Constant pool should share values

[BruceForstall]

Doubles added to constant pool are not shared: the same constant appearing multiple times appears multiple times in the constant pool.

Using the example from #35257:

private static double Process(double n)
{
  double res;
  res = 1;
  while (n > 0.0)
  {
    res *= n;
// User might not write such code, but here it is written merely to show that same constants are re-loaded in assembly code.
    n -= 1.0;
    n -= 2.0;
    n -= 1.0;
    n -= 2.0;
  }
  return res;
}

The generated x64 assembly is:

; Assembly listing for method BringUpTest:Process(double):double
; Emitting BLENDED_CODE for X64 CPU with AVX - Windows
; optimized code
; rsp based frame
; fully interruptible
; Final local variable assignments
;
;  V00 arg0         [V00,T00] ( 13, 43   )  double  ->  mm0
;  V01 loc0         [V01,T01] (  4, 10   )  double  ->  mm1
;# V02 OutArgs      [V02    ] (  1,  1   )  lclBlk ( 0) [rsp+0x00]   "OutgoingArgSpace"
;
; Lcl frame size = 0

G_M28499_IG01:
       C5F877               vzeroupper
                                                ;; bbWeight=1    PerfScore 1.00
G_M28499_IG02:
       C5FB100D4D000000     vmovsd   xmm1, qword ptr [reloc @RWD00]
       C5E857D2             vxorps   xmm2, xmm2
       C5F92EC2             vucomisd xmm0, xmm2
       7638                 jbe      SHORT G_M28499_IG04
                                                ;; bbWeight=1    PerfScore 4.33
G_M28499_IG03:
       C5F359C8             vmulsd   xmm1, xmm1, xmm0
       C5FB5C053F000000     vsubsd   xmm0, xmm0, qword ptr [reloc @RWD08]
       C5FB5C053F000000     vsubsd   xmm0, xmm0, qword ptr [reloc @RWD16]
       C5FB5C053F000000     vsubsd   xmm0, xmm0, qword ptr [reloc @RWD24]
       C5FB5C053F000000     vsubsd   xmm0, xmm0, qword ptr [reloc @RWD32]
       C5E857D2             vxorps   xmm2, xmm2
       C5F92EC2             vucomisd xmm0, xmm2
       77D2                 ja       SHORT G_M28499_IG03
                                                ;; bbWeight=4    PerfScore 101.33
G_M28499_IG04:
       C5F828C1             vmovaps  xmm0, xmm1
                                                ;; bbWeight=1    PerfScore 0.25
G_M28499_IG05:
       C3                   ret
                                                ;; bbWeight=1    PerfScore 1.00
RWD00  dq       3FF0000000000000h
RWD08  dq       3FF0000000000000h
RWD16  dq       4000000000000000h
RWD24  dq       3FF0000000000000h
RWD32  dq       4000000000000000h


; Total bytes of code 72, prolog size 3, PerfScore 116.22, (MethodHash=7c9e90ac) for method BringUpTest:Process(double):double
; ============================================================

In this case, RWD00, RWD08, and RWD24 are identical, and RWD16 and RWD32 are identical.

The constant pool is (conceptually) read-only (I believe), so we only need one of each unique value in the table.

category:cq
theme:constant-pool
skill-level:intermediate
cost:medium

[anthonycanino]

@BruceForstall

I think this issue may have been resolved. On my machine, the sample above generates the following code:

; Assembly listing for method Program:Process(double):double
; Emitting BLENDED_CODE for X64 CPU with AVX - Windows
; optimized code
; rsp based frame
; fully interruptible
; No PGO data
; Final local variable assignments
;
;  V00 arg0         [V00,T00] ( 13, 43   )  double  ->  mm0        
;  V01 loc0         [V01,T01] (  4, 10   )  double  ->  mm1        
;# V02 OutArgs      [V02    ] (  1,  1   )  lclBlk ( 0) [rsp+00H]   "OutgoingArgSpace"
;  V03 cse0         [V03,T02] (  3,  9   )  double  ->  mm2         "CSE - aggressive"
;
; Lcl frame size = 0

G_M22436_IG01:              ;; offset=0000H
       C5F877               vzeroupper 
						;; bbWeight=1    PerfScore 1.00
G_M22436_IG02:              ;; offset=0003H
       C5FB100D4D000000     vmovsd   xmm1, qword ptr [reloc @RWD00]
       C5E857D2             vxorps   xmm2, xmm2
       C5F92EC2             vucomisd xmm0, xmm2
       7636                 jbe      SHORT G_M22436_IG04
       C5FB101543000000     vmovsd   xmm2, qword ptr [reloc @RWD08]
                            align    [0 bytes]
						;; bbWeight=1    PerfScore 9.33
G_M22436_IG03:              ;; offset=001DH
       C5F359C8             vmulsd   xmm1, xmm1, xmm0
       C5FB5C052F000000     vsubsd   xmm0, xmm0, qword ptr [reloc @RWD00]
       C5FB5CC2             vsubsd   xmm0, xmm0, xmm2
       C5FB5C0523000000     vsubsd   xmm0, xmm0, qword ptr [reloc @RWD00]
       C5FB5CC2             vsubsd   xmm0, xmm0, xmm2
       C5E057DB             vxorps   xmm3, xmm3
       C5F92EC3             vucomisd xmm0, xmm3
       77DA                 ja       SHORT G_M22436_IG03
						;; bbWeight=4    PerfScore 89.33
G_M22436_IG04:              ;; offset=0043H
       C5F828C1             vmovaps  xmm0, xmm1
						;; bbWeight=1    PerfScore 0.25
G_M22436_IG05:              ;; offset=0047H
       C3                   ret      
						;; bbWeight=1    PerfScore 1.00
RWD00  	dq	3FF0000000000000h	;            1
RWD08  	dq	4000000000000000h	;            2

Looks like only two constants are created in the constant pool now. Perhaps this was solved with: #44419

[BruceForstall]

Thanks for checking. I'll go ahead and close this issue.