Commit 0011765b authored by Yukikaze-CZR's avatar Yukikaze-CZR
Browse files

Change ivfsq8 test report detailed version #255

parent 31aa89e1
Loading
Loading
Loading
Loading
+4 −4
Original line number Diff line number Diff line
@@ -28,13 +28,13 @@ Memory: 503GB

Docker version: 18.09

Nvidia Driver version: 430.34
NVIDIA Driver version: 430.34

Milvus version: 0.5.3

SDK interface: Python 3.6.8

Pymilvus version: 0.2.5
pymilvus version: 0.2.5



@@ -66,7 +66,7 @@ For details on this dataset, you can check : http://corpus-texmex.irisa.fr/ .

  > Note: In the query test of recall, we will test the following parameters with different values:
  >
  > nq - grouped by: [1, 5, 10,  200, 400, 600, 800, 1000], 
  > nq - grouped by: [10,  200, 400, 600, 800, 1000], 
  >
  > topk - grouped by: [1, 10, 100]

@@ -93,7 +93,7 @@ Milvus configuration
- gpu_cache_capacity: 6
- use_blas_threshold: 1100

You can check the definition of Milvus configuration on https://milvus.io/docs/en/reference/milvus_config/.
The definitions of Milvus configuration are on https://milvus.io/docs/en/reference/milvus_config/.

Test method

+5 −5
Original line number Diff line number Diff line
@@ -2,7 +2,7 @@

## 概述

本文描述了ivfsq8索引在milvus单机部署方式下的测试报告
本文描述了ivfsq8索引在milvus单机部署方式下的测试结果



@@ -28,13 +28,13 @@ GPU1: GeForce GTX 1080

Docker版本: 18.09

Nvidia Driver版本: 430.34
NVIDIA Driver版本: 430.34

Milvus版本: 0.5.3

SDK接口: Python 3.6.8

Pymilvus版本: 0.2.5
pymilvus版本: 0.2.5



@@ -66,7 +66,7 @@ Pymilvus版本: 0.2.5

  > 备注:在向量准确性测试中,我们会测试下面参数不同的取值来观察结果:
  >
  > 被查询向量的数量nq将按照 [1, 5, 10,  200, 400, 600, 800, 1000]的数量分组,
  > 被查询向量的数量nq将按照 [10,  200, 400, 600, 800, 1000]的数量分组,
  >
  > 单条查询中最相似的K个结果topk将按照[1, 10, 100]的数量分组。

@@ -93,7 +93,7 @@ Milvus设置:
- gpu_cache_capacity: 6
- use_blas_threshold: 1100

你可以在 https://milvus.io/docs/en/reference/milvus_config/上查询Milvus设置的详细定义
Milvus设置的详细定义可以参考 https://milvus.io/docs/en/reference/milvus_config/ 

测试方法