MonarchX;n9946571 said:
Thank you for your input SigilFey, but I was looking at this from testing perspective, not troubleshooting. GPU's got faster, drivers changed and thus I was curious how W3 would perform today with PhysX on GPU rather than CPU.
BTW, I disagree about the whole driver thing, especially when it comes to GPU drivers because, for example, running a game 20% slower with an older driver may not render the driver broken, but it would still be underperforming if newer driver was optimized to run the same game 20% faster. Sometimes you are not even aware of the issue until it gets fixed, making an obvious difference (or barely notice-able one). Nonetheless, it comes.down to user skills, experience, and the rig I update everything from drivers for all hardware to motherboard BIOS, videocard BIOS, as well as, use advanced tools like UBU to update specific motherboard firmware/ROM's/OROM's (Intel Management Engine Interface, Rapid Storage Tech., CPU microcode, LAN). I only had one minor issue this year with NVidia WHQL driver that was fixed promptly by NVidia's Hotfix driver.
If you want to play newly-released games, it makes far more sense.to.keep GPU drivers updated to the latest due to fixes and optimizations for those games than to stick with "if not broken, don't fix it" rule.
All that was so off-topic I guess?
Not off-topic at all! And, of course, it all comes down to personal decisions in the end. What is true, though, is the modern habit of staying current to-the-minute with programs, drivers, and OS updates is responsible for a
vast number of issues. (It's unfortunate that so many OS updates, especially, are nothing more than intrusive "security" features that will only benefit a tiny percentage of users, but will negatively impact performance and/or ease-of-use for all users. Other updates may intentionally limit user functionality for...questionable purposes. Add to this the human error inherent in all code, and many companies with "corporate incentives" to release "one update every week/month/season!" for example, and it's easy to see how constantly updating is basically rolling dice.)
For testing, go nuts! Especially with modding (which often hooks or overrides vanilla game functions), those
added features and processes may benefit from more current drivers than the older versions the vanilla game was optimized for. That is a valid consideration!
Across the board, however, it always works as follows: A program is originally written for driver version A. Over time, that program is updated to work with driver version B, C, D, and E -- but it works BEST with C. And it always will. If version C is installed, and I'm not having any issues with other programs, there is 0% reason to change the operating environment unless I encounter an issue. It risks introducing issues where there were none for (normally) insignificant gains. The
power-user approach (squeezing every last ounce of performance out of my system) is the polar opposite of the
optimization approach. If I'm optimizing, I'm trying to get 100% functionality with the
minimum percentage of resources possible. (So, if I can get a solid 60 FPS with my GPU running at only 30% -- that's
amazing.) Conversely, power-usage means that I'm going to drive my system to its limits to achieve 168% functionality while all reactors are firing at 115%. Yup, performance may be blazing, but it's next to impossible to maintain stability. (Plus, how long can I reasonably expect the hardware will hold out under that constant strain?) Any changes to the operating environment to squeeze more power out of a system is generally a direct path to introducing issues, not solving them.
But there are times when there
is a reason. It's extremely rare for a driver update to increase a games' performance by 20%...but such situations have occurred. (I remember an Nvidia patch coming out when Supreme Commander [...or maybe it was Ground Control...some fancy, 3D RTS...] was first released that literally
doubled performance during large battles.) It does happen. Although, it's very, very rare. If a driver update creates such drastic disparity in performance (especially if it affects more than a single title), that's normally indicative of a problem with the prior driver installation, not that the new drivers are that much better.
In the end, my point here (
why this is very much on-topic), is that by changing a driver version to forcibly enable something that the developers have
intentionally disabled, it directly introduces an operating environment that was known to be specifically problematic. This can also happen
unintentionally through an unnecessary update: if flags within the driver happen to change in the new version, it may allow players to activate features that prior driver versions shielded them from with particular titles. The "newest and best" thing may introduce an old problem...one that was
already solved by prior versions.
Overall, it's a relatively inefficient way of trying to test, problem-solve, or create a foundation for introducing something new in an older title. I know this might seem like a complicated and round-about way of explaining the point, but I'm not sure there's a way of making sense of it, otherwise. (And of course, I'm just sharing info -- not trying to dissuade you from fiddling with the PhysX thing.
Worst case scenario is: it simply won't work that well.)