Given the recent interest in stutter from the HD 7xx0 range of cards I decided to put some effort into testing for myself. I found some very interesting (at least I think so) results.
This is using an un-modded Skyrim and the benchmark run lasted around 1 minute for each test. I used an outside part of the map, raining, near a wooded area with a tower structure. Here is an image of the location with me at the edge of the woods. I would walk towards a clump of bushes to the left of the tower, turn around in front of the tower and walk back towards the woods before doing a slow 180 turn and heading back for the tower along a the foot of the hill at the left. In total it took around 1 minute.
System config as follows.
Intel i7 2600K @ 4.7 GHz
16 GB DDR 1886 RAM
HD 7970 VTX3D X-Edition @ 1050/1425
Windows 7 64 bit Pro
12.11 b11 with Cap 2
Res: 2560x1600
Skyrim graphics all settings at max/ultra. 8X MSAA, FXAA on, 16X AF.
I ran three different tests with different settings using Radeonpro 1.1.1.0
Test 1. Vsync off, no triple buffering, flip queue size = default.
From test 1 it is clear that with these settings the stutter is unacceptable. When playing at these settings Skyrim is unplayable. This would I assume be the default setting used in any reviews and mirrors the findings of TechReport form their HD 7950 test.
Test 2. Vsync off, no triple buffering, flip queue size = 1.
Using RadeonPro I changed the flip queue size from default to 1. Immediately it is obvious how much of a difference this one single setting makes to the smoothness of the game. For me it was instantly noticeable how much smoother the experience was. Judge for yourselves.
Test 3. Vsync on, triple buffering on, flip queue size = 1.
Finally I tested after turning on vsync and triple buffering along with flip queue size = 1. As can be seen smoothness did improve by another small amount. Though it was not so dramatic that it was perceptible in game. There are some minor spikes but they aren't noticeable and could be simply the textures being loaded to VRAM (at a guess).
I was adamant that this whole stutter issue was all a load of BS because my own experience did not echo that of TechReport. In hindsight the subconscious reason for my stance was that I never use CCC to setup my game profiles for my AMD cards. I now realise in hindsight that is a fallacious data source to base my judgement on. I usually set vsync with triple buffering for my games with a flip queue size = 1. To be honest I do the same thing using NvidiaInspector on my GTX 680 system. It is just a force of habit I picked up over the years. This in turn meant I was circumventing any serious stutter issues before I noticed them. So my experience on the GTX 680 and HD 7970 was identical in regards to smoothness.
This of course does not take away anything from the fact that out of the box my HD 7970 is not smooth for Skyrim. I did a brief test using 12.7 drivers and got similar results. The only fix for me in Skyrim is to set a flip queue size to 1 using 3rd party software.
Disclaimer I only tested Skryim, this is not a blanket statement that all games will be the same.
This is using an un-modded Skyrim and the benchmark run lasted around 1 minute for each test. I used an outside part of the map, raining, near a wooded area with a tower structure. Here is an image of the location with me at the edge of the woods. I would walk towards a clump of bushes to the left of the tower, turn around in front of the tower and walk back towards the woods before doing a slow 180 turn and heading back for the tower along a the foot of the hill at the left. In total it took around 1 minute.
System config as follows.
Intel i7 2600K @ 4.7 GHz
16 GB DDR 1886 RAM
HD 7970 VTX3D X-Edition @ 1050/1425
Windows 7 64 bit Pro
12.11 b11 with Cap 2
Res: 2560x1600
Skyrim graphics all settings at max/ultra. 8X MSAA, FXAA on, 16X AF.
I ran three different tests with different settings using Radeonpro 1.1.1.0
Test 1. Vsync off, no triple buffering, flip queue size = default.
From test 1 it is clear that with these settings the stutter is unacceptable. When playing at these settings Skyrim is unplayable. This would I assume be the default setting used in any reviews and mirrors the findings of TechReport form their HD 7950 test.
Test 2. Vsync off, no triple buffering, flip queue size = 1.
Using RadeonPro I changed the flip queue size from default to 1. Immediately it is obvious how much of a difference this one single setting makes to the smoothness of the game. For me it was instantly noticeable how much smoother the experience was. Judge for yourselves.
Test 3. Vsync on, triple buffering on, flip queue size = 1.
Finally I tested after turning on vsync and triple buffering along with flip queue size = 1. As can be seen smoothness did improve by another small amount. Though it was not so dramatic that it was perceptible in game. There are some minor spikes but they aren't noticeable and could be simply the textures being loaded to VRAM (at a guess).
I was adamant that this whole stutter issue was all a load of BS because my own experience did not echo that of TechReport. In hindsight the subconscious reason for my stance was that I never use CCC to setup my game profiles for my AMD cards. I now realise in hindsight that is a fallacious data source to base my judgement on. I usually set vsync with triple buffering for my games with a flip queue size = 1. To be honest I do the same thing using NvidiaInspector on my GTX 680 system. It is just a force of habit I picked up over the years. This in turn meant I was circumventing any serious stutter issues before I noticed them. So my experience on the GTX 680 and HD 7970 was identical in regards to smoothness.
This of course does not take away anything from the fact that out of the box my HD 7970 is not smooth for Skyrim. I did a brief test using 12.7 drivers and got similar results. The only fix for me in Skyrim is to set a flip queue size to 1 using 3rd party software.
Disclaimer I only tested Skryim, this is not a blanket statement that all games will be the same.
Last edited: