GURU 3D latency tests between GTX1060 and RX480

May 11, 2008
19,300
1,129
126
I have been reading about this, but what i find odd is that with one scene test, The data does not fit in the graph. Why would they do that with this single test. Is there something interesting that they kind of blanked out by limiting the graph scale ?


http://www.guru3d.com/articles-pages/radeon-rx-480-vs-gtx-1060-fcat-frametime-analysis-review,7.html

index.php


I do have to say, in all tests, both cards perform almost identical (7 directx titles + 4 directx 12 titles)

What do you guys and gals think ?

Complete article.

http://www.guru3d.com/articles_pages/radeon_rx_480_vs_gtx_1060_fcat_frametime_analysis_review,1.html
 
Feb 19, 2009
10,457
10
76
Hitman's benchmark changes scene often, thus the spike. It is not in-game stutter. I play the game myself, it's all smooth.
 
May 11, 2008
19,300
1,129
126
Hitman's benchmark changes scene often, thus the spike. It is not in-game stutter. I play the game myself, it's all smooth.

Yeah, that is what guru3d mentioned as well. That the spikes come from scene changes. I just think it is weird to show a graph where the relevant information falls outside of the scale.
 

greatnoob

Senior member
Jan 6, 2014
968
395
136
Yeah, that is what guru3d mentioned as well. That the spikes come from scene changes. I just think it is weird to show a graph where the relevant information falls outside of the scale.

Anything above 33.33ms is not relevant because framerate will be < 30fps, so, unplayable regardless of 35ms, 50ms or 100ms. Setting the ceiling at 50ms shows all relevant data.
 

Bacon1

Diamond Member
Feb 14, 2016
3,430
1,018
91
Heh they need to rename their charts... Scene frame time in milliseconds. I dunno about you but >10 second average frame time sounds terrible :D
 

bystander36

Diamond Member
Apr 1, 2013
5,154
132
106
Yeah, that is what guru3d mentioned as well. That the spikes come from scene changes. I just think it is weird to show a graph where the relevant information falls outside of the scale.

The point people are trying to make is that the spikes are scene changes, so they are not relevant.
 
May 11, 2008
19,300
1,129
126
The point people are trying to make is that the spikes are scene changes, so they are not relevant.

I understand that. I was just interested in those spikes.
The reason why i am interested, is that with a scene change, textures and shaders need to be loaded again. And maybe those spikes can say something about how both architectures cope with filling up that 6GB/8GB of gddr5 DRAM. Especially with hitman with all the strange texture effects, people have been seeing recently.
 

redzo

Senior member
Nov 21, 2007
547
5
81
Hard to tell, probably impossible. In the end, overall experience is what matters.
Carefully planned spikes from time to time is not that big of a deal. Now, a drop in performance while intense gameplay, that's an issue.

EDIT:

For example, check 1:50 bellow. I take this as an issue.
https://www.youtube.com/watch?v=Y2LV2FGvEBE
 
Last edited: