VectorDBBench 0.0.7 Release Notes

What's Changed

• Improving the performance of PGVectors by @yugeeklab in #235
• add new search_params for zillzcloud: level by @alwayslove2013 in #274
• add new index_type for milvus: ivf_sq8 by @alwayslove2013 in #277
• add dev container support by @zhjwpku in #280
• enhance: Support for windows by @XuanYang-cn in #290

Bug fixes

• test: fix UT unable to iter local cohere 10M by @XuanYang-cn in #266
• Fix normalization on data by @dangkhoasdc in #259
• Make Docker Images for VectorDB Bench by @ljqcodelove in #244
• issue 276 by @s-gruneberg in #278
• fix: tell setuptools to find package by vectordb_bench by @XuanYang-cn in #282
• fix: Unable to find module "vectordb_bench" by @XuanYang-cn in #289
• fix: Unable to download some files from AliyunOSS by @XuanYang-cn in #287
• enhance: Remove etag checks by @XuanYang-cn in #291

New Contributors

• @yugeeklab made their first contribution in #235
• @dangkhoasdc made their first contribution in #259
• @ljqcodelove made their first contribution in #244
• @zhjwpku made their first contribution in #280

Full Changelog: v0.0.6...v0.0.7

VectorDBBench 0.0.6 Release Notes

New Benchmark Results

• add test results of zilliz_cloud_beta_202401 by @alwayslove2013 in #258

New DB client

• feat: add new db pgvecto.rs by @cutecutecat in #242

New features

• Import database client while in use by @XuanYang-cn in #223
• [Milvus client]support gpu_index_type for milvus by @alwayslove2013 in #256 #261 #264
• ui support setting drop_old and dataset_address by @alwayslove2013 in #260
• Enable aliyun OSS by @XuanYang-cn in #262

Bug fixes

• [Milvus client]Wait for index done by @XuanYang-cn in #214
• fix_spelling by @alwayslove2013 in #226
• fix qp_dollar calculation by @alwayslove2013 in #230
• Fix milvus capacity test unable to load by @XuanYang-cn in #239
• fix bugs: milvus _preload by @alwayslove2013 in #246
• Correcting SIFT datasize for capacity case by @devanshsa5 in #252
• [Redis]Send password, batch pipeline by @ewpreston in #247
• update the scoring rules of latency by @alwayslove2013 in #251

New Contributors

• @cutecutecat made their first contribution in #242
• @ewpreston made their first contribution in #247
• @devanshsa5 made their first contribution in #252

Full Changelog: v0.0.5...v0.0.6

VectorDBBench 0.0.5 Release Notes

What's Changed

Fix some data points lost issues

Full Changelog: v0.0.4...v0.0.5

VectorDBBench 0.0.4 Release Notes

1. Add Chroma and Redis Client
2. Add 6 more cases and results

cases	dataset	filter	description
Search Performance Case	OpenAI generated 500K vectors, 1536 dimensions	N/A	Index building time, recall, latency, maximum QPS
Search Performance Case	OpenAI generated 5M vectors, 1536 dimensions	N/A	Index building time, recall, latency, maximum QPS
Filtering Search Performance Case	OpenAI generated 500K vectors, 1536 dimensions	1% vectors	Index building time, recall, latency, maximum QPS
Filtering Search Performance Case	OpenAI generated 5M vectors, 1536 dimensions	1% vectors	Index building time, recall, latency, maximum QPS
Filtering Search Performance Case	OpenAI generated 500K vectors, 1536 dimensions	99% vectors	Index building time, recall, latency, maximum QPS
Filtering Search Performance Case	OpenAI generated 5M vectors, 1536 dimensions	99% vectors	Index building time, recall, latency, maximum QPS

VectorDBBench 0.0.3 Release Notes

What's New

1. Add pgvector client and benchmark results
2. Support Leaderborad: https://zilliz.com/benchmark
3. Retest and update qdrant, pinecone, milvus, and zillizcloud results
4. Add Timeout for all cases

Enhancement

1. Sync all process before multi-processing search starts.
2. Refine DataSet

VectorDBBench 0.0.2 Release Notes

• Add Pinecone results
• Remove results of 100K
• Enable to save results as one image
• Enable to select all, deselect all filters