PhysX Performance Update: City of Villains
by Ryan Smith on September 7, 2006 6:00 AM EST- Posted in
- GPUs
When Things Go Wrong & The Test
It's worth noting that at one point this article had a very different tone to it, based on the benchmark results we had gotten. We recently replaced the test rig we run the PhysX articles on with a newer machine, and as the component of note, threw in an ATI Radeon X1900XTX as it's generally the fastest single-slot card we have that isn't an SLI card (i.e. the GeForce 7950 GX2). City of Heroes/Villains is a game we long ago established was CPU limited, so the choice in video cards is largely academic, or so we thought.
It turns out that ATI's latest drivers have a problem with City of Heroes/Villains where the performance of the game chokes when using some of the advanced rendering features. Fortunately, we caught this issue, but for a while we were wondering why the PhysX card wasn't helping as much as expected. It's always interesting to discover where the bottlenecks are in different benchmarks. City of Heroes is a great testing components since it's an OpenGL title that isn't built on the Doom3 engine. Unfortunately, this is bad timing for ATI, given their recent OpenGL improvements on games that do use the Doom3 engine.
Due to the issues with the ATI card, we switched to testing with a 7950 GX2 instead. Here are the details of our test setup.
It's worth noting that at one point this article had a very different tone to it, based on the benchmark results we had gotten. We recently replaced the test rig we run the PhysX articles on with a newer machine, and as the component of note, threw in an ATI Radeon X1900XTX as it's generally the fastest single-slot card we have that isn't an SLI card (i.e. the GeForce 7950 GX2). City of Heroes/Villains is a game we long ago established was CPU limited, so the choice in video cards is largely academic, or so we thought.
It turns out that ATI's latest drivers have a problem with City of Heroes/Villains where the performance of the game chokes when using some of the advanced rendering features. Fortunately, we caught this issue, but for a while we were wondering why the PhysX card wasn't helping as much as expected. It's always interesting to discover where the bottlenecks are in different benchmarks. City of Heroes is a great testing components since it's an OpenGL title that isn't built on the Doom3 engine. Unfortunately, this is bad timing for ATI, given their recent OpenGL improvements on games that do use the Doom3 engine.
Due to the issues with the ATI card, we switched to testing with a 7950 GX2 instead. Here are the details of our test setup.
PhysX Testbed Configuration | |
CPU: | Intel Core 2 Extreme X6800 (2.93GHz/4MB) |
Motherboard: | Intel D975XBX (LGA-775) |
Chipset: | Intel 975X |
Chipset Drivers: | Intel 7.2.2.1007 (Intel) |
Hard Disk: | Seagate 7200.7 160GB SATA |
Memory: | Corsair XMS2 DDR2-800 4-4-4-12 (1GB x 2) |
Video Card: | NVIDIA GeForce 7950GX2 |
Video Drivers: | NVIDIA ForceWare 91.33 |
OS: | Windows XP Professional SP2 |
31 Comments
View All Comments
shank15217 - Saturday, September 16, 2006 - link
All this review shows is how good the AGEIA software engine is capable of using a second core to it's full advantage. Infact their goals are slightly two sided. Its true the hardware is probably a magnitude faster than cpus but cpus are reaching 4 core and beyond with shared caches and very wide busses. Unless AGEIA puts their hardware implementation on a PCI Express / HTX slot its not gonna outrun even a 4 core core 2 duo or an athlon 64 for that matter.Vergil - Monday, September 11, 2006 - link
Ageia's PPU should be integrated with the motherboard to for maximum results ladies and gentzCan you imagine having the raw horsepower of a motherboard with a local GPU thats powerful as a 7800 GTX? Local components work better with any CPU(meaning have a better data transfer/flow rate) than any serial bus. Motherboard bus speeds(PCI/AGP/PCIe) limit the performance of most video & other cards(even the high end ones) to some degree. However, PCI-e is the next best thing without having it on the die with the CPU. But with chipsets that help with physics, it has to catch it BEFORE it hits the CPU, which is why there is a major performance hits to City of Villians. The CPU has to collect all of the data before everything else gets it providing thats its not integrated. Then it has th elaborious task of dealing with data it isnt prepared to use.
soulshagga - Saturday, September 9, 2006 - link
The author obviously hasn't written a physics engine before.The problem is n! in complexity -- the reason being is each entity interacts with each other. You can't just blindly update n entities independently of each other (consider collision detection for instance).
I won't elaborate here because it is beyond the scope of this article, not to mention these forum posts, but just think about what would happen if you just blindly update the physical attributes of n entities... you would lose information. If you don't get what I'm saying, don't worry about it -- just take my word for it. :)
Please, don't make bold statements like that if you are unsure.
Thanks.
ojingoh - Thursday, September 7, 2006 - link
sorry for the confusion -- i see you in fact did run the game unacellerated, thanks for thatojingoh - Thursday, September 7, 2006 - link
what i meant by the minimum physics -- this card does physical calculations, not just particles. things like collisions and deformations etc.ojingoh - Thursday, September 7, 2006 - link
i have an issue with your review. your testing doesnt include screenshots of the things tested.please explain "slightly lower quality physics mode" -- what setting was this? 75%? 50%? without screenshots it's hard to identify what you claim. the slider states "up to" implying that some frames will have the maximum but some have far less. is it hard to tell the diffrence between 500 and 1000 particles? what would be a reasonable number of particles?
also you exclude any test showing what happens when you have the card installed but run the game at minimum phisics, i think in cov it's 100 particles (idk if this is the case, this is also not tested.)
Regs - Thursday, September 7, 2006 - link
Dual core actually helps physics processing? I guess this is only in games that support dual core.PrinceGaz - Friday, September 8, 2006 - link
I imagine that most games which have support for offloading work to the PhysX card will also support offloading the same work to a seperate "software-physics" thread. If they've already made the effort to seperate the physics work from the main application so it can be sent to the PhysX card, it should be trivial to run that work in a second (or multiple) threads. Therefore games that make use of the PhysX card will most likely also be able to use a dual-core processor to good effect, as is the case with CoV.Calin - Friday, September 8, 2006 - link
I assume more games support dual cores than PPU right now (and the trend probably favours the dual cores rather than the PPU)Anyway, a dual core will help with short processor-intensive (even if very short) tasks that appear "out of the blue" - antivirus, some operating system tasks/schedules/other activities
Missing Ghost - Thursday, September 7, 2006 - link
As soon as there is a PCIe 1x/4x version available, I will buy it.