Ageia Physx vs. Havoc

What are they?

Are they only used for games?

Advantages and Disadvantages of each?

How do they help in daily computing (besides in gaming)?

How do they help engineering/scientific computing (3D simulation, data analysis, editing/modifying/hacking/analyzing)?

The only reason I ask is because if Ageia Physx is better i'd opt for a laptop with a Gefore. If Havoc is good then I can opt for A Radeon.

 

They are both physics engines, created to model classical mechanics for 3d games.

Pretty much, although they could be used for simulating physics for really any reason.

I really don't know, and it depends on what you would call an advantage or disadvantage. Physx can be accelerated by Nvidia CUDA GPU's.

Neither of the two really do except in very specialized tasks.

Physics engines are pretty much a requirement to model any sort of classical mechanics on a computer. What sort of physics engine you choose or design to use depends entirely on what you will be doing and what is supported by the software you are using. There are plenty of them out there for all sorts of simulations.

No, please don't make any sort of decision on it. It is seriously not a big deal or any deal at all. GPU assisted acceleration of either engine means very little in the grand scheme of things. There is an infinite amount of better reasons to base a purchase on.

 

What Trottel said. There are only 16 apps that do use Hardware PhysX acceleration anyway, even less when you count games.

 

Nvidia has made a much larger impact with their hardware accelerated physX, if that's all you're going by then get Nvidia, but s said before, there aren't enough games that make use of it for it to be much of a selling point. However, if you're wanting to do engineering then Nvidia all the way, CUDA is much better documented than whatever it is ATI have to offer.

 

Actually, they can be used for physics simulation as well, at least for PhysX as I did do some on it. However, you won't need a physX hardware GPU to use physX library, I'm using 5870MR and it runs fine, physX enabled GPU is only for the acceleration.

 

Oh so by what criterion should i go for buying a GPU from nvidia or Ati by?

What about OpenCL? Anyway most of the apps I'm going to be using in my education and in the future are going to be using CUDA for some performance, but if OpenCL is better, then I can still go for a Radeon GPU.

acceleration for the processing right? it takes the load off the CPU correct?

 

IE the whole point of physX.

 

Hmm technically my preference goes to nvidia (for cuda) which also helps in photoshop, all cyberlink software and much much more, but, there is a big but
Nvidia CPU's tend to die pretty quickly in notebooks (or the 7-8-9 range did) i dunno about the newer ones.
If i'd buy a new laptop today, i'd want one with ATI Gpu's or One with NVIDIA GPU's and an insane amount of warranty on it !!!

 

So funky monkey thanks for the answer, so as i stated before CUDA will help me, but I'm here OpenCL is great to, care to elaborate =)

 

Really? Well my nvidia gpu for my desktop died cause that fan's bearing was []..I don't know about laptops. What about OpenCl? How does that compare to CUDA?

 

So what exactly are you going to be doing with this laptop where an issue is choosing between using Nvidia's CUDA or ATI's Stream?

 

Ah a good question indeed. You see my laptop will multitask, but as people from engineering know, some projects may require scientific computing or 3D modeling. So the use of matlab, 3DS Max and such are going to be also a primary. Also I'd like to do some research on the laptop like data analysis (which idk if CUDA will help) and I'd love to do some video/audio analysis and modification. I'd love to be able to alter my system and learn hacking. So this laptop is also gonna be a learning tool of a type. And of course gaming (this will be my main console for my school year)

 

If you are serious about learning computer hardware, hacking, and modding, a desktop is by far the best option. If you break something it is easily replaceable, modifying anything is far simpler, and there is lots of documentation on how to do stuff.

If you are serious about crunching data, a desktop is also by far the best option. With a desktop, you could have both a Nvidia and ATI GPU to develop and implement with both Cuda and Stream. Also it can be much, much more powerful for computational work and is easily upgradeable.

Going with a laptop if you are serious about modding or doing computational work is just shooting yourself in the foot and in the wallet.

 

im going to residence and I don't feel like carrying around a desktop, that's why. Though I will have access to desktops and super computers in my university. So the laptop will be a side thing you know. I will do some of those stuff on it.

 

So then none of this is a big deal and there is no point to this thread?

 

Well i wanted to learn the difference between Havoc and Physx in the beginning. Now you see I love performance..thats why. The tasks i stated above, well i will need a laptop to be able to do though's to sometimes. Like say if I have a project on something, I wanna make some adjustments and the lab is closed. So i would need to boot up my laptop and say matlab to be able to do the work, so you know thats why i need a performance laptop

 

Ok. For anything like that then it is irrelevant what GPU you have since the processor can do just fine without the gpu maybe possibly helping sometimes in select scenarios. Just get whatever is best for gaming at the price point you are looking at.

 

NICE THANKS

 

I've been working with OpenCL (as a student) for the past few months and from what I've encountered, OpenCL and CUDA are very similar in many respects. From my experience though, OpenCL will be much more powerful in utilizing many different computing devices at once (CPUs, GPUs, APUs) based on it's do-everything nature vs CUDA and it's GPU does everything nature. The nod I do give to CUDA is based on simplicity for a newbie based simply on the fact that it constrains code to GPUs vs everything (OpenCL). In this example, CUDA takes care of all the context creation, etc. since everything is running on just a GPU, while OpenCL leaves it to the developer to create the context (CPU, GPU or APU) and set everything up. But as I said earlier, the actual CUDA and OpenCL code are almost the same. On the plus side, using AMD's OpenCL SDK, you can run OpenCL code on most newer CPUs (running just fine on my i7-720qm) vs. CUDA emulation on a CPU. So for testing, if you don't have a GPU, OpenCL will destroy CUDA (since it isn't emulating anything).

Since I've been working this for a while. Let me know if you have any questions.

 

Wow, thanks very much

So what other advantage does Open Cl have? From what I've read on your post, CUDA coes are emulated while Open CL are not? how