GTX580 runs faster than Tesla C2050

Accelerated Computing CUDA CUDA Programming and Performance
#1

Hi,

I’m wondering if someone can tell why my GTX580 runs my kernel almost twice as fast as my Telsa C2050. Both cards are in the same PC, where the GTX580 is also being used as the primary display.

My understanding was the the Telsa is suppose to be quicker at running kernels? Am I wrong with this assumption?

Regards,

Dave

#2

Hi,

I’m wondering if someone can tell why my GTX580 runs my kernel almost twice as fast as my Telsa C2050. Both cards are in the same PC, where the GTX580 is also being used as the primary display.

My understanding was the the Telsa is suppose to be quicker at running kernels? Am I wrong with this assumption?

Regards,

Dave

#3

The GTX 580 is supposed to be faster than the Tesla C2050, as can be deduced from the specs. The GeForce however should not be twice as fast. If that is the case, you either have a memory bandwidth bound kernel and ECC turned on on the Tesla (which the GeForce isn’t capable of), or something else is going on (the Tesla might be sitting in a slot that has only half the PCIe bandwidth).

The Tesla cards have other advantages than speed.

#4

The GTX 580 is supposed to be faster than the Tesla C2050, as can be deduced from the specs. The GeForce however should not be twice as fast. If that is the case, you either have a memory bandwidth bound kernel and ECC turned on on the Tesla (which the GeForce isn’t capable of), or something else is going on (the Tesla might be sitting in a slot that has only half the PCIe bandwidth).

The Tesla cards have other advantages than speed.

#5

Thanks Tera, that helps answer my question.

#6

Thanks Tera, that helps answer my question.