A Captivating Experience.
Though it relied on a custom codec, entitled FPS1, a single minute of fraps’ footage when captured at 30 frames per second in full HD 1080p, typically devoured a sliver over 3.5 gigabytes of disk space, when raising the frame rate to 60, this figure increased to just under 4.5GBs and some conventional hard drives began struggling to sustain the necessary write speeds.
Thankfully, the program was designed to dynamically adjust your frame quota in favour of a consistently smooth recording, though it’s worth nothing variable frame rates were not supported by several respected editing applications and that those which endorsed them, were also known to introduce universally exasperating tempo irregularities between the audio and video. To achieve optimum results for smoothness, FPS and resolution, a pair of fast HDDs (preferably enterprise class) configured in a RAID 0 array was all but compulsory.
Unilke FRAPS, DXStory was not inextricably bound to a natively engineered codec and permitted storing video in a selection of effective, high compression alternatives, entirely enforceable and customisable by the user. Built-in AVI multiplexing and Raw capture conversion tools were also offered as well as several menus sporting advanced tweaks that ostensibly resolved common problems or improved performance on specific systems.
In summary, both applications were admirably original, effective but ultimately expensive solutions for those wishing to extract the very best of what each could deliver. Some of their shortcomings were and continue to be shared by rivals, the most crucial of which was an innate necessity for every software based game grabber, namely that all have to be run on the same computer as the target application, consuming general system bandwidth and precious resources from the CPU and video card(s) that might otherwise be apportioned to the game itself.
SLI/Crossfire Compatibility and Setup.
A notable hindrance frequently declared to be inherent in every software recorder was their detrimental influence over multiple video cards running in SLI or Crossfire. Stories of success have been reported by fellow chroniclers, but when considering that FRAPS, and presumably all its competitors, can only intercept the data from one video card at a time, it would be logical to predict that the maximum recordable frame rate would be approximately 50% of the host system’s true optimum, assuming a dual card configuration.
To test this theory, I fired up Metro Last Light’s built in demo along with my heavily modified installation of Skyrim on the following system.
- Motherboard: Asus Rampage iV Gene
- Graphics Card(s): 2x Nvidia GTX 680s
- Hard Drive(s): 2x Samsung 256GB SSDs in RAID 0 – Dedicated recording array.
- 1x Western Digital Black 500GB – Dedicated game drive.
- Memory: 16GB Kingston HyperX 1600mhz (running at 1333mhz).
Testing Methodology.
Four batches of four tests were then conducted using our two most acclaimed screen snatchers, Fraps (no surprise) along with the more recent and equally praised DXtory.
The process was a follows. The games were loaded with each app configured and minimized in the system tray. A frame rate analysis of exactly one minute was then completed using FRAPS’s benchmarking utility (activated via the F11 key). The first of Metro Last Light’s semi-synthetic showcase and the second of a manually controlled amble through some of Skyrim’s opulent flora and fauna!
The analysis was run sixteen times in total, which broke down down as eight tests a piece for FRAPS and DXStory and two pairs of four tests per game with and without VSync and in single card and SLI modes. The results served as a baseline reference, demonstrating the video cards(s) true rendering power when neither program was carrying out a recording.
The complete cycle of 16 tests was then repeated, though this time a recording of exactly one minute was made concurrently with the frame rate analysis to assess the encumbrance each program placed upon the GPU(s) whilst capturing and the consequential reduction of speed and fluidity in both games.
The notion behind this exercise was to establish how much of a performance determent was caused by each application whilst a recording was in progress and avoid any further influence it might have over the outcome.
Consequently, for the Skyrim benchmarks when vsync was disabled, the capture rate was set to 120 fps since the video cards, whether in solo or harmony proved strong enough to sustain a pace in excess of 60fps for the entire duration. When bearing in mind fraps continuously regulates its recording frame rate in approximate accordance with the target one, there was a chance that this might also preclude the GPUs from achieving their optimum velocity, leading to skewed results and erroneous conclusions.
One minor imperfection was that FRAPS had to be running throughout both its own tests and those for DXTory since the latter lacked any facility to log frame data.