
This Little Puppy has a Big Bite
Hey guys Kashi here. I know a long time ago I talked about Nvidia’s tech Physx, and today I’m here to talk about what makes Physx work on your computers. More specifically, I want to compare the gaming standard years ago, the 9800 GT, and the upcoming standard for gaming today, the 660 Ti. I’m also going to add my old card, the 550 Ti, to the comparison as a good in between.
I will be up front about this, I vastly prefer Nvidia over ATI if I have the choice between the two products. Don’t get me wrong, ATI cards do have many great points, but if you want the best bang for your buck, go with Nvidia (for the same reason you go Intel over AMD). In fact, the origin of AriaBlarg.TV came from an Intel/Nvidia computer, and my own streaming rig is a Intel/Nvidia computer (my gaming rig, however is an AMD/Nvidia combo). One of the largest reasons that I think Nvidia is better is due to the fact that developers have access to all of the Nvidia technology, which includes CUDA, 3D Surround, SLI, and Physx. AMD, on the other hand, only has Crossfire and Eyefinity.
Enough of all the long talk though, let’s get down to the meat of things. The test rig I’m using to test all the graphics cards is an AMD motherboard, the Gigabyte 78LMT-S2P with an AMD Phenom (t) II x6 1045T clocked at 2.7 GHz, and a relatively fresh Windows 7 installation. After installing the 9800GT, I checked out its Windows Experience Index and while it wasn’t anything astounding, it does offer a baseline to compare other systems to. The 9800GT was at a 6.9 on graphic processing while both my 550 Ti and 660 Ti both hit a 7.2 and 7.3, respectively. Next, I wanted to run 3D Mark11 on all of cards. 3DMark 11 (if you do not know) is a software developed by Futuremark (a company that develops benchmarking software), that pushes the limits of your hardware and software, and aggregates that performance into a score. The test was performed at 720p, which is pretty much the standard for broadcasting. 3Dmark11 requires DirectX11 compatibility with the card, which the 9800 GT did not have, so it couldn’t compete in the test, but the 550 TI scored a P2618 overall with a graphic score of 2404, Physics score of 4725, and a combined score of 2622, while the 660 TI vastly beat the 550 TI with a score of P6976, a graphic score of 8205, physics score of 4846, and a combined score of 4767.
| Test Rig |
| Motherboard: |
Gigabyte 78LMT-S2P AMD3+ |
| CPU: |
AMD Phenom (t) II x6 [email protected] |
| RAM: |
Crucial Ballistix Sport 4GBx2 1333 Mhz |
| OS: |
Windows 7 SP1 |
| Testing Results |
| Graphic Card |
Window Experience Index |
3D Mark11 Overall |
Graphic |
Physic |
Combined |
| 9800 GT |
6.9 |
Could Not Test |
Could Not Test |
Could Not Test |
Could Not Test |
| 550Ti GTX |
7.2 |
P2618 |
2404 |
4725 |
2622 |
| 660 Ti GTX |
7.3 |
P6979 |
8205 |
4846 |
4767 |
For my personal tests, I used Borderlands 2, a game that I had previously written an article about (the article focused on the Physx aspect of the game). Part of the test was going into an area with all the settings as high as possible at 720p, with Adaptive V Sync on or capped at 60 frames, depending on the card. The benchmark I use during the in-game sessions was fraps, to measure how many frames the video card is producing while I play. In the game while I’m in the area previously mentioned, I unloaded the entire clip of shotgun shells on the ground forcing the card to spew out rocks that Physx can understand. Then, I used an explosion grenade to toss all the particles away, forcing the card to calculate all the projectiles, which could be seen as a miniature stress test using the borderlands 2 engine. The scores below shows how well each card did in their average, highest, and lowest FPS.
| Borderlands 2 FPS Result |
| Graphics Card |
Average FPS |
Max FPS |
Min Fps |
| 9800 GT |
32.391 |
49 |
25 |
| 550 Ti GTX |
45.453 |
61 |
36 |
| 660 Ti GTX |
55.555 |
62 |
41 |
I mentioned above that there was something called Adaptive VSync, which is included in the 660 Ti. Basically, what AVSync does is that it helps the video card properly give out all the frames correctly. For example, if it over performs and produces more than 60 frames, it will cap it at 60 frames, and likewise if it under-performs it will turn off VSync. Why should this matter to you? VSync, if you don’t have it on, (and if you graphics card is very powerful) can cause what we call tearing, where the game produces so may frames that the display might accidentally split the display with half of one frame and half of another. On the flip-side, if you have VSync on but your computer under-performs, it sets the game to a frame rate divisible by 60 (or whatever the frame rate of your monitor is). This means if it’s perfect it could run at 60 frames, but if it drops anywhere below that, then it goes to 30 frames and vice versa. Both choices have their positives and negatives but now there is a nice in between with Adaptive VSync.
Now that we got all the technical stuff away, it’s time to talk about my personal preference for the cards. While stress testing in Borderlands 2 with the 9800 GT, the card’s fan ran so loud that it would’ve been noticeable with my microphone, making it unusable during streaming under heavy gaming. Also, it was completely unable to complete the 3D Mark11 test, which really shows its age (although to be fair, it is a 5 year old card). With the 660 Ti, I have more room for multiple monitors while still keeping my setup for streaming in 720p. Now, the only thing that the 660 Ti has left to prove is that it can keep up with the games coming out in the future, which I believe it will not only be able to do, but be able to do so well that it will become the new standard.