MLT working - 9000 series or higher only Tweet

[YoonKyung]

yes,

we had to redo the multi-gpu using a different technique, so we launched 2.1 without it,
moved multi-gpu to the upcoming 2.2, and moved MLT to the 2.3 after that.

currently we're dug deep into the multi-gpu, which is nearly done.
then we break our heads over finishing MLT.

Radiance

A different technique? Could you let us know it?

[radiance]

it's not different for users, it's just a different internal architecture / way of doing it in the code.
the usage remains the same, and the fact that you can render larger resolutions with more GPUs is still true.

Radiance

[juan]

so gpus are sharing memory or something?