^^ Get a APC 600VA UPS @ 1.8k
Change that PSU and get FSP SAGA 400W @ 2k or 500W @ 2.2k
For GFX card HD 5750 is around ~7k and GTS 450 is around ~7.5k.
@ Cilus - I got your point buddy. Actually I've confused things a little bit. I thought you are going to get a complete new rig. I appreciate your efforts for gathering so much infos about physx and all.
But toms guys has used only one game to show physX performance and Mafia 2 is using some physx SDK 2.x version as far as I know. So I think the physx performance comparison is not future proof enough. And those system builders though made some really good builds but everything always can't be perfect.
I've suggested GTS 450 for low power consumption, smaller size and in physX it will deliver performances neck to neck compared to GTX 260 or may even outperform it - in 3DMark vantage it has outperformed GTX 260 with less SP count as GTS 450 has improved architecture.
Also GTS 450 has a feature called concurrent kernel execution which will be used in future physx SDK ( say from version 3.x ) so I thought GTS 450 is the best as a future physx card.
Source
Read more ..
So future games and apps which will use future physX SDK ( and I think they will for sure ) will benefit more from GTS 450 in physX.
Change that PSU and get FSP SAGA 400W @ 2k or 500W @ 2.2k
For GFX card HD 5750 is around ~7k and GTS 450 is around ~7.5k.
@ Cilus - I got your point buddy. Actually I've confused things a little bit. I thought you are going to get a complete new rig. I appreciate your efforts for gathering so much infos about physx and all.
But toms guys has used only one game to show physX performance and Mafia 2 is using some physx SDK 2.x version as far as I know. So I think the physx performance comparison is not future proof enough. And those system builders though made some really good builds but everything always can't be perfect.
I've suggested GTS 450 for low power consumption, smaller size and in physX it will deliver performances neck to neck compared to GTX 260 or may even outperform it - in 3DMark vantage it has outperformed GTX 260 with less SP count as GTS 450 has improved architecture.
Also GTS 450 has a feature called concurrent kernel execution which will be used in future physx SDK ( say from version 3.x ) so I thought GTS 450 is the best as a future physx card.
Concurrent Kernel Execution and Application Content Switching
With graphics operations, PhysX execution, CUDA and general purpose computing, the GPU is no longer a simple-minded, single-function amassment of transistors but has to be flexible to switch between the different applications and tasks on the fly. Each category of application has its own kernel or number of micro-kernels and, particularly in applications that are intertwined like graphics processing with PhysX, it is necessary to switch as quickly and efficiently between different kernels.
In the G80 and GT200 architecture, kernel execution was strictly serial, that is one kernel was executing and only after completion of the task would the processor be capable to load the next kernel. In the GF100 architecture this problem is overcome by enabling concurrent kernel execution, meaning that simultaneously different kernels (for example PhysX and graphics) can be running with the sole limitation that this is only possible if the different kernels are from the same application. If the different kernels are from different applications, they will have to run sequentially.
Source
One is the ability to run multiple, independent "kernels" or small programs on different thread groups simultaneously. Although graphics tends to involve very large batches of things like pixels, other applications may not happen on such a grand scale. Indeed, Nvidia admits that some kernels may operate on data grids smaller than a GPU like Fermi, as illustrated in the diagram above. Some of the jobs are smaller than the GPU's width, so a portion of the chip sits idle as the rest processes each kernel. Fermi avoids this inefficiency by executing up to 16 different kernels concurrently, including multiple kernels on the same SM. The limitation here is that the different kernels must come from the same CUDA context—so the GPU could process, say, multiple PhysX solvers at once, if needed, but it could not intermix PhysX with OpenCL.
Read more ..
So future games and apps which will use future physX SDK ( and I think they will for sure ) will benefit more from GTS 450 in physX.