# cutGetMaxGflopsDeviceId

**URL:** <https://forums.developer.nvidia.com/t/cutgetmaxgflopsdeviceid/15723>\
**Category:** CUDA Programming and Performance\
**Created:** [March 26, 2010, 8:39am UTC](https://forums.developer.nvidia.com/t/cutgetmaxgflopsdeviceid/15723 "2010-03-26T08:39:20Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![You7878](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@You7878](https://forums.developer.nvidia.com/u/You7878)\
**Post date:** [March 26, 2010, 8:39am UTC](https://forums.developer.nvidia.com/t/cutgetmaxgflopsdeviceid/15723/1 "2010-03-26T08:39:20Z")

</div>

it looks like cutGetMaxGflopsDeviceId will work incorrect for fermi.  
it calculates gflops as: max\_gflops = device\_properties.multiProcessorCount \* device\_properties.clockRate;  
actually this comparison is legal only if number of cores in SM is equal and that’s true only for G80 and GT200.  
does anyone know whether this function is rewritten in 3.0?