AnsweredAssumed Answered

bandwidth of texture units in gcn

Question asked by dmeiser on Feb 23, 2012
Latest reply on Feb 23, 2012 by dmeiser



I've been trying to find the theoretical peak bandwidth of the texture units on radeon 7970 gpus but haven't been able to come up with something.  Could somebody point to a spec sheet that has that number?


The reason why I ask is that in the past it was often recommended to employ the texture units for reading from global memory for bandwidth limited kernels (e.g. on a 5870 the aggregate texture bandwidth is ~1TB/s, about 5 times more than bandwidth from global memory).  Is this still a recommended optimization for gcn gpus?