With the 8500 GT aimed at a more mainstream market, I decided to use a slightly older system for testing the card in order to examine the card within a more "real-world" typical system layout.
Test System
Intel Pentium 4 560 (3.6 GHz) Processor
Gigabyte GA-8GPNXP Duo (Intel 915)
Mushkin 2GB (1GB x2) XP2 DDR-II Memory
PC Power & Cooling 510W ATX-PFC Power Supply
Seagate 250 GB SATA HDD
Dell 2405FPW LCD Monitor
XFX GeForce 8500 GT
NVIDIA ForceWare 158.22 Drivers at Quality Image Setting
Windows XP Professional 32-Bit with Service Pack 2 / DirectX 9.0c
Test Applications
Doom 3 - v1.3.1
F.E.A.R. - v1.08
Half-Life 2: Lost Coast
ForceWare Driver Hiccup
Driver Memory Reporting Issue
One thing noted with the 158.22 drivers is that it recognized the card as having 512MB of onboard memory, when the card sports half of that. This yielded some small, but avoidable problems with applications that scan the system to suggest settings. For example, a scan performed within Doom 3 recommended using the Ultra setting, which is reserved typically for card with 512 MB of onboard memory due to the larger textures used.
F.E.A.R. PERFORMANCE
Benchmark results in F.E.A.R. were based on the built-in performance test. Physics and graphics settings are documented in the table below.
F.E.A.R. Settings
F.E.A.R.'s built-in auto-detect functionality set the settings rather high for the card. I wonder if the memory reporting issue had anything to do with a few of the graphical settings set to maximum rather then lower levels.
F.E.A.R. Performance Results
F.E.A.R. performance was examined at 1024x768. Multiple settings had little effect on the performance of the card at 1024x768. Performance at 800x600 was extremely similar to the posted scores. The average frame rate stayed the same and the minimum frame rate was only 1 or 2 fps higher, depending on the settings. It's for this reason that the results were not included in the review. In situations such as this, the user might as well benefit from the free antialiasing offered by the card.
