As the orchestrated launch of the Turing GPU family and related video cards with NVIDIA continues to work well, we are back again this month with the next piece of the GeForce product stack. Last month was, of course, the GeForce GTX 1660 Ti; and if you know something about naming NVIDIA, then you know that NVIDIA never makes an independent Ti card. As a suffix that indicates higher performance, if there is a Ti card, there must also be a normal card. And today NVIDIA is delivering just that with the vanilla GeForce GTX 1660.
For the launch of the Turing family, NVIDIA followed a very simple down-to-top video card launch, and the current launch of GeForce GTX 1660 continues this scheme. With the GTX 1660 Ti arriving at the top end of the mainstream market with a price of $ 279, NVIDIA is now ready to launch its low-end $ 21
Looking at the new NVIDIA card, within the NVIDIA Turing GeForce product stacks the GTX The 1660 is essentially a reduced GTX 1660 Ti and serves as the GeForce GTX 1060 3GB version of this generation. Which means it's a card that uses the same GPU in a slightly reduced configuration – in this case the same TU116 introduced for the GTX 1660 Ti – while instead it makes a greater compromise in memory to bring the price of the card down. Gone is GDDR6 in favor of cheaper and more widely available GDDR5, and even better if you get 6GB complete.
Equally important, however, NVIDIA has (largely) halted the names of shenanigans for this generation by not using the memory capacity to indicate the overall performance of the GPU. Even though I'm not a fan of suffixes yet (since they tend to be unwittingly cut off), this situation is vastly better than the GTX 1060 naming system. So I'll credit NVIDIA for not having taken the same consumer-hostile decision twice below.
Set today's launch so, from a consumer perspective, the GTX 1660 could be the most important launch of the NVIDIA video card of the year. Being essentially the $ 200 video card from NVIDIA – with NVIDIA adding another $ 20 – we're entering NVIDIA's high-volume desktop products. These are the traditional cards that make up the bulk of NVIDIA's desktop shipments, so this is the card that will mark the pace of the mainstream market for the world. But first, let's see if it's good.
|Comparison of NVIDIA GeForce specifications|
|GTX 1660||GTX 1660 Ti||GTX 1060 3GB||GTX 1060 6GB|
|CUDA Cores  1408||1536||1152  1280|
|Core Clock||1530MHz||1500MHz||1506MHz||1506MHz  Boost Clock||1785MHz||1770MHz||1708MHz||1708MHz|
|Memory Clock||8Gbps GDDR5||12Gbps GDDR6||8Gbps GDDR5||8Gbps GDDR5 (X)|
|Memory bus width||192-bit||192-bit||192-bit||192-bit|
|VRAM||6 GB||6 GB||3 GB||6GB|
|Single Precision Perf.||5 TFLOPS||5.5 TFLOPS||3.9 TFLOPs||4.4 TFLOPs|
|"RTX-OPS"||N / A||N / A||N / A||N / A|
|TDP||120W||120W  120 W||120 W|
(284 mm2 )
|Production process||TSMC 12nm " FFN "||TSMC 12nm" FFN "||TSMC 16nm||TSMC 16nm|
|Launch date||14/03/2019||2/22 / 2019||18 / 08/2016||19/07/2016|
|Launch price||$ 219||$ 279||$ 199|| MSR P: $ 249
FE: $ 299
Immersing yourself in the specifications of the GeForce GTX 1660, what we find is a very familiar echo of the GTX 1660 Ti. NVIDIA obviously uses the same TU116 GPU as before, making the GTX 1660 the company's second TU116 card. However, rather than a fully enabled GPU, the TU116 is slightly reduced, eliminating 2 of the 24 SMs from the GPU, leaving 22 enabled for a total of 1408 CUDA cores.
Otherwise there are no other changes on the front of the GPU; in particular all the ROPs are enabled, and the clock has actually reached only a little, with the official boost clock rated for 1785 MHz. So, unlike the high-performance NVIDIA products, where the steps in GPU configurations are much larger, the GTX 1660 family keeps things close, giving NVIDIA a way to recover chips but not creating a large gap between them. As a result, on the card the GTX 1660 has 92% of its largest level of shading / texturing / throughput of the big brother's geometry and 101% of its ROP throughput. But don't be fooled; how the TU116 was cut sounds in the background compared to memory changes.
As I said at the beginning of this article, as before the GTX 1060 3GB, the modification of NVIDIA for this card is not the GPU, but rather the memory. For the GTX 1060 3GB, the change was throwing away half the memory, resulting in a 3GB GDDR5 card that I think today was a short-sighted decision. Fortunately, for the vanilla GTX 1660, NVIDIA is doing something different: swapping GDDR6 at the cutting edge for the real backbone of the video card industry that is GDDR5. GDDR5 is obviously not as fast as GDDR6 – and, in particular, the GTX 1660 loses much of its performance compared to GTX 1660 Ti, but still offers a good amount of memory bandwidth. And even better, it means that NVIDIA is sending the card with a more appropriate 6 GB of VRAM.
From the numbers then, the GTX 1660 comes with 6 GB of GDDR5, which is connected via a 192 bit memory bus and runs at 8 Gbps. In this case, this means that the GTX 1660 has exactly the same amount of bandwidth as the memory of GeForce GTX 1060 6GB and 3GB; therefore on a generational basis, NVIDIA needs to get more performance from the same amount of memory bandwidth.
Or, to put things in the context of the Turing generation, this is only 2/3 of the data transmission speed of the GTX 1660 Ti 12 Gbps GDDR6, so the reduction in bandwidth is significant, going from 288 GB / sec at 192 GB / s. However, as we will see here for the GTX 1660 – and as we have seen before with other video cards supplied with more memory speeds – the performance of the video card is much less than 1 to 1 with memory speed. So, although this means that the GTX 1660 falls behind the GTX 1660 Ti with a decent gap, the TU116 is not completely hindered by GDDR5.
Finally, to design a final parallel to the GTX 1060 3GB, like its predecessor NVIDIA is targeting the same 120W TDP. The GTX 1060 cards are all supplied with the same TDP, as well as for the GTX 1660 cards. The net result is that the Vanilla GTX 1660 is essentially a little less efficient than its Ti brother, offering less performance for the same energy consumption. And while NVIDIA's TDP choice is arbitrarily working, this gives the company an outlet for the marginal TU116 chips that are a little more attached to the juice (the best chips, of course, go into laptops). The result is that since the GTX 1660 has fewer SMs competing for power, it means it can increase slightly more, giving the card a higher average speed. This also helps keep the two GTX 1660 cards a little closer to performance than specified by the specs otherwise – at least when the GTX GTX 1660 is not tied to memory bandwidth.
Price, Product Positioning and Competition  In the Pascal generation, GTX 1060 3GB was NVIDIA's $ 199 "sweet spot" video card. However, like the rest of the new Turing generation, the GTX 1660 is subject to price inflation. The GTX 1660 Ti came to $ 279 – which was $ 30 higher than before – and similarly the GTX 1660 gets a price of $ 20 to $ 219. This means that even if NVIDIA is not hitting the weak point, the GTX 1660 is essentially their jack on a $ 200 video card.
Like the launch of last month's GTX 1660 Ti, the launch of the GTX 1660 is a pure virtual launch for NVIDIA and its on-board partners. which means that NVIDIA is not doing any kind of retail card here, and all the cards that hit the shelves are custom cards for sellers. In practice, expect most (if not all) of these cards to look like the GTX 1660 Ti cards that have just hit the market; with the same GPU running on the same TDP, it is fast and efficient for card manufacturers to reuse their projects, even with the change of memory types. TU116 is naturally compatible with pins, although I'm still not sure that GDDR6 and GDDR5 can use the same identical PCBs, since the new memory has a different pin number.
NVIDIA is calling today's launch as a difficult launch, although the truth is that I am not convinced it will be difficult enough to launch like last month's GTX 1660 Ti. In fact, we had to look at some on-board partners before we could guarantee a sample, as other card vendors didn't have samples available (thanks, EVGA!). So I have the distinct impression that the local offices and distributors were cutting them pretty close to receiving their cards before today's launch date. However, by the time this article is published, we should have a better idea of how well this launch is provided.
Looking at the product stacks, inside the NVIDIA product stack this card is set to replace the GTX 1060 3GB, with the previous GTX 1660 Ti doing the same with the GTX 1060 6GB. In practice, the channel was mostly out of 6 GB GTX 1060 cards, but any paper left has just been made obsolete for retail purposes, since the GTX 1660 is all round better at the same price or less . Meanwhile the remaining 3 GB GTX 1060 cards – which were mostly under $ 200 in the beginning – will soon meet the same fate.
Otherwise, NVIDIA is once again playing this bundle launch. There will be no game packages for any of the GTX 1660 series, so the value of the product is the value of the video card itself. Outgoing GTX 1060 cards instead qualify for NVIDIA's Fortnite package, although it is obvious that Fortnite is a free game to begin with.
Meanwhile in terms of the expected market, like the GTX 1660 Ti before it, NVIDIA is targeting the Vanilla GTX 1660 in the mainstream market, and is particularly keen on upgrading from the GTX 960, GTX 760, R9 380 and from other mainstream video cards from ~ $ 200 since before this decade. Turing cards were not a real generation upgrade compared to their predecessors Pascal and GTX 1660 is no different; the new card is about 28% faster than the GTX 1060 3GB it replaces.
Regarding NVIDIA's loyal opposition, the launch of the GTX 1660 will continue to put pressure on AMD's old Polaris video cards. The GTX 1660 is faster than all, including the faster RX 590, so AMD and its partners will have little choice but to cut prices. This means that AMD can try to place its cards as spoilers on the GTX 1660 – and the $ 219 RX 590 cards are already popping up – but they can't take NVIDIA in terms of performance or energy efficiency. AMD will have a bundle-in advantage, while they are continuing to offer their Raise the Game Bundle, which for the RX 590 is the complete, 3 game pack.