[nextpage title=”Introduction”]
The new GeForce GTX 670, which is being released today, is a more affordable version of the GeForce GTX 680, costing USD 400 and, therefore, competing with AMD’s Radeon HD 7950. Let’s see how the SuperClocked model from EVGA fared in our tests.
The GeForce GTX 670 is based on the same architecture used by the GeForce GTX 680, called “Kepler,” is manufactured under the new 28 nm processor, has a dynamic overclocking technology, and supports the PCI Express 3.0 specification. The only differences between the GeForce GTX 670 and the GeForce GTX 680 are the core clock (915 MHz vs. 1,006 MHz) and the number of graphics processors (1,344 vs. 1,536). The memory configuration is the same.
EVGA is releasing five models based on the GeForce GTX 670: The stock model (USD 400), the SuperClocked model (USD 420), the FTW model (USD 440), the stock model with 4 GB of memory (USD 470), and the SuperClocked model with 4 GB of memory (USD 490).
In the following table, we compare the main specifications of the video cards included in our review. The prices do not include rebates and are for the models with the clock and memory configurations listed below. Prices were researched at Newegg.com on the day we published this review, except for the GeForce GTX 670 models, which are the prices advertised by EVGA and NVIDIA. We also included the GeForce GTX 580 in this review, as its price dropped from USD 500 to USD 380 – USD 430 in the past few weeks with the release of the GeForce GTX 680. NVIDIA also offers a 3 GB version of the GeForce GTX 580 starting at USD 490. The second clock listed under “core clock” is the maximum clock the GPU can reach using its dynamic overclocking technology (i.e., “boost clock”).
| Video Card | Core Clock | Shader Clock | Memory Clock (Effective) | Memory Interface | Memory Transfer Rate | Memory | Shaders | DirectX | Price |
| GeForce GTX 670 | 915 MHz/980 MHz | NA | 6,008 MHz | 256-bit | 192.3 GB/s | 2 GB GDDR5 | 1,344 | 11 | USD 400 |
| EVGA GeForce GTX 670 SC | 967 MHz/1,046 MHz | NA | 6,210 MHz | 256-bit | 198.7 GB/s | 2 GB GDDR5 | 1,344 | 11 | USD 420 |
| GeForce GTX 580 | 772 MHz | 1,544 MHz | 4,008 MHz | 384-bit | 192.4 GB/s | 1.5 GB GDDR5 | 512 | 11 | USD 380 – 430 |
| Radeon HD 7950 | 925 MHz | 925 MHz | 5 GHz | 384-bit | 240 GB/s | 3 GB GDDR5 | 1,792 | 11.1 | USD 380 – 425 |
You can compare the specs of these video cards with other video cards by taking a look at our “AMD ATI Chips Comparison Table” and “NVIDIA Chips Comparison Table” tutorials.
Today, only the LGA2011 Core i7 (“Sandy Bridge-E”) and the “Ivy Bridge” processors have a PCI Express 3.0 controller. Therefore, we tested the video cards using a Core i7-3960X processor on a motherboard based on the Intel X79 chipset. In our Radeon HD 7970 review, we discovered that, at this time, there is no difference between using a PCI Express 2.0 or a PCI Express 3.0 connection. We also discovered that if you are using a high-end video card, the CPU doesn’t affect gaming performance.
Now let’s take a complete look at the EVGA GeForce GTX 670 SuperClocked.
[nextpage title=”The EVGA GeForce GTX 670 SuperClocked”]
Below we have an overall look at the EVGA GeForce GTX 670 SuperClocked. It requires two six-pin auxiliary power connectors.
Figure 1: EVGA GeForce GTX 670 SuperClocked
Figure 2: EVGA GeForce GTX 670 SuperClocked
This video card comes with one DVI-I, one DVI-D, one HDMI, and one DisplayPort connector, supporting the use of four video monitors at the same time. The product comes with a DVI-to-VGA adapter, so you can convert the DVI-I connector (which is at the bottom in Figure 3) into a VGA one.
Figure 3: Video connectors
EVGA GeForce GTX 670 SuperClocked uses a relatively small heatsink, with copper base and aluminum fins. This heatsink is cooled by a 60 mm radial fan that is placed at its side. The GeForce GTX 670 has a Thermal Design Power (TDP) of 170 W.
Figure 4: Heatsink
Figure 5: Heatsink
[nextpage title=”EVGA GeForce GTX 670 SuperClocked (Cont’d)”]
In Figure 6, you can see the video card with its cooler removed. We were impressed by how short the printed circuit board is, at 6.7” (170 mm). The video card, however, is a little longer, at 9.5” (240 mm), because of the fan.
It uses a voltage regulator with four phases for the GPU and two phases for the memory chips. The voltage regulator circuit uses a digital design and is controlled by an NCP5392P chip. All coils use ferrite cores and all capacitors are solid.
Figure 6: EVGA GeForce GTX 670 SuperClocked
Figure 7: Voltage regulator
The reviewed video card uses eight Hynix H5GQ2H24AFR-R0C GDDR5 chips, each one storing 2 Gbit of data, making the 2 GB of memory available on this video card. Each chip is connected to the GPU through a 32-bit lane, creating the 256-bit datapath that is available. These chips can run up to 6 GHz. On this video card, they are accessed at 6,210 MHz,
which is already above their labeled clock rate. Of course, you can always try to push the memory clock above its specs.
Figure 8: Memory chips
In Figure 9, you can see the accessories that come with this video card.
Figure 9: Accessories
Before seeing the performance results, let’s recap the main features of this video card.
[nextpage title=”Main Specifications”]
The main specifications for the EVGA GeForce GTX 670 SuperClocked include:
- Graphics chip: GeForce GTX 670 running at 967 MHz
- Memory: 2 GB GDDR5 memory (256-bit interface) running at 6,210 MHz QDR (Hynix H5GQ2H24AFR-R0C chips)
- Bus type: PCI Express 3.0 x16
- Video Connectors: One DVI-D, one DVI-I, one HDMI, and one DisplayPort
- Video Capture (VIVO): No
- Cables and adapters that come with this board: One DVI-to-VGA adapter and two power cable adapters
- Number of CDs/DVDs that come with this board: One
- Games included: None
- Programs included: Driver and overclocking utility
- More information: https://www.evga.com
- MSRP in the U.S.: USD 420.00
[nextpage title=”How We Tested”]
During our benchmarking sessions, we used the configuration listed below. Between our benchmarking sessions, the only variable was the video card being tested.
Hardware Configuration
- CPU: Core i7-3960X (3.3 GHz)
- Motherboard: Intel DX79SI (0460 BIOS)
- Memories: 16 GB DDR3-2133/PC3-1700, four G.Skill Ripjaws Z F3-17000CL9Q-16GBBZH memory modules
- Hard disk drive: Western Digital VelociRaptor WD3000GLFS (300 GB, SATA-300, 10,000 rpm, 16 MB cache)
- Video monitor: Samsung SyncMaster 305T (30” LCD, 2560×1600)
- Power Supply: Antec TruePower New 750 W
- CPU Cooler: Intel Liquid Cooling
Software Configuration
- Windows 7 Ultimate 64-bit
- Video resolution: 2560×1600 @ 60 Hz
Driver Versions
- AMD video driver version: Catalyst 12.4
- NVIDIA video driver version (GeForce GTX 580): 296.10
- NVIDIA video driver version (GeForce GTX 670): 301.33
- Intel Inf driver version: 9.2.3.1022
Software Used
- 3DMark 11 Professional 1.0.3
- Aliens vs. Predator + Benchmark Tool
- Battlefield 3
- Deus Ex: Human Revolution
- DiRT3
- Far Cry 2 – Patch 1.03
- Media Espresso 6.5
- StarCraft II: Wings of Liberty – Patch 1.4.3
Error Margin
We adopted a 3% error margin. Thus, differences below 3% cannot be considered relevant. In other words, products with a performance difference below 3% should be considered as having similar performance.
[nextpage title=”StarCraft II: Wings of Liberty”]
StarCraft II: Wings of Liberty is a very popular DirectX 9 game that was released in 2010. Though this game uses an old version of DirectX, the number of textures that can be represented on one screen can push most of the top-end graphics cards to their limits. StarCraft II: Wings of Liberty uses its own physics engine that is bound to the CPU and thus does not benefit from PhysX.
We tested this game at 1920×1200 and 2560×1600. The quality of the game was set to the “extreme” preset. We then used FRAPS to collect the frame rate of a replay on the “Unit Testing” custom map. We used a battle between very large armies to stress the video cards.
| Starcraft II: Wings of Liberty | 1920×1200 | Difference |
| EVGA GeForce GTX 670 SuperClocked | 218.8 | 3% |
| EVGA GeForce GTX 670 | 212.4 | |
| AMD Radeon HD 7950 | 206.8 | 3% |
| NVIDIA GeForce GTX 580 | 185.9 | 14% |
| Starcraft II: Wings of Liberty | 2560×1600 | Difference |
| EVGA GeForce GTX 670 SuperClocked | 178.5 | 3% |
| EVGA GeForce GTX 670 | 172.5 | |
| AMD Radeon HD 7950 | 167.7 | 3% |
| NVIDIA GeForce GTX 580 | 158.4 | 9% |
[nextpage title=”Far Cry 2″]
Released in 2008, Far Cry 2 is based on a game engine called Dunia, which is DirectX 10. We used the benchmarking utility that comes with this game at 1920×1200 and 2560×1600, setting overall quality to “ultra high,” maximizing all image quality settings, adjusting anti-aliasing to “8x,” and running the “Ranch Long” demo three times. The results below are expressed in frames per second and are an arithmetic average of the three results collected.
| FarCry 2 | 1920×1200 | Difference |
| EVGA GeForce GTX 670 SuperClocked | 121.0 | 1% |
| EVGA GeForce GTX 670 | 119.9 | |
| NVIDIA GeForce GTX 580 | 104.7 | 15% |
| AMD Radeon HD 7950 | 84.8 | 41% |
| FarCry 2 | 2560×1600 | Difference |
| EVGA GeForce GTX 670 SuperClocked | 83.8 | 4% |
| EVGA GeForce GTX 670 | 80.8 | |
| NVIDIA GeForce GTX 580 | 71.0 | 14% |
| AMD Radeon HD 7950 | 61.1 | 32% |
[nextpage title=”Aliens vs. Predator”]
Aliens vs. Predator is a DirectX 11 game that makes full use of tessellation and advanced shadow rendering. We used the Aliens vs. Predator Benchmark Tool developed by Rebellion. This program reads its configuration from a text file. (Our configuration files can be found here.) We ran this program at 1920×1200 and 2560×1600, with texture set at “very high,” shadows set at “medium,” anisotropic filtering set at “8x,” and anti-aliasing set at “2x.”
| Aliens vs. Predator | 1920×1200 | Difference |
| EVGA GeForce GTX 670 SuperClocked | 64.4 | 4% |
| EVGA GeForce GTX 670 |
61.9 |
|
| AMD Radeon HD 7950 |
59.9 |
3% |
| NVIDIA GeForce GTX 580 | 52.5 | 18% |
| Aliens vs. Predator | 2560×1600 | Difference |
| EVGA GeForce GTX 670 SuperClocked | 38.5 | 3% |
| EVGA GeForce GTX 670 |
37.2 |
|
| AMD Radeon HD 7950 |
36.8 |
1% |
| NVIDIA GeForce GTX 580 | 32.5 | 14% |
[nextpage title=”DiRT3″]
DiRT3 is another DirectX 11 game. We measured performance using this game by running a race and then playing it back using FRAPS. We ran this game at 1920×1200 and 2560×1536 with image quality set to “ultra,” and with anti-aliasing set at “8xMSAA.”
| DiRT3 | 1920×1200 | Difference |
| EVGA GeForce GTX 670 SuperClocked | 88.5 | 2% |
| EVGA GeForce GTX 670 | 87.0 | |
| NVIDIA GeForce GTX 580 | 68.8 | 26% |
| AMD Radeon HD 7950 | 66.2 | 32% |
| Dirt 3 | 2560×1600 | Difference |
| EVGA GeForce GTX 670 SuperClocked | 60.8 | 6% |
| EVGA GeForce GTX 670 | 57.3 | |
| NVIDIA GeForce GTX 580 | 46.7 | 23% |
| AMD Radeon HD 7950 | 46.0 | 24% |
[nextpage title=”Deus Ex: Human Revolution”]
Deus Ex: Human Revolution is a DirectX 11 game. We used the in-game introduction to measure the number of frames per second, using FRAPS. We ran the introduction in two resolutions, 1920×1200 and 2560×1536, maximizing all image quality settings, configuring anti-aliasing as “MLAA Mode” and anisotropic filtering at “16x.”
| Deus Ex: Human Revolution | 1920×1200 | Difference |
| EVGA GeForce GTX 670 SuperClocked | 154.7 | 4% |
| EVGA GeForce GTX 670 | 149.1 | |
| AMD Radeon HD 7950 | 138.3 | 8% |
| NVIDIA GeForce GTX 580 | 115.8 | 29% |
| Deus Ex: Human Revolution | 2560×1600 | Difference |
| EVGA GeForce GTX 670 SuperClocked | 104.5 | 2% |
| EVGA GeForce GTX 670 | 102.3 | |
| AMD Radeon HD 7950 | 99.6 | 3% |
| NVIDIA GeForce GTX 580 | 93.1 | 10% |
[nextpage title=”Battlefield 3″]
Battlefield 3 is the latest installment in the Battlefield franchise released in 2011. It is based on the Frostbite 2 engine, which is DirectX 11. In order to measure performance using this game, we walked our way through the first half of the “Operation Swordbreaker” mission, measuring the number of frames per second using FRAPS. We ran this game at 1920×1200 and 2560×1536, maximizing all image quality settings, configuring anti-aliasing as “4xMSAA” and anisotropic filtering at “16x.”
| Battlefield 3 | 1920×1200 | Difference |
| EVGA GeForce GTX 670 SuperClocked | 68.5 | 4% |
| EVGA GeForce GTX 670 | 66.1 | |
| AMD Radeon HD 7950 | 64.4 | 3% |
| NVIDIA GeForce GTX 580 | 59.8 | 10% |
| Battlefield 3 | 2560×1600 | Difference |
| EVGA GeForce GTX 670 SuperClocked | 58.9 | 9% |
| EVGA GeForce GTX 670 | 54.3 | |
| AMD Radeon HD 7950 | 53.6 | 1% |
| NVIDIA GeForce GTX 580 | 51.7 | 5% |
[nextpage title=”3DMark 11 Professional”]
3DMark 11 Professional measures Shader 5.0 (i.e., DirectX 11) performance. We ran this program at 1920×1200 and 2560×1600, selecting the four graphics tests available and deselecting the other tests available. We used two image quality settings, “performance” and “extreme,” both at their default settings. The results being compared are the “GPU Score” achieved by each video card.
| 3DMark 11 – Performance | 1920×1200 | Difference |
| EVGA GeForce GTX 670 SuperClocked |
4429 |
4% |
| EVGA GeForce GTX 670 |
4243 |
|
| AMD Radeon HD 7950 |
3472 |
22% |
| NVIDIA GeForce GTX 580 |
3184 |
33% |
| 3DMark 11 – Performance | 2560×1600 | Difference |
| EVGA GeForce GTX 670 SuperClocked |
2568 |
4% |
| EVGA GeForce GTX 670 |
2459 |
|
| AMD Radeon HD 7950 |
2083 |
18% |
| NVIDIA GeForce GTX 580 |
1848 |
33% |
| 3DMark 11 – Extreme | 1920×1200 | Difference |
| EVGA GeForce GTX 670 SuperClocked |
2735 |
5% |
| EVGA GeForce GTX 670 |
2617 |
|
| AMD Radeon HD 7950 |
2060 |
27% |
| NVIDIA GeForce GTX 580 |
1901 |
38% |
| 3DMark 11 – Extreme | 2560×1600 | Difference |
| EVGA GeForce GTX 670 SuperClocked |
1633 |
4% |
| EVGA GeForce GTX 670 |
1572 |
|
| AMD Radeon HD 7950 |
1299 |
21% |
| NVIDIA GeForce GTX 580 |
1156 |
36% |
[nextpage title=”Media Espresso 6.5″]
Media Espresso is a video conversion program that uses the graphics processing unit of the video card to speed up the conversion process. We converted a 449 MB, 1920x1080i, 18,884 kbps, MPEG2 video file to a smaller 640×360, H.264, .MPG4 file for viewing on a portable device such as an iPhone or iPod Touch. We also ran this test on our Core i7-3960X CPU in order to compare the difference in performance by using a high-end CPU and a high-end GPU to transcode video.
| Media Espresso 6.5 | Seconds | Difference |
| AMD Radeon HD 7950 |
30 |
9% |
| EVGA GeForce GTX 670 SuperClocked |
32 |
3% |
| EVGA GeForce GTX 670 |
33 |
|
| NVIDIA GeForce GTX 580 |
44 |
25% |
| Core i7-3960X |
46 |
28% |
[nextpage title=”Conclusions”]
The performance advantage of the new GeForce GTX 670 over the Radeon HD 7950 can be impressive. Depending on the game and resolution used, the GeForce GTX 670 can be up to 41% faster than its main competitor. However, the new video card achieved the same performance level as the Radeon HD 7950 on some games (Aliens vs. Predator, StarCraft II, and Deus Ex at 2560×1600). Since the Radeon HD 7950 wasn’t faster than the GeForce GTX 670 in any game, and they both cost the same, the new GeForce GTX 670 is, today, your best choice if you have USD 400 to spend on a video card.
The SuperClocked model from EVGA was up to 9% faster than the stock GeForce GTX 670 in our tests. Since it costs 5% more than the standard GeForce GTX 570, the price is right if you want to spend a little more and get a few extra frames per second on your favorite game.
For processing “regular” programs, however, the Radeon HD 7950 is faster. Using Media Espresso, the Radeon HD 7950 was 9% faster than the standard GeForce GTX 670, with the difference dropping to 3% on the EVGA GeForce GTX 670 SuperClocked.