OV - attention-gcc-gnr-256 - Loop 29

0x402ae0 VMAXPS	(%RAX),%YMM0,%YMM0    [1]

0x402ae4 ADD	$0x20,%RAX

0x402ae8 CMP	%RAX,%RDX

0x402aeb JNE	402ae0

/home/eoseret/llm-attention/attention_v2.cpp: 47 - 48

--------------------------------------------------------------------------------

47:       for (int idx = 0; idx <= row; ++idx) // vectorised

48:          if (S_row[idx] > max_val) max_val = S_row[idx];

Coverage (%)	Name	Source Location	Module
○100.00	main	attention_v2.cpp:283	attention-gcc-gnr-256

min	med	avg	max

Percentile Index	10	20	30	40	50	60	70	80	90	100
Value

min	med	avg	max

Percentile Index	10	20	30	40	50	60	70	80	90	100
Value

Path /

Metric	Value
CQA speedup if no scalar integer	1.00
CQA speedup if FP arith vectorized	1.00
CQA speedup if fully vectorized	2.00
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	4.00
Bottlenecks
Function	softmax(float const, float, float*, int)
Source	attention_v2.cpp:47-48
Source loop unroll info	not unrolled or unrolled with no peel/tail loop
Source loop unroll confidence level	max
Unroll/vectorization loop type	NA
Unroll factor	NA
CQA cycles	4.00
CQA cycles if no scalar integer	4.00
CQA cycles if FP arith vectorized	4.00
CQA cycles if fully vectorized	2.00
Front-end cycles	0.50
P0 cycles	1.00
P1 cycles	0.50
P2 cycles	0.33
P3 cycles	0.33
P4 cycles	0.00
P5 cycles	0.00
P6 cycles	0.50
P7 cycles	0.00
P8 cycles	0.00
P9 cycles	0.00
P10 cycles	0.00
P11 cycles	0.33
DIV/SQRT cycles	0.00
Inter-iter dependencies cycles	4
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	4.00
Nb uops	3.00
Nb loads	1.00
Nb stores	0.00
Nb stack references	0.00
FLOP/cycle	0.00
Nb FLOP add-sub	0.00
Nb FLOP mul	0.00
Nb FLOP fma	0.00
Nb FLOP div	0.00
Nb FLOP rcp	0.00
Nb FLOP sqrt	0.00
Nb FLOP rsqrt	0.00
Bytes/cycle	8.00
Bytes prefetched	0.00
Bytes loaded	32.00
Bytes stored	0.00
Stride 0	0.00
Stride 1	1.00
Stride n	0.00
Stride unknown	0.00
Stride indirect	0.00
Vectorization ratio all	100.00
Vectorization ratio load	100.00
Vectorization ratio store	NA
Vectorization ratio mul	NA
Vectorization ratio add_sub	NA
Vectorization ratio fma	NA
Vectorization ratio div_sqrt	NA
Vectorization ratio other	100.00
Vector-efficiency ratio all	50.00
Vector-efficiency ratio load	50.00
Vector-efficiency ratio store	NA
Vector-efficiency ratio mul	NA
Vector-efficiency ratio add_sub	NA
Vector-efficiency ratio fma	NA
Vector-efficiency ratio div_sqrt	NA
Vector-efficiency ratio other	50.00

Metric	Value
CQA speedup if no scalar integer	1.00
CQA speedup if FP arith vectorized	1.00
CQA speedup if fully vectorized	2.00
CQA speedup if no inter-iteration dependency	NA
CQA speedup if next bottleneck killed	4.00
Bottlenecks
Function	softmax(float const, float, float*, int)
Source	attention_v2.cpp:47-48
Source loop unroll info	not unrolled or unrolled with no peel/tail loop
Source loop unroll confidence level	max
Unroll/vectorization loop type	NA
Unroll factor	NA
CQA cycles	4.00
CQA cycles if no scalar integer	4.00
CQA cycles if FP arith vectorized	4.00
CQA cycles if fully vectorized	2.00
Front-end cycles	0.50
P0 cycles	1.00
P1 cycles	0.50
P2 cycles	0.33
P3 cycles	0.33
P4 cycles	0.00
P5 cycles	0.00
P6 cycles	0.50
P7 cycles	0.00
P8 cycles	0.00
P9 cycles	0.00
P10 cycles	0.00
P11 cycles	0.33
DIV/SQRT cycles	0.00
Inter-iter dependencies cycles	4
FE+BE cycles (UFS)	NA
Stall cycles (UFS)	NA
Nb insns	4.00
Nb uops	3.00
Nb loads	1.00
Nb stores	0.00
Nb stack references	0.00
FLOP/cycle	0.00
Nb FLOP add-sub	0.00
Nb FLOP mul	0.00
Nb FLOP fma	0.00
Nb FLOP div	0.00
Nb FLOP rcp	0.00
Nb FLOP sqrt	0.00
Nb FLOP rsqrt	0.00
Bytes/cycle	8.00
Bytes prefetched	0.00
Bytes loaded	32.00
Bytes stored	0.00
Stride 0	0.00
Stride 1	1.00
Stride n	0.00
Stride unknown	0.00
Stride indirect	0.00
Vectorization ratio all	100.00
Vectorization ratio load	100.00
Vectorization ratio store	NA
Vectorization ratio mul	NA
Vectorization ratio add_sub	NA
Vectorization ratio fma	NA
Vectorization ratio div_sqrt	NA
Vectorization ratio other	100.00
Vector-efficiency ratio all	50.00
Vector-efficiency ratio load	50.00
Vector-efficiency ratio store	NA
Vector-efficiency ratio mul	NA
Vector-efficiency ratio add_sub	NA
Vector-efficiency ratio fma	NA
Vector-efficiency ratio div_sqrt	NA
Vector-efficiency ratio other	50.00

Path /

Average path: Display a virtual path defined by average values of all real paths

Function	softmax(float const, float, float*, int)
Source file and lines	attention_v2.cpp:47-48
Module	attention-gcc-gnr-256

The loop is defined in /home/eoseret/llm-attention/attention_v2.cpp:47-48.

The related source loop is not unrolled or unrolled with no peel/tail loop.

gain
potential
hint
expert

Vectorization

Your loop is vectorized, but using only 256 out of 512 bits (AVX/AVX2 instructions on AVX-512 processors).

Details

All SSE/AVX instructions are used in vector version (process two or more data elements in vector registers).

Workaround

Read the "512-bits vectorization" report at "Potential" confidence level.

Execution units bottlenecks

Found no such bottlenecks but see expert reports for more complex bottlenecks.

512-bits vectorization

On some x86 processors supporting 512-bits vectorization, compilers are often too conservative and limit vectorization to 256 bits. Performance can then be improved by enforcing 512-bits vectorization, especially with many vectorized and high trip count loops. 512-bits vectorization performance overhead (compared to 256-bits) is generally lower on newer processors.

Workaround

Recompile with -mprefer-vector-width=512

Type of elements and instruction set

1 AVX instructions are processing arithmetic or math operations on single precision FP elements in vector mode (eight at a time).

Matching between your loop (in the source code) and the binary loop

The binary loop does not contain any FP arithmetical operations. The binary loop is loading 32 bytes (8 single precision FP elements).

Unroll opportunity

Loop body is too small to efficiently use resources.

Workaround

Unroll your loop if trip count is significantly higher than target unroll factor. This can be done manually. Or by recompiling with -funroll-loops and/or -floop-unroll-and-jam. Or with the unroll (resp. unroll_and_jam) directive on top of the inner (resp. surrounding) loop. You can enforce an unroll factor: #pragma GCC unroll N

General properties

nb instructions	4
nb uops	3
loop length	13
used x86 registers	2
used mmx registers	0
used xmm registers	0
used ymm registers	1
used zmm registers	0
nb stack references	0

Front-end

ASSUMED MACRO FUSION FIT IN UOP CACHE

micro-operation queue	0.50 cycles
front end	0.50 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11
uops	1.00	0.50	0.33	0.33	0.00	0.00	0.50	0.00	0.00	0.00	0.00	0.33
cycles	1.00	0.50	0.33	0.33	0.00	0.00	0.50	0.00	0.00	0.00	0.00	0.33

Execution ports to units layout:

P0 (256 bits): VPU, BRU, ALU, DIV/SQRT
P1 (256 bits): ALU, VPU
P2 (512 bits): load
P3 (512 bits): load
P4 (512 bits): store data
P5 (512 bits): ALU, VPU
P6: ALU, BRU
P7: store address
P8: store address
P9 (512 bits): store data
P10: ALU
P11 (512 bits): load

Cycles executing div or sqrt instructions	NA
Longest recurrence chain latency (RecMII)	4.00

Cycles summary

Front-end	0.50
Dispatch	1.00
Data deps.	4.00
Overall L1	4.00

Vectorization ratios

all	100%
load	100%
store	NA (no store vectorizable/vectorized instructions)
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	NA (no add-sub vectorizable/vectorized instructions)
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	100%

Vector efficiency ratios

all	50%
load	50%
store	NA (no store vectorizable/vectorized instructions)
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	NA (no add-sub vectorizable/vectorized instructions)
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	50%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each iteration of the binary loop takes 4.00 cycles. At this rate:

4% of peak load performance is reached (8.00 out of 192.00 bytes loaded per cycle (GB/s @ 1GHz))

Front-end bottlenecks

Found no such bottlenecks.

ASM code

In the binary file, the address of the loop is: 402ae0

Instruction	Nb FU	P0	P1	P2	P3	P5	P6	P10	P11	Latency	Recip. throughput	Vectorization
VMAXPS (%RAX),%YMM0,%YMM0	1	0.50	0.50	0.33	0.33	0	0	0	0.33	4	0.50	vect (50.0%)
ADD $0x20,%RAX	1	0	0	0	0	0	0	0	0	1	0.17	N/A
CMP %RAX,%RDX	1	0.20	0.20	0	0	0.20	0.20	0.20	0	1	0.20	N/A
JNE 402ae0 <_Z7softmaxPKfPfS1_i+0xb0>	1	0.50	0	0	0	0	0.50	0	0	0	0.50	N/A

Function	softmax(float const, float, float*, int)
Source file and lines	attention_v2.cpp:47-48
Module	attention-gcc-gnr-256

The loop is defined in /home/eoseret/llm-attention/attention_v2.cpp:47-48.

The related source loop is not unrolled or unrolled with no peel/tail loop.

gain
potential
hint
expert

Vectorization

Your loop is vectorized, but using only 256 out of 512 bits (AVX/AVX2 instructions on AVX-512 processors).

Details

All SSE/AVX instructions are used in vector version (process two or more data elements in vector registers).

Workaround

Read the "512-bits vectorization" report at "Potential" confidence level.

Execution units bottlenecks

Found no such bottlenecks but see expert reports for more complex bottlenecks.

512-bits vectorization

On some x86 processors supporting 512-bits vectorization, compilers are often too conservative and limit vectorization to 256 bits. Performance can then be improved by enforcing 512-bits vectorization, especially with many vectorized and high trip count loops. 512-bits vectorization performance overhead (compared to 256-bits) is generally lower on newer processors.

Workaround

Recompile with -mprefer-vector-width=512

Type of elements and instruction set

1 AVX instructions are processing arithmetic or math operations on single precision FP elements in vector mode (eight at a time).

Matching between your loop (in the source code) and the binary loop

The binary loop does not contain any FP arithmetical operations. The binary loop is loading 32 bytes (8 single precision FP elements).

Unroll opportunity

Loop body is too small to efficiently use resources.

Workaround

Unroll your loop if trip count is significantly higher than target unroll factor. This can be done manually. Or by recompiling with -funroll-loops and/or -floop-unroll-and-jam. Or with the unroll (resp. unroll_and_jam) directive on top of the inner (resp. surrounding) loop. You can enforce an unroll factor: #pragma GCC unroll N

General properties

nb instructions	4
nb uops	3
loop length	13
used x86 registers	2
used mmx registers	0
used xmm registers	0
used ymm registers	1
used zmm registers	0
nb stack references	0

Front-end

ASSUMED MACRO FUSION FIT IN UOP CACHE

micro-operation queue	0.50 cycles
front end	0.50 cycles

Back-end

	P0	P1	P2	P3	P4	P5	P6	P7	P8	P9	P10	P11
uops	1.00	0.50	0.33	0.33	0.00	0.00	0.50	0.00	0.00	0.00	0.00	0.33
cycles	1.00	0.50	0.33	0.33	0.00	0.00	0.50	0.00	0.00	0.00	0.00	0.33

Execution ports to units layout:

P0 (256 bits): VPU, BRU, ALU, DIV/SQRT
P1 (256 bits): ALU, VPU
P2 (512 bits): load
P3 (512 bits): load
P4 (512 bits): store data
P5 (512 bits): ALU, VPU
P6: ALU, BRU
P7: store address
P8: store address
P9 (512 bits): store data
P10: ALU
P11 (512 bits): load

Cycles executing div or sqrt instructions	NA
Longest recurrence chain latency (RecMII)	4.00

Cycles summary

Front-end	0.50
Dispatch	1.00
Data deps.	4.00
Overall L1	4.00

Vectorization ratios

all	100%
load	100%
store	NA (no store vectorizable/vectorized instructions)
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	NA (no add-sub vectorizable/vectorized instructions)
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	100%

Vector efficiency ratios

all	50%
load	50%
store	NA (no store vectorizable/vectorized instructions)
mul	NA (no mul vectorizable/vectorized instructions)
add-sub	NA (no add-sub vectorizable/vectorized instructions)
fma	NA (no fma vectorizable/vectorized instructions)
div/sqrt	NA (no div/sqrt vectorizable/vectorized instructions)
other	50%

Cycles and memory resources usage

Assuming all data fit into the L1 cache, each iteration of the binary loop takes 4.00 cycles. At this rate:

4% of peak load performance is reached (8.00 out of 192.00 bytes loaded per cycle (GB/s @ 1GHz))

Front-end bottlenecks

Found no such bottlenecks.

ASM code

In the binary file, the address of the loop is: 402ae0

Instruction	Nb FU	P0	P1	P2	P3	P5	P6	P10	P11	Latency	Recip. throughput	Vectorization
VMAXPS (%RAX),%YMM0,%YMM0	1	0.50	0.50	0.33	0.33	0	0	0	0.33	4	0.50	vect (50.0%)
ADD $0x20,%RAX	1	0	0	0	0	0	0	0	0	1	0.17	N/A
CMP %RAX,%RDX	1	0.20	0.20	0	0	0.20	0.20	0.20	0	1	0.20	N/A
JNE 402ae0 <_Z7softmaxPKfPfS1_i+0xb0>	1	0.50	0	0	0	0	0.50	0	0	0	0.50	N/A

Report Configuration

Vectorization

Details

Workaround

Execution units bottlenecks

512-bits vectorization

Workaround

Type of elements and instruction set

Matching between your loop (in the source code) and the binary loop

Unroll opportunity

Workaround

General properties

Front-end

Back-end

Cycles summary

Vectorization ratios

Vector efficiency ratios

Cycles and memory resources usage

Front-end bottlenecks

ASM code

Vectorization

Details

Workaround

Execution units bottlenecks

512-bits vectorization

Workaround

Type of elements and instruction set

Matching between your loop (in the source code) and the binary loop

Unroll opportunity

Workaround

General properties

Front-end

Back-end

Cycles summary

Vectorization ratios

Vector efficiency ratios

Cycles and memory resources usage

Front-end bottlenecks

ASM code