Kepler, Maxwell and Pascal performance over time

Thanks for this link, it's useful. I bought an RX470 in November and while color seemed great, there was always a weird thing with games in movement. This, paired to the fact that I didn't qualify for the Hitman promotion despite buying it in the acceptable window, led me to return it. Fast forward a few weeks, and I've been reading about frametimes and 0.1 percentile frame drops. Low and behold, I'm seeing that the RX 470 and 480 sometimes (more than I'm willing to accept) drop like whoa creating stutter at specific games (maybe the ones I happen to play, the one review I recall checking is Gamers Nexus GTX1060 3GB vs RX 480 4GB, where drops were easily 10-20fps less bad on Nvidia). Now I know that's exactly what was bothering me, as I'm back on my puny GTX 770 and movement seems way more fluid. That's a 3 year old card, and it's working very well - other than the 2GB VRAM is killing me with low-end textures being the only real possibility.

That realization, plus the link you just shared, have pushed me back to consider a 1060 after I upgrade the full system this March. Or, at that point, wait for the next batch of Nvidia cards. I'll keep an eye on Vega, but I've now learned that FPS is woefully inadequate to judge game performance. I'll surely be looking at both AMD/Nvidia's frametimes and frame pacing. I had no idea this would be such a big issue until I tried my first AMD card and I was not willing to put with the stuttering. (I'm sure there's many AMD cards that don't stutter, I'm not here to promote brand wars, but this was simply my experience).
 
I think all the confusion comes from what architectures developers target.
 
Thanks for this link, it's useful. I bought an RX470 in November and while color seemed great, there was always a weird thing with games in movement. This, paired to the fact that I didn't qualify for the Hitman promotion despite buying it in the acceptable window, led me to return it. Fast forward a few weeks, and I've been reading about frametimes and 0.1 percentile frame drops. Low and behold, I'm seeing that the RX 470 and 480 sometimes (more than I'm willing to accept) drop like whoa creating stutter at specific games (maybe the ones I happen to play, the one review I recall checking is Gamers Nexus GTX1060 3GB vs RX 480 4GB, where drops were easily 10-20fps less bad on Nvidia). Now I know that's exactly what was bothering me, as I'm back on my puny GTX 770 and movement seems way more fluid. That's a 3 year old card, and it's working very well - other than the 2GB VRAM is killing me with low-end textures being the only real possibility.

That realization, plus the link you just shared, have pushed me back to consider a 1060 after I upgrade the full system this March. Or, at that point, wait for the next batch of Nvidia cards. I'll keep an eye on Vega, but I've now learned that FPS is woefully inadequate to judge game performance. I'll surely be looking at both AMD/Nvidia's frametimes and frame pacing. I had no idea this would be such a big issue until I tried my first AMD card and I was not willing to put with the stuttering. (I'm sure there's many AMD cards that don't stutter, I'm not here to promote brand wars, but this was simply my experience).
To be clear I know of plenty of people with Polaris cards who don't suffer from this issue there's probably something about your configuration that causes it.
 
Seriously, what company would gimp their own cards to make new cards look better? What don't reviewers still compare to opponents cards too? Sheep go to great lengths to make things up......
 
oh and there's the ashes of the singularity improvementy of around ~8% i mentioned ages ago, came with the pascal drivers. More pronounced on kepler
 
Back
Top