Came for this and not disappointed. Isn't this like AMD's first thousand dollar card?
I've been chasing price > performance which to me has been the ##70 series. So for example I want to upgrade while my 2070 still commands $300 or more on the used market and upgrade to a 3070, which if a 6700XT was $399 I might make the switch. As it is... I might be shopping 3070's.... if they ever ship.
Can I also just say my computer BLAZES! My god this thing is fast. and this is on a 6th gen.... 6th!!! i7. The video card is just king. Takes so much pressure off the the processor. Still... astonishing... how fast it loads webpages. I love this thing.
Yeah once you have a certain level of CPU and start gaming at 1440p or 4k, the CPU doesn't matter so much like it would at 1080p gaming. My 4790k has done fine over the last 6 years (can't say the same for my 980 once I got a 1440p monitor though heh).
Personally I am not a fan of the 3070 in it's current iteration. To me the 8GB in 2020, almost 21 is planned obsolescence for higher resolution gaming. And while the FE cards might be $499 if you can find one, all the AIB models I saw for sale at my local Micro Center today were all $530 - $550. Remember, we're talking an xx70 class card here. If the 3070 was $399-$450 I would see it more favorably. It's overpriced.
Here's a breakdown of how you got less for more over time in this class, similar to the x80 tier.
- GTX 470 (2010): 1.28GB GDDR5 320-bit bus, GF100 chip (cut down highest end Fermi) - $349
- GTX 570 (2010): 1.28GB GDDR5 320-bit bus, GF110 chip (cut down highest end Fermi refresh) - $349
- GTX 670 (2012): 2GB GDDR5 256-bit bus, GK104 chip (cut down mid-range Kepler chip) - $400
- GTX 770 (2013): 2GB GDDR5 256-bit bus, GK104 chip (rebrand of the GTX 680) - $399
- GTX 970 (2014): 3.5+0.5GB GDDR5 224+32-bit bus, GM204 chip (cut down mid-range Maxwell chip) - $329 (see #9 below for clarification on mem config)
- GTX 1070 (2016): 8GB GDDR5 256-bit bus, GP204 chip (cut down mid-range Pascal chip) - $449 (Founder's Edition), $379+ (AIB models)
- GTX 1070 Ti (2017): 8GB GDDR5 256-bit bus, GP104 chip (functionally same mid-range Pascal chip with only one less SM than the GTX 1080 vs 5 on the 1070 - Released to beat the Vega 56) - $449
- RTX 2070 (2018): 8GB GDDR6 256-bit bus, TU106 chip (Not even the mid-range chip anymore, now shares same chip as xx60 class) - $599 (Founder's Edition), $499+ (AIB models)
- RTX 2070 Super (2019): 8GB GDDR6 256-bit bus, TU104 chip (Notice, this version has same mid-range Turing chip as 2080 and 2080 Super now, only cut down vs those chips. Released to combat 5700XT) - $499+
- RTX 3070 (2020): 8GB GDDR6 256-bit bus, GA104 chip (currently only GA104 card as the 3080 moved back up a tier in chip) - $499 (Founder Edition), $499+ (AIB models)
From this list we can extrapolate a few trends.
- Prior to the GTX 670, the x80 and x70 class cards always utilized the top end chips (i.e. x80 is full chip (or fullest version of chip) with the x70 existing as a slightly cut down version of said chip. I only went back as far as Fermi, but it is historically true prior as well. With Kepler, both products would retain these product numbers associated with high-end GeForce, as well as pricing, but move to be mid-range chips/cards within the stack.
- A further sign that the x70 and x80 class cards were now firmly a tier lower than their previous slots within the product lineup can be seen by the move towards a 256-bit memory interface. Slower and problematic with the memory controllers/logic used for these cards for higher resolution gaming or high-res textures and texture mods. I can attest to that first hand with my GTX 670 in 2012 and modded Skyrim struggling hard on this card.
- With Kepler, Maxwell, and Pascal, VRAM capacity doubled each generation, as did their x80, x80 Ti, Titan counterparts. However, the Pascal Titan X (referred to as the Xp) did not over it's Maxwell Titan X predecessor and so the 1080 Ti was segmented from this Titan by removing 1 memory chip which made it 11GB vs 12GB and removes 32-bits of the memory bus making it a 352-bit card vs 384-bit. Basically, it started getting weird here when it came to VRAM configs.
- With Turing, the 2070 actually moved down another tier in chip where instead of sharing the same mid-range chip as the 2080, it instead is a higher binned/unlocked chip of the 2060. Yet the price increased as well. Also of note is that the entire stack retained the same memory configuration as their Pascal predecessors, with the exception of the Titan RTX moving to 24GB. That said, they did move from GDDR5 to GDDR6 with this launch, which did increase the bandwidth. That said, you will find titles that will max VRAM utilization on these cards at 4k.
- The 2070 was officially discontinued with the launch of the 2070 Super which price-point replaced it and was launched along with the 2060 Super in response to the launch of RX 5000 series. Of note, this returned the 2070 Super to the same tier chip as the 2080/2080 Super, and functionally the 2070 Super was a marginally slower 2080.
- The 3070 releases and performance does make it effectively a 2080 Ti for $500, except there's a few caveats. It's still only 8GB of VRAM (3 less than the 2080 Ti) which is seen as not very future proof as we move into next gen consoles, and higher resolution gaming. To Nvidia's credit, it does stay within the x04 chip tier instead of trying to bump it down again.
- Founders Edition bunged everything up starting with Pascal. Anytime you see an AIB price, you can almost guarantee the cards will actually cost more, and they always did.
- Nvidia stagnated certain specs such as memory since Pascal.
- EDIT: Oh and the GTX 970 is infamous for it's memory config. It is a 4GB card, but 3.5GB is really the usable portion over a 224-bit bus. The last 0.5GB portion existed separately being significantly slower (7x slower) on it's own 32-bit bus, to the point games reaching the slow RAM after 3.5GB would take a significant performance hit once reaching aboe 3.5GB VRAM utilization. Class action lawsuit. GTX 970 purchasers received $30 for each card purchased as card was marketed as having a 4GB 256-bit config, same as GTX 980.
Turing was definitely the biggest offender, but like the x80's, over the years you are philosophically speaking receiving less for more as it exists within the product stack. The 2070 is the absolute BIGGEST offender on this front when compared to all other x70 class cards generation to generation. Compare current lineup with Fermi and prior and what you would see since Kepler is that Titans
were the x80 product. The x80 Ti's
were the x70's. Current x80's (with the exception of Ampere),
were the x60's. And x70's
were the x50's.
Now, obviously functionally you get more performance each gen, so I am not trying to diminish that. But you are paying more for lower in the product stack, and since Pascal, Nvidia definitely stagnated certain elements such as memory which would seem like planned obsolescence for any card being released at the cusp of 2021.