SQL Server optimize code in Stored Procedure

Comparing two codes below,both do the same,but with slighty differences:

ALTER procedure [dbo].[SP_USUARIOS_UPDATE]
    @usu_ds varchar(100),
    @usu_dt_lst_log datetime,
    @usu_ds_senha varchar(255),
    @usu_ds_email varchar(100)
    as
    begin
    declare @usu_ID int;
    create table #TempUser
    (
        UsuID int,
        Senha varchar(255),
        Email varchar(100)
    )
    select Usuarios.usu_ID as UsuID,Usuarios.usu_ds_senha as Senha, 
    Usuarios.usu_ds_email as Email into #TempUser from Usuarios where Usuarios.usu_ds = @usu_ds
    if(@usu_ds_senha is null)
    begin
    set @usu_ds_senha = (select #TempUser.Senha from #TempUser);
    end
    if(@usu_ds_email is null)
    begin
    set @usu_ds_email = (select #TempUser.Email from #TempUser);
    end
    set @usu_ID = (select #TempUser.UsuID from #TempUser);
    update Usuarios set usu_dt_lst_log = 
    @usu_dt_lst_log,usu_ds_senha = @usu_ds_senha,usu_ds_email = @usu_ds_email where usu_ID = @usu_ID
    end

AND

ALTER procedure [dbo].[SP_USUARIOS_UPDATE]
@usu_ds varchar(100),
@usu_dt_lst_log datetime,
@usu_ds_senha varchar(255),
@usu_ds_email varchar(100)
as
begin
declare @usu_ID int;
if(@usu_ds_senha is null)
begin
set @usu_ds_senha = (select Usuarios.usu_ds_senha from Usuarios where Usuarios.usu_ds = @usu_ds);
end
if(@usu_ds_email is null)
begin
se开发者_开发技巧t @usu_ds_email = (select Usuarios.usu_ds_email from Usuarios where Usuarios.usu_ds = @usu_ds);
end
set @usu_ID = (select Usuarios.UsuID from Usuarios where Usuarios.usu_ds = @usu_ds);
update Usuarios set usu_dt_lst_log = 
@usu_dt_lst_log,usu_ds_senha = @usu_ds_senha,usu_ds_email = @usu_ds_email where usu_ID = @usu_ID
end

Do you think the first is faster than second in performance,i mean,first code use temp table (#TempUser) to store 3 fields from the real table.Second code,select all fields from the real table one by one.

What code is best optimized?

---

First -- if it's possible for your passed parameters to be null, you need to set defaults. For example:

@usu_ds_email varchar(100) = null
...

Otherwise, your null checks further down will never come into play -- the procedure will just fail.

Second -- just run a direct update. It seems like you're pushing a lot of data back and forth unnecessarily. E.g., you don't need to create a temp table from the table you're going to update, and then turn right around and update your table from the temp table you just created.

ALTER procedure [dbo].[SP_USUARIOS_UPDATE]
@usu_ds varchar(100),
@usu_dt_lst_log datetime,
@usu_ds_senha varchar(255) = null,
@usu_ds_email varchar(100) = null
as
begin

    update Usuarios 
    set usu_dt_lst_log = @usu_dt_lst_log,
        usu_ds_senha = isnull(@usu_ds_senha, usu_ds_senha),
        usu_ds_email = isnull(@usu_ds_email, usu_ds_email)
    where usu_ID = @usu_ds

end

---

Third way...

ALTER procedure [dbo].[SP_USUARIOS_UPDATE] 
@usu_ds varchar(100), 
@usu_dt_lst_log datetime, 
@usu_ds_senha varchar(255) = null, 
@usu_ds_email varchar(100) = null 
as 
begin 

update x 
set x.usu_dt_lst_log = @usu_dt_lst_log, 
    x.usu_ds_senha = ISNULL(@usu_ds_senha, x.usu_ds_senha), 
    x.usu_ds_email = ISNULL(@usu_ds_email, x.usu_ds_email)
from Usuarios x where x.usu_ds = @usu_ds
end 

---

Which one runs faster? Set a profiler, run both, and get some real data: then you'll know.

However, based on my own previous experiences, the Temp Table has always been faster for me.