Options are:
  *  Leave it completely up to the database (AMM)
then just set memory_max_size and memory_target (and sga_max_size and sga_target to 0)\\
This allows the database to manipulate both sga and pga as it feels necessary. Latest recommendation is to use this only if the total amount of memory available for the database is less than 4G.
  *  Set maximum values for SGA and PGA (ASMM)
then just set sga_max_size and sga_target (and memory_max_size and memory_target to 0)

Set pga_aggregate_target and eventually ringfence it with pga_aggregate_limit.

Setting any of the pool sizes means this value become a minimum.

==== Queries for the sga ====
<code>
select * from v$sga; 

select sum(value) from v$sga; 

select sum(bytes) from v$sgastat;

select sum(current_size) from v$sga_dynamic_components;

select * from v$sga_dynamic_free_memory;

select component, current_size from v$sga_dynamic_components;

select pool, sum(bytes)/1024/1024 from v$sgastat group by pool;

alter system flush buffer_cache;

alter system flush shared_pool;

select 
  component, 
  oper_type, 
  oper_mode, 
  initial_size/1024/1024 "Initial", 
  TARGET_SIZE/1024/1024 "Target", 
  FINAL_SIZE/1024/1024 "Final", 
  status 
from 
  v$sga_resize_ops order by component;

select 
  component, 
  current_size/1024/1024 "CURRENT_SIZE", 
  min_size/1024/1024 "MIN_SIZE",
  user_specified_size/1024/1024 "USER_SPECIFIED_SIZE", 
  last_oper_type "TYPE" 
from 
  v$sga_dynamic_components order by component;

select * from v$sgainfo where name = 'Granule Size';
</code>
==== Show PGA Memory Allocation/Usage for Database Sessions ====
<code>
SET PAGESIZE 60
SET LINESIZE 300
 
COLUMN username FORMAT A20
COLUMN module FORMAT A50
COLUMN program FORMAT A50
 
SELECT NVL(a.username,'(oracle)') AS username,
       a.module,
       a.program,
       Trunc(b.value/1024) AS Memory_KB
FROM   v$session a,
       v$sesstat b,
       v$statname c
WHERE  a.sid = b.sid
AND    b.statistic# = c.statistic#
AND    c.name = 'session pga memory'
AND    a.program IS NOT NULL
ORDER BY b.value DESC
/
</code>
=== Handy views to look at ===
<code>
V$SGAINFO
V$SGASTAT
V$SGA_DYNAMIC_COMPONENTS
</code>

<code>
col name for a40
select * from v$sgainfo
</code>
produces something like this...
<code>
NAME                                          BYTES RESIZEABLE
---------------------------------------- ---------- ------------
Fixed SGA Size                              2163592 No
Redo Buffers                               17948672 No
Buffer Cache Size                        1493172224 Yes
Shared Pool Size                         1644167168 Yes
Large Pool Size                            16777216 Yes
Java Pool Size                             16777216 Yes
Streams Pool Size                          16777216 Yes
Shared IO Pool Size                               0 Yes
Granule Size                               16777216 No
Maximum SGA Size                         3207790592 No
Startup overhead in Shared Pool           365946152 No
Free SGA Memory Available                         0

12 rows selected.
</code>
or
<code>
select name, (bytes/1024/1024) megs,resizeable from v$sgainfo;

NAME                                      MEGS RESIZEABLE
----------------------------------- ---------- ------------
Fixed SGA Size                       2.1346283 No
Redo Buffers                         381.84375 No
Buffer Cache Size                         4608 Yes
Shared Pool Size                         18432 Yes
Large Pool Size                            512 Yes
Java Pool Size                             512 Yes
Streams Pool Size                         1024 Yes
Shared IO Pool Size                          0 Yes
Granule Size                               512 No
Maximum SGA Size                    764799.984 No
Startup overhead in Shared Pool     4617.96307 No
Free SGA Memory Available               739328

12 rows selected.
</code>

=== How memory is used (assuming conventional path reads - not direct path and dedicated server) ===
A user runs a query which either finds the blocks it needs already in the SGA or fetches the blocks\\
needed from disk into the SGA.\\
The desired rows/data is taken from those blocks in the SGA and pulled over to the PGA for sorting and/or grouping.\\
If more than a "reasonable amount" of memory is required then the sorting and/or grouping spills over to TEMP space on disk and ultimately returns the rows to the user.\\

=== References ===
[[http://asktom.oracle.com/pls/asktom/f?p=100:11:0::::P11_QUESTION_ID:1137385000346840516|asktom]]