diff --git a/MdePkg/Library/BaseMemoryLibOptDxe/X64/CopyMem.S b/MdePkg/Library/BaseMemoryLibOptDxe/X64/CopyMem.S
index 426a602286..8fbc90f0e4 100644
--- a/MdePkg/Library/BaseMemoryLibOptDxe/X64/CopyMem.S
+++ b/MdePkg/Library/BaseMemoryLibOptDxe/X64/CopyMem.S
@@ -24,7 +24,6 @@
 #
 #------------------------------------------------------------------------------
 
-
 #------------------------------------------------------------------------------
 #  VOID *
 #  EFIAPI
@@ -33,43 +32,52 @@
 #    IN VOID   *Source,
 #    IN UINTN  Count
 #    )
-#------------------------------------------------------------------------------
-.intel_syntax noprefix
-.globl ASM_PFX(InternalMemCopyMem)
-ASM_PFX(InternalMemCopyMem):
-    push    rsi
-    push    rdi
-    mov     rsi, rdx                    # rsi <- Source
-    mov     rdi, rcx                    # rdi <- Destination
-    lea     r9, [rsi + r8 - 1]          # r9 <- End of Source
-    cmp     rsi, rdi
-    mov     rax, rdi                    # rax <- Destination as return value
-    jae     L0
-    cmp     r9, rdi
-    jae     L_CopyBackward              # Copy backward if overlapped
-L0:
-    mov     rcx, r8
-    and     r8, 7
-    shr     rcx, 3                      # rcx <- # of Qwords to copy
-    jz      L_CopyBytes
-    movd    r10, mm0                    # (Save mm0 in r10)
-L1:
-    movq    mm0, [rsi]
-    movntq  [rdi], mm0
-    add     rsi, 8
-    add     rdi, 8
-    loop    L1
-    mfence
-    movd    mm0, r10                    # (Restore mm0)
-    jmp     L_CopyBytes
-L_CopyBackward:
-    mov     rsi, r9                     # rsi <- End of Source
-    lea     rdi, [rdi + r8 - 1]         # rdi <- End of Destination
-    std                                 # set direction flag
-L_CopyBytes:
-    mov     rcx, r8
-    rep     movsb                       # Copy bytes backward
-    cld
-    pop     rdi
-    pop     rsi
-    ret
+#------------------------------------------------------------------------------
+.intel_syntax noprefix
+.globl ASM_PFX(InternalMemCopyMem)
+ASM_PFX(InternalMemCopyMem):
+    push    rsi
+    push    rdi
+    mov     rsi, rdx                    # rsi <- Source
+    mov     rdi, rcx                    # rdi <- Destination
+    lea     r9, [rsi + r8 - 1]          # r9 <- Last byte of Source
+    cmp     rsi, rdi
+    mov     rax, rdi                    # rax <- Destination as return value
+    jae     L0                          # Copy forward if Source > Destination
+    cmp     r9, rdi                     # Overlapped?
+    jae     @CopyBackward               # Copy backward if overlapped
+L0:
+    xor     rcx, rcx
+    sub     rcx, rdi                    # rcx <- -rdi
+    and     rcx, 15                     # rcx + rsi should be 16 bytes aligned
+    jz      L1                          # skip if rcx == 0
+    cmp     rcx, r8
+    cmova   rcx, r8
+    sub     r8, rcx
+    rep     movsb
+L1:
+    mov     rcx, r8
+    and     r8, 15
+    shr     rcx, 4                      # rcx <- # of DQwords to copy
+    jz      L_CopyBytes
+    movdqa  [rsp + 0x18], xmm0           # save xmm0 on stack
+L2:
+    movdqu  xmm0, [rsi]                 # rsi may not be 16-byte aligned
+    movntdq [rdi], xmm0                 # rdi should be 16-byte aligned
+    add     rsi, 16
+    add     rdi, 16
+    loop    L2
+    mfence
+    movdqa  xmm0, [rsp + 0x18]           # restore xmm0
+    jmp     L_CopyBytes                  # copy remaining bytes
+L_CopyBackward:
+    mov     rsi, r9                     # rsi <- Last byte of Source
+    lea     rdi, [rdi + r8 - 1]         # rdi <- Last byte of Destination
+    std
+L_CopyBytes:
+    mov     rcx, r8
+    rep     movsb
+    cld
+    pop     rdi
+    pop     rsi
+    ret
diff --git a/MdePkg/Library/BaseMemoryLibOptDxe/X64/ZeroMem.S b/MdePkg/Library/BaseMemoryLibOptDxe/X64/ZeroMem.S
index 97c3130709..9c6be9c97e 100644
--- a/MdePkg/Library/BaseMemoryLibOptDxe/X64/ZeroMem.S
+++ b/MdePkg/Library/BaseMemoryLibOptDxe/X64/ZeroMem.S
@@ -41,9 +41,10 @@ ASM_PFX(InternalMemZeroMem):
     mov     rdi, rcx
     mov     rcx, rdx
     shr     rcx, 3
-    and     rdx, 7
+    and     rdx, 7
+    cld
     rep     stosq
-    mov     ecx, edx
+    mov     rcx, rdx
     rep     stosb
     pop     rax
     pop     rdi