C# 操作mongodb 多次查询快还是使用管道查询速度快

最近特殊原因，需要查询表数据，分别两张表，结构大概如下：
 

 public partial class ItemRelationPO : DbExtField
 {
     /// 

     /// 道具ID 唯一
     /// 
     [BsonId]
     [BsonElement("ItemOid")]
     public ulong ItemOid { get; set; }

     /// 

     /// 父节点
     /// 
     public ulong Parent { get; set; }

     /// 

     /// 所有者ID
     /// 
     public ulong Owner { get; set; }

     /// 

     /// 关联字段
     /// 
     public ulong Associated { get; set; }

     /// 

     /// 祖先节点
     /// 
    // public string AncestorsStr { get; set; }
 }
 
 
 public partial class GameItemPO : DbExtField
{
    /// 

    /// 组件ID
    /// 
    [BsonId]
    [BsonElement("ItemOid")]
    public ulong ItemOid { get; set; }

    /// 

    /// 模板ID
    /// 
    public long TemplateId { get; set; }

    /// 

    /// 数量
    /// 
    public long Count { get; set; }

    /// 

    /// NFT 信息
    /// 
    public NftInfoPO NftInfo { get; set; }

    /// 

    /// 是否是容器
    /// 
    public bool IsContainer { get; set; }

    /// 

    /// 容器容量
    /// 
    public long Capacity { get; set; }

    /// 

    /// CD 信息
    /// 
    public ItemCDlInfoPO CdInfo { get; set; }

    /// 

    /// 道具失效时间----
    /// 
    [BsonRepresentation(BsonType.String)]
    public DateTimeOffset ExpireEndTime { get; set; } = DateTimeOffset.MinValue;

}

两张表数据大概在1000W，我需要查询两张表数据，

方式1：分别两次查询，然后拼接

Stopwatch stopwatch = Stopwatch.StartNew();
var relationFilter = Builders.Filter.In(a => a.Owner, playerIds);
var data = await ItemRelationDb.Find(relationFilter).ToListAsync();
var ids = data.Select(a => a.ItemOid).ToList();

var filter = Builders.Filter.In(o => o.ItemOid, ids);
var listItems = await GameItemDb.Find(filter).ToListAsync();
//拼接成大类
var result = ItemRepositoryUtility.ConvertPOToDomain(listItems, _mapper, data);
stopwatch.Stop();
_logger.LogInformation("原始查询：{ml}", stopwatch.ElapsedMilliseconds);

方式2：使用管道查询

stopwatch.Restart();
var playerIdsBson = playerIds.Select(id => new BsonInt64((long)id)).ToList();
_logger.LogInformation("Player IDs Bson: " + string.Join(", ", playerIdsBson));
var relationFilter2 = Builders.Filter.In(a => (long)a.Owner, playerIdsBson);

_logger.LogInformation("Filter: " + relationFilter2.ToBsonDocument().ToJson());
var filterBson = new BsonDocument
{
    { "Owner", new BsonDocument { { "$in", new BsonArray(playerIdsBson) } } }
};
var pipeline = new BsonDocument[]
    {
        new("$match", filterBson),
        new ("$lookup",new BsonDocument{
            { "from", GameItemDb.CollectionNamespace.CollectionName },
            { "localField", "_id" },
            { "foreignField", "_id" },
            { "as", "gameItems" }

        }),
        new("$unwind", "$gameItems"),
        new BsonDocument("$project", new BsonDocument
            {
                { "_id", 0 },
                { "ItemOid", 1 },
                { "Parent", 1 },
                { "Owner", 1 },
                { "Associated", 1 },
                //{"item","$gameItems" }
                { "TemplateId", "$gameItems.TemplateId" },
                { "Count", "$gameItems.Count" },
                { "NftInfo", "$gameItems.NftInfo" },
               // { "Types","$gameItems._t"}
            })

    };

var result2 = await ItemRelationDb.Aggregate(pipeline)
      .ToListAsync();

测试结果：发现，两次访问所需时间比使用管道查询性能要高