Comments on: PhysX running on 16 CPU cores in FluidMark

By: Tweak-Venetica

Tweak-Venetica — Tue, 09 Mar 2010 12:37:16 +0000

Quite a curious act this , i’v blamed the opposing side vast amounts of times , finding custom edits for each app to be capable of utilizing more than a instance of 25% / 1core cpu utilization. what bothers me the most is that i done vast amounts of research concerning this topic , iv even spoken to the editor of guru 3d about this and funny i blamed Nvidia , now i see they arnt to blame . it’s the programmers , but out of a curious perspective , why would the programmers limit us ? ? hence the usage of physX , any game using this addressing particle AI has a limitation , odd don’t you think ? but i’v read the email various times , and curious m id like to quote why say that vantage has the ability of addressing 12 threads ? if he then says test the cpu yourself with a dedicated physX card and CPU of your choice ? does this not defeat the object ? second , sure it can address 12 threads , but correct me , a single hd 4870 has 10 units ? now before you ask why i use a different none physX card as an example , think , if the card has 10 threads , then th opposing nvidia card has 10 threads ? oke , when you use a dedicated physX card you get full physX out of the threads odd because why address all the 3d cards core threads / units , but only one of the cpu ? as soon as software mode is enabled the threads only detect 1 ,, sorry something seems odd .. sorry if i make no sense , i might even be wrong , but hey , thats my opinion.

By: MicrO

MicrO — Thu, 04 Mar 2010 19:40:11 +0000

i think GPU is faster ( more FPS ) ..if FPS on picture is rendering by CPU, not by GT220

By: Zogrim

Zogrim — Thu, 04 Mar 2010 19:13:10 +0000

MicroO
I understand what you’re saying – I’m asking why you’re thinking that 16 CPU cores are slower than GPU ?
If you trying to compare new Fluidmark with GPU PhysX results in current one (1.1), take a look at number of particles firstly.

By: MicrO

MicrO — Thu, 04 Mar 2010 18:49:03 +0000

Zogrim – my english is bad … sorry man
mareknr can translate it
,,lol, na 16CPU jede fluidmark hůř než na jednom GPU čipu ? … wow
jak jsi říkal – skvělá práce

By: mareknr

mareknr — Thu, 04 Mar 2010 17:01:15 +0000

I’m interesting in difference between CPU@PhysX and GPU@PhysX performance. New Fluid Mark can provide many answers on credibility of using GPU for PhysX computing.

By: Zogrim

Zogrim — Thu, 04 Mar 2010 13:25:30 +0000

MicrO
What ? I don’t see point in your post – explain

By: MicrO

MicrO — Thu, 04 Mar 2010 12:57:09 +0000

lol, 16CPU Cores run fluidmark worst then one GPU core ? … wow
how u say – amazing work ! xD