Question

In a slight mental tangle and I expect this is easier than I imagine.
Got the following tables:

create table #x
(
handid int,
cardid int
)
insert into #x
values
(1,2),(1,3),(1,4),(1,5),(1,6),(1,7),(1,8),
(2,2),(2,3),(2,4),(2,300),(2,400),(2,500),(2,8),
(3,2),(3,3),(3,4),(3,300),(3,400),(3,7),(3,8),
(4,2),(4,300),(4,400),(4,500),(4,6),(4,7),(4,8)


create table #winners(cardid int)
insert into #winners values(300),(400),(500)

select a.* 
from 
        #x a 
        inner join #winners b
            on
            a.cardid = b.cardid 

This returns the following:

I only want this query to return the rows when all of the three cardids exists for a handid. So the desired result set would not include handid 3.

This is a model of reality.
In reality #x contains 500 mill records.

EDIT

Ok – there are actually winners that are made up of sets of data from #winners which have a variable number of records. So amending the original code to the following the result set should not include handId 1 or handId 3. I am also getting some unwanted duplicate records in the result set:

create table #x
(
handid int,
cardid int
)
insert into #x
values
(1,2),(1,3),(1,4),(1,5),(1,6),(1,7),(1,8000),
(2,2),(2,3),(2,4),(2,300),(2,400),(2,500),(2,8),
(3,2),(3,3),(3,4),(3,300),(3,400),(3,7),(3,8),
(4,2),(4,300),(4,400),(4,500),(4,6),(4,7),(4,8)


create table #winners(winningComb char(1), cardid int)
insert into #winners values('A',300),('A',400),('A',500),('B',8000),('B',400)

select a.* 
from 
        #x a 
        inner join #winners b
            on
            a.cardid = b.cardid 

Answer 1

You can use something like this:

select handid
from #x  
where cardid in (select cardid from #winners)
group by handid
having count(handid) = (select count(distinct cardid)
                         from #winners);

See SQL Fiddle with Demo

Result:

| HANDID |
----------
|      2 |
|      4 |

Based on your edit, here is an attempt that returns the correct result however I am not sure if it will work with the larger dataset that you have:

;with cte as
(   
    select w1.cardid, w1.winningComb, w2.ComboCardCount
    from winners w1
    inner join
    (
        select COUNT(*) ComboCardCount, winningComb
        from winners
        group by winningComb
    ) w2
        on w1.winningComb = w2.winningComb
) 
select a.handid
from x a
inner join cte b
    on a.cardid = b.cardid
where a.cardid in (select cardid from cte)
group by handid, b.ComboCardCount
having COUNT(a.handid) = b.ComboCardCount

See SQL Fiddle with Demo

Result:

| HANDID |
----------
|      2 |
|      4 |