Terracotta Discussion Forums (LEGACY READ-ONLY ARCHIVE)

We are thaining big AI models. They are generally looks like big arrays of objects with weights and links and should be kept in RAM (possibly virtually) while training.

We aro looking a way to scale with cluster along the size of model increases...

Yes, thanks, I've just removed this switch and it worked!

Are there any terracotta products/technologies, which allow to process big objects in memory transparently, probably sharing it between many computers?

There are many mentions of grid computing by terracotta in old documents, but most of link are broken or obsolete.

Is it able to perform grid computing with terracotta nowadays? Where I can find any samples or guides?

BigMemory samples use UseCompressOops option which is for 64bit JVM only. Do I need 64bit Java to execute samples or I can use 32bit just removing this key? I.e. does 64bit is absolutely required?

thanks!

Hi! There are lot of samples on Terracotta including usage of dso-java script to run samples. But this script is absent in downloaded terracotta. Where is it now? Thanks.