by Sergey Lepilov
05/16/2014 | 03:13 AM
We’re going to test yet another original GeForce GTX 780 Ti today. This time around, it is an ASUS product called GeForce GTX 780 Ti DirectCU II OC.
Besides our standard testing, we will see how the performance of a GeForce GTX 780 Ti scales out depending on its graphics memory frequency. Today we’ve got a good opportunity for such a test as you will see shortly.
The specs and price of the ASUS GeForce GTX 780 Ti DirectCU II OC are listed in the following table in comparison with those of the reference Nvidia GeForce GTX 780 Ti, Nvidia GeForce GTX 780 and AMD Radeon R9 290X.
The graphics card comes in ASUS’s traditional retail packaging with claw marks on the front of the box. You can also see a picture of the cooling system there.
The card’s video outputs, key features and system requirements are detailed on the back of the box. There’s a black cardboard box inside the colorful external wrapper. The card is placed in its center amidst pieces of foam rubber. Above it, there is a small flat box with accessories.
The accessories to the ASUS GeForce GTX 780 Ti DirectCU II OC include a power adapter, a brief installation guide, a CD with drivers and overclocking utility, and two sets of stickers for the card’s cooler.
There are no video adapters or games among the accessories.
Manufactured in China, the ASUS GeForce GTX 780 Ti DirectCU II OC comes to retail for $699, which is the recommended price of the reference GTX 780 Ti. Original versions of this card from other brands cost considerably more. The product is shipped with a 3-year warranty.
The ASUS GeForce GTX 780 Ti DirectCU II OC has a rather unusual appearance because its sculpted cooler casing has no coloring. The card looks somewhat dull as the consequence.
Well, that’s not a problem, actually. Included with the card are two sets of colored stickers which let you customize its appearance if you want to.
The card measures 287x133x40 millimeters, so it is 20 and 33 mm larger than the reference GTX 780 Ti in length and height, respectively.
Like the reference card, the ASUS version has dual-link DVI-I and DVI-D outputs, one HDMI 1.4a and one DisplayPort 1.2. The video connectors are protected from dust with plastic caps.
The cooling system being larger than the PCB, the SLI connectors have to be implemented via an additional small PCB.
This solution doesn’t look pretty but you will hardly have problems connecting your SLI configuration. The card has two 8-pin power connectors with red and green LED indicators. The peak power draw is specified at 375 watts. A 600-watt PSU is recommended for a PC with one such graphics card inside.
Although massive, the cooler is secured with only four screws around the GPU, so it was easy to take off:
To dismantle the card further, we unfastened the metallic back-plate and the small aluminum heatsink mounted on the power components. Now we can have a closer look at the card’s PCB:
Like all other ASUS products of this class, the card features Super Alloy Power technology. Comprising premium components and a custom PCB, it ensures high energy efficiency (by up to 15% on this specific graphics card model), lower interference and longer service life (by up to 2.5 times).
The power system incorporates 10 phases: 8 for the GPU and 2 for the graphics memory and PLL.
Tantalum capacitors and ferrite-core chokes help ensure stability at high loads.
The GPU voltage regulator is based on a DIGI+ ASP1212 controller:
The 28nm GK110-425-B1 graphics chip was manufactured in Taiwan on the 44th week of 2013 (late October or early November).
Its base clock rate in 3D mode is increased to 954 MHz (by 8.9%) compared to the reference card’s. That’s the smallest factory overclocking we’ve seen so far in our tests of original GTX 780 Ti products. The peak boost frequency is 1020 MHz. According to our monitoring tools, the GPU voltage is 1.175 volts in 3D mode and 0.887 volts in 2D mode. The ASIC quality of our GPU sample is 69.0%:
The ASUS GeForce GTX 780 Ti DirectCU II OC is equipped with 3 gigabytes of GDDR5 memory in 12 chips from SK Hynix (marked as H5GQ2H24AFR-R2C).
The memory chips are rated for 7000 MHz and are indeed clocked at that frequency. So, the memory is not pre-overclocked. With a 384-bit bus, the peak memory bandwidth is 336 GB/s.
The latest version of the GPU-Z utility provides a summary of the graphics card’s specs:
Now we can proceed to checking out its cooling system.
The graphics card is equipped with the DirectCU II cooling system which is supposed to be 20% more efficient and 3 times quieter than the reference GTX 780 Ti cooler.
It is a large and massive thing that consists of an aluminum heatsink (with heat pipes and direct-touch technology) and a couple of fans (in a metallic casing).
The cooler has as many as five heat pipes, two of which are 6 mm in diameter. Two more pipes are 8 mm and there’s also a 10mm pipe here. The thermal grease imprint indicates that the GPU only contacts with the three central pipes.
The outermost 6mm pipes transfer heat from the sides of the adjacent pipes.
The pipes and heatsink fins are soldered to each other.
The fans are 95 mm in diameter. One of them implements the CoolTech technology, combining blower and axial fan features.
We saw the same Everflow fan in our review of the ASUS Radeon R9 290 DirectCU II.
The fans are PWM-regulated in a speed range of 1000 to 3050 RPM.
On the reverse side of the card, there is a protective plate with insulating film and perforation. It is supposed to facilitate the card’s ventilation.
The power components are cooled by a small aluminum heatsink via thermal pads.
To measure the temperature of the ASUS GeForce GTX 780 Ti DirectCU II OC we ran Aliens vs. Predator (2010) five times with maximum visual quality settings, at a resolution of 2560x1440 pixels, with 16x anisotropic filtering and with 4x MSAA.
We used MSI Afterburner 3.0.0 beta 19 and GPU-Z version 0.7.7 to monitor temperatures inside the closed computer case. The computer’s configuration is detailed in the following section of our review. All tests were performed at 25°C room temperature.
With the fans regulated automatically, the GPU is 79°C hot while the fans rotate at 1980 RPM.
Auto fan speed mode
This is no record for original GTX 780 Ti cards, yet a good enough result to call the cooler efficient. At the maximum speed of 3050 RPM the peak GPU temperature would drop by 9°C.
Maximum fan speed mode
As for the power system components, their temperature was 72°C in the first test and 67°C in the second (max fan speed) test.
We measured the level of noise using an electronic noise-level meter CENTER-321 in a closed and quiet room about 20 sq. meters large. The noise-level meter was set on a tripod at a distance of 15 centimeters from the graphics card which was installed on an open testbed. The mainboard with the graphics card was placed at an edge of a desk on a foam-rubber tray. The bottom limit of our noise-level meter is 29.8 dBA whereas the subjectively comfortable (not low, but comfortable) level of noise when measured from that distance is about 36 dBA. The speed of the graphics card’s fans was being adjusted by means of a controller that changed the supply voltage in steps of 0.5 V.
The diagram and table below help compare the noisiness of the ASUS GeForce GTX 780 Ti DirectCU II OC with the original GTX 780 Ti cards we tested earlier: Zotac GeForce GTX 780 Ti AMP! Edition, MSI GeForce GTX 780 Ti Gaming, Inno3D iChill GeForce GTX 780 Ti HerculeZ X3 Ultra, Palit GeForce GTX 780 Ti JetStream and Gainward GeForce GTX 780 Ti Phantom. We’ve also added reference Nvidia GeForce GTX 780 Ti and AMD Radeon R9 290X cards for the sake of comparison.
The vertical dotted lines mark the top speed of the fans in the automatic regulation mode. Here are the results:
Interestingly, the noise level graph of the ASUS GeForce GTX 780 Ti DirectCU II OC almost coincides with the graph of the MSI GeForce GTX 780 Ti Gaming, but the ASUS is louder in the automatic fan regulation mode as its fans work at 1980 RPM whereas the MSI's, at 1500 RPM.
Compared to the other GTX 780 Ti versions, the ASUS is noisier than the Inno3D and Zotac but quieter than the Palit and Gainward. It is quieter than the reference coolers from Nvidia and AMD, too. Overall, the ASUS DirectCU II cooler is average in terms of noise level. Its fans didn’t rattle or anything throughout the entire speed range. In 2D mode, with the fans working at 1000 RPM or lower, the card was virtually silent.
Despite the smallest factory overclocking among all original GTX 780 Ti cards we’ve tested so far, the ASUS version has poor overclocking potential in terms of its GPU. We could only increase its clock rate by 40 MHz. Adjusting GPU voltage didn’t help because it didn’t work, actually. Enthusiasts may want to edit the card’s BIOS for that purpose, but we don’t do that in our overclocking experiments. So we have to be content with the modest +40 MHz.
On the other hand, we managed to overclock the card’s memory by as much as 1000 MHz, which is the best result we’ve seen so far. The final clock rates were 994-1060/8000 MHz.
The overclocked card had a GPU temperature of 83°C while its fans accelerated to 2120 RPM.
Since we’ve achieved such a success overclocking its memory, we’re going to see the effect the memory frequency has on the card’s performance in our tests.
Here is the list of components we use in our testbed.
First of all, we will be comparing the original card from ASUS with the reference Nvidia GeForce GTX 780 Ti and AMD Radeon R9 290X at their default clock rates, with their fans regulated automatically.
We also include two original cards into this testing: a Gainward GeForce GTX 780 Ti Phantom, as one of the fastest single-GPU cards available, at its default clock rates of 980-1033/6008 MHz and an MSI Radeon R9 290X Lightning at 1080/5000 MHz.
We set Power Limit at its maximum on each graphics card.
In order to lower the dependence of the graphics cards’ performance on the overall platform speed, we overclocked our 32nm six-core CPU to 4.8 GHz by setting its frequency multiplier at x48 and enabling Load-Line Calibration. The CPU voltage was increased to 1.38 volts in the mainboard’s BIOS.
Hyper-Threading was turned on. We used 32 GB of system memory at 2.133 GHz with timings of 9-11-11-20_CR1 and voltage of 1.6125 volts.
The testbed ran Microsoft Windows 7 Ultimate x64 SP1 with all critical updates installed. We used the following drivers:
We benchmarked the graphics cards’ performance at two display resolutions: 1920x1080 and 2560x1440 pixels. There were two visual quality modes: “Quality+AF16x” means the default texturing quality in the drivers + 16x anisotropic filtering whereas “Quality+ AF16x+MSAA 4x(8x)” means 16x anisotropic filtering and 4x or 8x antialiasing. In some games we use antialiasing algorithms other than MSAA as indicated below and in the diagrams. We enabled anisotropic filtering and full-screen antialiasing from the game’s menu. If the corresponding options were missing, we changed these settings in the Control Panels of the Catalyst and GeForce drivers. We also disabled Vsync there. There were no other changes in the driver settings.
The graphics cards were tested in two benchmarks and 14 games updated to the latest versions.
We publish the bottom frame rate for games that report it. Each test was run twice, the final result being the best of the two if they differed by less than 1%. If we had a larger difference, we reran the test at least once again to get repeatable results.
The test results of the Nvidia GeForce GTX 780 Ti are colored light-green in the diagrams. The color of the Radeon R9 290X and MSI Radeon R9 290X Lightning is AMD's traditional red. The two original GTX 780 Ti cards from Gainward and ASUS are turquoise. The results of the ASUS with overclocked memory are a darker shade of turquoise.
Since we already analyzed the other data in our previous reviews, we will focus on the effect from overclocking the graphics memory on the ASUS GeForce GTX 780 Ti DirectCU II OC.
In the first 3DMark 2013 test the 1000MHz increase in memory frequency only makes the ASUS GeForce GTX 780 Ti DirectCU II OC faster by 2.2%. In the second test the performance gain is only 100 points or less than 0.3%.
We’ve got the same picture in Unigine Valley:
With MSAA enabled, the 1000MHz difference in memory frequency only translates into a 2.3% difference in speed. With MSAA turned off, we can hardly see any performance benefits from our overclocking at all.
Our memory overclocking isn’t rewarding in this game, either.
The ASUS GeForce GTX 780 Ti DirectCU II OC gets a mere 1.3-2.1% faster after a 1000MHz increase in its memory frequency with antialiasing turned on. It means just a single extra frame per second, which can hardly affect the gamer’s experience.
This test shows a bigger effect from our memory overclocking:
Our memory overclocking improves the frame rate by 3.5% at 2560x1440 with antialiasing. The bottom speed is also increased in each test mode.
Like most other games, Sleeping Dogs doesn’t speed up much when we overclock graphics memory.
There’s an even smaller difference in this game:
We can see the effect from our memory overclocking at nearly any settings:
The maximum performance gain is 6.3% at 2560x1440 with antialiasing, yet it only means an additional 2 fps.
The performance benefits are smaller here:
The extra 1GHz of memory frequency means a 1.1 to 4.3% increase in frame rate.
BioShock Infinite doesn’t speed up much after our memory overclocking.
This game is more responsive to the higher memory frequency both with Advanced PhysX enabled...
…and with Advanced PhysX disabled.
The maximum performance gain can be seen at 2560x1440 with antialiasing. It amounts to 5.9%.
You can enjoy GRID 2 on a GeForce GTX 780 Ti without overclocking its graphics memory, especially as an additional 1000 MHz to its clock rate only leads to a 3.4% increase in speed.
This game reacts to our memory overclocking in the same way:
Finally we’ve reached a game where the overclocked memory does ensure a substantial performance boost for the ASUS GeForce GTX 780 Ti DirectCU II OC.
The performance benefits can be observed in Total War: Rome II at any settings. They amount to 6.7-8.3%. Although not proportional to the memory frequency growth (+14.3%), it is still better than what we’ve seen so far.
Alas, the performance gains from our memory overclocking are small again:
Interestingly, the maximum performance boost in this game can be seen in the easy test mode, i.e. at 1920x1080 without antialiasing.
Our memory overclocking has little practical worth in this game, too.
The 1GHz increase in the clock rate of GDDR5 translates into a 2.5-3.4% increase in frame rate.
The ASUS GeForce GTX 780 Ti DirectCU II OC doesn’t speed up much from our memory overclocking in this game, either:
Here is a table with full test results.
Now we can move on to our summary charts.
Since we’ve focused on exploring the effect of memory frequency on the performance of a GeForce GTX 780 Ti, the first pair of our summary charts is about that topic:
As we’ve noted during our tests, our memory overclocking had the biggest effect in Total War: Rome II. The overall trend is that the performance gain is higher with enabled antialiasing and lower without it: 2.0 to 2.1% on average without AA and 2.5 to 3.7% on average with AA. Anyway, it means that overclocking a GTX 780 Ti's memory doesn't lead to a substantial increase in speed in today's games.
The next pair of charts compares the ASUS GeForce GTX 780 Ti DirectCU II OC with the reference GeForce GTX 780 Ti:
The ASUS version is 4.5-5.1% ahead of the reference GTX 780 Ti at 1920x1080 and 4.1-5.3% ahead at 2560x1440. The original GTX 780 Ti cards we tested earlier have a higher GPU clock rate and thus can offer slightly higher performance.
Here is a comparison of the ASUS with one such original GTX 780 Ti (it is Gainward’s GeForce GTX 780 Ti Phantom):
The gap is negligible: 1.3-2.3% at 1920x1080 and 2.1-2.7% at 2560x1440 pixels. It’s impossible to feel this difference while actually playing games.
We measured the power consumption of computer systems with different graphics cards using a multifunctional panel Zalman ZM-MFC3 which can report how much power a computer (the monitor not included) draws from a wall socket. There were two test modes: 2D (editing documents in Microsoft Word and web surfing) and 3D (the intro scene of the Swamp level from Crysis 3 running four times in a loop at 2560x1440 with maximum visual quality settings but without MSAA). Here are the results:
Despite the smaller factory GPU overclocking, the configuration with ASUS GeForce GTX 780 Ti DirectCU II OC consumes 13 watts more power than the configuration with Gainward GeForce GTX 780 Ti Phantom. The ASUS also needs 17 watts more in 2D mode. With the graphics memory overclocked by 1000 MHz, the total power consumption of the ASUS configuration grew by 4 watts at peak load. A 600-watt PSU is going to be sufficient for any of these setups.
The ASUS card may be the last original GTX 780 Ti we get to test, so we want to compile a summary on all the products we’ve tested so far. The following table shows the ASIC quality of their GPUs, their default clock rates (including the actual clock rates according to our monitoring tools) and their overclockability results. The best numbers are colored red and the worst are blue. The cards are listed in the order of our testing them.
The reference Nvidia GeForce GTX 780 Ti has the highest-quality GPU chip (77.8%). The MSI GeForce GTX 780 Ti Gaming is the fastest card by default as its GPU clock rate can get as high as 1163 MHz. The fastest memory is installed on the Zotac and Inno3D but, as we’ve seen in our today’s tests, this factor doesn’t influence performance much.
As for overclockability, the Zotac has the highest GPU frequency result but the reference card has the highest frequency growth (+240 MHz). According to our monitoring tools, the highest GPU clock rate was achieved by the Inno3D, though. The ASUS card we've tested today is the best in terms of memory overclocking.
The next diagram shows the results of our temperature and noise level tests at the maximum speed of the fans and in the automatic fan regulation mode. The results are sorted in the order of ascending top GPU temperature.
The Inno3D wins in both fan regulation modes thanks to its fantastically efficient cooler. It is followed by the Zotac and Palit, although the latter is too noisy. The coolers from MSI and ASUS are comparable in efficiency but the MSI is quieter. The Gainward and Nvidia cards have the worst results in this comparison.
And the next diagram shows temperature and noise data for when the cards are overclocked. The fan speed of the reference Nvidia GeForce GTX 780 Ti was set manually at 85% (3530 RPM) for the sake of stability. We also increased the fan speed of the MSI card to 65% or 1920 RPM.
We’ve got the same leader here: the Inno3D enjoys a large lead over the second-best Zotac both in temperature and noise level. The Palit is third, but it has lower overclocking results than the Inno3D and Zotac cards and produces more noise. Next goes the MSI with manual fan speed regulation, followed by the cards from Gainward and Nvidia. The ASUS takes last place at overclocking but it is quieter than the Gainward, Nvidia and Palit.
If in the near future we get elite products from MSI’s Lightning, EVGA’s Classified K|NGP|N Edition and ASUS’s Matrix (MATRIX-GTX780TI-P-3GD5) series or this Gigabyte model, we’ll update this summary with new test results.
The ASUS GeForce GTX 780 Ti DirectCU II OC is an interesting product, even though it can hardly claim to be the best single-GPU Nvidia-based solution available. We've mentioned its downsides in the course of this review including its rather noisy cooler, unimpressive factory overclocking and low frequency potential. Otherwise, it is a well-made product with low operating temperatures, an original PCB with enhanced power system, informative and reliable packaging, and certain customization options concerning its appearance. The card is less expensive than other original GeForce GTX 780 Ti versions.
And one more fact we’ve learned from today’s testing is that overclocking a GTX 780 Ti’s graphics memory has but a very small effect on its performance.