After a break following the launch of NVIDIA's GeForce RTX 20 Series Super Cards and AMD's new RD Architecture based Radeon RX 5000 series, the graphics card market is recovering in time for critical holiday shopping. AMD wants to fill the main market share and is already shipping its Radeon RX 5500 series cards to OEMs – retailing is coming soon. In the meantime, NVIDIA is developing its own devices for the market.
The real hardware for this holiday is NVIDIA, which announces its GeForce GTX 1660 Super. This is a relatively small, but interesting, review of the GTX 1660 family that adds a 1660-SKU (vanilla) with faster GDDR6 memory for improved performance. Along with the GeForce GTX 1650 Super (delivered in late November), these two maps are the backbone of NVIDIA's efforts to end the year. And while the other GTX 1
The GTX 1660 Super is the fourth super card NVIDIA brings to the market. It's a bit harder to read than the high-end cards that hit the market last summer. In terms of performance, the card is just short of the GTX 1660 Ti and is priced just north of the GTX 1660 (vanilla). But unlike the other supercards, this introduction is not the same kind of transparent price shuffle as the previous introduction. NVIDIA is not issuing any cards this time, and the shelves will be shared by all three GTX 1660 Series cards over the coming months. Still, this latest card has at least the same basic role as a mid-range kicker – even though the original GTX 1660 was released just seven months ago.
|NVIDIA GeForce GTX 1660 Series|
|GTX 1660 Ti|| Full TU116
with 12 Gbps GDDR6
| GTX 1660 Super (New) [19659009SuperTU116cutback
with 14 Gbps GDDR6
|GTX 1660|| TU116 cutback
at 8 Gb / s GDDR5
Here is often the placeholder factor case is AMD. The company has already announced some of its RX 5500 products. However, it is not clear what to expect from the charts in terms of performance, especially since AMD has defined them in the lower price segment compared to the NVIDIA GTX 1650. So, at first glance, the GTX 1660 Super seems to be an NVIDIA counterpart to the upcoming AMD. Outside the test labs of AMD and NVIDIA, it remains to be seen where things end.
However, it is clear that this may well be an actual price cut for the NVIDIA GeForce GTX 1660 Ti. The GTX 1660 Super is expected to be released for $ 50 less than the fastest GTX 1660 card from NVIDIA and is at 96 % of their performance only slightly slower than the GTX 1660 Ti – just outside our regular run-to-run variants. When NVIDIA reduced the GTX 1660 Ti to create the original GTX 1660, GDDR6 was replaced with a cheaper GDDR5. Now that they've effectively undone this, and much more besides, the majority of the power that NVIDIA has abolished for the GTX 1660 is due to the cheaper supercard. The result is that while the GTX 1660 Super does not offer anything in terms of performance that we have not seen before, it does lower this Turing performance to a lower price. NVIDIA hopes this will be more enticing for GTX's 10- and 9-Series owners, who were deterred by the relatively high prices of the latest cards.
|NVIDIA GeForce specification comparison|
|GTX 1660 Ti||GTX 1660 Super||GTX 1660||GTX 1060 6 GB|
|CUDA cores||1536||1408||1408 [ROS||48||48||48||48|
|Core Clock||1500 MHz||1530 MHz||1530 MHz||1506 MHz  Boost Clock||1770 MHz||1785 MHz||1785 MHz||Memory||1708 MHz [19659025D 14 Gb / s GDDR6||8 Gb / s GDDR5||8Gb / s GDDR5 (X)|
|Memory Bus Width||192-bit||192-bit|
|VRAM||6GB||6GB  6 GB||6 GB|
|Single Precision Perf.||5.5 TFLOPS||5 TFLOPS||5 TFLOPS [1965902744TFLOPs|
|TGP||120 W||125 W||120 W||120 W|
|Manufacturing process||TSMC 12nm "FFN"||TSMC 12nm "FFN"||TSMC 12nm "FFN"||TSMC 16nm|
|Start Price||279 USD||] $ 229||$ 219|| MSRP: $ 249
FE: $ 299
What exactly is the GeForce GTX 1660 Super? In short, take a GTX 1660 (vanilla), replace the 8 Gb / s GDDR5 with the 14 Gb / s GDDR6, select the TDP a few watts high to meet the increased energy needs, and you have a GTX 1660 Super.
On the GPU front nothing has changed compared to the GTX 1660. It is the same TU116 GPU in the same partially disabled configuration with the same clock rates. To be honest, I'm not even sure if NVIDIA's GPU binning operations for this product have changed or if partners have plugged TU116-300 GPUs into a GDDR6-wired circuit board. In both cases, performance on the paper did not change over the GTX 1660, as the GPU itself can handle as many pixels, texels and data elements as before.
Instead, memory support has changed, and here is the vast majority of the performance enhancements of the GTX 1660 Super. While the original GTX 1660 was shipped with 8Gb / s GDDR5 to cut costs and differentiate / segment the card from the full GTX 1660 Ti, the GTX 1660 Super has all the memory bandwidth it can handle. NVIDIA has provided it with 6 GB of 14 Gb / s GDDR6, currently the fastest available memory. This gives the GTX 1660 Super a memory bandwidth of 336 GB / s, 75% more bandwidth than the original GTX 1660 or the same as the RTX 2060. And this enormous bandwidth is almost exclusively responsible for the 12% performance increases we see The original card.
In fact, the GTX 1660 Super has more memory bandwidth than the high-end GTX 1660 Ti. This card comes with a slightly slower (and cheaper) 12 Gb / s GDDR6, giving the GTX 1660 Super a bandwidth advantage, which further negates the low lead of the GTX 1660 Ti. At present, GDDR6 is almost overwhelmed by GDDR6, which is why the GTX 1660 Super does not overtake the GTX 1660 Ti despite its advantage. At this performance, the extra memory bandwidth is not nearly as important as the additional two SMs the GTX 1660 Ti receives.
Finally, as a member of the GTX 1660, the newest card receives a very similar but not quite identical TGP of 125W. This is a small 5W increase over the GTX 1660 and 1660 Ti (as well as the GTX 1060/960 families) and seems to be mainly here to cope with the increased power requirements of shipping with fast GDDR6 memory – GTX 1660 Super would be different. Compared to the original GTX 1660, this is a small change Board vendors should easily be able to meet the increased power requirements with their existing GTX 1660 Ti designs, so that GTX 1660 Super Cards can be easily built with current components. The only drawback in this case is that the higher TDP and lower power compared to the GTX 1660 Ti mean that the GTX 1660 Super is not as energy-efficient as the card, which otherwise makes it redundant.
The chaotic state of the NVIDIA product stack  If NVIDIA released many GeForce graphics cards last year, it's because they did. In the 13 months since the introduction of the Turing architecture and the GeForce RTX 20 card series, NVIDIA has now launched 12 different GeForce SKUs. And even if we throw away the RTX 2080 and RTX 2070, which have been officially discontinued but are still on some store shelves, the current NVIDIA product stack is still 10 cards deep.
|NVIDIA GeForce 20/16 Series (Turing) Product Stack  RTX 20 Series||GTX 16 Series|
|RTX 2080 Ti||GTX 1660 Ti|
|RTX 2080 Super||GTX 1660 Super|
|RTX 2080 (discontinued)||GTX 1660  RTX 2070 Super||GTX 1650 Super|
|RTX 2070 (Discontinued)||GTX 1650|
|RTX 2060 Super|| RTX 2060|
Compared to this, the GeForce The product stack of the series was only 8 cards deep at its most messy point, and that was after the GTX 1070 Ti was introduced 16 months later.
To be fair to NVIDIA, this is partly because the price of the GeForce product stack is higher than before, especially with the $ 1100 RTX 2080 Ti. Nevertheless, NVIDIA has brought more cards to market faster than anything Time in the last half decade. In particular, the supercards have contributed significantly to this figure as NVIDIA does not deliver a mid-generation upgrade to its product line less than a year after its launch. Compared to the GTX 9 and GTX 10 series, where NVIDIA went through different product cycles for two years, with very little filler introduced in between, the RTX 20 and GTX 16 series have essentially led to annual card introductions to one Period of two years were designed GPU cycles.
Ultimately, this means that there are a very large number of NVIDA graphics card options on the market, with NVIDIA offering 5 SKUs under $ 300. Keep in mind that the board vendors are running their own factory overclocks for their premium cards, and we're looking for a GeForce for every price. As a sales strategy, market saturation is definitely a viable way to maximize sales – make sure that there is always a card available at a price that someone is willing to spend – but this confuses NVIDIA's product stack a bit, especially now with three different GTX 1660 cards.
Product Positioning and Competition
Let's talk about the hardware behind the launch today when switching. Since the new GTX 1660 Super cards are a mid-range kicker on a mainstream card, there are no reference cards to talk about here. Instead, it is a pure virtual market launch where all the cards on the market are partner designs.
At this point, partners have been shipping TU116-based maps for more than 6 months so they can improve their abilities in the GTX 1660 family of designs. This means they are up-to-date with their existing card designs for the GTX 1660 Super. At this point I have not seen any new designs specifically for the super card and do not expect any. So these cards will look and work the same as the GTX 1660 cards already on the market.
Unsurprisingly, the GTX 1660 Super occupies the same general slot in the NVIDIA product stack as the simpler GTX 1660 cards. This means that NVIDIA addresses the same audience as before for both system builders and upgrades. The fact that the GTX 1660 Super performs better for the price is definitely an improvement, and it's likely to make sure that NVIDIA is always one step ahead of AMD cards.
Upgraders, however, are likely to be more demanding. In terms of performance, Turing is not a generation ahead of the Pascal-based GTX 10, and the GTX 1660 Super does not change that. While the GTX 1660 Super is about 35-40% faster than a GTX 1060 6GB, it will be much more attractive for GTX 960 owners and the like as the new card offers more than twice the performance.
I mentioned at the beginning of this article that NVIDIA is releasing the new GTX 1660 Super Card today, but they will not be retiring any of their other cards. On paper, the GTX 1660 Super is positioned very closely between the GTX 1660 Ti and the GTX 1660 (vanilla).
At current prices, the GTX 1660 Super is $ 50 cheaper than the GTX 1660 Ti for $ 229. And since the card is only about 4% behind the top 1660 card from NVIDIA, I would argue that the GTX 1660 Ti has become obsolete at this time. The GTX 1660 Ti is not far enough ahead for valuable results. The next real step after the GTX 1660 Super is the GTX 2060, which is 22% faster but about $ 100 more expensive. Otherwise, the vanilla GTX 1660 from the other direction can still be found regularly at $ 219. With a price difference of just $ 10 to the GTX 1660 Super, the super card is currently the obvious choice as 12% more power is worth $ 10 more. Despite all this, NVIDIA insists that she and her partners continue to sell the Vanilla GTX 1660. So it'll be interesting to see what NVIDIA is actually doing here, both in terms of card supply and pricing. The GTX 1660 must fall in price to have a future. and if that happens, it could make for a decent spoiler under $ 200.
It's also worth noting that NVIDIA does not make any bundles or promotions for GTX 16 Series cards, so you see what you get with the new cards.
After all, the competitive situation for the GTX 1660 Super is mixed. AMD is still doing what they can, in which case the Radeon RX 590 is the closest competitor to the GTX 1660 Super. The admittedly awkward Polaris late-cycle card was originally designed to outperform the GTX 1060 6GB (throwing energy-efficiency out the window), which it undoubtedly did. However, this means that the GTX 1660 Super is superior by around 20% (and a lot less power), which is why AMD uses the card instead to secure its claim to the $ 199 price. Buyers who can not exceed $ 200 have the option. Currently, however, the GTX 1660 Super is not exposed to much competition.