Re: Gradual Growth of Memory Use?

Martin informs me that my Memory Dump did not make it through the email system, being too large. So here is the front of it:


(#S(CLASS-COUNTER :NAME CONS
                  :COUNT 7434476
                  :TOTAL-SIZE 118951616
                  :CUMULATIVE-SIZE 288117792)
 #S(CLASS-COUNTER :NAME FUNCTION
                  :COUNT 1160246
                  :TOTAL-SIZE 61906744
                  :CUMULATIVE-SIZE 169166176)
 #S(CLASS-COUNTER :NAME SIMPLE-VECTOR
                  :COUNT 1768914
                  :TOTAL-SIZE 58177208
                  :CUMULATIVE-SIZE 107259432)
 #S(CLASS-COUNTER :NAME VECTOR
                  :COUNT 3161
                  :TOTAL-SIZE 13120960
                  :CUMULATIVE-SIZE 49082224)
 #S(CLASS-COUNTER :NAME COM.RAL.ACTORS:ACTOR
                  :COUNT 342304
                  :TOTAL-SIZE 8215296
                  :CUMULATIVE-SIZE 35961264)
 #S(CLASS-COUNTER :NAME CLOS::SYSTEM-OBJECT
                  :COUNT 128033
                  :TOTAL-SIZE 7296456
                  :CUMULATIVE-SIZE 27745968)
 #S(CLASS-COUNTER :NAME SIMPLE-STRING
                  :COUNT 22434
                  :TOTAL-SIZE 5240472
                  :CUMULATIVE-SIZE 20449512)
 #S(CLASS-COUNTER :NAME SYMBOL
                  :COUNT 103354
                  :TOTAL-SIZE 4960992
                  :CUMULATIVE-SIZE 15209040)
 #S(CLASS-COUNTER :NAME SIMPLE-BASE-STRING
                  :COUNT 126134
                  :TOTAL-SIZE 3815736
                  :CUMULATIVE-SIZE 10248048)
 #S(CLASS-COUNTER :NAME DOUBLE-FLOAT
                  :COUNT 104678
                  :TOTAL-SIZE 1674848
                  :CUMULATIVE-SIZE 6432312)
 #S(CLASS-COUNTER :NAME COMPLEX
                  :COUNT 21622
                  :TOTAL-SIZE 518928
                  :CUMULATIVE-SIZE 4757464)
 #S(CLASS-COUNTER :NAME HASH-TABLE
                  :COUNT 2039
                  :TOTAL-SIZE 407800
                  :CUMULATIVE-SIZE 4238536)
 #S(CLASS-COUNTER :NAME STANDARD-GENERIC-FUNCTION
                  :COUNT 6811
                  :TOTAL-SIZE 381416
                  :CUMULATIVE-SIZE 3830736)
 #S(CLASS-COUNTER :NAME CLOS::CACHE-INFO
                  :COUNT 4692
                  :TOTAL-SIZE 337824
                  :CUMULATIVE-SIZE 3449320)
 #S(CLASS-COUNTER :NAME STANDARD-METHOD
                  :COUNT 9784
                  :TOTAL-SIZE 234816
                  :CUMULATIVE-SIZE 3111496)
 #S(CLASS-COUNTER :NAME EDITOR::TEXT-PROPERTY-POINT
                  :COUNT 2871
                  :TOTAL-SIZE 229680
                  :CUMULATIVE-SIZE 2876680)
 #S(CLASS-COUNTER :NAME STANDARD-EFFECTIVE-SLOT-DEFINITION
                  :COUNT 9262
                  :TOTAL-SIZE 222288
                  :CUMULATIVE-SIZE 2647000)



> On Mar 19, 2025, at 06:11, David McClain <[email protected]> wrote:
> 
> The system is under no overt memory pressure. Total running image on the machine is around 6 GB out of 32 GB total RAM. No overt GC calls anywhere, unless something down in LWW does that beneath the table.
> 
> Explorer.exe is reasonably well behaved at the moment with an allocation of 200 MB. Later, it will grow to several GB for unknown reasons. It never gives back and must be restarted from cold to reclaim memory.
> 
> The dump did not include Gen0.
> 
> I see the dump indicating that I have 342,000 Actors allocated. These are 1-slot structs with a pointer to a functional closure. I didn’t define that many Actors, but copies can be CREATEd at any time by an Actor behavior, many are anonymous continuation Actors. But clearly, the GC has not discarded most of them. At any one time, I would expect the extant live Actors to be no more than a few hundred. Most will be used once and discarded by non-reference, awaiting GC.
> 
> Since I don’t call GC, and I don’t know what it considers a higher generation collection pressure, the GC might well have not done much except to graduate objects from Gen0.
> 
> So I don’t know how to view the dump in a meaningful way, except, perhaps, to come back in a few hours and see how it has grown.
> 
> - DM
> 
>> On Mar 19, 2025, at 05:29, David McClain <[email protected]> wrote:
>> 
>> That was a dump of the system, having run about 4 hours, now up to 300 MB, which started from about 120 MB.
>> 
>>> On Mar 19, 2025, at 05:22, David McClain <[email protected]> wrote:
>>> 
>>> Thanks Tim, what does this tell you?
>>> 
>>> 
>>> (#S(CLASS-COUNTER :NAME CONS
>>>                  :COUNT 7434476
>>>                  :TOTAL-SIZE 118951616
>>>                  :CUMULATIVE-SIZE 288117792)
>>> #S(CLASS-COUNTER :NAME FUNCTION
>>>                  :COUNT 1160246
>>>                  :TOTAL-SIZE 61906744
>>>                  :CUMULATIVE-SIZE 169166176)
>>> #S(CLASS-COUNTER :NAME SIMPLE-VECTOR
>>>                  :COUNT 1768914
>>>                  :TOTAL-SIZE 58177208
>>>                  :CUMULATIVE-SIZE 107259432)
>>> #S(CLASS-COUNTER :NAME VECTOR
>>>                  :COUNT 3161
>>>                  :TOTAL-SIZE 13120960
>>>                  :CUMULATIVE-SIZE 49082224)
>>> #S(CLASS-COUNTER :NAME COM.RAL.ACTORS:ACTOR
>>>                  :COUNT 342304
>>>                  :TOTAL-SIZE 8215296
>>>                  :CUMULATIVE-SIZE 35961264)
>>> #S(CLASS-COUNTER :NAME CLOS::SYSTEM-OBJECT
>>>                  :COUNT 128033
>>>                  :TOTAL-SIZE 7296456
>>>                  :CUMULATIVE-SIZE 27745968)
>