Graphic Card, nvidia

Review of the Nvidia GeForce GTX 1660 Super.

Pros

  • Good 1080p performance
  • Better value than 1660 Ti
  • Proven Turing architecture

Cons

  • Missing RTX features
  • Crowded price point

Earlier this year, Nvidia introduced its new GTX 16-series, which includes the GTX 1660 Ti, GTX 1660, and GTX 1650, aiming to establish itself among the leading graphics cards. The absence of ray tracing capabilities marks a notable shift from the previous year’s RTX 20-series, yet this decision appears logical for the mid-range and budget segments. When Nvidia updated the RTX lineup with the Super variants, I did not anticipate this expansion into the GTX series; however, the GTX 1660 Super has indeed emerged. It utilizes the same Turing architecture and TU116 GPU, which leaves me somewhat perplexed. I suppose this is all part of the competitive landscape.

While I do not intend to imply that the card is inferior, it does exhibit significant overlap with the GTX 1660 Ti in various aspects. It retains the GPU core counts and clock speeds characteristic of the standard GTX 1660, yet it is now equipped with GDDR6 memory—previously a key distinction between the 1660 and the 1660 Ti. Interestingly, the GDDR6 memory operates at a speed of 14Gbps, surpassing the 12Gbps rate found in the 1660 Ti. This situation presents a peculiar scenario where a lower-tier card can be perceived as both superior and inferior to a higher-tier counterpart. On a positive note, the pricing is more accessible, leading us to the official specifications overview.

The primary specifications are as expected. It features the same TU116 GPU, retaining 22 SMs enabled out of a possible 24, now paired with 14Gbps GDDR6 memory. I previously raised concerns regarding the absence of 14Gbps GDDR6 in the 1660 Ti upon its release, presuming it was largely unnecessary. However, it would have been more logical to equip that model with the faster VRAM and reserve the 12Gbps for the 1660 Super. Nevertheless, this was not the outcome, resulting in an increasingly congested selection of Nvidia’s budget to mid-range offerings.

It is essential to understand that the 1660 Super, along with the 1650 Super, is designed to complement the existing GTX 16-series GPUs rather than to replace any of them. Consequently, Nvidia now offers five models within the GTX 16-series, an increase from the previous three. In the accompanying table, I have included details regarding the GTX 1650 Super, as well as the 2060 and 2060 Super, to illustrate the enhancements available in the next tier above the GTX series. Additionally, it is noteworthy that the 1650 now utilizes the TU116 chip; this change was necessary to incorporate the upgraded NVENC hardware, which was not present in the TU117.

The introduction of the new GTX Super cards is undoubtedly motivated by AMD’s recent activities. AMD has just released the OEM version of the Radeon RX 5500, with consumer variants anticipated to follow soon. Coupled with the decline in prices of the existing RX 590 cards over the last six months, it appears that Nvidia is seeking to enhance the competitive positioning of its products against AMD’s current and forthcoming models. This is my interpretation of the situation; Nvidia would likely assert that this strategy was part of their original plan.

Another noteworthy feature of the GTX 16-series is its absence of hardware ray tracing capabilities. While there is driver support for DirectX Raytracing (DXR), performance typically does not exceed 30 frames per second at 1080p with the GTX 1660 models. Given that the RTX 2060 is now available starting at approximately, with some sales reducing the price to as low as, the GTX 1660 Ti faces significant competition. I would generally recommend that gamers consider upgrading to the RTX 2060 instead of the 1660 Ti, as the price difference is only about. Additionally, with the GTX 1660 Super priced, the cost difference is substantial enough to make the Super a compelling choice.

The issue of the previous generation graphics cards persists, alongside the current availability of the 1660 models. Most of Nvidia’s GTX 10-series components are now predominantly found in the second-hand market. For individuals who already possess a GTX 1070, 1070 Ti, or 1080, there is little incentive to downgrade to the 1660 Super. This sentiment is similarly applicable to AMD graphics cards, particularly those at the Vega 56 level or above. Consequently, the mid-range offerings of the current generation find themselves in competition with both the previous generation’s mid-range and high-end GPUs, often failing to outperform the latter.

Currently, the official prices for the majority of Nvidia’s Turing graphics cards are approximately 10 percent higher than the retail prices. Consequently, a 1660 Super priced the least expensive GTX 1660 Ti. With Black Friday sales approaching, it is anticipated that some GTX 1660 Ti models may actually be priced lower than the newly introduced GTX 1660 Super. It should be evident that the 1660 Ti is typically the superior option when prices are relatively comparable. Furthermore, if there are concerns regarding the 12Gbps GDDR6 memory in comparison to the 14Gbps on the Super, rest assured—the 1660 Ti cards I have evaluated are capable of overclocking the VRAM to 14Gbps, and frequently even 15Gbps.

Introducing the EVGA GTX 1660 Super Ultra SC.

This establishes the foundation for our benchmarks and evaluation of the GTX 1660 Super. For this assessment, Nvidia provided the EVGA GTX 1660 Super Ultra SC, which features a slight factory overclock. Specifically, it boasts a boost clock of 1835MHz, in contrast to the reference clock of 1785MHz. This enhancement results in a performance increase of a few percentage points in real-world scenarios; although the difference may not be readily apparent during gaming, it will certainly be reflected in the benchmark results.

The design of this EVGA model is quite conventional, featuring dual cooling fans and a modest heatsink. With a thermal design power (TDP) of 125W, there is no necessity for an elaborate cooling system, and increasing the price would contradict the essence of a mid-range product. Rather than investing an additional $20-$50 for enhanced cooling and a higher factory overclock, it is more prudent to upgrade to the next performance level, which in this instance is the GTX 1660 Ti. However, it is worth noting that the card is not particularly silent, registering 60 dB(A) under load from a distance of 5 cm, although this allows the GPU to maintain a relatively cool maximum core temperature of 74C.

In a comparative analysis, the Sapphire RX 5700 Pulse recorded a sound level of only 45 dB(A) under identical testing conditions, while the RX 5700 XT reference card reached 57 dB(A). Additionally, the 5700 XT operates at approximately 10 degrees Celsius higher, stabilizing at 84 degrees Celsius. The elevated noise levels can be attributed to the fact that EVGA has opted for a less substantial heatsink, depending primarily on fan speeds and airflow for thermal management. Despite the 1660 Super consuming less than half the power of a 2080 Ti, its smaller heatsink and absence of heatpipes result in a notable amount of noise generation.

Configuration of the Test Environment

The majority of the other graphics cards are reference clocked models; however, the RTX 2070, 2080, and 2080 Ti are Founders Edition cards that feature a factory overclock of 90MHz. Similarly, AMD’s RX 590, 580, and 570 cards are also factory overclocked. Nevertheless, I typically do not regard performance differences of less than 5 percent as significant when making comparisons.

I am still utilizing my standard GPU testbed, which I established approximately two years ago. It features a Core i7-8700K processor overclocked to 5.0GHz, ensuring that the CPU does not become a limiting factor during testing. While there are occasions when an i7-9700K or i9-9900K may exhibit slightly improved performance in gaming scenarios, the variance is generally minimal—often less than a few percentage points, particularly at resolutions of 1440p and 4K. This discrepancy is even less pronounced when evaluating mid-range graphics options such as the 1660 Super. Additionally, the system is equipped with high-speed DDR4-3200 CL14 G.Skill memory and SSD storage, further ensuring that the GPU remains the primary determinant of performance.

I am currently utilizing Nvidia’s pre-release 441.07 drivers on the GTX 1660 Super and have conducted spot checks on various other Nvidia graphics cards to confirm the absence of any irregularities. The majority of the findings are consistent; however, several games, notably Forza Horizon 4, necessitated retesting with the most recent drivers. The results for the AMD GPUs are based on version 19.9.2, with the slightly more recent 19.10.1 drivers applied in The Outer Worlds.

The gaming performance of the GeForce GTX 1660 Super.

In alignment with my established practice of incorporating new games into the testing suite as they are released and deemed significant, I have revised the list following the Sapphire RX 5700 review. Notably, I have removed Fortnite and included The Outer Worlds, primarily due to the frequent changes in Fortnite, which render the creation and upkeep of a reliable benchmark exceedingly challenging. This update offers insights into the performance of both recent titles and some well-known, albeit slightly older, games, with the most dated title currently on the list being Total War: Warhammer 2, released in late 2017.

The eleven games under evaluation comprise a balanced selection of titles endorsed by both AMD and Nvidia, with a current predominance of AMD-supported games. In most instances, DirectX 12 is employed when accessible; however, I assessed Borderlands 3 using DX11 mode, as well as Total War: Warhammer 2 in DX11 on Nvidia graphics cards. This choice was made due to the superior performance of the DX11 configuration in both scenarios.

Each graphics card undergoes testing across four configurations: 1080p medium or its equivalent, as well as 1080p, 1440p, and 4K ultra, unless specified otherwise. The charts presented below include the 4K gaming results primarily to illustrate the performance of other GPUs, as it is unlikely that a consumer purchasing a GTX 1660 Super would consider 4K ultra gaming. Each setting is evaluated multiple times to guarantee the reliability of the results, and the highest score is utilized. The minimum frames per second (FPS) is determined by aggregating all frame times that exceed the 97th percentile and dividing this total by the number of frames, thereby providing a more accurate depiction of the lower performance spectrum, rather than focusing solely on the single lowest framerate recorded during a benchmark test.

The following presents the gaming performance outcomes for the EVGA GTX 1660 Super SC Ultra.

It is likely not unexpected that the GTX 1660 Super demonstrates slightly lower performance compared to the GTX 1660 Ti across nearly all tests conducted. On average, the 1660 Super is approximately 3-5 percent slower than the 1660 Ti, while it outperforms the standard GTX 1660 by about 15-20 percent. Despite the enhanced GDDR6 memory, the new model did not secure any advantages, which is understandable given its 9 percent shortfall in GPU cores.

In practical terms, the real-world performance between the 1660 Ti and the 1660 Super is nearly indistinguishable. Without conducting benchmarks, it is unlikely that anyone would notice a mere 5 percent difference, and adjusting a few settings slightly could easily compensate for that shortfall. While a significantly overclocked 1660 Ti may demonstrate a more pronounced advantage over a standard 1660 Super, a substantially overclocked 1660 Super would likely outperform a 1660 Ti operating at reference clock speeds.

A more relevant inquiry pertains to the rationale behind Nvidia’s release of the 1660 Super. Is it a response to the competitive landscape posed by AMD’s RX 5500, or perhaps a reaction to the reduced pricing of the RX 590? The answer remains uncertain. It is also plausible that Nvidia has secured GDDR6 memory at costs comparable to GDDR5, allowing for a transition in memory technology without adversely affecting profit margins. Regardless of the underlying motivations, it is evident that the 1660 Super offers superior value. Priced officially at  less than the 1660 Ti—an 18 percent reduction—its performance is only marginally lower, within a 5 percent range.

The GTX 1660 Super may not secure the highest position in the GPU value rankings, but it is certainly competitive. It is likely that both AMD and Nvidia will welcome the eventual depletion of the RX 570 stock, as it has been available at discounted prices for the majority of the past year. From the standpoint of an entire system, the 1660 Super and 1660 Ti are largely equivalent, as the expenses associated with other components generally surpass those of the GPU.

These value charts are unable to capture the entirety of potential buyers’ preferences, as many individuals prioritize raw performance over price. This explains why certain gamers opt for the RTX 2080 Ti, despite its low ranking on the GPU value charts and its mid-tier position on the system value charts. I advise gamers to first determine their desired resolution and settings, as well as the level of performance they seek. Only after establishing these criteria should they consult the value charts, which will then provide more relevant insights. While the RX 570 may appear to offer excellent value, I personally prefer a card capable of handling at least 1440p with high to ultra settings.

GeForce GTX 1660 Super: comparable performance at a lower cost.

I must confess that when the rumors began to emerge several months ago regarding a ‘Super’ refresh of the 1660 and 1650, I was skeptical. The GTX 16-series had only been released approximately six months prior, and the addition of GDDR6 to the GTX 1660 essentially transforms it into a 1660 Ti. While I was correct in my assessment of the performance, I misjudged the validity of the rumors. However, I remain unconvinced by Nvidia’s Super branding. It strikes me as something a child might invent. “Look, Dad, I just created a more powerful robot. I named it the super robot!” Children may find it appealing, or perhaps I am simply aging and have lost my connection to the excitement surrounding superheroes.

It is essential to evaluate the positioning of AMD’s RX 5500 graphics cards in terms of both pricing and performance. Based on specifications and preliminary performance assessments, it appears that the GTX 1650 Super will compete with the RX 5500, which consequently renders the GeForce GTX 1660 Super somewhat unusual in this context.

It is important to evaluate whether a graphics card priced at will meet all your requirements. This card offers performance comparable to that of the GTX 1070, which was released three years ago and typically retailed for  prior to the fluctuations in pricing caused by the cryptocurrency boom. While the GTX 1070 remains adequate for 1080p gaming, the impending release of next-generation consoles featuring ray tracing capabilities suggests that the RTX 2060 and RTX 2060 Super may represent more prudent long-term investments.

While a price difference may seem negligible, especially in the high-end and extreme markets, it becomes significant in mainstream and budget builds where every dollar matters. In this context, the GTX 1660 Super stands out as a highly recommended option. It offers performance nearly on par with its more expensive counterpart while being more affordable, and it surpasses any AMD offerings in the same price bracket, boasting approximately 20 percent greater performance than the RX 590. Additionally, it achieves this superior performance with lower power consumption. For those looking to upgrade their graphics card without exceeding a budget and currently using a GPU that is several generations outdated, the GTX 1660 Super is an excellent choice.