Jason Lowe

The ArrayIndexOutOfBoundsException is a bug in BatchWithPartitionDataUtils.splitColumnarBatch where it's not handling a row-count-only columnar batch properly. Filed #11155 to track that bug.

> It seems that it is the https://github.com/NVIDIA/spark-rapids/discussions/9966 that I have already reported before. Yes, this is likely the same issue. I assume the [workaround you noted before](https://github.com/NVIDIA/spark-rapids/discussions/9966#discussioncomment-7769087) still works?

Glad to hear you have a path forward with the workaround. Did I understand correctly that running the 24.08 snapshot with the fix from #11155 resolved this issue? We can...

> Does Spark-RAPIDS use LIBCUDF_NVCOMP_POLICY option of STABLE ? Indirectly, yes. We don't set the LIBCUDF_NVCOMP_POLICY environment variable, and STABLE is the default.

After chatting with the Nsight Systems team, we can probably accomplish most of the tracing needs we want by leveraging the [cupti](https://developer.nvidia.com/cupti) toolkit. This won't generate a qdrep file, but...

I'm personally OK if a spark.rapids.memory.gpu.allocSize setting causes a crash if it goes above the max fraction (or otherwise conflicts with other settings). Essentially you'd have to increase max alloc...

Tracked down this issue, it's caused by the Parquet scan mutating the same Hadoop configuration instance that was received from the broadcast. When files don't have the same schema, it...

Currently the time range and job/stage range profiling configs are mutually exclusive. Even if they were allowed to be combined, it wouldn't quite do what is being requested here. IIUC...

Regarding JDK17+JDK8 in the same jar, I don't think that's technically impossible. jars are just zip files, and each class file separately contains metadata with what JDK it was compiled...

> Do you think this solution is reasonable? The answer depends on how well it performs in practice. The problems we've seen in the past with PERFILE are aggravated by...