Username:
Password:
Remember Me?
   Lost your password?
Search



GTA 4 (PC) Benchmarks

GTA 4 Patch 1.0.3.0 tested: Noticeable performance benefits

In the release notes of Patch 1.0.3.0 for GTA 4 Rockstar promises performance optimizations. Pc Games Hardware tests 15 graphics cards in order to check the differences to version 1.0.2.0.
GTA 4 Patch 1.0.3.0 tested
 
GTA 4 Patch 1.0.3.0 tested [Source: view picture gallery]
For this benchmark review of the GTA 4 Patch 1.0.3.0 we use a test system based on an X58 chipset, 3x 2 GiByte DDR3 memory and a Core i7-920 running at 3.5 GHz. This way we try to prevent a CPU limitation of the benchmarks. For further information about the CPU performance see our GTA 4 (PC): CPU benchmark review with 13 processors.

GTA 4 Patch 1.0.3.0 reviewed: That's what the update delivers
GTA 4 v1.020 (100% range of view, maximal details)
 
GTA 4 v1.020 (100% range of view, maximal details) [Source: view picture gallery]
GTA 4 v1.030 (100% range of view, maximal details)
 
GTA 4 v1.030 (100% range of view, maximal details) [Source: view picture gallery]
The full change log can be found in our article GTA 4 (PC): Version 1.0.3.0 is ready for download. For this benchmark review the most interesting features are the promised performance optimizations which apparently don't affect the visual quality - compare the screenshots on the left and right side. Since the game does not offer a quick save option it is difficult to receive 100 percent equal comparative screenshots thus the position of the persons is slightly different. One thing worth noticing is that the patch enables the use of 2 GiByte graphics cards.

Excerpt of the Release Notes:
GRAPHICS:
* Interior Lights for ATI 1900 series cards:
- All light sources appear for users with 1900 series ATI cards. (ie. Pool Mini Game)
- ALT-TAB fixed
* Graphics Settings:
- Added menu option " OFF " for Reflection quality
- Added menu option " OFF " for Shadow quality

PERFORMANCE:
* Physics optimizations to improve situations with heavy physics calculations (ie. 32 player races)
* Graphical Performance Optimizations


GTA 4 Patch 1.0.3.0 reviewed: Benchmark methods
GTA 4 v1.030 (100% range of view, maximal details): No problem with 2 GiByte VRAM
 
GTA 4 v1.030 (100% range of view, maximal details): No problem with 2 GiByte VRAM [Source: view picture gallery]
For comparable results it is not enough to go to the GTA 4 video options an put all sliders to the limit, because the game couples the maximal details to the available video memory of the graphics card. A 512 MiByte card is not allowed to display the same view distance as a card with 768 or 1,024 MiByte. This restriction can be worked around by editing the commandline. But now another problem arises: Although you can set the video options to the maximal values now, not all details are displayed in the game. This is caused by the lack of video memory because of which the game doesn't stream all the data. But this can again be countered by combining the command "availablevidmem” with a digit. With this tewak you convince the game that more video memory is available than there I really is.

With 1,024 MiByte VRAM textures are missing or popping up rarely only and because of that we use this as a basic value for our benchmarks. For graphics cards that have this amount of video memory we only maximize the settings in the video options with the view distance at 50 percent. The blur filter is deactivated. Cards with less than 1,024 MiByte are tweaked with the following values in order to make the game believe that more than 1,000 MiByte are available.
512 MiB cards: availablevidmem 2.1
Geforce 8800 GTX/Ultra: availablevidmem 1.4
Geforce GTX 260: availablevidmem 1.2

This way we ensure that all cards display the same details. This costs a lot of performance for 512 MiByte cards since they have to swap a lot of data to the system memory.

GTA 4 Patch 1.0.3.0 reviewed: Test sequence
GTA 4 (PC): PCGH benchmark sequence
 
GTA 4 (PC): PCGH benchmark sequence [Source: view picture gallery]
We use our "Promenade” benchmark for this review but this time we run the tests at in-game sunset because of the different lighting effects this scene is more challenging for the graphics cards than it would be at in-game high noon. We sprint down the pavement for 20 seconds and take the average results of two runs.

The benchmarks are run at 1,280 x 1,024, 1,680 x 1,050, 1,920 x 1,200 and 2,560 x 1,600 pixels. 16:1 AF is always activated while FSAA is still not possible. The Radeon cards are powered by the Catalyst 9.3 WHQL and the Geforces use the Geforce 182.08 WHQL.

Picture gallery  (enlarge to view source)



--
Author: Raffael Vötter (Mar 24, 2009)






Advertisement
Copyright © 2014 by Computec Media GmbH      About/Imprint  •  Terms/Conditions