Karthikeyan
Karthikeyan
**Describe the bug** Due to limitations of certain algorithms such `cub::ExclusiveScan`, and json tree implementation, the input json can not exceed limit 2GiB. (INT_MAX) The json_read_data_type INTEGRAL benchmark fail to...
## Description This PR adds optional column order to enforce column order in the output. This feature is useful for spark from_json. TBD. ## Checklist - [x] I am familiar...
There are 2 bugs reported in the issue. When I loaded the hidream_e1 from Comfy UI templates, it showed many missing models. So, I opened the Model Manager and checked...
[cuDF] Decouple CPU and GPU Driver Parallelism to Avoid CPU Underutilization with cuDF DriverAdapter
### Description When executing GPU operators via the cuDF `DriverAdapter`, the number of Velox drivers (threads) must be kept relatively small (typically 1–8) to efficiently saturate the GPU. However, Velox...
### Description The decimal type is not supported in cudf GPU operators. The conversion fails because interop API (toArrow and FromArrow) had a bug. So, decimal support was never added...
When running Presto TPC-DS queries with scale factor 100 with a single Velox worker and the cuDF backend enabled and CPU fallback enabled, some queries falls back to CPU in...
This PR adds an example to use libucxx and libcudf together. This example is based on https://github.com/rapidsai/ucxx/blob/main/cpp/examples/basic.cpp The build is done in 2 ways conda and docker. Both does not...