nvdisasm fatal value sm61 is not defined for option binary
nvdisasm accepts a single input file each time it's run. The bones usage is as following:
nvdisasm [options] <input cubin file>
Hither'south a sample output of nvdisasm:
.headerflags @"EF_CUDA_TEXMODE_UNIFIED EF_CUDA_64BIT_ADDRESS EF_CUDA_SM70 EF_CUDA_VIRTUAL_SM(EF_CUDA_SM70)" .elftype @"ET_EXEC" //--------------------- .nv.info -------------------------- .department .nv.info,"",@"SHT_CUDA_INFO" .align 4 ...... //--------------------- .text._Z9acos_main10acosParams -------------------------- .section .text._Z9acos_main10acosParams,"ax",@progbits .sectioninfo @"SHI_REGISTERS=14" .align 128 .global _Z9acos_main10acosParams .type _Z9acos_main10acosParams,@function .size _Z9acos_main10acosParams,(.L_21 - _Z9acos_main10acosParams) .other _Z9acos_main10acosParams,@"STO_CUDA_ENTRY STV_DEFAULT" _Z9acos_main10acosParams: .text._Z9acos_main10acosParams: /*0000*/ MOV R1, c[0x0][0x28] ; /*0010*/ NOP; /*0020*/ S2R R0, SR_CTAID.X ; /*0030*/ S2R R3, SR_TID.X ; /*0040*/ IMAD R0, R0, c[0x0][0x0], R3 ; /*0050*/ ISETP.GE.AND P0, PT, R0, c[0x0][0x170], PT ; /*0060*/ @P0 Get out ; .L_1: /*0070*/ MOV R11, 0x4 ; /*0080*/ IMAD.WIDE R2, R0, R11, c[0x0][0x160] ; /*0090*/ LDG.E.SYS R2, [R2] ; /*00a0*/ MOV R7, 0x3d53f941 ; /*00b0*/ FADD.FTZ R4, |R2|.reuse, -RZ ; /*00c0*/ FSETP.GT.FTZ.AND P0, PT, |R2|.reuse, 0.5699, PT ; /*00d0*/ FSETP.GEU.FTZ.AND P1, PT, R2, RZ, PT ; /*00e0*/ FADD.FTZ R5, -R4, 1 ; /*00f0*/ IMAD.Broad R2, R0, R11, c[0x0][0x168] ; /*0100*/ FMUL.FTZ R5, R5, 0.5 ; /*0110*/ @P0 MUFU.SQRT R4, R5 ; /*0120*/ MOV R5, c[0x0][0x0] ; /*0130*/ IMAD R0, R5, c[0x0][0xc], R0 ; /*0140*/ FMUL.FTZ R6, R4, R4 ; /*0150*/ FFMA.FTZ R7, R6, R7, 0.018166976049542427063 ; /*0160*/ FFMA.FTZ R7, R6, R7, 0.046756859868764877319 ; /*0170*/ FFMA.FTZ R7, R6, R7, 0.074846573173999786377 ; /*0180*/ FFMA.FTZ R7, R6, R7, 0.16667014360427856445 ; /*0190*/ FMUL.FTZ R7, R6, R7 ; /*01a0*/ FFMA.FTZ R7, R4, R7, R4 ; /*01b0*/ FADD.FTZ R9, R7, R7 ; /*01c0*/ @!P0 FADD.FTZ R9, -R7, 1.5707963705062866211 ; /*01d0*/ ISETP.GE.AND P0, PT, R0, c[0x0][0x170], PT ; /*01e0*/ @!P1 FADD.FTZ R9, -R9, iii.1415927410125732422 ; /*01f0*/ STG.East.SYS [R2], R9 ; /*0200*/ @!P0 BRA `(.L_1) ; /*0210*/ Exit ; .L_2: /*0220*/ BRA `(.L_2); .L_21:
To get the control flow graph of a kernel, use the following:
nvdisasm -cfg <input cubin file>
nvdisasm is capable of generating control flow of CUDA assembly in the format of DOT graph clarification linguistic communication. The output of the control flow from nvdisasm can be direct imported to a DOT graph visualization tool such as Graphviz.
Here's how you tin can generate a PNG image (cfg.png) of the control flow of the above cubin (a.cubin) with nvdisasm and Graphviz:
nvdisasm -cfg a.cubin | dot -ocfg.png -Tpng
Here's the generated graph:
Figure 1. Control Menstruum Graph
To generate a PNG image (bbcfg.png) of the basic cake control flow of the above cubin (a.cubin) with nvdisasm and Graphviz:
nvdisasm -bbcfg a.cubin | dot -obbcfg.png -Tpng
Here's the generated graph:
Figure 2. Basic Block Control Flow Graph
nvdisasm is capable of showing the register (full general and predicate) liveness range information. For each line of CUDA assembly, nvdisasm displays whether a given device register was assigned, accessed, alive or re-assigned. It also shows the total number of registers used. This is useful if the user is interested in the life range of any item register, or annals usage in general.
Here's a sample output (output is pruned for brevity):
// +-----------------+------+ // | GPR | PRED | // | | | // | | | // | 000000000011 | | // | # 012345678901 | # 01 | // +-----------------+------+ .global acos // | | | .type acos,@role // | | | .size acos,(.L_21 - acos) // | | | .other acos,@"STO_CUDA_ENTRY STV_DEFAULT" // | | | acos: // | | | .text.acos: // | | | MOV R1, c[0x0][0x28] ; // | one ^ | | NOP; // | 1 ^ | | S2R R0, SR_CTAID.X ; // | 2 ^: | | S2R R3, SR_TID.X ; // | 3 :: ^ | | IMAD R0, R0, c[0x0][0x0], R3 ; // | 3 x: v | | ISETP.GE.AND P0, PT, R0, c[0x0][0x170], PT ; // | 2 v: | 1 ^ | @P0 Exit ; // | 2 :: | 1 5 | .L_1: // | ii :: | | MOV R11, 0x4 ; // | 3 :: ^ | | IMAD.WIDE R2, R0, R11, c[0x0][0x160] ; // | 5 v:^^ v | | LDG.Due east.SYS R2, [R2] ; // | 4 ::^ : | | MOV R7, 0x3d53f941 ; // | 5 ::: ^ : | | FADD.FTZ R4, |R2|.reuse, -RZ ; // | half dozen ::five ^ : : | | FSETP.GT.FTZ.AND P0, PT, |R2|.reuse, 0.5699, PT; // | six ::v : : : | 1 ^ | FSETP.GEU.FTZ.AND P1, PT, R2, RZ, PT ; // | half-dozen ::v : : : | 2 :^ | FADD.FTZ R5, -R4, i ; // | 6 :: v^ : : | 2 :: | IMAD.WIDE R2, R0, R11, c[0x0][0x168] ; // | eight v:^^:: : v | 2 :: | FMUL.FTZ R5, R5, 0.5 ; // | 5 :: :x : | 2 :: | @P0 MUFU.SQRT R4, R5 ; // | five :: ^5 : | 2 five: | MOV R5, c[0x0][0x0] ; // | 5 :: :^ : | 2 :: | IMAD R0, R5, c[0x0][0xc], R0 ; // | v 10: :v : | ii :: | FMUL.FTZ R6, R4, R4 ; // | v :: v ^: | 2 :: | FFMA.FTZ R7, R6, R7, 0.018166976049542427063 ; // | 5 :: : vx | two :: | FFMA.FTZ R7, R6, R7, 0.046756859868764877319 ; // | v :: : vx | ii :: | FFMA.FTZ R7, R6, R7, 0.074846573173999786377 ; // | 5 :: : vx | 2 :: | FFMA.FTZ R7, R6, R7, 0.16667014360427856445 ; // | 5 :: : vx | 2 :: | FMUL.FTZ R7, R6, R7 ; // | 5 :: : vx | 2 :: | FFMA.FTZ R7, R4, R7, R4 ; // | iv :: v x | 2 :: | FADD.FTZ R9, R7, R7 ; // | 4 :: v ^ | ii :: | @!P0 FADD.FTZ R9, -R7, one.5707963705062866211 ; // | 4 :: v ^ | two v: | ISETP.GE.AND P0, PT, R0, c[0x0][0x170], PT ; // | iii v: : | 2 ^: | @!P1 FADD.FTZ R9, -R9, 3.1415927410125732422 ; // | 3 :: x | 2 :v | STG.Due east.SYS [R2], R9 ; // | three :: five | 1 : | @!P0 BRA `(.L_1) ; // | 2 :: | 1 v | Get out ; // | i : | | .L_2: // +.................+......+ BRA `(.L_2); // | | | .L_21: // +-----------------+------+ // Legend: // ^ : Register assignment // 5 : Register usage // x : Register usage and reassignment // : : Register in employ // <infinite> : Annals non in use // # : Number of occupied registers
nvdisasm is capable of showing line number information of the CUDA source file which can be useful for debugging.
To get the line-info of a kernel, use the post-obit:
nvdisasm -thousand <input cubin file>
Here'southward a sample output of a kernel using nvdisasm -g command:
//--------------------- .text._Z6kernali -------------------------- .section .text._Z6kernali,"ax",@progbits .sectioninfo @"SHI_REGISTERS=24" .marshal 128 .global _Z6kernali .type _Z6kernali,@function .size _Z6kernali,(.L_4 - _Z6kernali) .other _Z6kernali,@"STO_CUDA_ENTRY STV_DEFAULT" _Z6kernali: .text._Z6kernali: /*0000*/ MOV R1, c[0x0][0x28] ; /*0010*/ NOP; //## File "/home/user/cuda/sample/sample.cu", line 25 /*0020*/ MOV R0, 0x160 ; /*0030*/ LDC R0, c[0x0][R0] ; /*0040*/ MOV R0, R0 ; /*0050*/ MOV R2, R0 ; //## File "/habitation/user/cuda/sample/sample.cu", line 26 /*0060*/ MOV R4, R2 ; /*0070*/ MOV R20, 32@lo((_Z6kernali + .L_1@srel)) ; /*0080*/ MOV R21, 32@hi((_Z6kernali + .L_1@srel)) ; /*0090*/ CALL.ABS.NOINC `(_Z3fooi) ; .L_1: /*00a0*/ MOV R0, R4 ; /*00b0*/ MOV R4, R2 ; /*00c0*/ MOV R2, R0 ; /*00d0*/ MOV R20, 32@lo((_Z6kernali + .L_2@srel)) ; /*00e0*/ MOV R21, 32@hello((_Z6kernali + .L_2@srel)) ; /*00f0*/ Phone call.ABS.NOINC `(_Z3bari) ; .L_2: /*0100*/ MOV R4, R4 ; /*0110*/ IADD3 R4, R2, R4, RZ ; /*0120*/ MOV R2, 32@lo(arr) ; /*0130*/ MOV R3, 32@howdy(arr) ; /*0140*/ MOV R2, R2 ; /*0150*/ MOV R3, R3 ; /*0160*/ ST.E.SYS [R2], R4 ; //## File "/dwelling/user/cuda/sample/sample.cu", line 27 /*0170*/ ERRBAR ; /*0180*/ Leave ; .L_3: /*0190*/ BRA `(.L_3); .L_4:
nvdisasm is capable of showing line number information with additional office inlining info (if any). In absence of whatsoever function inlining the output is aforementioned as the one with nvdisasm -g command.
Here'south a sample output of a kernel using nvdisasm -gi command:
//--------------------- .text._Z6kernali -------------------------- .department .text._Z6kernali,"ax",@progbits .sectioninfo @"SHI_REGISTERS=16" .align 128 .global _Z6kernali .type _Z6kernali,@part .size _Z6kernali,(.L_18 - _Z6kernali) .other _Z6kernali,@"STO_CUDA_ENTRY STV_DEFAULT" _Z6kernali: .text._Z6kernali: /*0000*/ IMAD.MOV.U32 R1, RZ, RZ, c[0x0][0x28] ; //## File "/home/user/cuda/inline.cu", line 17 inlined at "/home/user/cuda/inline.cu", line 23 //## File "/home/user/cuda/inline.cu", line 23 /*0010*/ UMOV UR4, 32@lo(arr) ; /*0020*/ UMOV UR5, 32@howdy(arr) ; /*0030*/ IMAD.U32 R2, RZ, RZ, UR4 ; /*0040*/ MOV R3, UR5 ; /*0050*/ ULDC.64 UR4, c[0x0][0x118] ; //## File "/dwelling house/user/cuda/inline.cu", line 10 inlined at "/home/user/cuda/inline.cu", line 17 //## File "/dwelling/user/cuda/inline.cu", line 17 inlined at "/home/user/cuda/inline.cu", line 23 //## File "/home/user/cuda/inline.cu", line 23 /*0060*/ LDG.E R4, [R2.64] ; /*0070*/ LDG.E R5, [R2.64+0x4] ; //## File "/home/user/cuda/inline.cu", line 17 inlined at "/habitation/user/cuda/inline.cu", line 23 //## File "/home/user/cuda/inline.cu", line 23 /*0080*/ LDG.E R0, [R2.64+0x8] ; //## File "/domicile/user/cuda/inline.cu", line 23 /*0090*/ UMOV UR6, 32@lo(ans) ; /*00a0*/ UMOV UR7, 32@hi(ans) ; //## File "/domicile/user/cuda/inline.cu", line x inlined at "/dwelling house/user/cuda/inline.cu", line 17 //## File "/dwelling house/user/cuda/inline.cu", line 17 inlined at "/home/user/cuda/inline.cu", line 23 //## File "/home/user/cuda/inline.cu", line 23 /*00b0*/ IADD3 R7, R4, c[0x0][0x160], RZ ; //## File "/domicile/user/cuda/inline.cu", line 23 /*00c0*/ IMAD.U32 R4, RZ, RZ, UR6 ; //## File "/abode/user/cuda/inline.cu", line x inlined at "/home/user/cuda/inline.cu", line 17 //## File "/abode/user/cuda/inline.cu", line 17 inlined at "/home/user/cuda/inline.cu", line 23 //## File "/abode/user/cuda/inline.cu", line 23 /*00d0*/ IADD3 R9, R5, c[0x0][0x160], RZ ; //## File "/abode/user/cuda/inline.cu", line 23 /*00e0*/ MOV R5, UR7 ; //## File "/domicile/user/cuda/inline.cu", line x inlined at "/home/user/cuda/inline.cu", line 17 //## File "/home/user/cuda/inline.cu", line 17 inlined at "/home/user/cuda/inline.cu", line 23 //## File "/domicile/user/cuda/inline.cu", line 23 /*00f0*/ IADD3 R11, R0.reuse, c[0x0][0x160], RZ ; //## File "/dwelling house/user/cuda/inline.cu", line 17 inlined at "/home/user/cuda/inline.cu", line 23 //## File "/home/user/cuda/inline.cu", line 23 /*0100*/ IMAD.IADD R13, R0, 0x1, R7 ; //## File "/home/user/cuda/inline.cu", line 10 inlined at "/home/user/cuda/inline.cu", line 17 //## File "/home/user/cuda/inline.cu", line 17 inlined at "/home/user/cuda/inline.cu", line 23 //## File "/domicile/user/cuda/inline.cu", line 23 /*0110*/ STG.E [R2.64+0x4], R9 ; /*0120*/ STG.E [R2.64], R7 ; /*0130*/ STG.E [R2.64+0x8], R11 ; //## File "/home/user/cuda/inline.cu", line 23 /*0140*/ STG.E [R4.64], R13 ; //## File "/home/user/cuda/inline.cu", line 24 /*0150*/ Get out ; .L_3: /*0160*/ BRA `(.L_3); .L_18:
Source: https://docs.nvidia.com/cuda/cuda-binary-utilities/index.html
Posted by: marlowesirstee1955.blogspot.com
0 Response to "nvdisasm fatal value sm61 is not defined for option binary"
Post a Comment