. Conversely, the poorest choice would be at the bottom right.
In a nutshell, our scatter plots provide a visual representation of the value curve, which should help locate the most interesting combinations of pricing and performance. The performance-per-dollar bar charts come as complements to these plots, laying out the same data (more or less) in a purely numerical fashion.
In our view, the best deals often lie where either performance stops rising substantially while prices keep rising or where prices suddenly shoot up without performance following suit. You’ll see what we mean once we get into our comparisons.
Although we used the results from our Socket AM3 Phenom II review, we did a little hedge trimming and left out several configurations. The Core 2 Extreme QX9775 “Skulltrail” config and Core 2 Extreme QX9770 were the first to go, since the QX9770 is discontinued now, and the QX9775 setup’s high asking price would make our scatter plots harder to read. Besides, we really wouldn’t recommend blowing three large on a pair of Core 2 processors today, since the Core i7 is available.
We also removed our DDR3-powered Phenom II X4 810 setup for consistency’s sake. That shouldn’t be particularly consequential, because the X4 810 performed almost identically in our real-world tests regardless of the memory type we used. If you’d like to see the DDR3 performance results anyway, feel free to check out our review.
As for our remaining setups, we carried over several little testing peculiarities: our Core 2 Quad Q8200 is actually an underclocked Q8300, the Phenom II X4 920 is an underclocked X4 940, and the Core 2 Quad Q9550 is an underclocked Core 2 Extreme QX9650. We expect the performance of these “simulated” CPUs to be identical to the real things, but we kept some of them out of our power consumption testing because we do anticipate power use would vary slightly from the actual products.
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.
Our test systems were configured like so:
|Processor||Core 2 Quad Q6600
|Core 2 Duo E8400
Core 2 Duo E8600
Core 2 Quad Q8200 2.33 GHz
Core 2 Quad Q9300 2.5 GHz
Core 2 Quad Q9400 2.66 GHz
Core 2 Quad Q9550 2.83 GHz
|Core i7-940 2.66 GHz
Core i7-940 2.93 GHz
Extreme 3.2 GHz
|Athlon 64 X2 6400+
|Phenom X3 8750
| Phenom II X4 920
Phenom II X4 940
| Phenom X4 9950
Black 2.6 GHz
|Phenom II X3 720
Phenom II X4 810
|System bus||1066 MT/s
|QPI 4.8 GT/s
|QPI 6.4 GT/s
|HT 2.0 GT/s
|HT 3.6 GT/s (1.8 GHz)||HT 3.6 GT/s (1.8 GHz)|
|HT 4.0 GT/s (2.0 GHz)||HT 4.0 GT/s (2.0 GHz)|
|Motherboard||Asus P5E3 Premium||Asus P5E3 Premium||Intel DX58SO||Intel DX58SO||Asus M3A79-T Deluxe||Asus M3A79-T Deluxe||MSI DKA790GX Platinum|
|North bridge||X48 Express MCH||X48 Express MCH||X58 IOH||X58 IOH||790FX||790FX||790GX|
|Chipset drivers||INF Update 22.214.171.1248
Matrix Storage Manager 126.96.36.1992
|INF Update 188.8.131.528
Matrix Storage Manager 184.108.40.2062
|INF update 220.127.116.117
Matrix Storage Manager 18.104.22.1682
|INF update 22.214.171.1247
Matrix Storage Manager 126.96.36.1992
|AHCI controller 3.1.1540.61||AHCI controller 3.1.1540.61||AHCI controller 3.1.1540.61|
|Memory size||4GB (2 DIMMs)||4GB (2 DIMMs)||6GB (3 DIMMs)||6GB (3 DIMMs)||4GB (2 DIMMs)||4GB (2 DIMMs)||4GB (2 DIMMs)|
|Memory type||Corsair TW3X4G1800C8DF
|Memory speed (Effective)||1066 MHz||1333 MHz||1066 MHz||1600 MHz||800 MHz||1066 MHz||1066 MHz|
|CAS latency (CL)||7||8||7||8||4||5||5|
|RAS to CAS delay (tRCD)||7||8||7||8||4||5||5|
|RAS precharge (tRP)||7||8||7||8||4||5||5|
|Cycle time (tRAS)||20||20||20||24||12||15||15|
with SoundMAX 188.8.131.5280 drivers
with SoundMAX 184.108.40.20680 drivers
with Realtek 220.127.116.1104 drivers
with Realtek 18.104.22.16804 drivers
with SoundMAX 22.214.171.12480 drivers
with SoundMAX 126.96.36.19980 drivers
with Realtek 188.8.131.5204 drivers
|Hard drive||WD Caviar SE16 320GB SATA|
|Graphics||Radeon HD 4870 512MB PCIe with Catalyst 8.55.4-081009a-070794E-ATI drivers|
|OS||Windows Vista Ultimate x64 Edition|
|OS updates||Service Pack 1, DirectX redist update August 2008|
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.
Our single-socket test systems were powered by OCZ GameXStream 700W power supply units. The dual-socket system was powered by a PC Power & Cooling Turbo-Cool 1KW-SR power supply. Thanks to OCZ for providing these units for our use in testing.
Also, the folks at NCIXUS.com hooked us up with a nice deal on the WD Caviar SE16 drives used in our test rigs. NCIX now sells to U.S. customers, so check them out.
The test systems’ Windows desktops were set at 1600×1200 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:
- WorldBench 6 beta 2
- Half-Life 2: Episode Two
- Crysis Warhead
- Far Cry 2
- Unreal Tournament 3 1.3
- Valve VRAD map build benchmark
- Valve Source Engine particle simulation benchmark
- Cinebench R10 64-bit Edition
- POV-Ray for Windows 3.7 beta 29 64-bit
- notfred’s Folding benchmark CD 9/28/08 revision
- The Panorama Factory 5.2 x64 Edition
- Windows Media Encoder 9 x64 Edition
- x264 HD benchmark 2.0 with x264 version 0.59.819
- LAME MT 3.97a 64-bit
The tests and methods we employ are usually publicly available and reproducible. If you have questions about our methods, hit our forums to talk with us about them.
We measured Warhead performance using the FRAPS frame-rate recording tool and playing over the same 60-second section of the game five times on each processor. 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.
We tested at relatively modest graphics settings, 1024×768 resolution with the game’s “Mainstream” quality settings, because we didn’t want our graphics card to be the performance-limiting factor. This is, after all, a CPU test.
A cursory look at the scatter plot and perf-per-dollar chart tells us the Core 2 Duo E8400 is among the better-placed solutions here. It performs almost as well as the Core i7-940 for considerably less dough, and getting much higher performance involves shelling out over $100 more. That only gets you a handful of extra frames per second, too.
With that said, we should take a step back and qualify these observations. As we’ve just noted, we tested in a non-GPU-limited scenario in order to highlight CPU performance. Should you run the game at a higher resolution or higher visual quality settings, the graphics processor(s) could easily become the primary bottleneck, somewhat mitigating the benefits of a faster CPU.
Then again, our results give a very clear idea of what you can buy today to avoid running into CPU bottlenecks, and you can see that the slower processors in the bunch do tend to struggle with this game. Pay special attention to the median low frame rates we reported when thinking about avoiding CPU-based slowdowns. Although we haven’t graphed those results on a per-dollar basis, the CPUs whose frame rates bottom out in the teens and low twenties are questionable performers in Warhead, which is one of the most processor-intensive PC games today.
Far Cry 2
We decided to test Far Cry 2 by recording frame rates during the jeep ride sequence at the very beginning of the game. We found that frame rates during this sequence were generally similar to those when running around elsewhere in the game, and after all, playing Far Cry 2 involves quite a bit of driving around. Since this sequence was repeatable, we just captured results from three 90-second sessions.
Again, we didn’t want the graphics card to be our primary performance constraint, so although we tested at fairly high visual quality levels, we used a relatively low 1024×768 display resolution and DirectX 9.
The Core 2 Duo E8400 does well here, too, but AMD’s Phenom II X3 720 looks even better positioned.
Of course, even at this relatively low resolution, CPU performance doesn’t seem to impact average frame rates quite as much as in Crysis Warhead. Even the Core i7-940 only leads the X3 720 by 2.8 FPS overall, although it does generate significantly higher “median low” frame rates.
UT3 normally isn’t quite as CPU-limited as the titles on the previous page, so we concocted an interesting scenario by setting up a 24-player CTF bot match on the epic Facing Worlds map. We racked up frags like mad while capturing five 60-second gameplay sessions for each processor. The screen resolution was set to 1280×1024 for testing, with UT3’s default quality options and “framerate smoothing” disabled.
The Phenom II X 720 comes out on top yet again, this time beating our cheapest processor in the perf-per-dollar charta rare occurrenceand lying awfully close to the top left of our scatter plot.
Admittedly, you’re probably not going to notice the difference between 76 and 133 FPS on a typical LCD monitor. However, it’s worth pointing out that the “median low” frame rates these processors generate aren’t all above the 60 FPS limitso you may notice choppiness in busier fights with a slower processor. Regular UT3 games without a gazillion bots will probably be less demanding, however.
Half Life 2: Episode Two
Our next test is a good, old custom-recorded in-game timedemo, precisely repeatable.
Here, we have a virtual toss-up between the Core 2 Duo E8400 and Phenom II X3 720. Considering the E8600 is almost neck-and-neck with the Core i7-965, we’ll have to conclude Half-Life 2: Episode Two is more sensitive to high clock speeds and single-threaded performance than to parallel number-crunching power.
Source engine particle simulation
Next up is a test we picked up during a visit to Valve Software, the developers of the Half-Life games. They had been working to incorporate support for multi-core processors into their Source game engine, and they cooked up some benchmarks to demonstrate the benefits of multithreading.
This test 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 are limited by the available computing horsepower. Valve’s particle system distributes the load across multiple CPU cores.
Valve’s particle simulator really seems to take advantage of extra cores and threadsso much so that the Core i7-920 ended up third in our perf-per-dollar chart despite its high price tag. We should probably point out that buying a Core i7 involves blowing at least $200 on an X58 motherboard and paying a premium for triple-channel DDR3 memory kits, though. (We’ll look at full system prices in a little while.)
WorldBench’s overall score is a pretty decent indication of general-use performance for desktop computers. This benchmark uses scripting to step through a series of tasks in common Windows applications and then produces an overall score for comparison. WorldBench also records individual results for its component application tests, allowing us to compare performance in each.
We’ll look at the overall score, and then we’ll show individual application results alongside the results from some of our own application tests. For the sake of conciseness, however, we won’t be look at all individual testswe’ve left out Nero 7 Ultra Edition, Roxio VideoWave Movie Creator 1.5, Firefox 2.0, and 3ds max 8 DirectX performance.
The Core i7-920 looks far less appealing in WorldBench’s overall rankings, where the $163 Core 2 Duo E8400 finishes just nine points (7%) behind.
If our scatter plot shows anything, it’s that general desktop productivity tasks in the WorldBench suite aren’t terribly CPU-bound overall. That’s why so many points are clumped together. We’d see even more clumping together if it weren’t for AMD’s south bridge chips, whose inadequate support for Native Command Queuing leads to poor performance in WorldBench’s WinZip, Nero, and Photoshop CS2 tests. Have a look at page six of our Socket AM3 Phenom II review for more details.
Productivity and general use software
MS Office productivity
Multitasking – Firefox and Windows Media Encoder
Office is a perfect example of an application where higher CPU performance doesn’t do all that much: going from our baseline Athlon X2 6400+ to a Core i7-965 only reduces the test run-time by about 11%. Coincidentally, the i7-965 is also around 11 times more expensive than the 6400+.
Our multitasking test shows more notable differences and, naturally, gives the advantage to triple- and quad-core CPUs. The Phenom II X3 720 does particularly well there, although the Core 2 Duo E8400 isn’t all that far behind. Jumping on the Core i7 bandwagon in this instance only yields substantial performance dividends with the $999 Core i7-965 Extreme Edition.
With AMD processors at a disadvantage because of the NCQ issue we mentioned, Intel’s offerings reign supreme in Photoshop CS2. The Core 2 Duo E8400 looks particularly strong here, almost nipping at the Core i7-920’s heels. Then again, a newer version of Photoshop or a different mix of filters might be more multithreaded and thus better able to take advantage of additional processor cores.
The Panorama Factory photo stitching
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 widely multithreaded. We 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, but we’ll be looking at the total operation time today.
Extra cores matter a lot more in The Panorama Factory, where the Core 2 Quad Q8200 looks compelling despite being clocked a whole gigahertz below the Core 2 Duo E8600. The Core 2 Quad Q9400 isn’t a bad alternative, although the extra performance it brings will cost you an extra $50.
WinZip file compression
Same deal as with Photoshop CS2: this is an older release that seems to favor dual-core CPUs, and Phenoms perform poorly here because of the NCQ problem from which AMD’s south bridges suffer.
x264 HD benchmark
This benchmark tests performance with one of the most popular H.264 video encoders, the open-source x264. The results come in two parts, for the two passes the encoder makes through the video file. These scores come from the newer, faster version 0.59.819 of the x264 executable.
We’ve chosen to report results for the individual passes, then report an aggregate result based on a sum of encoding times for both passes. We’ll be basing our value calculations on that aggregate.
Multiple cores shine in recent video encoding applications like this, so it’s no surprise to see the Phenom II X3 720 being (apparently) penalized for its lack of a fourth core. Like in The Panorama Factory, the Core 2 Quad Q8200 is one of the most attractive deals here in spite of its low clock speed.
Windows Media Encoder x64 Edition video encoding
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. Unfortunately, it doesn’t appear to use more than four threads, even on an eight-core system. For this test, we 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, we instead used the “Multichannel audio” codec. Both audio codecs have a variable bitrate peak of 192Kbps.
Windows Media Encoder delivers results similar to those of the x264 HD benchmark, although the AMD processors perform relatively better. The Core 2 Quad Q8200 and Phenom II X4 810 are practically neck-and-neck on the value scale here, the latter being both faster and more expensive.
LAME MT audio 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 are encoding a massive 10-minute, 6-second 101MB WAV file. Our results are from a 64-bit version of LAME MT built using an Intel compiler. We left out results from the build made with a Microsoft compiler since this version produces lower encoding times with both Intel and AMD processors (and it apparently doesn’t penalize the AMD CPUs).
Surprise, surprise: the Core 2 Duo E8400 looks extremely well-positioned in a benchmark that doesn’t benefit from more than two cores.
Graphics is a classic example of a computing problem that’s easily parallelizable, so it’s no wonder 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 (or threads, in CPUs with multiple hardware threads per core) are available. We’ll be looking at the multithreaded results here.
The Core i7-920 really shines in this test, outpacing similarly priced quad-core processors quite noticeably. That said, we should once again note that a Core i7 platform costs morethat’s because LGA1366 motherboards and triple-channel DDR3 memory kits are still somewhat expensive. For more modest budgets, the Phenom II X4 810 and Phenom II X4 940 both look interesting (more so than the Intel equivalents).
We’re using the latest beta version of POV-Ray 3.7 that includes native multithreading and 64-bit support. Some of the beta 64-bit executables have been quite a bit slower than the 3.6 release, but this should give us a decent look at comparative performance, regardless.
3ds max rendering
Here are two more 3D rendering benchmarks where the Core i7-920 really distances itself from previous-generation products. If you’re shopping for cheaper alternatives, the AMD options clearly perform better in POV-Ray, while 3ds max 8 awards higher marks to Core 2 Quad processors like the Q8200.
Valve VRAD map compilation
This next test processes a map from Half-Life 2 using Valve’s VRAD lighting tool. Valve uses VRAD to pre-compute lighting that goes into games like Half-Life 2.
For at least part of the Source engine map building process, Intel’s Core 2 Quad Q8200 is more tantalizing than triple- and quad-core offerings in the same price range. It’s also a better deal than pricier Core 2 Quads. The Core i7-920 does quite well here, too, although these charts still don’t account for platform costs.
Next, we have a slick little [email protected] benchmark CD created by notfred, one of the members of Team TR, our excellent Folding team. For the unfamiliar, [email protected] 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. We’d encourage you to visit our distributed computing forum and consider joining our team if you haven’t already joined one.
The [email protected] 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, [email protected] 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. We’ve isolated that last number for this exercise, but you can view detailed results in our Socket AM3 Phenom II review.
This may be a somewhat quirky method of estimating overall performance, but it generally ought to work. We’ve discussed some potential reservations about how it works here, for those who are interested.
Our [email protected] scatter plot looks almost the same as those on the previous page: the Core 2 Quad Q8200 does very well, and the Core i7-920 is ahead of the pack by a considerable margin. Clearly, the Q8200’s 2.33GHz clock speed doesn’t hamper its ability to perform competitively in apps that put its four cores to work.
Power consumption and efficiency
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, graphics solution, hard drives, and anything else plugged into the power supply unit. (We plugged the computer monitor into a separate outlet, though.) We measured how each of our test systems used power across a set time period, during which time we ran 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 via Windows Vista’s “Balanced” power options profile.
Although we don’t usually include “simulated” CPU speed grades in our power results, we’ve made an exception for the Q8200 out of sheer curiosity.
You can see detailed results in our Socket AM3 Phenom II review, but for this article, we decided to look only at the amount of energy used to render the scene. Since the different systems completed the render at different speeds, we isolated the render period for each system. We then 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.
In effect, we’ll be looking at power efficiency per dollar. The 1/microjoules value in our power efficiency per dollar graph is really 1/(watt-seconds/1000000), or 1,000,000 m-2 kg-1 s2. That’s a little obscure, but it quantifies power efficiency in a readable fashion based on the source data, which is in joules.
Since the cheap triple- and quad-core processors completed the test more quickly than the dual-cores, we’re not surprised to see these CPUs score better on our efficiency-per-dollar chart. The Core i7 does deliver the best energy efficiency overall, which leaves the i7-920 in an enviable position on our value scatter plot. You’ll pay dearly for the additional efficiency of the higher end Core i7 processors, though.
If you can’t be bothered to study all of our charts and scatter plots… well, you probably should anyway. Our synthetic average-performance-versus-price plot is seducingly straightforward, but it’s very much an artifact of our test suite, which probably consists of more multi-threaded apps than what your typical enthusiast might run.
To create a synthetic “performance” score, we computed an unweighed average for results in all 22 of our benchmarks (including the WorldBench overall score). As our baseline, the Athlon X2 6400+ gets a 100% score. Other scores are all relative to it.
Since our aggregate performance scores are somewhat biased toward multithreaded tests, the Core 2 Quad Q8200 looks like a better overall deal than the Phenom II X3 720even though the X3 720 leads in non-multithreaded tasks.
Looking at the rest of the plot, we can spot several other trends. The Core 2 Quad Q9400 and Phenom II X4 940 are neck-and-neck, but the X4 920 outperforms the Q9300. That means the 920 may be a better overall deal than the Core 2 Quad Q8300, which has the same clock speed but less cache than the Q9300. (The 920 and Q8300 are both priced almost identically, although the AMD offering is slightly cheaper at Newegg.) In plainer terms, AMD’s high-end quad-core offerings are refreshingly competitive.
Looking further north, the Core i7-920 distances itself from the pack quite a bit overall. This is probably a good time to start looking at how full system pricing factors into the Core i7 formula, though. Let’s do that now.
Below, you’ll see another scatter plot with the same aggregate performance scale on the Y axis. To get our pricing numbers for the X axis, we’ve added the cost of a motherboard, memory kit, graphics card, and hard drive to that of our processors. Wherever it made sense, we picked components from our latest system guide. Also, we got all our prices from Newegg. Here’s a complete breakdown:
|Intel LGA775 platform||AMD Socket AM2+ platform||Intel Core i7 platform|
|Gigabyte GA-EP45-UD3P||$135||Gigabyte GA-MA790X-UD4||$115||Gigabyte GA-EX58-UD3R||$210|
|4GB Kingston DDR2-800||$41||4GB Kingston DDR2-800||$41||6GB Kingston DDR3-1333||$92|
|Sapphire Radeon HD 4870 512MB||$180||Sapphire Radeon HD 4870 512MB||$180||Sapphire Radeon HD 4870 512MB||$180|
|Western Digital Caviar Black 640GB||$80||Western Digital Caviar Black 640GB||$80||Western Digital Caviar Black 640GB||$80|
So, does the value picture change at all when we do this?
The Core i7-920 does look less appealing when we factor in the cost of other parts, but it still seems like a worthwhile step up. Looking at the lower portion of the graph, it’s also clear that cheaping out on a processor doesn’t impact the cost of a full system as much as one might think. At least, going from our baseline Athlon to a Phenom II X4 940 only raises the full system price from $506 to $641. That’s a 27% price increase (less once you throw in an enclosure, monitor, and all that jazz), but at least here, you get a 56% overall performance increase in return.
Speaking of the Phenom II X4 940, we didn’t just pick it as a random examplethis processor really looks particularly compelling. It’s a somewhat better deal than the Core 2 Quad Q9550, and it’s not only as fast as the Q9400, but also slightly cheaper when we account for the lower prices of AMD enthusiast motherboards. (Actually, the X4 940 is 30 bucks cheaper than the Q9400 on Newegg right now, even though official price lists give the advantage to the Q9400.) You might find the X4 920 more enticing because of its even lower price, but the 940 has an unlocked upper multiplier, which can make overclocking considerably easier.
We can’t really condense 64 different graphs or 11 pages of analysis and peripheral information into a short conclusion. However, we spotted several noteworthy constants in all of these graphs.
Generally speaking, AMD’s Phenom II X4 processors appear to be slightly better deals than the Intel Core 2 Quad equivalents. Not only are they great performers for the money, but the Socket AM2+ and AM3 platform has a better upgrade path than Intel’s soon-to-be-retired LGA775 platform. The Phenom II X3 720 is more of a mixed bag, since it’s the top performer neither in single-threaded tasks nor in heavily multithreaded ones. However, the 720 is still a good middle ground between cheap quad-cores and high-end dual-core CPUs.
Also, the Core i7-920 really distances itself from other processors in multithreaded tasks, but without giving much ground to dual-core chips in other tasks. Pricey X58 motherboards and DDR3 memory make the i7-920 an expensive step up, but clearly, you’re not throwing your money out the window. The i7-920’s position bodes well for upcoming mainstream Nehalem derivatives (Lynnfield), provided they’re not dramatically slower.
We should probably give a few pointers to folks who are shopping for a new CPU right now, as well. To start off, while this article contains many useful nuggets of information, we strongly recommend you also read our system guides and performance-oriented processor reviews to get a more complete picture of the market.
If you’re still unsure of what to buy, then take a step back and think about all of the applications you run. Looking at the relevant tests here should help you determine which processor can do the best job overall for the money. You’d be ill-advised to pick a CPU strictly based on our average-performance scatter plots on the previous page, because our test suite may not reflect what you’d run in your day-to-day activities. If you use your PC for nothing but games, for example, then a dual- or triple-core processor with a high clock speed may be a better investment than a part like the Core 2 Quad Q8200.
Otherwise, your processor choice should depend in large part on your overall budget. We’ve established that going from a cheap duallie to a decent quad-core processor doesn’t break the bank when one considers the cost of an entire system. However, that doesn’t mean you should allocate a disproportionate portion of cash to a fast processor and neglect other pieces like the graphics processor and memory. Our system guides should help you make that call better than this article can.