diff --git a/src/arch/x86/isa/decoder/two_byte_opcodes.isa b/src/arch/x86/isa/decoder/two_byte_opcodes.isa
index b8ccec760..24349231c 100644
--- a/src/arch/x86/isa/decoder/two_byte_opcodes.isa
+++ b/src/arch/x86/isa/decoder/two_byte_opcodes.isa
@@ -508,7 +508,7 @@
                     0x1: mulps_Vo_Wo();
                     0x2: cvtps2pd_Vo_Wq();
                     0x3: cvtdq2ps_Vo_Wo();
-                    0x4: subps_Vo_Wo();
+                    0x4: Inst::SUBPS(Vo,Wo);
                     0x5: Inst::MINPS(Vo,Wo);
                     0x6: divps_Vo_Wo();
                     0x7: Inst::MAXPS(Vo,Wo);
@@ -519,7 +519,7 @@
                     0x1: mulss_Vd_Wd();
                     0x2: cvtss2sd_Vq_Wd();
                     0x3: cvttps2dq_Vo_Wo();
-                    0x4: subss_Vd_Wd();
+                    0x4: Inst::SUBSS(Vd,Wd);
                     0x5: Inst::MINSS(Vd,Wd);
                     0x6: divss_Vd_Wd();
                     0x7: Inst::MAXSS(Vd,Wd);
@@ -530,7 +530,7 @@
                     0x1: mulpd_Vo_Wo();
                     0x2: cvtpd2ps_Vo_Wo();
                     0x3: cvtps2dq_Vo_Wo();
-                    0x4: subpd_Vo_Wo();
+                    0x4: Inst::SUBPD(Vo,Wo);
                     0x5: Inst::MINPD(Vo,Wo);
                     0x6: divpd_Vo_Wo();
                     0x7: Inst::MAXPD(Vo,Wo);
@@ -915,7 +915,7 @@
                     0x0: decode OPCODE_OP_BOTTOM3 {
                         0x2: cmpccps_Vo_Wo_Ib();
                         0x3: cvtdq2ps_Vo_Wo();
-                        0x4: subps_Vo_Wo();
+                        0x4: Inst::SUBPS(Vo,Wo);
                         0x5: Inst::MINPS(Vo,Wo);
                         0x6: divps_Vo_Wo();
                     }
@@ -927,7 +927,7 @@
                     // operand size (0x66)
                     0x1: decode OPCODE_OP_BOTTOM3 {
                         0x2: cmpccpd_Vo_Wo_Ib();
-                        0x4: subpd_Vo_Wo();
+                        0x4: Inst::SUBPD(Vo,Wo);
                         0x5: Inst::MINPD(Vo,Wo);
                         0x6: divpd_Vo_Wo();
                         default: Inst::UD2();
diff --git a/src/arch/x86/isa/insts/simd128/floating_point/arithmetic/subtraction.py b/src/arch/x86/isa/insts/simd128/floating_point/arithmetic/subtraction.py
index 10669136b..378abc070 100644
--- a/src/arch/x86/isa/insts/simd128/floating_point/arithmetic/subtraction.py
+++ b/src/arch/x86/isa/insts/simd128/floating_point/arithmetic/subtraction.py
@@ -54,22 +54,73 @@
 # Authors: Gabe Black
 
 microcode = '''
-# SUBPS
-# SUBPD
-# SUBSS
+def macroop SUBSS_XMM_XMM {
+    msubf xmml, xmml, xmmlm, size=4, ext=1
+};
+
+def macroop SUBSS_XMM_M {
+    ldfp ufp1, seg, sib, disp, dataSize=8
+    msubf xmml, xmml, ufp1, size=4, ext=1
+};
+
+def macroop SUBSS_XMM_P {
+    rdip t7
+    ldfp ufp1, seg, riprel, disp, dataSize=8
+    msubf xmml, xmml, ufp1, size=4, ext=1
+};
 
 def macroop SUBSD_XMM_XMM {
-    subfp xmml, xmml, xmmlm
+    msubf xmml, xmml, xmmlm, size=8, ext=1
 };
 
 def macroop SUBSD_XMM_M {
     ldfp ufp1, seg, sib, disp, dataSize=8
-    subfp xmml, xmml, ufp1
+    msubf xmml, xmml, ufp1, size=8, ext=1
 };
 
 def macroop SUBSD_XMM_P {
     rdip t7
     ldfp ufp1, seg, riprel, disp, dataSize=8
-    subfp xmml, xmml, ufp1
+    msubf xmml, xmml, ufp1, size=8, ext=1
+};
+
+def macroop SUBPS_XMM_XMM {
+    msubf xmml, xmml, xmmlm, size=4, ext=0
+    msubf xmmh, xmmh, xmmhm, size=4, ext=0
+};
+
+def macroop SUBPS_XMM_M {
+    ldfp ufp1, seg, sib, "DISPLACEMENT", dataSize=8
+    ldfp ufp2, seg, sib, "DISPLACEMENT + 8", dataSize=8
+    msubf xmml, xmml, ufp1, size=4, ext=0
+    msubf xmmh, xmmh, ufp2, size=4, ext=0
+};
+
+def macroop SUBPS_XMM_P {
+    rdip t7
+    ldfp ufp1, seg, riprel, "DISPLACEMENT", dataSize=8
+    ldfp ufp2, seg, riprel, "DISPLACEMENT + 8", dataSize=8
+    msubf xmml, xmml, ufp1, size=4, ext=0
+    msubf xmmh, xmmh, ufp2, size=4, ext=0
+};
+
+def macroop SUBPD_XMM_XMM {
+    msubf xmml, xmml, xmmlm, size=8, ext=0
+    msubf xmmh, xmmh, xmmhm, size=8, ext=0
+};
+
+def macroop SUBPD_XMM_M {
+    ldfp ufp1, seg, sib, "DISPLACEMENT", dataSize=8
+    ldfp ufp2, seg, sib, "DISPLACEMENT + 8", dataSize=8
+    msubf xmml, xmml, ufp1, size=8, ext=0
+    msubf xmmh, xmmh, ufp2, size=8, ext=0
+};
+
+def macroop SUBPD_XMM_P {
+    rdip t7
+    ldfp ufp1, seg, riprel, "DISPLACEMENT", dataSize=8
+    ldfp ufp2, seg, riprel, "DISPLACEMENT + 8", dataSize=8
+    msubf xmml, xmml, ufp1, size=8, ext=0
+    msubf xmmh, xmmh, ufp2, size=8, ext=0
 };
 '''
diff --git a/src/arch/x86/isa/insts/simd64/floating_point/arithmetic/subtraction.py b/src/arch/x86/isa/insts/simd64/floating_point/arithmetic/subtraction.py
index 6e4c1804d..363794411 100644
--- a/src/arch/x86/isa/insts/simd64/floating_point/arithmetic/subtraction.py
+++ b/src/arch/x86/isa/insts/simd64/floating_point/arithmetic/subtraction.py
@@ -54,6 +54,33 @@
 # Authors: Gabe Black
 
 microcode = '''
-# PFSUB
-# PFSUBR
+def macroop PFSUB_MMX_MMX {
+    msubf mmx, mmx, mmxm, size=4, ext=0
+};
+
+def macroop PFSUB_MMX_M {
+    ldfp ufp1, seg, sib, disp, dataSize=8
+    msubf mmx, mmx, ufp1, size=4, ext=0
+};
+
+def macroop PFSUB_MMX_P {
+    rdip t7
+    ldfp ufp1, seg, riprel, disp, dataSize=8
+    msubf mmx, mmx, ufp1, size=4, ext=0
+};
+
+def macroop PFSUBR_MMX_MMX {
+    msubf mmx, mmxm, mmx, size=4, ext=0
+};
+
+def macroop PFSUBR_MMX_M {
+    ldfp ufp1, seg, sib, disp, dataSize=8
+    msubf mmx, ufp1, mmx, size=4, ext=0
+};
+
+def macroop PFSUBR_MMX_P {
+    rdip t7
+    ldfp ufp1, seg, riprel, disp, dataSize=8
+    msubf mmx, ufp1, mmx, size=4, ext=0
+};
 '''