yeah. its going to be a tough fight. the 2000 series was a bit disappointing (first gen new tech generally is) but this looks like a proper monster.
when measuring up which to buy i usually put a bit of extra weight on the AMD side because of open source drivers and rooting for the underdog (though the latter may be shifted to intel whe i am looking at a GPU purchase again...poor intel). but everything still gets investigated.
But the RTX 3000 series has really given a great first impression. the 3090 is obscene and stupid for gaming (but i wants it!!). but the 3080 and 3070 look to be great packages. a lot of weight has been added to the 3080 side of the scales.
As for AMD it will be tough. RDNA 1 was a big step in the right direction.....huge improvement over the vega 64 in terms of efficiency. performs better, costs less to make, generates less heat, uses less power and there is no downside for gaming compared to V64. it was just uarch wins compared to where they were. the drivers really let it down but it was doing quite well at launch (this cant happen again for RDNA 2).
so how much more of a leap will they have made for RDNA2? will it scale well past 40CUs? how is the RTX performance? can they implement something like DLSS using the CUs rather than something like a tensor core? do they have something like a tensor core (everything at the moment points to no)? what is their answer for RTXIO (i mean..they have worked with MS on this for the XSX....the should have something in RDNA2 on the PC surely)?
the 3090 is not really a concern (though i would like to see them take a crack at it. 16GB of vram and everything RDNA2 can muster for 999....try to pull the rug from under it.) but taking the wind out of the 3080s and 3070s sails.....thats going to be very hard.
Log in to comment