DX11 vs DX9 - Comments

[placidus33]

Today I tried to start TLD in DX9 just for curiosity. I went to the root directory of the game and simply double-clicked on the batch file kindly provided by the devs.

My card is fully capable to use DX11 libraries, and I always played this way but I wanted to see the difference with my eyes.

As expected, the graphics were less refined in DX9, but I found out an unexpected thing:

This game is less performant when using DX9!!! I had a significant drop in FPS, about 5-6 frames per second....

I went in a couple of places in ML where my card goes at about 50-52 FPS in DX11, expecting to see maybe a 60 stable, but it was 45-46.

I've seen many times here in the forums people suggesting to owners of relatively low-end computers to switch the game in DX9. Maybe it is not the case if you have a graphic card capable of running in DX11?
It is only my opinion. Maybe can someone else confirm or deny what I saw?
Who knows.... could be helpful to many people looking for a game experience a little more fluid.

[IFIYGD]

My toaster's 740GT is DX11 capable. And I run the game in DX11. In previous versions ofthe game, I have switched back and forh to see which gave me better performance, and DX9 did give me better performance overall, but yeah, the graphics were not as nice.

Haven't switched to DX9 in this version to test it. I'll do that in a bit after I am done playing my current session. Will let you know when I see what I see.

[hinterland_community]

Moving this thread to the technical subforum for discussion.

[Manel84]

the instructions of the dx9 are archaic for your graphic card, I could guess a lot of things that aren't suppose to be used for a incompatible design but I'm not an expert and neither anyone will read this thread.
But you can think every case is a case, so that's why you need to experiment, it's quicker than to do all the maths and knowledge.

[IFIYGD]

Okay so...

Running the game in DX11, I get between 45-60 ps, an occasional hitch or microstutter, but so brief it really does no affect my gamplay. (Makes it seem like my character had a suc dden twitch or shrug goiug on, and only happens rarley.) Graphics look really nice in DX11, no artifact, no odd effects.

In DX9- I got between 55-70 fps, my GPU heated up a titch more, but not even close to overheating. (45°C). Graphics had no artifacts or odd effects, but looked worse, blurrier perhaps, and less crispness than what I am used to. However, I had no microstutter or hitching, dong the same things in the same places as I usually see them.

Both versions has the same fps spikes and drops, as I ran both in Mystery Lake, both on the Southern River Access and the climb up to the Lake Overlook and the cave there. I am used to some wonliness and textures rapidy chage during climbs or travels over the ice in that area.

Relatively mixed results with one HUGE exception- In DX( my game stopped responding on quitting from the main menu. Every time. No CTD, No BSoD, no black screen. Just "tld.exe has stopped responding" and had to use the "close program" button on the notification pop up to close the game out. DX11 ran and closed out fromthe main menu perfectly each time.

Checked to be sure I had al updated drivers and redists, and I do. Not entirely sure what to make of it, I have a few suspicions, but am not able to extensively test for repro steps rg ight now to try to diagnose it.

Pastebin of the Windows Event log- https://pastebin.com/42bVqnkn

Pastebin of the output_log - https://pastebin.com/qTEVhcWD

(And I'll go log these into Atlassian in a minute as well.)

[placidus33]

Originally posted by IFIYGD:

Relatively mixed results with one HUGE exception- In DX( my game stopped responding on quitting from the main menu. Every time. No CTD, No BSoD, no black screen. Just "tld.exe has stopped responding" and had to use the "close program" button on the notification pop up to close the game out.

Yep. It happened to me too, but in DX9 I ran the game directly from the batch file, and I thought it was an issue more related to Steam that doesn't like when you do things avoiding to use its GUI.



Now I have some chores... later I'll do some more digging with a more scientific approach. I'll put FRAPS on screen, set my drivers to use only the graphis settings of the game and turn off V-Synch and I want to see the real comparable difference taking some screenshots inside Camp Office (save point), and a couple outside locations where I have high FPS and low FPS

[placidus33]

All right.... I did some more trials.

I began to check as written in the previous post, but weather patterns are so complex and so random (so real) that is impossible to repeat exactly the same situation in exteriors. Everytime I loaded the same savegame in the same position at the same time I had different weather evolution. In interiors repeatability is slightly better but not enough to be 100% objective.

The only way to check the overall performance of a particular rig/settings/overclocking/DX9vsDX11 is the loading screen!!!
And it is the fastest too!

With the same graphics settings and the same rig, same resolution, same everything blah blah, my graphic card (Nvidia GTX 780) performs better with DX11.

DX11 loading screen FPS the first seconds after loading the game = 93-94 (borderless or full screen doesn't change a bit)

DX9 loading screen FPS the first seconds after loading the game = 86-87 (borderless or full screen no change)

Other considerations off-topic:

Using the -force-d3dxx parameter always result with a crash of the game on exit. It doesn't matter if you use the batch file or if you use Steam GUI.

The game engine for some reason unknown to me doesn't go higher than 121 FPS. In my case, when I lowered the game settings to "high" I had 115 FPS in DX11 and 105 FPS in DX9. Then I lowered to "medium" settings and I had 121 both in DX11 and DX9. Same goes for "low" settings: 121 FPS both in DX9 and DX11.

To Overclockers: if you keep game settings at max is better to slightly downclock the card or keep standard clocks: Power consumption goes down, the card can shift to higher clocks vs voltage and practical/playable FPS goes up. I had 52 FPS outside Camp Office in high OC (+156 MHz). With standard clock changed dynamically without exiting the game the FPS went up to 59. By downclocking the card at -26 MHz no change. After a downclock of -52 MHz I began to see the FPS decreasing again.