Are you shy, Ruckus?

There we go again, this time a post by Ruckus:

No. The PhysX on CPU is not the same as PhysX on GPU. PhysX hardware accelerated looks and performs different than on CPU does. Ask anyone with Batman with PhysX hardware acceleration. So the comparisons can’t really be made like that. I’ve read other more credible articles that say the performance is hampered by 50% or more. I don’t even read XtremeSystems Forums (more like XtremeRageStupidity) since it’s so full of rubbish, it’s too hard to read through the junk and even find one credible post.
In the end it doesn’t change the fact that PhysX is crap for CPU. It’s not optimized for ANY CPU made in the last 10 years. Whether the blog idiot wants to argue that or not, he’s simply wrong on that point. There is no excuse why PhysX is not SSE 4+ optimized and with full multi-thread support to take advantage of every logical core available from Intel and later AMD’s Bulldozer CPUs. The fact the blog idiot is still using a junk dual core and compiled it in visual studio means this PhysX compile has no optimization for any modern CPU today with multi-threading. He has no arguement either.

I don’t believe the hype for GPU accelerated anything. Rendering is faster on CPU sometimes, but definitely the complexity and quality is far superior than anything from CUDA.

Even supposedly where GPU accelerated encoding is supposed to be faster it’s not. x264 is faster than any CUDA encoder and anything from Intel’s Quick Sync while maintaining a higher level of quality.

I see no reason why Physics would be different if it’s done right. I’d much rather have a game running Havok than PhysX or Bullet.

Sure, you can be tough and insult me there, but you don’t dare to argue directly with me, do you? Go ahead, show me your “more credible sources” (lol). What can be more credible than various people on various forums independently recompiling an open source project in both SSE and x87, and independently reaching the same results? (And no, not all of them used dual cores).

Hampered more than 50%? Pah, you just drank the SSE koolaid. The people who actually bothered to recompile Bullet are actual developers, who have actual first-hand experience with SSE (one of the people who verified my results on his own machine was Andrew Lauritzen, he actually works for Intel, how’s that for credibility).

This entry was posted in Uncategorized. Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s