The highly anticipated GeForce RTX 20 series is here and with all the hype around this new generation and all the new technology to come with it.
One has to be quite skeptical about the silence from Nvidia’s end combined with little to nothing to show to warrant a pre-order. Followed by alleged shameful paid articles like the Tom’s Hardware “Just Buy it!” article.
Today we’re doing a review recap to determine whether or not these new cards were worth the wait.
Spoiler Alert: Not in the fucking slightest.
Official Performance Numbers? Gone for a reason.
Nvidia has taken a firm stance on this whole ray tracing charade, it was obvious during the announcement event as Nvidia refused to showcase any meaningful performance metrics instead just pumping up more RTX Bullshit.
Having no actual gaming performance figures at all raised some concern, leaving the consumers out in the dark for a whole month up leading up towards launch day.
However it wasn’t until just a few days before the review embargo was lifted that we finally got something…. an overview of the Turing architecture followed by more blank and meaningless claims like the one below.
So What’s the Performance Really Like?
2080 = 2X 1080? Nope.
According to most review outlets the RTX 2080 is around 30-40% faster than the GTX 1080 which is below the claim.
However in terms of pricing the RTX 2080 should be seen as a replacement for the 1080Ti with the 2080Ti taking the role of the Titan, the RTX 2080 compared to a Founder’s Edition 1080Ti is sort of a hit or miss, but overall the 1080Ti wins out in terms of longevity, price and performance.
As for the RTX 2080 Ti, it’s ~30% faster than the GTX 1080 Ti.
Where is the “2X 1080” performance you may ask? well it was achieved with DLSS (Deep Learning Super Sampling) which basically reduces image quality in favor of perfomance and that’s the short version that you need to know.
Nvidia desperately needs to make their RTX cards look better and are willing to do so by reducing performance of any non RTX cards.
Nvidia is rather inconsistent when it comes to pricing, sometimes they charge you up the ass and sometimes they go the “gamer’s champion” route.
Can you guess which way did they go this time? well.. it’s the crazy route.
The prices for this generation are the most expensive we’ve ever seen on each segment.
They are doing this as they are planning to sell Pascal alongside Turing.
Nvidia are charging $1200 for a Titan masquerading as the 2080Ti, $800 for the 2080 and $600 for the 2070. Once more, the highest we’ve ever seen from each segment.
But if only with the 2080 that the performance actually made the price actually worthwhile… sad to say that this won’t be the case as the 1080Ti performs better in most cases (stock), has more VRAM and costs over $100 less.
A New Controversy?
It seems that Nvidia has been quite sneaky as news broke just days before the embargo was lifted that Nvidia will have two variants of each SKU, An “A” variant and a regular non “A” variant.
The difference being that the A variant is binned for overclocking and is allowed to be pre-overclocked by add in board partners while the non A variant isn’t and AIB partners are not allowed to pre-overclock it.
Here is an example
We can see the RTX 2080 Ti die the TU102 and next to the code name we see 300(A) so in retrospect this chip would be the OC binned variant that can be pre-overclocked by AIBs
The RTX 2070 performance is still unknown to this day, with a certain few people hoping that it would challenge the 1080Ti and with a price tag of $600 you might just assume as much right? Considering the 2080 is shown to be an inferior competitor to the 1080Ti while costing more it’s speculated that the RTX 2070 would only trade blows with the GTX 1080 and for a price tag like that, it’s downright pathetic.
So far we have concluded that the RTX 2080 is overpriced and under powered by the cheaper GTX 1080Ti despite coming to us with Nvidia’s custom 12nm node in collaboration with TSMC, the die sizes for Turing are monstrous and despite TU104 having a much larger die size than GP102 the performance is lackluster and adding on GDDR6 isn’t helping.
Not a good start, though I’m more than confident that the performance differences between the two will sway in favor of the RTX 2080 with gimmicks like DLSS and actual RTX which makes the pill all that much easier to swallow. I honestly feel bad for those who were suckered into pre-ordering it.
I have better hopes for the RTX 2070 but that’s not saying much. I’m expecting performance equaling to that of the GTX 1080 but taking into consideration the die size difference between TU106 and GP104, Turing is massive. The RTX 2070 features the die we’d expect to come on a X60 product and the lack of NVLink support strengthens this. While performing similar the added cost does not make the RTX 2070 worthwhile at all.
The only card worthwhile from this series would be the RTX 2080Ti. The golden child of the lineup for sure, even though the price tag will for sure turn most people off one cannot deny the performance this card is something to behold.
Overall, the Turing launch was a complete utter failure. Being that the only cards that actually offer some value are either previous generation hardware or a $1200 Titan passed off as a Ti.
Consumers would be wise to avoid anything aside from the 2080Ti at all costs.