There's just one gpu_available_frac, which is the fraction of time when BOINC is running that GPUs are available to compute. With the recent change, this is used in the estimation of whether GPU jobs will be completed by their deadline.
Work requests (#seconds, #instances) are sent separately for each GPU type. But that's something different. -- David On 20-Mar-2014 2:14 PM, Jon Sonntag wrote:
Just curious... Does it maintain values for each type, nVidia and AMD? e.g. If a machine has both and it has an over abundance of one, will it go into panic mode on both GPUs?
_______________________________________________ boinc_dev mailing list [email protected] http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev To unsubscribe, visit the above URL and (near bottom of page) enter your email address.
