Apple M1 Microarchitecture Research by Dougall Johnson

Firestorm: Overview | Base Instructions | SIMD and FP Instructions
Icestorm:  Overview | Base Instructions | SIMD and FP Instructions

MOVI (vector, 4H, zero)

Test 1: uops

Code:

  movi v0.4h, #0

(no loop instructions)

1000 unrolls and 1 iteration

Retires: 1.000

Issues: 0.000

Integer unit issues: 0.000

Load/store unit issues: 0.000

SIMD/FP unit issues: 0.000

retire (01)cycle (02)031e3f5160696d6escheduler rewind (75)scheduler stall (76)8283pipeline redirect (84)85inst all (8c)inst neon or fp (9a)a8cfd5d6ddinst fetch restart (de)e0? fp/simd (ee)f5f6f7f8fd
1004152102715001341521523101521521110011000074215111491000153153153153153
1004152102715001341521523101521521110011000074115111491000153153153153153
1004152102715001341521523101521521110011000074115111491000153153153153153
1004152102715001341521523101521521110011000074115111491000153153153153153
1004152102715001341521523101521521110011000074115111491000153153153153153
1004152102715001341521523101521521110011000074115111491000153153153153153
1004152102715001341521523101521521110011000074115111491000153153153153153
1004152102715011341521523101521521110011000074115111491000153153153153153
1004152102715011341521523101521521110011000074115111491000153153153153153
1004152102715001341521523101521521110011000074115111491000153153153153153

Test 2: throughput

Count: 8

Code:

  movi v0.4h, #0
  movi v1.4h, #0
  movi v2.4h, #0
  movi v3.4h, #0
  movi v4.4h, #0
  movi v5.4h, #0
  movi v6.4h, #0
  movi v7.4h, #0

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code divided by count): 0.1258

retire (01)cycle (02)03081e3a3f51inst issue (52)~issue int (53)~dispatch int (56)huge thing int (59)60696d6escheduler rewind (75)scheduler stall (76)~dispatch op (78)~map op int (7c)~map lookup int (7f)8283pipeline redirect (84)85inst all (8c)inst branch (8d)inst branch taken (90)inst b.cc (94)inst integer (97)inst neon or fp (9a)9fa1a6a8acc2c5branch mispredict (cb)cdcfd5d6ddinst fetch restart (de)e0? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
802041009078015028956110010010050011004210064100646101002002001006410064118020110099100100800001000000000051133163310057800001001006110061100611006110061
80204100607800139955410010010050011003410060100603181002002001006010060118020110099100100800001000000000051133163310057800001001006110061100611006110061
8020410060780258139955410010010050001003410060100603181002002001006010060118020110099100100800001000000000051133163310057800001001006110061100611006110061
80204100607800139955410010010050011003410060100603181002002001006010060118020110099100100800001000000000051133163310057800001001006110061100611006110061
8020410060780360139955410010010050011003410060100603181002002001006010060118020110099100100800001000000000051133163310057800001001006110061100611006110061
80204100607700139955410010010050011003410060100603181002002001006010060118020110099100100800001000000000051133163310057800001001006110061100611006110061
8020410060780129139955410010010050001003410060100603181002002001006010060118020110099100100800001000000000051133163310057800001001006110061100611006110061
80204100608106195956110010010050001004210064100646821002002001006410064118020110099100100800001000010011151191161110061800001001006510065100651006510065
80204100647800035955410010010050011003410060100603181002002001006010060118020110099100100800001000000000051133163310057800001001006110061100611006110061
802041006078045035955410010010050011003410060100603181002002001006010060118020110099100100800001000000000051133163310057800001001006110061100611013410061

1000 unrolls and 10 iterations

Result (median cycles for code divided by count): 0.1255

retire (01)cycle (02)03080b191e1f3f51inst issue (52)~issue int (53)~dispatch int (56)huge thing int (59)60696d6escheduler rewind (75)scheduler stall (76)~dispatch op (78)~map op int (7c)~map lookup int (7f)8283pipeline redirect (84)85inst all (8c)inst branch (8d)inst branch taken (90)inst b.cc (94)inst integer (97)inst neon or fp (9a)9fa1a6a8accfd0d5d6ddinst fetch restart (de)e0ea? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
800241005478000321035998212101050110012100381003831810202010038100381180021109101080000100000502404162410035080000101003910039100391003910039
8002410038780000035998210101050110012100381003831810202010038100381180021109101080000100030502402162410035080000101003910039100391003910039
8002410038780000035998210101050110012100381003831810202010038100383180021109101080000109000502402164210035080000101003910039100391003910039
8002410038770000035998210101050110012100381003831810202010038100381180021109101080000100000502404162510035080000101003910039100391003910039
8002410038780003035998210101050110012100381003831810202010038100381180021109101080000100020502302162410035080000101003910039100391003910039
80024100387800027035998210101050110012100381003831810202010038100381180021109101080000104000502204162410035080000101003910039100391003910039
8002410038770000035998210101050110012100381003831810202010038100384180021109101080000100000502204164210035080000101003910039100391003910039
8002410038800007503599821010105011001210038100883181020201003810042218002110910108000010001405024021642100351880000101003910039100391003910039
8002410038780000035998210101050010012100381003831810202010038100381180021109101080000100000502405162410035080000101003910039100391003910039
8002410038780000035998210101050010012100381003831810202010038100381180021109101080000100000502404167510035080000101003910039100391003910039