17m readI wrote this a while ago. Just ran on a MacBook Pro Mid 2015, 2.8 GHz Intel Core i7.
It computed the first 1000 digits in less than 1 minute, 2000 digits in 5 minute, 3000 digits, in 14 minute, 4000 digits in 32 minute, 5000 digits in 60 minute...
PI ; SRS 2011-07-18
; Compute PI using Plouffe and Bellard algorithm, based upon "C" code
; found at <http://iweb.dl.sourceforge.net/project/projectpi/
; Digit%20Extraction%20Methods/Plouffe%20and%20Bellard%20v1/pi1_f.c>.
KILL ^PI,^PRIME
SET ^PI=-1
FOR I=0:1:9 JOB C($INCREMENT(^PI))
QUIT
C(p) NEW $ETRAP SET $ETRAP="DO ^%ETN"
SET n=p*9+1
SET bign=+$TRANSLATE(n+20*$ZLN(10)/$ZLN(2),".","!")
SET sum=0
SET a=3 FOR QUIT:a>(2*bign) DO SET a=$$NP(a)
. SET vmax=+$PIECE($ZLN(2*bign)/$ZLN(a),".")
. SET av=1 FOR i=1:1:vmax SET av=av*a
. SET s=0,num=1,den=1,v=0,kq=1,kq2=1
. FOR k=1:1:bign DO
. . SET t=k
. . IF kq'<a FOR SET t=t\a,v=v-1 IF t#a SET kq=0 QUIT
. . SET kq=kq+1,num=num*t#av
. . SET t=2*k-1
. . DO:kq2'<a
. . . IF kq2=a FOR SET t=t\a,v=v+1 QUIT:t#a
. . . SET kq2=kq2-a
. . SET den=den*t#av,kq2=kq2+2
. . DO:v>0
. . . SET t=$$IM(den,av),t=t*num#av,t=t*k#av
. . . FOR i=v+1:1:vmax SET t=t*a#av
. . . SET s=s+t
. . . SET:s'<av s=s-av
. SET t=$$PM(10,n-1,av),s=s*t#av
. SET sum=sum+(s/av),sum=+("."_$PIECE(sum,".",2))
SET ^PI(p)=$EXTRACT($PIECE(sum,".",2)_"000000000",1,9)
JOB C($INCREMENT(^PI))
QUIT
NP(a) NEW (a) FOR SET r=$ORDER(^PRIME(a)) QUIT:r'="" DO QUIT:r'=""
. LOCK ^PRIME
. SET r=$ORDER(^PRIME(a)) IF r'="" LOCK QUIT
. IF $DATA(^PRIME)=0 SET ^PRIME=3,^PRIME(2)="",^PRIME(3)="" LOCK QUIT
. FOR r=^PRIME:2 DO IF pr SET ^PRIME(r)="" QUIT:r>a
. . SET pr=1 FOR p=3:2:$ZSQR(r) IF r#p=0 SET pr=0 QUIT
. SET ^PRIME=r
. LOCK
QUIT r
IM(x,y) NEW (x,y)
SET u=x,v=y,c=1,a=0
FOR SET q=v\u,t=c,c=a-(q*c),a=t,t=u,u=v-(q*u),v=t QUIT:u=0
SET a=a#y
SET:a<0 a=a+y
QUIT a
PM(a,b,m) NEW (a,b,m)
SET r=1,aa=a
FOR SET:b#2 r=r*aa#m SET b=b\2 QUIT:b=0 SET aa=aa*aa#m
QUIT r
In most recent versions of Caché, we install a directory dev/cache/callout/demo. This contains demonstration code for writing $ZF() functions. One of the examples is tzchange(), and it works by setting a environment variable in the current process, and it should be easy enough to adapt to your needs. Start by reading czf.pdf.
Social networks
As a rule of thumb disks allocate files in blocks larger than the I/O block. This is called the allocation size or cluster size. I believe the default for large Windows disks is 64 kio. This page has some explanation: <https://docs.microsoft.com/en-us/Exchange/plan-and-deploy/deployment-ref..., search for NTFS allocation unit size.