Eric Lam
Eric Lam
I think the right thing to read whole file content is to read by multiple chunks via datalink(offset, size) just like HNSW did. select 0, cast('stage_url?offset=0&size=64000' as datalink); select 1,...
this feature is already done. please verify and close
The PR draft link is https://github.com/unum-cloud/usearch/pull/638 HardwareAcceleration is serial To compile the usearch library. (cd usearch-2.17.1 && cmake -B build -DUSEARCH_BUILD_BENCH_CPP=OFF -DUSEARCH_BUILD_TEST_C=OFF -DUSEARCH_BUILD_TEST_CPP=OFF -DUSEARCH_BUILD_LIB_C=ON -DCMAKE_CXX_FLAGS=-I/home/ubuntu/p/matrixone/thirdparties/install/include -DUSEARCH_USE_OPENMP=1 && cmake --build build...