MSI Gaming GeForce GTX 1660 review

It might be seen by some in the PC gaming community that Nvidia’s first GTX-branded “Turing” GPU (versus RTX-branded) was an admission that the world had not yet reached the point where its RTX line could be used for raytracing. Despite the fact that video cards are planned out for years, this graphics card does take RTX technology backwards. This was done by removing some of its future-looking flagship features. That was the plan from the start.
Consequently, this Turing-architecture card emphasizes value to players today by being chopped-down and smoothed out. A muscle card costs $500-plus. However, but for gamers who need high refresh rates (online multiplayer gamers), the GTX 1660 Ti is suitable. It can also compete with older AAA titles that are slightly more expensive, such as GeForce RTX 2060. The RTX 2060 Founders Edition, also excellent, is priced nearly as much as, but more discretely than, the $309 MSI version we tested.
Pros
1080p gaming at an excellent price-to-performance ratio.
At both ends of the price spectrum, it’s better than previous generation GTX cards.
A cooling system of exceptional quality.
The potential for overclocking is solid.
It is one of the best GPU for video editing
Cons
Playing games in 4K is not recommended.
With MSI’s upclocked card, your GTX 1660 Ti card will cost less than most RTX 2060 Founders Edition cards.
Design of MSI Gaming GeForce GTX 1660

With dimensions of 9.7 x 5 x 1.6-inches, the MSI GTX 1660 Super Gaming X is a dual-slot card. Due to the fact that it is a two-slot solution, whatever plugging in there is not going to protrude above the slot line. Why? The space in between will be limited. There is no doubt that this card is large. It extends almost to the width of an ATX motherboard. What is more is that its height barely over the PCIe expansion slots.
But it’s compact enough that it should fit in most non-SFF systems. Be sure, however, to check the chassis specifications prior to making a purchase as always.
Read more on Best GPU For Intel Core i7-13700K in 2023
Two large fans
Black shrouds surround the two large fans on the Twin Frozr 7 cooler on this card. In addition to the sharp angles and chevron pattern of the plastic cover, there are four RGB LEDs on the front, above and below the fans. A simple backplate protects the card’s back, which illuminates the MSI Twin Frozr 7 writing and the gaming dragon. In addition, the top of the card has RGB lighting which illuminates the MSI Twin Frozr 7 writing. This makes MSI a good GPU for mining.
With a black-and-grey design, the card blends in with most build themes. It also and offers a decent amount of style and RGB glow.

In our testing, we were able to keep the Twin Frozr 7 cooler than EVGA’s solution. This was surprisingly quiet during the testing process, while keeping this card cooler than EVGA’s tougher 2+ slot cooler.
Twin Frozr 7 is equipped with dual 90mm fans. As a result of the Torx Fan 3.0, MSI claims airflow is accelerated and increased by the specially curved blades. The fans are kept off during idle and low load conditions using MSI’s Zero Frozr technology, resulting in complete silence on the desktop and light use.
Heat sink

Under the fans lies a large, dense heatsink, which runs most of the card’s length — and most of its height. There are three s-shaped heat pipes above the base plate, each passing twice through the fin stack. A nickel-plated copper base plate mates to the die to the heatsink. A black baseplate cools the memory and VRMs.
Gaming X uses a 4+2 power delivery system, in which the GPU is divided into four phases and the memory is divided into two. There are eight channels in the OnSemi NCP81610 controller to control the GPU voltage. The MOSFETs are also NCP302045 DrMOS from OnSemi, and 225W of in-spec power will be delivered by the single 8-pin PCIe connector. In combination with the PCIe slot, 225W of in-spec power will be delivered to the capable (but not overbuilt) VRMs.
Read more on best GPU for video editing.
How We Tested MSI’s GTX 1660 Super Gaming X
Our GPU test system was recently updated from a six-core to an eight-core Core i7-8086K processor with a new platform. It is powered by a Corsair DDR4 3200 MHz CL16 processor paired with two 16GB Corsair DDR4 memory (CMK32GX4M2B3200C16) on an MSI Z390 MEG Ace motherboard. In addition to the Corsair H150i Pro RGB AIO, our test system is equipped with a 120mm Sharkoon fan, which provides general airflow across the system. A Kingston KC2000 NVMe PCIe 3.0 x4 drive stores our OS and gaming suite.
Currently, the motherboard is running a BIOS version 7B12v16 (at this time). After setting up the system using optimized defaults, the XMP profile of the memory was enabled to achieve the memory’s rated 3200 MHz CL16 speed. As of December 2019, Windows 10 (1909) is fully updated and was used on the system. No other performance enhancements were enabled.
Our database of results is constantly being updated based on this test system, as we continue to build it up. In the meantime, we are including GPUs that are very close to the card under review. As part of this review, we will include the Zotac GTX 1660 Amp, the EVGA GTX 1660 Super SC Ultra, and the EVGA GTX 1660 Ti XC cards, while AMD will be represented by the XFX RX 590 Fat Boy, and the Gigabyte RX 5500 XT Gaming OC 8G.
Read more best graphics card for mining
Architecture of MSI Gaming GeForce GTX 1660

GeForce GTX 1660 uses the same 12 nm “TU116” silicon as GTX 1660 Ti. The GTX 1660 is a new GPU from NVIDIA that features8 GB of GDDR5 memory instead of twelve gigabytes of GDDR6 memory and disables two of the 24 streaming multiprocessors (SMs). When it comes to memory bus width is concerned, it is unchanged at 192 bits.
However, memory bandwidth has been reduced by a third to 192 GB/s instead of 288 GB/s. As opposed to the GTX 1660 Ti, which has 1500/1770 MHz GPU clock speeds, the GTX 1670 Ti has 1530/1770 MHz GPU clock speeds. Msi geforce gtx 1660 super gaming x 6gb is slightly lower that this device because of small memory and other features.
Read more on Best GPU for PC 2022
No RT cores and Tensor Cores
As a result of reengineering the Graphics Processing Clusters (GPCs) of the silicon, NVIDIA has eliminated RT cores and Tensor Cores. The chip’s hierarchy is similar to that of other Turing GPUs. In addition to PCI-Express 3.0 x16 host and 192-bit GDDR5 memory interfaces, the GPU’s GigaThread Engine and L2 cache serve as town squares for the GPU.
There are four TPCs (Texture Processing Clusters) in each GPC, which share a Polymorph Engine with two streaming multiprocessors (SM). The Turing SMs contain 64 CUDA cores, so the TPC has 128 CUDA cores, the GPC 512, and the silicon has 1,536 CUDA cores in total. With 22 out of 24 SMs (or 11 out of 12 TPCs) enabled, the GTX 1660 has 1,408 CUDA cores. Despite the fact that it’s still more than 1,280, the Turing architecture offers better IPC.
In addition to improving Turing’s concurrent-execution capabilities, NVIDIA has developed a number of CUDA core-specific innovations. This is not the same as asynchronous computing, but the concepts are not too far apart. In contrast to older architectures, such as Pascal, which can only execute one kind of instruction at a time, Turing CUDA cores can execute both integers and floating-point instructions per clock cycle in parallel. In terms of GPU computing, asynchronous compute refers to its capacity to handle various graphics and computation workloads simultaneously.
Read more one GPU for gaming laptops
L1 caches
L1 caches provide improved protection to CUDA cores. These caches are enlarged threefold and have a fourfold increase in load/store bandwidth. Caches can be configured on the fly as either two 32 KB partitions per SM or a single 64 KB block per TPC. A dedicated FP16 core per SM has also replaced tensor cores. In addition to 64 FP32 and 64 INT32 cores per SM, these components are physically separate components that execute FP16 twice as fast as FP32 cores do. Accordingly, the tensor cores on the RTX 2060 are configured to handle FP16 operations at a tremendous rate. This is true even though there are no dedicated FP16 cores per SM.
On the GTX 1660, NVIDIA is deploying GDDR5 memory, a slower and older generation of memory. The memory is clocked at an 8 Gbps data rate, as opposed to the 1660 Ti, which has 12 Gbps GDDR6 memory. There is a drastic 33 percent decrease in memory bandwidth compared with the GTX 1660 Ti. It is exactly the same as the GTX 1060 6 GB memory. The amount of memory remains the same at 6 GB.
Read more one 11 best GPU for gaming PC
Features of MSI Gaming GeForce GTX 1660
First of all, let’s discuss the elephant in the room. The GTX 1660 does not support real-time raytracing or Deep Learning Systems because it does not possess RT cores or tensor cores. However, the GTX 1660 does have Variable Rate Shading capability.
A feature called Adaptive Shading (aka variable-rate shading) that was introduced by Turing is still available on the GTX 1660. It can be used to create content-aware shaders (CAS) as well as motion-adaptive shading (MAS). As a result, CAS minimizes the repetition of shading details in a scene in a pursuit to increase detail where it really matters. As a result, MAS minimizes the repetition of shading details to maximize performance as opposed to minimizing it in favor of performance to increase details.
Read more on Best graphics cards under 100
Thermals: Keeping It Cool and (Mostly) Hushed
If you had decent headphones on, and the fan noise was at a lower volume, you probably wouldn’t notice it. If you have speakers, you may need to turn them up just a bit more.
Despite that, we deliberately turned up MSI Afterburner to 100 percent for the purpose of testing. The fans never reached that point when set on “auto” in any of our game tests or 3DMark tests. In other words, this problem is unlikely to occur unless you specifically try to cool the card down (for example, during the summer).
Throughout the thermal test, we tracked the progress of the temperatures using 3DMark Fire Strike Ultra Stress Test. Almost all of the heat dispersion was generated by the fan. The fan was vertically positioned closest to the back of the case. In spite of the fact that this isn’t ideal (we would’ve preferred a little heat to dump directly out the back), it helps you determine how best to redirect heat based on the style of your case.
Read more on best GPU under 200
1080p Benchmark Performance Thoughts
In 1920 x 1080 on MSI’s new GeForce GTX 1660 SUPER GAMING X, the GTX 1660 SUPER and GTX 1660 Ti are neck and neck. They show amazing performance right out of the box.
In Shadow of War at 1080p, we average 89 frames per second. In the GTX 1660 Ti averages 87 frames per second and the RTX 2060 averages 100 frames per second. Due to its faster memory bandwidth courtesy of GDDR6, NVIDIA’s GTX 1660 SUPER zooms ahead of AMD’s RX 570/580/590 in Shadow of War.
NVIDIA’s GeForce RTX 2060 SUPER has passed the magic 1080p 120FPS mark. This puts it far ahead of AMD’s Radeon RX 5700 and Radeon RX 5700 XT graphics cards. AMD’s Radeon RX 5700 displays 121FPS and 127FPS averages, respectively.
Read more on best graphics cards for warzone game
Also related: Best gpu with ryzen 9 5950x
Final Thoughts MSI Gaming GeForce GTX 1660
In terms of hardware and performance, the new NVIDIA GeForce GTX 1660 SUPER has a reputation. It is known as one of the most powerful GPUs on the market. The card comes with the NULL technology, ReShade filters and other new features.
There is no difference in performance between the GTX 1660 SUPER and the GeForce GTX 1660 Ti. In addition, depending on the specs we are looking at, it can vary between being a little better, a little worse, and a lot worse than that. The GTX 1660 SUPER offers the same 6GB of GDDR6 on-board as the regular GTX 1660 Ti. However, but its clock speeds are 14Gbps instead of 12Gbps.
Since the Turing GPU hasn’t been changed much, NVIDIA made some important changes. NVIDIA needed to differentiate the GTX 1660 SUPER from the GTX 1660 Ti. This means the GTX 1660 SUPER can shine in memory bandwidth intensive games. There are two fewer SMs on the GTX 1660 SUPER.
GeForce GTX 1660 has 6GB of VRAM. However, the new GTX 1660 SUPER provides a significant upgrade over the older (and still Turing-based) card. A GTX 1660 SUPER is equipped with 6GB GDDR5 VRAM. However, the VRAM is only clocked at 8Gbps. The GTX 1660 is clocked at 14Gbps. Therefore, the GTX 1660 SUPER boasts 192.1GB/sec of memory bandwidth, compared to a massive 336GB/sec on the GTX 1660. That’s a 75% increase in memory bandwidth.