Apple M1 Microarchitecture Research by Dougall Johnson

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

SQRSHL (scalar, B)

Test 1: uops

Code:

  sqrshl b0, b0, b1
  movi v0.16b, 1
  movi v1.16b, 2

(no loop instructions)

1000 unrolls and 1 iteration

Retires: 1.000

Issues: 1.000

Integer unit issues: 0.000

Load/store unit issues: 0.000

SIMD/FP unit issues: 1.000

retire (01)cycle (02)0308181e3f4e51inst issue (52)~issue fp/simd (54)~dispatch fp/simd (57)huge thing fp/simd (5b)60696d6edispatch stall (70)scheduler rewind (75)scheduler stall (76)~dispatch op (78)~map op fp/simd (7e)~map lookup fp/simd (81)8283pipeline redirect (84)85inst all (8c)inst neon or fp (9a)a8accfd5d6ddinst fetch restart (de)e0? fp/simd (ee)f5f6f7f8fd
100430372300016825482510001000100039831313018303730372415328951000100020003037303711100110001073216222630100030383038303830383038
10043037220096125482510001000100039831313018303730372415328951000100020003037303711100110000073216222630100030383038303830383038
10043037230036125482510001000100039831303018303730372415328951000100020003037303711100110000073216222630100030383038303830383038
10043037220096125482510001000100039831303018303730372415328951000100020003037303711100110000073216222630100030383038303830383038
10043037230006125482510001000100039831303018303730372415328951000100020003037303711100110000073216222630100030383038303830383038
10043037230036125482510001000100039831303018303730372415328951000100020003037303711100110000073216222630100030383038303830383038
10043037230006125482510001000100039831303018303730372415328951000100020003037303711100110000073216222630100030383038303830383038
10043037220096125482510001000100039831303018303730372415328951000100020003037303711100110000073216222630100030383038303830383038
10043037220006125482510001000100039831303018303730372415328951000100020003037303711100110000073216222630100030383038303830383038
10043037220006125482510001000100039831303018303730372415328951000100020003037303711100110000073216222630100030383038303830383038

Test 2: Latency 1->2

Code:

  sqrshl b0, b0, b1
  movi v0.16b, 1
  movi v1.16b, 2

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code): 3.0037

retire (01)cycle (02)0308090b181e1f3a3f4e51inst issue (52)~issue int (53)~issue fp/simd (54)~dispatch int (56)~dispatch fp/simd (57)huge thing int (59)huge thing fp/simd (5b)60696d6edispatch stall (70)scheduler rewind (75)scheduler stall (76)~dispatch op (78)~map op int (7c)~map op fp/simd (7e)~map lookup int (7f)~map lookup fp/simd (81)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)9fa0a1a6a8a9acc2cfd5d6ddinst fetch restart (de)e0? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
10204300372250000000612954825101001001000010010000500427731303001830037300372826532874510100200100002002000030037300371110201100991001001000010000000007102161129634100001003003830038300383003830038
10204300372250000000612954825101001001000010010000500427731303001830037300372826532874510100200100002002000030037300371110201100991001001000010000000007101161129634100001003003830038300383003830038
10204301802250000000612954825101001001000010010000500427731303001830037300372826532874510100200100002002000030037300371110201100991001001000010000000007101161129634100001003003830038300383003830038
10204300372250000000612954825101001001000010010000500427731303001830037300372826532874510100200100002002000030037300371110201100991001001000010000000007101161129634100001003003830038300383003830038
10204300372250000000612954825101001001000010010000500427731303001830037300372826532874510100200100002002000030037300371110201100991001001000010000000007101161129634100001003003830038300383003830038
10204300372240000000612954825101001001000010010000500427731303001830037300372826532874510100200100002002000030037300371110201100991001001000010000000007101161129634100001003003830038300383003830038
102043003722500000006129548251010010010000100100005004277313030018300373003728265242874510100200100002002000030037300371110201100991001001000010000000007101161129634100001003003830038300383003830085
10204300372250000000612954825101001001000010010000500427731303001830037300372826532874510100200100002002000030037300371110201100991001001000010000000007101161129634100001003003830038300383003830038
10204300372250000000612954825101001001000010010000500427731303001830037300372826532874510100200100002002000030037300371110201100991001001000010000000007101161129634100001003003830038300383003830038
10204300372250000000612954825101001001000010010000500427731303001830037300372826532874510100200100002002000030037300371110201100991001001000010000010007101161129634100001003003830038300383003830038

1000 unrolls and 10 iterations

Result (median cycles for code): 3.0037

retire (01)cycle (02)0308090b18191e1f3f4c4e51inst issue (52)~issue int (53)~issue fp/simd (54)~dispatch int (56)~dispatch fp/simd (57)huge thing int (59)huge thing fp/simd (5b)60696d6edispatch stall (70)scheduler rewind (75)scheduler stall (76)~dispatch op (78)~map op int (7c)~map op fp/simd (7e)~map lookup int (7f)~map lookup fp/simd (81)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)9fa0a1a8acc2cfd5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
10024300372251012160010302954825100101010008101000050427731303001830037300372828732876710010201032320200003003730037111002110910101000010000006402162229630010000103003830085300853003830038
1002430037225000000061029548251001010100001210000504277313030018300373003728287328767100122010000202000030037300371110021109101010000100001506402162229630010000103003830038300383003830038
100243003722500000006102954825100101010000101000050427731313001830037300372828732876710010201000020200003003730037111002110910101000010000006402162229630010000103003830038300383003830038
100243003722500000006102954825100101010000101000050427731303001830037300372828732876710010201000020200003003730037111002110910101000010000006402162229630010000103003830038300383003830038
100243003722500000006102954825100101010000101000050427731303001830037300372828732876710010201000020200003003730037111002110910101000010000006502162229630010000103003830038300383003830038
100243003722400000006102954825100101210000121000060427731313001830037300372828732876710012201000020200003003730037111002110910101000010000006404164229632210000103003830038300383003830038
100243008422500000006102954825100101010000101000050427731313001830037300372828732876710010201000020200003003730037111002110910101000010000006402162229630010000103003830038300383003830038
100243003722500000006102954825100101010000101000050427731313001830037300372828732876710010201000020200003003730037111002110910101000010000006402162229630010000103003830038300383003830038
1002430037224000000061029548251001010100001210000504277313030054300373027428292328767109072410988242196430321303228110021109101010000100202830464691209930143410000103003830038300383003830038
1002430037224000020114461029530273101191510104191202792428138403001830037300372828732882310010201000020200003003730037111002110910101000010000009352162229630710000103003830038310213065930038

Test 3: Latency 1->3

Code:

  sqrshl b0, b1, b0
  movi v0.16b, 1
  movi v1.16b, 2

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code): 3.0037

retire (01)cycle (02)0308181e1f3f4e51inst issue (52)~issue int (53)~issue fp/simd (54)~dispatch int (56)~dispatch fp/simd (57)huge thing int (59)huge thing fp/simd (5b)60696d6edispatch stall (70)scheduler rewind (75)scheduler stall (76)~dispatch op (78)~map op int (7c)~map op fp/simd (7e)~map lookup int (7f)~map lookup fp/simd (81)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)9fa8accfd5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
1020430037225000048162954825101001001000010010000500427731303001830037300372826532874510100200100002002000030037300371110201100991001001000010031371011611296340100001003003830038300383003830038
10204300372250000612954825101001001000010010000500427731303001830037300372826532874510100200100002002000030037300371110201100991001001000010025971011611296340100001003003830038300383003830038
10204300372250000612954825101001001000010010000500427731303001830037300372826532874510100200100002002000030037300371110201100991001001000010027371011611296340100001003003830038300383003830038
102043003722500006129548251010010010000100100005004277313130018300373003728265328745101002001000020020000300373003711102011009910010010000100181871011611296340100001003003830038300383003830038
10204300862250000612954825101001001000010010000500427731313001830037300372826532874510100200100002002000030037300371110201100991001001000010029371011611296340100001003003830038300383003830038
10204300372250000612954825101001001000010010000500427731313001830037300372826532874510100200100002002000030037300371110201100991001001000010071371011611296340100001003003830038300383003830038
10204300372250000612954825101001001000010010000500427731313001830037300372826532874510100200100002002000030037300371110201100991001001000010027371011611296340100001003003830038300383003830038
10204300372250000612954825101001001000010010000500427731303001830037300372826532874510100200100002002000030037300371110201100991001001000010027671011611296340100001003003830038300383003830038
1020430037225000061295482510100100100001001000050042773130300183003730037282653287451010020010000200200003003730037111020110099100100100001003071011611296340100001003003830038300383003830038
10204300372250000612954825101001001000010010000500427731313001830037300372826532874510100200100002002000030037300371110201100991001001000010031371011611296340100001003003830038300383003830038

1000 unrolls and 10 iterations

Result (median cycles for code): 3.0037

retire (01)cycle (02)03071e3f4e5051inst issue (52)~issue int (53)~issue fp/simd (54)~dispatch int (56)~dispatch fp/simd (57)huge thing int (59)huge thing fp/simd (5b)60696d6edispatch stall (70)scheduler rewind (75)scheduler stall (76)~dispatch op (78)~map op int (7c)~map op fp/simd (7e)~map lookup int (7f)~map lookup fp/simd (81)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)9fa8a9acc5cfd5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
100243003722500612954830021251001010100001010000504277313130018300373003728287328767100102010000202000030037300371110021109101010000100023706402162229630010000103003830038300383003830038
1002430037224006129548300212510010101000010100005042773131300183003730037282873287671001020100002020000300373003711100211091010100001000006402162229630010000103003830038300383003830038
100243003722500612954802510012121000010100005042773131300183003730037282873287671001020100002020000300373003711100211091010100001000006402164429632010000103003830038300383003830038
10024300372251366129548025100101010000101000050427731313001830037300372828732876710010201000020200003003730037111002110910101000010003006402162229630010000103003830038300383003830038
100243003722400612954802510010101000010100005042773131300183003730037282873287671001020100002020000300373003711100211091010100001000006402162229630010000103003830038300383003830038
100243003722500612954802510010101000010100005042773130300183003730037282873287671001020100002020000300373003711100211091010100001000006402162229630010000103003830038300383003830038
100243003722500612954802510010101000010100005042773131300183003730037282873287671001020100002020000300373003711100211091010100001000006402162229630010000103003830038300383003830038
100243003722500612954802510010101000010100005042773130300183003730037282873287671001020100002020000300373003711100211091010100001000006402162229630010000103003830038300383003830038
1002430037225006129548025100101010000101000050427731303001830037300372828720287671001020100002020000300373003711100211091010100001000006402162229630010000103003830038300383003830038
100243003722510612954802510010101000010100005042773131300183003730037282873287671001020100002020000300373003711100211091010100001004306402162229630010000103003830038300383003830038

Test 4: throughput

Count: 8

Code:

  sqrshl b0, b8, b9
  sqrshl b1, b8, b9
  sqrshl b2, b8, b9
  sqrshl b3, b8, b9
  sqrshl b4, b8, b9
  sqrshl b5, b8, b9
  sqrshl b6, b8, b9
  sqrshl b7, b8, b9
  movi v8.16b, 9
  movi v9.16b, 10

(fused SUBS/B.cc loop)

100 unrolls and 100 iterations

Result (median cycles for code divided by count): 0.2505

retire (01)cycle (02)03181e3f51inst issue (52)~issue int (53)~issue fp/simd (54)~dispatch int (56)~dispatch fp/simd (57)huge thing int (59)huge thing fp/simd (5b)60696d6edispatch stall (70)scheduler rewind (75)scheduler stall (76)~dispatch op (78)~map op int (7c)~map op fp/simd (7e)~map lookup int (7f)~map lookup fp/simd (81)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)9fa8a9accfd5d6ddinst fetch restart (de)e0? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
802042003915000412580100100800001008000050064000012002020039200399973399978010020080000200160000200392003911802011009910010080000100104051103161120036800001002004020040200402004020040
80204200391500041258010010080000100800005006400001200202003920039997339997801002008000020016000020039200391180201100991001008000010000051101161120036800001002004020040200402004020040
80204200391500041258010010080000100800005006400001200202003920039997339997801002008000020016000020039200391180201100991001008000010000051101161120036800001002004020040200402004020040
80204200391500041258010010080000100800005006400001200202003920039997339997801002008000020016000020039200391180201100991001008000010000051101161120036800001002004020040200402004020040
80204200391500041258010010080000100800005006400000200202003920039997339997801002008000020016000020039200391180201100991001008000010000051101161120036800001002004020040200402004020040
80204200391500041258010010080000100800005006400001200202003920039997339997801002008000020016000020039200391180201100991001008000010000141851101161120036800001002004020040200402004020040
80204200391500041258010010080000100800005006400001200202003920039997339997801002008000020016000020039200391180201100991001008000010000051101161120036800001002004020040200402004020040
80204200391500041258010010080000100800005006400001200202003920039997339997801002008000020016000020039200391180201100991001008000010000051101161120036800001002004020040200402004020040
80204200391500041258010010080000100800005006400001200202003920039997339997801002008000020016000020039200391180201100991001008000010000651101161220036800001002004020040200402004020040
80204200391500041258010010080000100800005006400001200202003920039997339997801002008000020016000020039200391180201100991001008000010020351101161120036800001002004020040200402004020040

1000 unrolls and 10 iterations

Result (median cycles for code divided by count): 0.2505

retire (01)cycle (02)03080a1e1f3f51inst issue (52)~issue int (53)~issue fp/simd (54)~dispatch int (56)~dispatch fp/simd (57)huge thing int (59)huge thing fp/simd (5b)60696d6edispatch stall (70)scheduler rewind (75)scheduler stall (76)~dispatch op (78)~map op int (7c)~map op fp/simd (7e)~map lookup int (7f)~map lookup fp/simd (81)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)9fa1a7a8acc2cfd5d6ddinst fetch restart (de)e0? int output thing (e9)? fp/simd (ee)gpr retires (ef)f5f6f7f8fd
8002420048150000040258001010800001080000506400000200202003920039999631001980010208000020160000200392003911800211091010800001000000502011168820036080000102004020040200402004020040
800242003915000004025800101080000108000050640000020020200392003999963100198001020800002016000020039200391180021109101080000100044135050207167720036080000102004020040200402004020040
800242003915000004025800101080000108000050640000120020200392003999963100198001020800002016000020039200391180021109101080000100000050207168820036080000102004020040200402004020040
80024200391500000402580010108000010800005064000002002020039200399996310019800102080000201600002003920039118002110910108000010003812050207167520036080000102004020040200402004020040
8002420039150000040258001010800001080000506400000200202003920039999631001980010208000020160000200392003911800211091010800001000024050207168620036080000102004020040200402004020040
800242003915000008225800101080000108000050640000020020200392003999963100198001020800002016000020039200391180021109101080000100000050207167520036080000102004020040200402004020040
800242003915000004025800101080000108000062640778020020200392003999963100198001020800002016000020039200391180021109101080000100000050207166820036080000102004020040200402004020040
8002420039150000051525800101080000108000050640000020020200392003999963100198001020800002016000020039200391180021109101080000100000050206165720036180000102004020040200402004020040
800242003915000004025800101080000108000050640000120020200392003999963100198001020800002016000020039200391180021109101080000100000050207167620036080000102004020040200402004020040
800242003915000004025800101080000108000050640000020020200392003999963100198001020800002016000020039200391180021109101080000100000050207167520036080000102004020040200402004020040