Similar GPU, Half the RAM: MSI GeForce GTX 1060 3GB Review
The new mid-range offerings from AMD and Nvidia haven’t quite delivered like we hoped they might. It’s not for a lack of performance, as both the RX 480 and GTX 1060 hold their own when gaming at 1080p and somewhat beyond. Instead, the issue stems from pricing and availability, the latter likely being responsible for the former.
Upon arrival, the Radeon RX 480 promised to provide previous-generation R9 390-like performance at a cost of just $200, while the GTX 1060 claimed GTX 980-level performance for $250.
As of writing, the cheapest RX 480 on Newegg is $250 for a custom MSI 4GB version and the rest are closer to $300. Things don’t look much better for the green team. Although the least expensive GTX 1060 can be had for $250, most are sitting at $300 or higher.
Last month, AMD announced another mid-range contender, though at $180 the RX 470 isn’t entirely differentiated. MSRP and actual street price are two different things especially at launch, so currently the cheapest RX 470 costs $190 with most above $200.
What about Nvidia’s new 3GB version of the GTX 1060, you ask?
For $200 — and we’ve been able to find at least one selling at that price — but as is the case with the RX 480, RX 470 and the 6GB GTX 1060, we find that most of the 3GB models are selling for above the official pricing.
Anyhow, assuming prices settle down shortly, there should be a $50 difference between the 3GB and 6GB versions of the GTX 1060, which would make the 3GB model 20% cheaper which is significant in sub-$300 graphics cards.
The 3GB frame buffer should be ample for 1080p gaming and might even make do at 1440p, though reduced VRAM isn’t the only downgrade made to the 3GB model. Despite keeping the GTX 1060 name, the 3GB model has an extra SM unit disabled.
With only nine of the ten SMs active, the core count drops from 1280 to 1152, and the texture units see a similar reduction from 80 to 72. Clock speeds remain unchanged at 1506 MHz with a boost to 1709MHz, but rated compute performance decreases by 11% (from 4. 4 to 3.9 TFLOPs). Although its core count has been cut, the TDP of the 3GB model remains at 120W.
Changing the core configuration for the GTX 1060 3GB means comparisons between the two GPUs now go beyond the included frame buffer. Nvidia claims the real-world performance drop from disabling one SM is around five percent, which is pretty reasonable considering the $50 saving.
Rather than take Nvidia’s word for it, we’re keen to see just how much of an impact the 3GB frame buffer and 10% reduction in core count has on performance in our benchmarks. There’s no Founders Edition 3GB model so on hand for testing we have the impressive looking MSI GTX 1060 Gaming X 3G.
The GTX 1060 3GB might only be a $200 graphics card, but that hasn’t stopped MSI from going all out and while the company’s Gaming X 3G is currently retailing for $240, we hope that price to dip down slightly over the coming weeks.
All up, the card measures 277mm long, 39mm wide and a very tall 140mm. MSI and Asus like to add a bit of extra height to their graphics cards and we have to admit it does make them look quite imposing.
The latest-generation Twin Frozr VI cooler has been installed featuring a pair of Torx 2.0 fans which boast a unique steeper curved blade design that is said to generate 22% more air pressure. The fans also stop spinning entirely when the card is operating below 60 degrees during light load.
Actually, it isn’t often the card will require the fans to spin up and certainly never to a noticeable level. This is helped not only by the efficient Pascal GPU but also the massive heatsink that is riddled with copper heatpipes measuring up to 8mm wide. The heatsink is capped off with a large nickel plated base that covers the entire GPU.
Under the heatsink is a huge heat shroud that covers much of the PCB and of course provides cooling for the memory and PWM. There is also a huge full-size backplate with a pretty cool looking design.
MSI has upgraded the single 6-pin PCIe connector to an 8-pin connector which essentially doubles the amount of juice that can be delivered here. Helping take advantage of that extra power is a 5+1 phase PWM design which is a considerable upgrade over the 6GB Founders Edition graphics card.
Out of the box, MSI has set the Gaming X 3G’s base clock at 1569MHz, resulting in a boost clock of 1784MHz. That’s a mild 4% overclock, but please note for the sake of consistency we have downclocked the GPU to Nvidia’s spec which calls for a 1506MHz base clock. We did this to ensure a fair comparison between the GTX 1060 6GB Founders Edition graphics card that we’ve already tested.
MSI GeForce GTX 1060 Armor review
48points
MSI GeForce GTX 1060 Armor
MSI GeForce GTX 1060 Armor
Why is MSI GeForce GTX 1060 Armor better than the average?
- GPU clock speed?
1506MHzvs1353.08MHz - Thermal Design Power (TDP)?
120Wvs199.53W - GPU memory speed?
2002MHzvs1571.13MHz - DisplayPort outputs?
3vs2. 07 - DVI outputs?
1vs0.73 - Multi-GPU?
4vs2.73
Which are the most popular comparisons?
MSI GeForce GTX 1060 Armor
vs
Nvidia GeForce MX550 Laptop
MSI GeForce GTX 1060 Armor
vs
MSI GeForce GTX 1060 Gaming X
MSI GeForce GTX 1060 Armor
vs
Asus Dual GeForce GTX 1060
MSI GeForce GTX 1060 Armor
vs
Zotac GeForce GTX 1060 AMP!
MSI GeForce GTX 1060 Armor
vs
MSI Radeon RX 570 Armor OC 8GB
MSI GeForce GTX 1060 Armor
vs
Asus ROG Strix GeForce GTX 1060
MSI GeForce GTX 1060 Armor
vs
Gigabyte GeForce RTX 2070 Super Gaming OC
MSI GeForce GTX 1060 Armor
vs
MSI GeForce GTX 1660 Armor OC
MSI GeForce GTX 1060 Armor
vs
Zotac GeForce GTX 1060 Mini
MSI GeForce GTX 1060 Armor
vs
Asus Dual GeForce RTX 2060 Evo OC 12GB
Price comparison
User reviews
Performance
GPU clock speed
1506MHz
The graphics processing unit (GPU) has a higher clock speed.
GPU turbo
1709MHz
When the GPU is running below its limitations, it can boost to a higher clock speed in order to give increased performance.
pixel rate
72.3 GPixel/s
The number of pixels that can be rendered to the screen every second.
floating-point performance
3.85 TFLOPS
Floating-point performance is a measurement of the raw processing power of the GPU.
texture rate
120.5 GTexels/s
The number of textured pixels that can be rendered to the screen every second.
GPU memory speed
2002MHz
The memory clock speed is one aspect that determines the memory bandwidth.
shading units
Shading units (or stream processors) are small processors within the graphics card that are responsible for processing different aspects of the image.
texture mapping units (TMUs)
TMUs take textures and map them to the geometry of a 3D scene. More TMUs will typically mean that texture information is processed faster.
render output units (ROPs)
The ROPs are responsible for some of the final steps of the rendering process, writing the final pixel data to memory and carrying out other tasks such as anti-aliasing to improve the look of graphics.
Memory
effective memory speed
2002MHz
The effective memory clock speed is calculated from the size and data rate of the memory. Higher clock speeds can give increased performance in games and other apps.
maximum memory bandwidth
192.2GB/s
This is the maximum rate that data can be read from or stored into memory.
VRAM (video RAM) is the dedicated memory of a graphics card. More VRAM generally allows you to run games at higher settings, especially for things like texture resolution.
memory bus width
192bit
A wider bus width means that it can carry more data per cycle. It is an important factor of memory performance, and therefore the general performance of the graphics card.
version of GDDR memory
Newer versions of GDDR memory offer improvements such as higher transfer rates that give increased performance.
Supports ECC memory
✖MSI GeForce GTX 1060 Armor
Error-correcting code memory can detect and correct data corruption. It is used when is it essential to avoid corruption, such as scientific computing or when running a server.
Features
DirectX version
DirectX is used in games, with newer versions supporting better graphics.
OpenGL version
OpenGL is used in games, with newer versions supporting better graphics.
OpenCL version
Some apps use OpenCL to apply the power of the graphics processing unit (GPU) for non-graphical computing. Newer versions introduce more functionality and better performance.
Supports multi-display technology
✔MSI GeForce GTX 1060 Armor
The graphics card supports multi-display technology. This allows you to configure multiple monitors in order to create a more immersive gaming experience, such as having a wider field of view.
load GPU temperature
Unknown. Help us by suggesting a value.
A lower load temperature means that the card produces less heat and its cooling system performs better.
supports ray tracing
✔MSI GeForce GTX 1060 Armor
Ray tracing is an advanced light rendering technique that provides more realistic lighting, shadows, and reflections in games.
Supports 3D
✔MSI GeForce GTX 1060 Armor
Allows you to view in 3D (if you have a 3D display and glasses).
supports DLSS
✖MSI GeForce GTX 1060 Armor
DLSS (Deep Learning Super Sampling) is an upscaling technology powered by AI. It allows the graphics card to render games at a lower resolution and upscale them to a higher resolution with near-native visual quality and increased performance. DLSS is only available on select games.
PassMark (G3D) result
Unknown. Help us by suggesting a value.
This benchmark measures the graphics performance of a video card. Source: PassMark.
Ports
has an HDMI output
✔MSI GeForce GTX 1060 Armor
Devices with a HDMI or mini HDMI port can transfer high definition video and audio to a display.
HDMI ports
More HDMI ports mean that you can simultaneously connect numerous devices, such as video game consoles and set-top boxes.
HDMI version
HDMI 2.0
Newer versions of HDMI support higher bandwidth, which allows for higher resolutions and frame rates.
DisplayPort outputs
Allows you to connect to a display using DisplayPort.
DVI outputs
Allows you to connect to a display using DVI.
mini DisplayPort outputs
Allows you to connect to a display using mini-DisplayPort.
Price comparison
Which are the best graphics cards?
MSI GeForce GTX 1060 Armor 6G Video Card Review and Test (page 3)
Test Bench
The MSI GeForce GTX 1060 Armor 6G was tested on the following configuration:
- Motherboard: ASUS Sabertooth X99;
- Processor: Intel Core i7-5930K @ 4200 MHz;
- Cooling System: Thermalright SilverArrow SB-E Extreme;
- Thermal interface: Prolimatech PK-1;
- RAM: G.Skill Ripjaws4 F4-3000C15Q-16GRR, 4 x 4GB, DDR4-3000 15-15-15-35, 1.35V;
- Hard drive: Western Digital Caviar Blue (WD500AAKS), 500 GB;
- Power Supply: Corsair CMPSU-750HX, 750 Watts;
- Housing: open stand.
Overclocking and temperature
As a test for warm-up and stability, we decided to use the HWBOT Unigine Heaven app in Xtreme Preset mode. This test is demanding both on the stability of the graphics processor/memory and on the cooling system of the video card, and unlike a purely synthetic load, it is as close as possible to real scenarios for using graphics accelerators.
Monitoring of frequencies and temperatures, as well as the speed of rotation of the impellers of the fans was carried out using the utilities MSI Afterburner 4.3.0 Beta 4 and GPU-Z 1.10.0. The temperature of the auxiliary elements of the video card was measured using four thermocouples, two of which were connected to the motherboard with Thermal Radar software. Readings of two more sensors were taken using a Mastech MY64 multimeter. Thermocouple locations:
As can be seen from the photo above, PCB textolite in four areas was chosen as points for measuring temperatures. One in the area of the GPU power converter, one in the area of the memory power converter, one in the area of the memory chips on top of the GPU (those chips that are in contact with the main cooling system), the last thermocouple is in the area of the memory chips to the right of the GPU (those chips that without additional cooling).
recommendations
Temperature test
To begin with, it was decided to check how the cooling system will perform in the factory mode, without user intervention.
In idle, the video card operates in power-saving mode with processor/memory frequencies of 139/810 MHz, at reduced voltages. In this mode, it completely stops the fans, so the system remains silent. The test stand was turned on for an hour without load, but the fans did not turn on, the results of measuring temperatures:
Sensor |
temperature value, degrees Celsius |
Thermocouple in GPU power converter area |
42 |
Thermocouple in the area of memory power converter |
42 |
Thermocouple in the area of the memory chips above the GPU |
33 |
Thermocouple in the area of the memory chips to the right of the GPU |
37 |
GPU, GPU-Z |
41 |
For the video card in the passive cooling mode, the results cannot but rejoice. Apparently, the dimensions of both the radiator of the cooling system and the PCB itself have an effect.
The next test (without changing any settings and without starting overclocking) was to go through several passes of Unigine Heaven, measuring temperatures after each:
Sensor |
First test iteration |
Second test iteration |
Third test iteration |
Thermocouple in the area of the GPU power converter |
58 |
60 |
62 |
Thermocouple in the area of memory power converter |
54 |
55 |
56 |
Thermocouple in the area of the memory chips on top of the GPU |
44 |
44 |
45 |
Thermocouple in the area of the memory chips to the right of the GPU |
63 |
64 |
65 |
GPU, GPU-Z |
64 |
65 |
65 |
After the third iteration of the test, the temperature of the MSI graphics cards stabilized and did not rise any more. In general, there are no complaints about them, except that one can note a significant difference in temperatures between memory chips located in different places.
As for the operation of the cooling system, the automatic adjustment of the cooler under load set the speed at 59%, which corresponds to 1351 rpm according to the tachometer. In the case of a 95 mm impeller, this is a low level; in most system units, the video card will not exceed the background noise level of other system components. Recording from a voice recorder located a couple of centimeters from the impellers.
In the maximum speed mode, the fan impellers spin up to about 2300 rpm, which is already quite noisy. Corresponding record from a voice recorder.
I note that during the operation of the turntables there are no overtones, chirping, etc., only the sound of “pumped” air is clearly audible. As for the sound emitted by the video card itself without the participation of fans, there is some squeaking of the throttles. In normal mode, it stands out slightly against the background of the fans and increases with further overclocking. The squeak level is unpleasant, but tolerable.
Here are the temperature readings during the test with the maximum fan speed.
Sensor |
temperature value, degrees Celsius |
Thermocouple in GPU power converter area |
52 |
Thermocouple in the area of memory power converter |
46 |
Thermocouple in the area of the memory chips on top of the GPU |
38 |
Thermocouple in the area of the memory chips to the right of the GPU |
54 |
GPU, GPU-Z |
56 |
By increasing the speed of the impellers, the temperatures of both the GPU and other elements of the printed circuit board have decreased. But the difference is not fundamental, especially against the background of the difference in noise level. Regular fan control works efficiently and does not require intervention.
Overclocking test
Before starting overclocking, let’s measure the voltage of the GPU.
Meaning |
Program monitoring |
multimeter |
Idle voltage, V |
0.625 |
0.626 |
Voltage under load, V |
1.05-1.062 |
1.085-1.095 |
Memory supply voltage: 1. 401 V idle and 1.56 V under load.
Measurements were taken with a Mastech MY64 multimeter. There is some deviation of the monitoring readings from the values measured under load, but the difference is not fundamental.
A minute of game. Yes, it is more convenient to measure voltages.
Unlike models of previous generations, overclocking of modern Nvidia graphics accelerators (at least without extreme sports) is extremely simplified. The video card itself controls the voltage and frequency of the GPU, focusing on the level of power consumption and temperature conditions. In the bottom line, we only have the opportunity to move the sliders and get one or another overclocking result. The Power limit level in the case of MSI GeForce GTX 1060 Armor 6G is limited to 108%, which does not add anything interesting.
So, as for the operating frequencies and the achieved overclocking results. In fully normal mode and without changing any settings, the frequency of the GPU during the test was 1823-1835 MHz. After setting the Power limit to 108%, other things being equal, the operating frequency of the GPU under load increased to 1860 MHz.
In order to eliminate the influence of third-party factors, the maximum fan speed was initially set, and with a gradual increase in the frequency slider, the result of overclocking was +161 MHz. A further increase in frequencies caused the test to hang, followed by a driver crash. According to monitoring, the GPU frequency under load in this overclocking mode was 2012-2025 MHz.
As for the memory, when the GPU was overclocked and the fans were at maximum speed, the +593 MHz mode became the maximum stable frequency, that is, the memory frequency was 9180 MHz. Intermediate overclocking result:
Taking into account the fact that the video card becomes noisy at such speeds, and most users will not use it in this mode, an attempt was made to overclock the MSI GeForce GTX 1060 Armor 6G with the default cooler settings.
From the point of view of overclocking the GPU, the settings did not have to be changed (but the video card itself under load lowered the frequency due to an increase in temperature), but the memory overclocking dropped to +444 MHz, the final memory frequency was 8900 MHz. Final overclocking result:
As a result, for the daily operation of the test instance, the final result can be called the 1987/8900 MHz mode.
In addition, I will note one of the aspects of the behavior of the review hero during memory overclocking: the test can also be passed at higher frequencies, around 9400-9500 MHz, and without performance drops. But in some subscenes of the test, the image starts to flicker, and in some artifacts appear.
Additional experiments
Not to say that the final overclocking result was good, so attempts were made to somehow influence the operation of the video card. First, the ASUS GPU Tweak utility was downloaded, which allows you to control the voltage of the GPU on some models. Indeed, there were settings in the range from 0 to +100 MV in the program.
Voltage measurement:
Meaning |
Program monitoring |
multimeter |
Idle/load voltage, +0 mode, V |
0.625 / 1.05-1.062 |
0.626 / 1.085-1.095 |
Idle/load voltage, +100 mode, V |
0.625 / 1.081-1.093 |
0.626 / 1.117-1.131 |
Let the voltage increase not by the level specified in the program, but less, but it still increases, providing at least some opportunity to influence overclocking. However, despite the increase in the GPU supply voltage, it was not possible to advance further than +161 MHz, so the main result of such a step can be called an increase in the temperature regime and the volume of the squeak of the throttles.
The second attempt to overclock the GPU above the results already achieved was to flash the BIOS version of the GeForce GTX 1060 Gaming X:
However, the only achievements from the BIOS firmware of the older model in the line were only the nominal frequencies, as well as a slight change in the cooler operation algorithms. If someone is interested in the firmware of both versions of MSI for their experiments, then here are the relevant links — 1060Armor.rom and 1060Gaming.rom.
That’s probably all.
Conclusion
The MSI GeForce GTX 1060 Armor 6G left a mixed impression after testing. There are no complaints about the video card in the “assembled the computer, installed the drivers and forgot” format: it works quietly, the temperature is normal, the fans turn off when idle, and in terms of cost, it compares favorably with some other GeForce GTX 1060.
As for overclocking results, the final figures, in my opinion, turned out to be not the highest. Perhaps this is a problem of a particular test instance and in another case it may be different. In addition, during acceleration, the squeak of the throttles increases, which is not particularly annoying in normal mode.
Advantages of MSI GeForce GTX 1060 Armor 6G:
- Efficient cooling system that provides a comfortable temperature regime for both the GPU and the power converter and memory chips;
- Smart fan control. The very case when there is no desire to make changes to the factory settings of the cooler;
- Not overpriced compared to other GeForce GTX 1060s.
Video card disadvantages:
- Not very good overclocking results.
Might not suit:
- Throttle squeal that increases with acceleration.
MSI GeForce GTX 1060 Armor 6G video card page on the manufacturer’s official website:
- Global mirror;
- Russian-language mirror.
Konev Ivan aka
Ivan_FCB
We would like to express our gratitude to:
- Nikolai Geek for providing the MSI GeForce GTX 1060 Armor 6G video card for testing.
- To Dushm@n for advice.
Subscribe to our channel in Yandex.Zen or telegram channel @overclockers_news — these are convenient ways to follow new materials on the site. With pictures, extended descriptions and no ads.
Test GTX 1060 3GB MSI Gaming X
This content has been written by a website visitor and has been rewarded.
The video card initially belongs to the flagship series with the same flagship cooling system, and therefore it already works at fairly high frequencies out of the box, so you should not expect a serious increase from overclocking. However, he is.
In stock, the video card boosts the frequency to an average of 1898-1924 MHz. Most often 1911 MHz. The cooler is configured for passive operation up to 63 degrees, when it reaches 63 degrees, the fans spin up to about 800 rpm and maintain such speed, keeping the temperature within 63 degrees.
recommendations
Overclocked in MSI AB through a curve with Unigine Heaven 4.0 running in the background.
As a result, we have: to video memory when writing leads to a serious drop in performance, and hence the relevance of this record as such. Therefore, in all games I had to throw off a number of settings, starting with textures, in order to somehow free up the memory of the video card. Moreover, the indications of monitoring do not give a transparent picture. The video recording is probably running data through the same VRAM. Therefore, for the video, all games were launched with minimal textures, because. they usually do not affect the overall performance, no more than the level of error, and this is the surest and most transparent way to free up video memory.
In order to make it more clear, I suggest that you familiarize yourself with examples of what happens if the video memory is clogged to the eyeballs and starts to climb into the RAM, pay attention to the difference in performance, the amount of video memory and RAM occupied:
Next test. And do not forget that recording in any case eats away part of the performance of the video card and without recording the results are somewhat higher. The latest version of NVENC new does not solve the problem, although in some cases there is a performance improvement.
Configuration:
- Ryzen 5 1600 @3.8 GHz 1.164V
- Gigabyte B450M Gaming
- DDR4 2×8 Samsung @3200 MHz 16-18-18-36-56 T1 1.3V (M378A1K43CB2-CRC)
- GTX 1060 3GB MSI Gaming X
- SSD ADATA SP580 120GB
- HDD WD10EZEX 1TB