AMX-COMPLEX support

-- TCMMIMFP16PS, TCMMRLFP16PS instructions
-- AMX.asm fix: Similar to GATHER instructions, 3-operand AMX instructions cannot have the same operand more than once
Checked with XED version: [v2025.06.08]

Author: InstLatx64

test/amx.asm

@@ -1,7 +1,9 @@
 	bits 64
 
-%macro amx 1
+%macro amx 3
   %define treg tmm %+ %1
+  %define treg2 tmm %+ %2
+  %define treg3 tmm %+ %3
 
 	ldtilecfg [rsi]
 	sttilecfg [rdi]
@@ -16,11 +18,14 @@
 	tileloaddt1 treg, [rax,rdx]
 	tileloaddt1 treg, [rax,rdx*2]
 
-	tdpbf16ps treg, treg, treg
-	tdpbssd treg, treg, treg
-	tdpbusd treg, treg, treg
-	tdpbsud treg, treg, treg
-	tdpbuud treg, treg, treg
+	tdpbf16ps treg, treg2, treg3
+	tdpbssd treg, treg2, treg3
+	tdpbusd treg, treg2, treg3
+	tdpbsud treg, treg2, treg3
+	tdpbuud treg, treg2, treg3
+	tdpfp16ps treg, treg2, treg3
+	tcmmimfp16ps treg, treg2, treg3
+	tcmmrlfp16ps treg, treg2, treg3
 
 	tilestored [rax], treg
 	tilestored [rax,rdx], treg
@@ -30,7 +35,11 @@
 %endmacro
 
 %assign n 0
+%assign m 1
+%assign l 2
   %rep 8
-	amx n
-    %assign n n+1
+	amx n, m, l
+    %assign n ((n+1) % 8)
+    %assign m ((m+1) % 8)
+    %assign l ((l+1) % 8)
   %endrep

x86/iflags.ph

@@ -157,6 +157,7 @@ if_("AMXTILE",           "AMX tile configuration instructions");
 if_("AMXBF16",           "AMX bfloat16 multiplication");
 if_("AMXFP16",           "AMX FP16 multiplication");
 if_("AMXINT8",           "AMX 8-bit integer multiplication");
+if_("AMXCOMPLEX",        "AMX float16 complex multiplication");
 if_("FRED",              "Flexible Return and Exception Delivery (FRED)");
 if_("RAOINT",		 "Remote atomic operations (RAO-INT)");
 if_("UINTR",		 "User interrupts");

x86/insns.dat

@@ -5420,6 +5420,8 @@ LDTILECFG	mem512				[m:	vex+.128.np.0f38.w0 49 /0]		AMXTILE,SZ,LONG
 STTILECFG	mem512				[m:	vex+.128.66.0f38.w0 49 /0]		AMXTILE,SZ,LONG
 TDPBF16PS	tmmreg,tmmreg,tmmreg		[rmv:	vex.128.f3.0f38.w0 5c /r]		AMXBF16,LONG
 TDPFP16PS	tmmreg,tmmreg,tmmreg		[rmv:	vex.128.f2.0f38.w0 5c /r]		AMXFP16,LONG
+TCMMIMFP16PS	tmmreg,tmmreg,tmmreg		[rmv:	vex.128.66.0f38.w0 6c /r]		AMXCOMPLEX,LONG
+TCMMRLFP16PS	tmmreg,tmmreg,tmmreg		[rmv:	vex.128.np.0f38.w0 6c /r]		AMXCOMPLEX,LONG
 TDPBSSD		tmmreg,tmmreg,tmmreg		[rmv:	vex.128.f2.0f38.w0 5e /r]		AMXINT8,LONG
 TDPBSUD		tmmreg,tmmreg,tmmreg		[rmv:	vex.128.f3.0f38.w0 5e /r]		AMXINT8,LONG
 TDPBUSD		tmmreg,tmmreg,tmmreg		[rmv:	vex.128.66.0f38.w0 5e /r]		AMXINT8,LONG
@@ -5442,8 +5444,6 @@ T2RPNTLVWZ0RST1	tmmreg,mem			[rm:	vex.128.np.map5.w0 f9 /r]		FUTURE,SIB
 T2RPNTLVWZ1RS	tmmreg,mem			[rm:	vex.128.66.map5.w0 f8 /r]		FUTURE,SIB
 T2RPNTLVWZ1RST1	tmmreg,mem			[rm:	vex.128.66.map5.w0 f9 /r]		FUTURE,SIB
 
-TCMMIMFP16PS	tmmreg,tmmreg,tmmreg		[rmv:	vex.128.66.0f38.w0 6c /r]		FUTURE
-TCMMRLFP16PS	tmmreg,tmmreg,tmmreg		[rmv:	vex.128.np.0f38.w0 6c /r]		FUTURE
 TCONJTCMMIMFP16PS tmmreg,tmmreg,tmmreg		[rmv:	vex.128.np.0f38.w0 6b /r]		FUTURE
 TCONJTFP16	tmmreg,tmmreg			[rm:	vex.128.66.0f38.w0 6b /r]		FUTURE
 TCVTROWD2PS	zmmreg,tmmreg,reg32		[rmv:	evex.512.f3.0f38.w0 4a /r]		FUTURE