Review of the NVIDIA GeForce GTX 1060 video adapter: the fight for the new generation

The transition of the discrete GPU industry to 14/16 nm photolithography continues to bear fruit. Both major manufacturers in this area - AMD and NVIDIA - presented the first samples of silicon for consumer video cards (Polaris and Pascal, respectively), produced according to these technological standards. AMD chose to try out a new process technology on chips with small to medium transistor budgets (Polaris 10, which formed the basis of the Radeon RX 480, and Polaris 11 as part of the upcoming Radeon RX 460) - and this, in fact, is not a new strategy, since the architecture GCN versions 1.1 and 1.2 also did not immediately appear on large GPUs. NVIDIA, on the contrary, continues the tradition of previous years, when the new architecture debuts in the most powerful models of the gaming line (in this case, the GeForce GTX 1070 and GTX 1080).

In the confrontation between the GeForce GTX 1060 and Radeon RX 480, these directions converged. Although AMD released its product first, and as you'd expect from a 14nm process, the RX 480 offered greater performance and lower power consumption than similarly priced previous-generation accelerators. However, not much time has passed since the announcement of the latter for AMD to take full advantage of the temporary lack of competition in the “productive” segment of discrete video adapters. Less than two weeks ago, NVIDIA announced that the video card number 60 in the GeForce 10 line is ready for release. From now on, the GeForce GTX 1060 goes on sale, and we can publish the results of its testing.

⇡#NVIDIA GP106

The third oldest GPU in the Pascal family is designed as half of the GP104. At least that's true for the GPU front-end, which includes two of the four GPCs (designs in recent NVIDIA architectures that contain most of the compute and graphics pipeline) present in the GP104, leaving the GTX 1060 and a half on the number of CUDA cores and texture mapping units that the GTX 1080 has.

However, if the GM206 (GeForce GTX 960) was in almost all respects half of the GM104 (GeForce GTX 980), then the GP106 in terms of back-end composition returns to the positions of the GK106 from the Kepler line, since the new chip is equipped with 192-bit, not 128-bit bit memory bus. Indeed, in terms of theoretical performance of shader ALUs, the new product is not that much different from the GeForce GTX 980, hence the increased requirements for memory bandwidth (BMB). NVIDIA could solve this problem using GDDR5X memory, but the latter is still far from widespread use, so the GP106 relies on a relatively wide RAM bus for NVIDIA GPUs in this class, and the more efficient delta color compression implemented in Pascal is negated residual difference between memory bandwidth in GTX 1060 and GTX 980.

Since each 32-bit memory controller in the NVIDIA architecture is associated with 8 ROP blocks and a L2 cache segment, the GP106 is also 75% of the GP104 in these parameters: 48 ROP and 1,536 MB L2.

Functionally, the GP106 is equivalent to the older gaming GPU of the Pascal family: the new architecture brings specific optimizations for the VR environment and the uncore section of the chip, which contains an updated set of image output interfaces and an H.265 video encoding/decoding unit. You can read more about this in our review of the GeForce GTX 1080. However, the main advantage of Pascal is its full-fledged work with parallel queues of rendering and calculation commands (Async Compute). We will release a separate article on this topic in the near future, but you can already see that DirectX 12 games that use this feature do not give Pascal the problems that Maxwell often encounters.

What's new?

According to NVIDIA, the new architecture is revolutionary, but is it really so? The most striking innovation in the case of PASCAL is the transition to a new 16 nm process technology. This made it possible to achieve a significant increase in the operating frequency of the GPU, which led to a noticeable increase in performance. Another important feature was the use of FinFET (Field Effect Transistor with Fin) - transistors, which resulted in a reduction in power consumption and an increase in efficiency. Despite this, the PASCAL architecture is more an evolution of Maxwell than something fundamentally new.

If we pay attention to the sphere of innovative technologies, it is important to note the presence of support for the advanced package of tools Nvidia VRWorks , which is designed to provide the most complete immersion in the VR environment.

This will be achieved by combining different player sensations, which will be provided by high-quality surround sound, correct reaction of the environment to certain actions and high-quality graphics.

For fans of multi-monitor configurations and virtual reality helmets, NVIDIA has prepared a technology for simultaneous multi-projection ( Simultaneous Multi-Projection ). In essence, this optimization is designed to simultaneously calculate geometric data for several pre-prepared projections. What is it for? Firstly, to eliminate various distortions when using curved monitors or multi-display configurations. Second, it improves performance for similar use cases on your system. Few of us had the opportunity to try NVIDIA 3D Vision at one time, but those who had the chance noted a noticeable lack of performance, since the video card was subject to double the load when drawing images for each eye. With the introduction of SMP, this problem is eliminated.

Also, one cannot fail to mention two more interesting innovations, which are entitled Ansel and Fast Sync.

Ansel is an invaluable tool for creating related gaming content, which will be useful for streamers and authors of various gaming blogs. During the game, you can press pause and adjust the camera to any desired angle to take a screenshot. In addition, you will have access to a number of effects to enhance the image, as well as the ability to take a photo with stunning clarity. In addition, it is possible to create 3D images, panoramic screenshots and 360° images. However, it should be noted that this will only be available in those games where appropriate support for this option will be introduced.

Fast Sync is a kind of addition to a function such as VSYNC. In general terms, activating this innovation will allow you to get rid of the delays that occur when you enable VSYNC. In addition, artifacts in the form of picture tearing are eliminated, which can be noticeable when vertical synchronization is turned off. This scenario is only valid if the FPS value exceeds the screen refresh rate threshold.

⇡#GeForce GTX 1060: technical specifications, price

The GPU in the GeForce GTX 1060 operates at a base frequency of 1,506 MHz - similar to the GeForce GTX 1070 - with the difference that the GTX 1060 has a higher Boost Clock, it is located in the gap between the corresponding characteristics of the GTX 1070 and GTX 1080.

The video card is equipped with 6 GB of GDDR5 SDRAM memory operating at a speed of 8 Gbit/s per contact. Contrary to rumors circulating shortly before the card's launch, the GTX 1060 specs do not include a 3GB RAM option. NVIDIA will be able to use this configuration in the proposed GTX 1050. Otherwise, the GTX 1050 will receive 2/4 GB of memory if the creators decide to cut the number of ROPs and memory controllers in the GP106 for this model.

The GeForce GTX 1060 board has three DisplayPort 1.3/1.4 connectors and one HDMI 2.0b, which allows you to output a signal with a resolution of 4K and 5K at 60 Hz (via HDMI and DisplayPort, respectively) with HDR support (now in HDMI 2.0b and in the future DisplayPort 1.4, which is currently an unfinished standard).

As you can see from the photos of the GTX 1060 that NVIDIA published in advance, the video card does not have an SLI connector. Obviously, the manufacturer wants to reserve this technology for top-end GPUs. Otherwise, the large amount of RAM and lower price of a pair of GTX 1060s compared to a single GTX 1080 could make a multi-GPU configuration more attractive than a flagship model. However, in this regard, not all is lost for the GTX 1060, since DirectX 12, among other things, allows you to combine the efforts of several GPUs without the help of a video card driver (although the loss of the “bridge” will certainly reduce performance in this case).

The GeForce GTX 1060 is designed for a 120 W thermal package - this is 30 W less than the TDP that the Radeon RX 480 (150 W) has.

The recommended price for the GeForce GTX 1060 is $299 for the reference version (Founder's Edition) and $249 for original design cards (US, excluding sales tax). The price for the Russian market is set at RUB 18,999. Partner cards are due to go on sale starting today, so the Founder's Edition can only be purchased directly from the NVIDIA website in select countries (which, unfortunately, does not include Russia).

Considering the appearance of partner versions of the GTX 1060 from the very first day of sales, to compare the new product with the Radeon RX 480, it is appropriate to rely on the price of $249. In this case, the GTX 1060 turns out to be $20–50 more expensive, depending on the configuration of the AMD product: with 4 or 8 GB of RAM. Since the GTX 1060 is inferior to its competitor in available memory but superior in price, NVIDIA should be confident in its superior performance. How much higher we will have to check in practice.

Testing the GeForce GTX 1060 using the MSI GTX 1060 Gaming X 6G video card as an example

The season of high-profile announcements continues. After the older graphics cards of the Pascal family, NVIDIA is introducing a mid-level model - the GeForce GTX 1060. The new product will be a response to the recent release of the Radeon RX 480 and will replace a number of mid-high-level models from the previous series.

GeForce GTX 1060

The GeForce GTX 1060 video card will receive all the benefits of the new Pascal architecture. It is based on the GP106 graphics processor, which is based on two graphics clusters of five TPC multiprocessors. The architecture of such clusters and multiprocessors is traditionally uniform within one series. Read more about them in a separate article about the architectural and technological features of Pascal.

Each multiprocessor has 128 CUDA stream processors and eight texture units. As a result, the GPU has only 1280 stream processors and 80 TMUs. Work with external memory is organized on a 192-bit bus through six 32-bit controllers. Support for advanced color data compression techniques improves information transfer efficiency and actual throughput. Improved work with asynchronous calculations. A special Simultaneous Multi-Projection block has been added to the Polymorph Engine, which allows you to simultaneously calculate geometry for several projections at the hardware level. This allows you to correctly display images on displays with complex shapes and create a panorama with the correct viewing angle in multi-monitor configurations. Additionally, multiple NVIDIA VRWorks technologies use multi-projection to optimize and speed up VR rendering.

In terms of its main characteristics, the GeForce GTX 1060 occupies an intermediate position between the GeForce GTX 960 and GeForce GTX 970. Slightly more computing units compared to the GeForce GTX 960, L2 cache memory has been increased from 1 MB to 1.5 MB, the memory bus has been increased from 128 bits up to 192 bits. Thanks to the new 16-nm process technology and optimization of the silicon core structure, frequencies have increased significantly. Taking into account the architectural improvements and significant increase in frequencies, one would expect the new product to replace the GeForce GTX 970, but the manufacturer promises the GeForce GTX 1060 to have the same level of performance as the GeForce GTX 980, which is impressive.

The base frequency of the GP106 GPU is set at 1506 MHz, and the average Boost Clock is 1708 MHz. This is close to the core frequency level of the GeForce GTX 1070, and Boost is even slightly higher. GDDR5 memory chips with a total capacity of 6 GB are used. Their effective communication frequency is 8 GHz. The TDP level is a modest 120 W, which is similar to the TDP of the GeForce GTX 960.

For a clear comparison of the GeForce GTX 1060 with previous models and the GeForce GTX 1070, let’s summarize their characteristics in one table.

Even at the time of the announcement of the GeForce GTX 1080, we wrote about the new Ansel software technology. Now it has acquired the status of a finished product. This is a platform for taking screenshots with a free camera and the ability to add additional visual effects. For now it only works in the game Mirror's Edge: Catalyst, soon The Witcher 3: Wild Hunt and other games will receive Ansel support.

Buyers of the GeForce GTX 1060 and owners of other current NVIDIA video cards will be able to try Ansel with the release of the next official driver package.

The first GeForce GTX 1060 on the market will follow the reference design. These models will be called Founders Edition and will be offered at a higher price. Later, non-standard versions will be released, the official price tag of which may be lower by $50.

GeForce GTX 1060 Founders Edition looks similar to older models, but is a little more compact and a little simpler. Instead of a transparent window, there is a plug on the cooling system.

Turbine-type cooling inherits the design of top-end cards. There is a massive base and two heatsinks - one on the GP106 chip and an additional heatsink to improve VRM cooling.

The board is as compact as possible, as was the case with the GeForce GTX 760 or GeForce GTX 660. The core has three-phase power.

Our practical acquaintance will immediately begin with the non-reference version. We will get acquainted with the new video card using the example of the MSI GeForce GTX 1060 Gaming X 6G, which will be tested at factory frequencies, at recommended frequencies (reference level) and overclocked.

MSI GeForce GTX 1060 Gaming X 6G

The video card comes in a medium-sized box. Red color dominates the design of the packaging and the card itself, which is unusual for NVIDIA products, but has already become a distinctive feature of MSI Gaming.

Externally, this model does not differ from more powerful solutions in this series, for example, MSI GeForce GTX 1070 Gaming X 8G. That is, despite the initial simplicity of the board and low TDP, MSI did not make any serious simplifications, releasing a version that is technically not much inferior to top solutions.

The new Gaming series graphics cards have an aggressive design with red teeth and an intricate casing shape. The impressive appearance is complemented by equally impressive large fans. The total length of the card is almost 28 cm.

The back of the board is protected by a metal plate with a dragon logo. There is partially decorative perforation to improve the convection of hot air. In the corner there is one eight-pin power connector.

There is an illuminated logo on the side. The red teeth are also illuminated. Intensity and color can be changed and customized via MSI software.

The rear panel has three DisplayPort, one HDMI and DVI connectors.

The design of the cooling system shows continuity with older MSI Gaming models. Same large heatsink, but fewer heat pipes. Contact with the surface of the graphics crystal is made through a massive base.

Two tubes with a regular diameter of 6 mm, one is thicker and more massive. They bend and penetrate the plate array twice, ensuring efficient heat distribution.

On the front side, two 100 mm fans are screwed onto the radiator. The special shape of the impeller in accordance with Torx 2.0 technology provides more powerful air flow and better dispersion for complete ventilation of the entire radiator.

Important elements on the board are covered with a plate, which acts as a heatsink and an additional element that gives rigidity to the structure.

Let's now take a look at the board itself. It is very large and does not at all resemble a standard “stub”.

The power system has increased the number of phases, although there are unsoldered places, perhaps for some top-end modification.

The GPU power subsystem has five phases, GDDR5 memory is powered from one phase.

Full markings of the GP106-400-A1 processor.

Six Samsung K4G80325FB-HC25 memory chips are soldered.

Our default instance ran at a base core frequency of 1595 MHz and a Boost Clock of 1810 MHz. Effective memory frequency 8100 MHz.

This frequency configuration corresponds to the OC Mode settings profile. Retail versions can run by default with the Gaming Mode profile, which provides frequencies of 1570/1785 MHz for the core and 8014 MHz for memory. Switching between these modes is done through the Gaming App, which at the same time allows you to adjust many other parameters, including customizing the backlight.

In addition to the mentioned modes, there is also a Silent Mode with frequencies of 1507/1722 MHz for the core and 8014 MHz for the memory. This configuration is almost identical to the recommended frequencies of the reference versions.

The video cards were operated on an open bench at 28 °C. Not the most favorable ambient temperature, but MSI's performance is excellent even in such conditions. When tested in Tom Clancy's The Division, the core heated up to 65–66 °C with fan speeds at 1000 rpm. Long-term load with the Metro: Last Light benchmark warmed up the core a degree higher at fan speeds of no more than 1100 rpm.

With a base frequency of 1595 MHz, the real Boost turned out to be at the level of 2000 MHz, and this frequency was stably maintained in all gaming tests. That is, MSI Gaming initially expanded the power limit, which allows you not to exceed these limits. And the temperature limit of 83 °C with such effective cooling is simply unattainable. The resulting noise is minimal.

The increased characteristics of the video card are good for the buyer, but for an objective picture we still need to test the GeForce GTX 1060 at the recommended frequencies or close to this level. In normal Silent mode, the Boost frequency remained stable at 1911 MHz. By the way, switching to this settings profile had minimal impact on the change in the final temperatures and fan speeds, which were initially very low.

Given the higher average Boost Clock, we decided to further lower the core frequency so as to limit the maximum Boost to 1860 MHz. Monitoring parameters in this mode when testing in Metro: Last Light is shown in the bottom screenshot.

What is the overclocking potential of the video card? At first, we settled on a frequency configuration that provided a stable peak Boost at 2126 MHz. The video card successfully withstood many difficult tests at this frequency, but in some places it failed. As a result, we had to settle on the base value of 1705 MHz with Boost 2101 MHz. This is slightly higher than the frequencies we received on the GeForce GTX 1070/1080. Well, it’s worth noting separately that this is not a floating frequency, but an absolutely stable value.

The final frequency looks impressive, but relative to the same 1860 MHz this is an increase of 13%. If we take the initial frequencies as a starting point, then the increase is only 5%. But the memory was amazing in that it was able to operate at 9548 MHz, which is 19% higher than the recommended level!

To maintain performance during overclocking, we fixed the fan speed at 1300 rpm. As a result, it was still quiet, and the heating was at the initial level.

Additionally, it should be noted that the power limit can only be raised by 8%. If the reference versions have the same small power reserve, then this may affect, if not the maximum level of Boost frequencies, then their stability during overclocking. Softvoltmod is not supported, but it is often not useful on higher cards either.

Characteristics of tested video cards

The reviewed video card will be compared with the Radeon RX 480 and GeForce GTX 970 from recent testing. At the same time, top-level models from the reviews of GeForce GTX 1080 Founders Edition and MSI GeForce GTX 1080 Gaming X 8G will be added to the comparison.

Test bench

The test bench configuration is as follows:

• processor: Intel Core i7-6950X (3, [email protected] ,1 GHz);
• cooler: Noctua NH-D15 (two NF-A15 PWM fans, 140 mm, 1300 rpm);
• motherboard: Gigabyte GA-X99P-SLI;
• memory: G.Skill F4-3200C14Q-32GTZ (4x8 GB, DDR4-3200, CL14-14-14-35);
• system disk: Intel SSD 520 Series 240GB (240 GB, SATA 6Gb/s);
• additional drive: Hitachi HDS721010CLA332 (1 TB, SATA 3Gb/s, 7200 rpm);
• power supply: Seasonic SS-750KM (750 W);
• monitor: ASUS PB278Q (2560x1440, 27″);

• operating system: Windows 10 Pro x64;
• GeForce GTX 1060 driver: NVIDIA GeForce 368.64;
• GeForce GTX 1080/1070 driver: NVIDIA GeForce 368.39;
• GeForce GTX 980 Ti driver: NVIDIA GeForce 368.22;
• Radeon R9 Fury X driver: AMD Grimson 16.5.3;
• Radeon RX 480 driver: AMD Grimson 16.6.2.

The test method described in one of the previous articles is taken as a basis. But since the settings configuration for top-end video cards was used, this comparison uses simpler parameters for some applications. Often, only the quality of anti-aliasing was reduced. Therefore, comparisons with high-level models are not present everywhere, but even the results presented are enough to compare the potential of the GeForce GTX 1060 with more expensive cards. As for the main three participants, they have been tested in all applications. If the final results at the main test resolution of 2560x1440 were unsatisfactory for practical use, additional tests were carried out at a resolution of 1920x1080.

Test results

Batman: Arkham Knight

The advantage of the GeForce GTX 1060 over the GeForce GTX 970 in Arkham Knight is 11–14%, a tiny lag behind the Radeon RX 480. Overclocking gives a speedup of 14–17%, allowing you to maintain a slight advantage over the forced Radeon RX 480.

Battlefield 4

The GeForce GTX 1060's position in Battlefield 4 is already stronger. The gap from the GeForce GTX 970 is 21–24% and up to 30% gap from the Radeon RX 480. MSI Gaming is 5% faster than the reference, further overclocking gives a 10% acceleration.

DiRT Rally

GeForce GTX 1060 at initial frequencies is 18–25% more productive than GeForce GTX 970. A small gap from the Radeon R9 290 and a more significant gap from the Radeon RX 480. Overclocking allows you to get closer to the performance of the Radeon R9 Fury X.

DOOM

In the new DOOM, the GeForce GTX 1060 and GeForce GTX 970 have similar performance at standard frequencies; the situation is similar in OpenGL and in older NVIDIA cards. The advantage over Radeon RX 480 is small. The game recently received an update that finally added the promised support for the Vulkan API. In this mode, we will compare opponents in the following articles.

Fallout 4

The GeForce GTX 1060 is 8-13% faster than the GeForce GTX 970 in Fallout 4 and has an even smaller difference with AMD competitors. Acceleration allows you to increase your lead over your opponents.

Far Cry Primal

A good lead over the GeForce GTX 970 in Far Cry Primal, the advantage over the Radeon RX 480 is more modest. Overclocking allows you to increase performance by 13–14%. The accelerated competitor from AMD is 14% weaker. If we talk about the difference between the GeForce GTX 1060 and GeForce GTX 1070, then it is about 39%, such a gap cannot be compensated for by overclocking

Gears of War: Ultimate Edition

Again, a good lead over the GeForce GTX 970. But the Radeon RX 480 this time is slightly faster than the GeForce GTX 1060. These competitors are equal in overclocking. It is interesting to note that the newcomer's performance level at initial frequencies is slightly higher than that of the overclocked GeForce GTX 970.

Grand Theft Auto 5

A good lead over its rivals in GTA 5. At standard frequencies, the newcomer shows performance on par with the overclocked GeForce GTX 970 and higher than the result of the boosted Radeon RX 480.

Just Cause 3

Less than 13% difference with GeForce GTX 970 in Just Cause 3 and no more than 5% difference with Radeon RX 480. MSI Gaming frequencies provide an additional advantage of 4-7%, further overclocking gives less than 9% speedup. Increasing frequencies allows us to minimize the lag behind the Radeon R9 Fury X, but with the GeForce GTX 1070 the difference is still more than 20%.

Metro: Last Light

Let's start with the results without the resource-intensive SSAA anti-aliasing.

At standard frequencies, the newcomer is 14% faster than the GeForce GTX 970 in Last Light. The difference with its competitor from AMD is about 19%, so even with overclocking, the Radeon RX 480 is not able to catch up with the GeForce GTX 1060.

The transition to SSAA increases the gap between the GeForce GTX 1060 and its closest competitors. The GeForce GTX 970 is 17% weaker, and the Radeon RX 480 loses 25%. The gap with the GeForce GTX 1070 is more than 41%. Overclocking allows you to gain up to 16% and reduce the gap from Fury X to 4%.

Quantum Break

The regular version of the GeForce GTX 1060 is faster than the GeForce GTX 970 overclocked in Quantum Break. But at the same time, the performance of the Radeon RX 480 is slightly higher, closer to this competitor MSI Gaming with factory overclocking. After overclocking their competitors, the GeForce GTX 1060 and Radeon RX 480 are equal.

Rise of the Tomb Raider

In Full HD mode, where younger participants show results acceptable for Rise of the Tomb Raider, the newcomer seriously outperforms the competition. The GeForce GTX 1060 is 31–37% faster than the GeForce GTX 970, the difference with the Radeon RX 480 is up to 26%. None of these competitors are able to catch up with the hero of the review due to their overclocking.

If you choose a higher resolution and turn all the parameters to the limit, the ratio remains almost unchanged. The higher result in comparison with the Radeon R9 Fury X is striking, but the reason for this is the insufficient memory capacity of such an opponent. But the gap with the GeForce GTX 1070 is 41–44%. The Radeon RX 480 again cannot catch up with the GeForce GTX 1060 even by increasing its frequencies.

The Witcher 3: Wild Hunt

In The Witcher 3, the GeForce GTX 1060 has a good advantage over the GeForce GTX 970, but its companion with increased frequencies is still a little stronger. Compared to the Radeon RX 480, the gain is mainly based on average fps.

The transition to a higher resolution reduces the difference in minimum performance between the GeForce GTX 1060 and GeForce GTX 970; in terms of average fps, the newcomer's advantage is more than 15%. The Radeon RX 480 lags behind only the average indicator - about 12%.

Tom Clancy's The Division

In The Division, the performance of the GeForce GTX 1060 is close to the level of the overclocked GeForce GTX 970 and is only 4-5% higher than the results of the Radeon RX 480.

Switching to a higher resolution slightly changes the situation in terms of minimum fps. Now we can even talk about some kind of parity with the Radeon RX 480, but after overclocking the GeForce GTX 1060 strengthens its position. When comparing the standard version with the GeForce GTX 1070, we can state a difference of 45%.

Total War: Warhammer

Unexpectedly, we see the same results from the GeForce GTX 1060 and GeForce GTX 970, both are weaker than the Radeon RX 480, and even the Radeon R9 290. When overclocked, the newcomer's performance scales better than its friend, and it is almost as good as its souped-up rival on Polaris 10.

XCOM 2

In XCOM 2, the newcomer's lead over the GeForce GTX 970 is 22-24%, the difference with the Radeon RX 480 is up to 12%. Both competitors cannot catch up with the GeForce GTX 1060 even after increasing their frequencies.

3DMark 11

The GeForce GTX 1060 wins about 20% over the GeForce GTX 970 and more than 26% over the Radeon RX 480. Increasing frequencies to 1705–2101 MHz/9548 MHz allows you to suddenly outperform the Radeon R9 Fury X.

3DMark Fire Strike

The ratio between NVIDIA's junior participants does not change, but the gap between the Radeon RX 480 and the GeForce GTX 1060 is now less than 13%.

Energy consumption

Everything is as expected. The system with the GeForce GTX 1060 has the lowest power consumption. But the new Radeon RX 480 is closer to the level of the GeForce GTX 970.

conclusions

Based on the testing results, we can state that the GeForce GTX 1060 shows performance at the level of the overclocked GeForce GTX 970. So the newcomer should really be close to the GeForce GTX 980. And this is with record low power consumption for this level of performance! In terms of performance and power consumption, this is the best mid-class video card at the moment. Impressive performance figures are ensured by the transition to the new Pascal architecture and thanks to the high initial frequencies that have become available to NVIDIA graphics accelerators within the 16nm process technology. The GeForce GTX 1060 did not cut down much on the bus, as was the case with the GeForce GTX 960. The combination of a 192-bit bus and high-speed GDDR5 8 GHz modules provides high throughput. The memory capacity is higher compared to the GeForce GTX 970 or GeForce GTX 980, which in some modern games can be quite in demand even at the popular 1920x1080 resolution. If we talk about higher resolutions or heavy graphics modes, then the GeForce GTX 1060 will receive even more significant advantages. As for comparison with the Radeon RX 480, the newcomer NVIDIA has a clear advantage. This advantage is not significant everywhere, but as frequencies increase it becomes more impressive.

The overclocking potential of the GP106 GPU is not much different from the results of the GP104 on older NVIDIA video cards. Based on our general experience with new accelerators, let's assume that MSI's core overclocking is average or close to average. But the memory overclocking is impressive. Moreover, the real increase in performance is often higher than the increase in core frequency, which indicates a clear benefit from overclocking the video buffer. In such conditions, much will depend on the memory capabilities of each individual video card. It’s still not possible to catch up with older Pascal models. The GeForce GTX 1070 is about 40% more powerful, which cannot be compensated for by overclocking the GeForce GTX 1060. The difference with the GeForce GTX 980 Ti is smaller, but even here it’s impossible to catch up with the older video card.

MSI GeForce GTX 1060 Gaming X 6G is an improved version of the GeForce GTX 1060, which is closer to older models in terms of manufacturing quality and cooling level. High factory overclocking will provide acceleration relative to conventional versions, and a powerful cooling system will ensure low operating temperatures with extremely low noise levels. As for overclocking, we have already noted the high potential of the memory. It’s doubly nice that higher frequencies are easily achievable while maintaining complete acoustic comfort thanks to the low fan speed. This video card is for those who want maximum performance from the GeForce GTX 1060 with no less maximum comfort. An excellent option for purchasing.

Separately, it should be noted that our article compared the non-reference version of the GeForce GTX 1060 and the standard Radeon RX 480 card. This fact plays an important role, because the weak cooling of the Radeon acts as a limiting factor during overclocking. And in nominal terms, due to exceeding the temperature limit, the overall performance decreases slightly. Non-reference versions of the Radeon RX 480 also promise good overclocking, and it is possible that such cards will perform better in confrontation with the GeForce GTX 1060. How true the assumptions are will be shown by a real comparison of non-reference rivals, which we will try to do in the future. Stay tuned for our reviews!

⇡#Design

Since the Founder's Edition of this video card will not hit the Russian market, few people here, besides journalists, will see it with their own eyes.

In terms of build quality and materials, the reference GeForce GTX 1060 is not inferior to the GTX 1080/1070. However, there is a difference in the design - the plastic insert that hides the radiator is opaque in the GTX 1060. In addition, the GeForce logo is not illuminated by LEDs, as is done in older cards.

The accelerator's printed circuit board is quite short, so that part of the cooling system protrudes beyond it. There is no protective plate on the back surface.

The cooling system itself consists of a radial fan and an aluminum radiator penetrated by two heat pipes. The vapor chamber, typical of reference versions of top-end NVIDIA video cards, is not necessary for the GTX 1060.

Water cooling for GTX 1060 graphics card

Water cooling is required when using the GTX 1060 6GB for mining. This will reduce the load on the system and extend its service life. The advantages of such cooling include:

1. Improving system performance;
2. Possibility of cooling several devices simultaneously;
3. Compact dimensions;

Minuses:

• Difficult to install and configure correctly;
• A specially prepared liquid is used as water, which costs money. Tap water will not work;
• There may be a leak, which will lead to a short circuit in the system;
• High price;

⇡#Pay

What immediately attracts attention on the GTX 1060 board are the two empty areas for RAM chips with traces of conductors from the GPU. There can be two explanations for this oddity: either the GP106 silicon actually contains a 256-bit memory controller, which was not fully used in the GTX 1060 (an unlikely scenario), or the graphics card PCB is designed for two different GPUs - GP106, as we see in this case, as well as the GP104 version with some memory controllers disabled by the manufacturer. The appearance of such GTX 1060 variants in OEM supplies is not excluded as soon as NVIDIA has a surplus of crystals that are unsuitable for use in the GTX 1070/1080.

Another unusual feature of the GeForce GTX 1060 is that the six-pin power connector is actually mounted on the cooling system housing and connected to the board with flexible conductors in order to maintain the design of the end of the video card. In principle, there are no practical complaints about such a solution.

The board's voltage converter includes three phases for powering the GPU and one for memory chips. The six-pin power connector, coupled with the lines in the PCI-Express slot, gives the video card a power reserve of 150 W. By the way, connecting the power with an intermediate conductor easily demonstrates that of the three 12 V circuits that can be used in the six-pin connector, the GTX 1060 uses only two (which is what it is formally designed for). However, taking into account the reserve cross-section of conductors that most power supplies are characterized by, the factor limiting the overclocking of the GTX 1060 reference version will not be this, but, as usual, the load on the power converter and the safe TDP limit programmed in the BIOS of the video card.

Printed circuit board

Before us is a compact printed circuit board made in black with a classic arrangement of components. In the central part there is a graphics chip around which memory chips are soldered. The power subsystem is located on the left side.

Note the lack of a protective frame around the GPU. This means that you must be extremely careful when removing or installing the cooling system so as not to accidentally damage the chip.

The total memory capacity is comprised of six chips labeled K4G80325FB manufactured by Samsung. The power subsystem has 4 phases, three of which are reserved for the GPU and one for memory. The uP9509P PWM controller manufactured by uPI Semiconductor is used to control the power. The transistors used are NTMFD4C85N manufactured by ON Semiconductor.

⇡#Test stand, testing methodology

The CPU runs at a constant frequency. In the NVIDIA driver settings, the CPU is selected as the processor for PhysX calculations. In the AMD driver settings, the Tesselation setting is moved from the AMD Optimized state to Use application settings.

⇡#Test participants

The following video cards took part in performance testing:

• AMD Radeon RX 480 (1266/8000 MHz, 8 GB);
• AMD Radeon R9 Fury (1000/1000 MHz, 4 GB);
• AMD Radeon R9 390X (1050/6000 MHz, 8 GB);
• NVIDIA GeForce GTX 1080 (1607/10008 MHz, 8 GB);
• NVIDIA GeForce GTX 1070 (1506/8000 MHz, 8 GB);
• NVIDIA GeForce GTX 1060 (1506/8000 MHz, 6 GB);
• NVIDIA GeForce GTX 980 (1126/7000 MHz, 4 GB);
• NVIDIA GeForce GTX 970 (1050/7010 MHz, 4 GB);
• NVIDIA GeForce GTX 960 (1126/7010 MHz, 2 GB).

⇡#Performance: 3DMark

The 3DMark benchmark makes it difficult to clearly give preference to the GeForce GTX 1060 or Radeon RX 480. In all tests of the Fire Strike package, the AMD product is inferior to its rival, but by no more than 6%, but it takes the lead in the Time Spy test under the DirectX 12 API.

Compared to the GTX 960, whose position the new product occupies in the Pascal family, NVIDIA provided an average twofold increase in performance.

The GeForce GTX 980 and Radeon R9 390X have the closest indicators to the GTX 1060 in 3DMark - they are only 5-7% faster.

⇡#Performance: games (1920 × 1080, 2560 × 1440)

Gaming benchmarks show the GeForce GTX 1060 in a more favorable light. In most tests, the GTX 1060's main rival, the Radeon RX 480, is inferior to the new product by 2 to 27% FPS, although the gap varies greatly from one game to another.

It is noteworthy that in two of the three games that use DirectX 12 and Async Compute - Ashes of the Singularity and Total War: WARHAMMER, the advantage of the GTX 1060 is greatest. The situation in Rise of the Tomb Raider is different, and we know what the reason may be - RotTR uses Async Compute in a way that should be done with Maxwell processors, not Pascal (more details in an upcoming article on this topic).

The only game where the GTX 1060 suffered a defeat, and a crushing one at that, was DOOM with a patch that introduced Vulkan API support into the engine.

Finally, in gaming tests, the GTX 1060 comes even closer to two powerful video cards of the previous generation - GeForce GTX 980 and Radeon R9 390X. Compared to the GTX 960, the younger Pascal has now more than doubled its performance.

Profitability and consumption of 1060 in mining

First of all, let's see how much the card is capable of producing on various algorithms, after that we will be able to determine the exact profitability of the 1060 6Gb in mining, as well as see its overclocking potential and how much it can add in mining ether and other coins.

Profitability and consumption of GTX 1060 6Gb

Yield 1060 6Gb:

Based on the table data, the profitability of 1060 6Gb at the end of March 2021 is as follows:

Profitability of 1060 6Gb in mining at the end of March 2022
As we can see, as of March 20, 2021, the card brings in about $2 per day. At the same time we consume 90 watts. Is it possible to increase the hashrate of the card using overclocking?

For mining on a GTX 1060 video card, we recommend the Binance pool, the most reliable and convenient, since it is both an exchange and a pool with minimal commissions. Read how to mine on Binance Pool on the website. As well as the Hive OS operating system. Link to register Hive OS (with the promotional code InsidePC you get $10 to your account).

For stable operation of video cards in the rig, make the correct settings on the motherboard for mining.

⇡#Performance: games (3840 × 2160)

Moving to a higher resolution slightly shifts the emphasis in the results. The rival Radeon RX 480 no longer receives such help from the Vulkan API in DOOM, but in other games, on the contrary, it has strengthened its position. The same applies to the “old timers” - GTX 980 and R9 390X. Overall, neither the RX 480 nor the GTX 1060 provide comfortable frame rates in this mode.

Game tests

Let's move on to gaming applications and focus on the testing methodology. FPS measurements were carried out using the FRAPS utility. All games were tested in the three most current resolutions: 1920x1080, 2560x1440 and 3840x2160. The following options are manually disabled:

• VSync (Vertical Sync)

All other settings, including anti-aliasing, were set to the maximum possible. For clarity, each game is provided with screenshots of the graphics settings menu; only the resolution was changed, all other parameters remained unchanged.

Based on the testing results, it can be noted that the performance of the GTX 1060 is approximately on the same level as the GTX 980. This video card is perfect for gaming at a resolution of 1920x1080, even if such a resource-intensive option as anti-aliasing is enabled. The exception in our case is games such as Assassin's Creed Syndicate and Metro Last Light; here it is recommended to disable this option. If you pay attention to the higher resolution of 2560x1440, then here too you can count on a decent level of performance, but only when you turn off anti-aliasing and parameters such as shadows, volumetric lighting, and object drawing distance. The video card is completely unsuitable for the newfangled 4K, and this is largely due to the lack of SLI support.

⇡#Performance: Computing

Pascal chips have the fastest H.264/H.265 decoder - this is undeniable. As for calculations performed on shader ALUs of the processor, the split in this area that began with the Maxwell generation continues - some tasks are more beneficial for the Pascal architecture, some for the GCN, although in general it is worth recognizing that the Radeon RX 480 and other AMD accelerators occupy on average a higher (and more predictable) position in non-graphics computing tests.

Note: since decoders within the same GPU line usually do not differ, the diagrams show one device from each family (or more if this rule is violated).

Which processor will open 1060

Any 4-core processor with a frequency of 3.5 Mhz is quite sufficient for this video card, the more the better. In theory, a Core i3 may be enough, but it is still recommended to take, for example, a Core i5 6400 (Skylake), and in principle, the Haswell generation will be able to perform well.

If we talk about exceptions, then, for example, GTA V may not fully load the video card, since it largely uses the processor for physics. For this video card, the frequency is more important, so it is better to overclock the same 6400 on the bus.

⇡#Clock frequencies, power consumption, temperature

In normal mode, the GPU clock speed reaches 1911 MHz in games, but more often does not exceed 1873 MHz (as we observed in Crysis 3, which we use as a load in power consumption tests). The maximum core supply voltage is 1.062 V. The cooling system operates quietly at 1950 rpm, keeping the GPU temperature within 71–72 ° C - an impressive result both for a video card of this performance class and for a closed-type reference cooler.

As for power consumption, although the GTX 1060 has the same TDP as the GTX 960, in practice the power of the new product is significantly higher. Radeon RX 480 and GeForce GTX 1060 share 28 W of power - almost the same as the specifications.

The reference GTX 1060 overclocks almost as well as the GTX 1070/1080. The stock BIOS allows you to raise the TDP limit by 16%, increase the peak voltage on the GPU to 1.093 V and run the turbine at 3200 rpm.

Together, these tools ensured that the GPU was overclocked to a base 1736 MHz, and in dynamics (Crysis 3) the frequency rose to an amazing 2101 MHz. The memory chips achieved an effective frequency of 9500 MHz. The cooling system controls the GPU temperature during overclocking, operating at maximum speed. But, as a rule, in such conditions we see a more significant decrease in temperature compared to the standard mode, so for everyday use (in terms of noise level) with such a large overclock, the Founder's Edition may not be as well suited as the original versions of the GTX 1060 with open-design coolers (although the cost of overclocking in terms of power was only 14 W in Crysis 3).

⇡#Performance: overclocking

The excellent overclocking of the GeForce GTX 1060 cannot be ignored in the test results. According to average indicators of synthetic and gaming benchmarks, the card's performance increases by 14% and 16–17%, respectively.