When Nvidia announced their RTX 20-Series GPUs, it was thought that the GTX series was being replaced. With advanced architecture and specialized RT and Tensor cores, it was assumed the RTX newcomers would outperform GTX in every conceivable way, but the 1080, Nvidia’s flagship GTX GPU, isn’t going easy.
Through benchmark testing, it’s already been established that the GTX 1080 released in 2016 still offers a much higher performance for gaming than the best the 2019 16-Series GPUs have to offer, but then it got gamers thinking. If the original GTX king is destroying newer GTX units, how would it fare against the entry-level RTX GPU?
Well, we’ve done the research and put in the work, comparing these two fantastic GPUs in a number of categories, and we have the answer for you!
ASUS ROG STRIX GTX 1080
MSI GAMING GeForce RTX 2060 6GB
The fundamental fissure between these two GPUs is their architecture. The GTX series were all designed using Pascal microarchitecture, but the new RTX line has state of the art Turing microarchitecture. Generally speaking, Turing architecture is much more impressive. By allowing an independent integer data path to collaborate on command execution with the floating-point datapath – teamwork that was blocked in Pascal – each CUDA core can deliver twice the performance of Pascal cores. Still, let’s examine the details of each GPU’s architecture more precisely.
Built with GP104-400-A1 variant of Pascal microarchitecture, the 1080 has a gargantuan total of 7.2 billion transistors spread across 2560 CUDA cores and features 2560 shading units, 160 textural units, and 64 rendering units. It supports PCI-e 3 and DirectX 12, and has a base clock speed of 1607MHz and boosted rate of 1733MHz
The RTX 2060 is a product of TU106-200A-KA-A1 Turing microarchitecture, which is the most basic Turing format besides the ones found in GTX 16-Series GPUs. It encompasses 10.8 billion transistors across a notable fewer 1920 CUDA cores, but remember, they’re twice as productive as Pascal cores. What’s more, it boasts a smaller 12nm process size, 1920 shading units, 120 textural units, 48 rendering units, and also supports PCI-e 3 and DirectX 12. It’s base clock peaks at 1365 but boosts to 1680MHz
So at this juncture, it’s hard to tell which GPU will perform better. The GTX card has brawny architecture specs but outdated tech. The RTX card has the more modest numbers overall, but it’s more advanced. The 1080 has some pumped clock speeds, but we should also pay close attention to the 2060’s significant boost in transistors, as more transistors typically amount to more efficient performance.
While the GTX 1080 may have a higher thermal capacity, maxing out at 94°C, it too, runs much hotter than the energy-efficient RTX 2060 GPU. There’s not much between them in terms of power suck. The GTX pulls 180 watts to the RTX’s 160 watts. That’s nothing compared to some AMD GPUs that pulls well over the 200-watt mark, but because the 1080 is a blower and decidedly older, it’s bound to run hotter.
The 1080 commonly reaches beyond 80°C during gaming, but we wouldn’t recommend an aftermarket cooling system unless you’re sticking with it for the foreseeable future, or perhaps you’re going to be pushing it to its limits day in day out. In that instance, it might be worth equipping some liquid.
The RTX 2060 is an open-air GPU that is great for keeping its local temperature way below capacity with bolstered CFM airflow, but it relies on your case’s exhaust system to exorcise that devilish heat altogether. So, you don’t need any aftermarket cooling for the 2060 itself, but you may want to augment your case’s general exhaust matrix to protect your other hardware.
In regards to die size, the 1080 is a little smaller. The Pascal architecture is built on a 314mm2 foundation. The 2060’s foundation measures in at 445mm2.
In terms of overall size, however, the RTX GPU is a touch smaller, measuring 4.435” (H) x 9” (L) x 2-slot (W), as opposed to the GTX 1080’s dimensions of 4.376” (H) x 10.5” (L) x 2-slot (W). So, as you can see, the 2060 is only a fraction of bigger in terms of height, but a whole inch and a half shorter in length.
These minute differences may be fairly negligible for most gamers, but to those of you who pride yourself on your diminutive setups, the RTX will probably have a larger appeal.
It’s hard to say which of these cards performs better from specs alone. It’s all theoretical conjecture until you bring it down to brass tax do some real world comparative gaming, so without further ado…
1080 is the GTX card’s bread and butter. Don’t get us wrong, both cards perform exceptionally well at this resolution, achieving average frame rates far beyond the 200 mark in Counter Strike: Global Offensive, but you’d be hard pressed to find a title that the 1080 doesn’t clean up on. Tested on over 50 games, the GTX 1080 came out with a slim 2% victory.
1440p brings their average frame rates way beneath the 100fps mark, but even though both of their performances drops, the GTX 1080 steals an even larger win with 6% higher average frame rates. Overall the RTX mustered an average of 76.9fps, while the 1080 champ pulled an average of 81.3fps.
Both cards really start to struggle at 4K, falling beneath the 50fps threshold, but the 1080 continues to dominate, and in fact, with a 7% win, finishes proportionally stronger than in any other resolution.
The 1080 may have won the quick draw, claiming all three resolutions and most games with raw frames per second speed, but graphics quality is a slightly different arena, one in which the RTX 2060 has a huge advantage.
Turing architecture bestowed ray tracing capabilities upon the 2060 thanks to dedicated RT cores. Not every game supports ray tracing, but for the ones that do, the 2060 will provide hyper-realistic light-based graphics.
There is an Nvidia driver that allows you to enjoy this enhanced global illumination on GTX cards too, but you can expect their general performance to crumple under the workload.
These Gpus have starkly contrasting memory configurations. The 1080 features an 8GB GDDR5X memory with a 256-bit bus interface and 320GBps bandwidth. The 2060 rolls with just 6GB of GDDR6 memory with a 192-bit bus interface, and 336GBps bandwidth.
The biggest difference between these memory types other than that GDDR6 is designed for compatibility with stacked chip GPUs is that GDDR6 is much more energy efficient, but the way that Turing architecture structures memory functionality changes everything.
The 2060 may have smaller memory capacity, but unlike the individual memory caches of Pascal architecture, the SM memory path conglomerates texture caching, shared memory, and memory load caching into a single unit. This amounts to wider bandwidth, an unburdened L1 cache, and 14GBps memory speed compared to the Pascal-hearted 1080’s 10GBps
ASUS ROG STRIX GTX 1080
MSI GAMING GeForce RTX 2060 6GB
It’s a shocking and illuminating result. The 1080 is practically geriatric in tech-years, but continues to brush newer more expensive GPUs off like they’re nothing at all. With higher average frame rates, and an ever widening percentage gap with each bump in resolution, the GTX 1080 is a smoother GPU and a great choice for affordable competitive gaming. But not everyone is interested in milking as many fps as graphically possible…
Even if the GTX 1080 is still fighting the good fight, and is technically faster than this RTX unit, because of Turing’s ray tracing and Tensor cores, for titles that support them, the 2060 is going to offer a much more realistic and impressive graphical performance.