Default Slowdown Threshold for Tesla C1060

Accelerated Computing CUDA CUDA Programming and Performance
1

Hi,

I have two Tesla workstations. The first one is a Dell precision T7400 with one NVS 290 and one Tesla C1060. The other one is build by ourself with one NVS 290 and three Tesla C1060 on the ASUS P6T7 motherboard. When I checked both workstation for the default value of slowdown threshold in the Nvidia driver, the Tesla C1060 in the Dell Precision T7400 has a value of 190C while the three Tesla cards inside the other workstation has a value of 110C.

I’d like to know whether it is normal that different Tesla configurations would result in different slowdown threshold value and if it is possible to change the slowdown threshold.

Thanks

2

yeah, I remember when this was first noticed. it’s just a reporting bug in one driver–the slowdown threshold is the same at 100C, with shutdown at 105C (or maybe 110).

3

will that cause any problem? and did you mean that the tesla card reporting 190 has bug while the three reporting 110 do not?

The three tesla workstation setting screen shows yellow, meaning we are using resources near the limit, while dell workstation shows green, both without running applications

on the cards.