Red Hawk
Diamond Member
- Jan 1, 2011
- 3,266
- 169
- 106
CUDA will be the last killer feature touted.
GPU transcoding in Rage!
CUDA will be the last killer feature touted.
CUDA will be the last killer feature touted.
Red Faction Guerilla has a great destruction model. Here is a quick look at some of the games that try to use realistic physics effects. Mafia II effects are not very good at all with excessive debris.
Game Technology: Best Destruction On Games, top 10
http://www.youtube.com/watch?v=377h3l6K3hk&feature=related
P.S. After we have exhausted HD7950's 3GB of VRAM, 32 ROPs, 384-bit bus, 30-40% overclocking, addressed that NV and AMD both have driver problems on both sides, the discussion has finally narrowed itself down to PhysX. It's been a while since we found ourselves revisiting this "killer" NV feature. Good times.
AMD did address that and was very welcomed.
Some overclocking results. Nothing real world, but something.
http://wccftech.com/geforce-gtx-660ti-benchmarks-leaked-overclocked-performance-exposed/
3DMark 11 Extreme Preset:
GeForce GTX 660 Ti (915/980/1502 MHz) – X2730
GeForce GTX 660 Ti (1015/1080/1502 MHz) – X2880
HD7970 GE (1050/6000 Mhz) - X2980
GeForce GTX 670 Stock (915/980/1502 Mhz) - X3002
GeForce GTX 680 (1006/1058/1502 Mhz) - X3276
GeForce GTX 680 (1111/1176/1502 MHz) – X3621
NV does really well in 3dMark11:
I just don't see the point in these discussions when the market is trumping everything being said, nobody really cares.
It actually shows weak OC gains for the 660ti. Pretty piss poor *if true*.
3DMark 11 Extreme Preset:
uld beat HD7970 GE.
The reality is GE beats it by 10% at 1080P and by 19% at 1600P.
GTX670 beats HD7970 by 10% in 3dMark11 in Extreme setting but a 925mhz 7970 beats it by 8% at 1600P. 3dMark 11 = worthless bench.
Think about it logically, a stock GTX670 has 33% more memory bandwidth and 33% more ROPs than a stock GTX660Ti. It's going to take a huge overclock for the 660Ti just to catch up to a stock 670, unless the game is shader bound where ROPs and memory bandwidth aren't that important.
I guess it is a fine line -- to offer compelling choice to consider for its intended price-point, but not to deflate too much the GTX 670 and GTX 680 premiums. To try to bring more compelling 28nm competition for gamers than just the HD 7870 and HD 7950.
No because I could get almost the same performance with an OC 7950 and I got it for $330.
Let's assume I did buy a $460 7970 and ran it at the same clock, I'm getting negligable performance increase thats actually pretty damn hard to notice outside of benchmarks getting a few % difference. Now, I would paid $130 extra for what? Negligable perf increase along with a sizable power use increase??
The aspect I like about nVidia is their pro-active nature and oddly, is what some gamers complain about in this forum. nVidia's developer relations, flexibility of settings, features and support like 3d Vision, GPU Physics, willing to spend resources for their customer base, while trying to improve the gaming experience for their customers and continued by offering adaptive V-sync, frame limiters, GPU Boost and TXAA. They're not the same to me.
Some may hear gamers clamoring on some of these features as gimmick, going to die, foolish, waste of resources but this pro-active nature as a whole is the reason for the strong Brand-name to me. Not because nVidia buyers are dumb.
