
    8hBS                       S SK Jr  S SKrS SKJrJrJr  S SKrS SKJ	r	  SSK
JrJr  SSKJrJrJrJr  SSKJrJrJrJr  SS	KJrJrJrJrJrJr  SS
KJ r   SSK!J"r"  \(       a  S SK#J$r$  SSKJ%r%  \RL                  " \'5      r(\RR                  RT                  r*\S 5       r+\S 5       r,S r-Sr. \" S\+S\.-   S-   \.-   S-   S9r/Sr0\" S\,S\0-   S-   \0-   S-   S9r1\" \Rd                  SS\*Rd                  Rf                  S9r4S  r5\" \5S5      r6 " S! S"\5      r7                    S)S# jr8S$ r9S% r:\" \*Rd                  5                        S*S& j5       r2\" \*Rv                  5      S' 5       r;S( r<\" \*Rd                  \<5        g)+    )annotationsN)OptionalTYPE_CHECKING	TypedDict)CKGroupedConvFwdTemplate   )configir)add_layout_constraintconstrain_to_fx_strides	loweringsregister_lowering)autotune_select_algorithmExternKernelChoiceSymbolicGridFnTritonTemplate)is_onesis_zerospad_listlikesympy_productuse_ck_conv_templateuse_triton_template)V   )mm_config_kwargs)Sequence)	TensorBoxc               B    U" X-  U-  US   5      U" XS   5      US   4$ NBLOCK_MBLOCK_NGROUPS )nchwmetacdivs         U/var/www/fran/franai/venv/lib/python3.13/site-packages/torch/_inductor/kernel/conv.pyconv2d_gridr+   .   s5     	QUQYY(QY X     c               H    U" X-  U-  U-  US   5      U" XS   5      US   4$ r   r#   )r$   r%   dr&   r'   r(   r)   s          r*   conv3d_gridr/   7   s9     	QUQY]DO,QY X r,   c                <    U S:  d  US:  d  US:  a  gX-  U-  S:  $ )N   Ti   r#   )mr$   ks      r*   _is_large_block_for_cpur4   @   s)    3w!c'QW519ur,   a  
        idx_x_h = i - PADDING_H + idx_y_h * STRIDE_H
        idx_x_w = j - PADDING_W + idx_y_w * STRIDE_W
        idx_x_c = tl.arange(0, BLOCK_K) + k

        x_ptrs = x_base + (
            (idx_x_h * stride_xh)[:, None]
            + (idx_x_w * stride_xw)[:, None]
            + (idx_x_c * stride_xc)[None, :]
        )
        mask_x = (
            (idx_n < BATCH)[:, None]
            & (idx_x_h >= 0)[:, None]
            & (idx_x_h < IN_H)[:, None]
            & (idx_x_w >= 0)[:, None]
            & (idx_x_w < IN_W)[:, None]
            & (idx_x_c < GROUP_IN_C)[None, :]
        )
        matrix_x = tl.load(x_ptrs, mask=mask_x, other=0.0)

        w_ptrs = w_base + (
            (idx_x_c * stride_wc_in)[:, None] + (i * stride_wh) + (j * stride_ww)
        )
        mask_w = (idx_x_c[:, None] < GROUP_IN_C) & (idx_y_c[None, :] < GROUP_OUT_C)
        matrix_w = tl.load(w_ptrs, mask=mask_w, other=0.0)
        acc += tl.dot(matrix_x, matrix_w, allow_tf32=ALLOW_TF32)
convolution2dag  
{{def_kernel("X", "W")}}
    # Tensor dimensions
    BATCH = {{size("X", 0)}}
    IN_C = {{size("X", 1)}}
    IN_H = {{size("X", 2)}}
    IN_W = {{size("X", 3)}}
    OUT_C = {{size(None, 1)}}
    OUT_H = {{size(None, 2)}}
    OUT_W = {{size(None, 3)}}

    # Strides:
    stride_xn = {{stride("X", 0)}}
    stride_xc = {{stride("X", 1)}}
    stride_xh = {{stride("X", 2)}}
    stride_xw = {{stride("X", 3)}}
    stride_wc_out = {{stride("W", 0)}}
    stride_wc_in = {{stride("W", 1)}}
    stride_wh = {{stride("W", 2)}}
    stride_ww = {{stride("W", 3)}}

    nhw = tl.program_id(0) * BLOCK_M + tl.arange(0, BLOCK_M)
    idx_y_w = nhw % OUT_W
    nh = nhw // OUT_W
    idx_y_h = nh % OUT_H
    idx_n = nh // OUT_H
    idx_y_c = tl.program_id(1) * BLOCK_N + tl.arange(0, BLOCK_N)

{% if GROUPS == 1 %}
    group = 0
    GROUP_IN_C = IN_C
    GROUP_OUT_C = OUT_C
{% else %}
    group = tl.program_id(2)
    GROUP_IN_C = IN_C // GROUPS
    GROUP_OUT_C = OUT_C // GROUPS
{% endif %}

    x_base = X + (group * stride_xc * GROUP_IN_C + idx_n * stride_xn)[:, None]
    w_base = (
        W + (group * stride_wc_out * GROUP_OUT_C + idx_y_c * stride_wc_out)[None, :]
    )

    acc = tl.zeros((BLOCK_M, BLOCK_N), dtype=tl.float32)

{% if UNROLL %}
{% for i in range(KERNEL_H) %}
{% for j in range(KERNEL_W) %}
    i = {{i}}
    j = {{j}}
    for k in range(0, GROUP_IN_C, BLOCK_K):
        a  
{% endfor %}
{% endfor %}
{% else %}
    # Could be simplified, but slightly slower:
    # for i in range(KERNEL_H):
    #     for j in range(KERNEL_W):
    #         for k in range(0, GROUP_IN_C, BLOCK_K):
    BLOCK_K_COUNT = (GROUP_IN_C + BLOCK_K - 1) // BLOCK_K
    for ijk in range(KERNEL_H * KERNEL_W * BLOCK_K_COUNT):
        k = (ijk % BLOCK_K_COUNT) * BLOCK_K
        ij = ijk // BLOCK_K_COUNT
        i = ij // KERNEL_W
        j = ij % KERNEL_W
        a  
{% endif %}

    mask = (
        (idx_n < BATCH)[:, None]
        & (idx_y_h < OUT_H)[:, None]
        & (idx_y_w < OUT_W)[:, None]
        & (idx_y_c < GROUP_OUT_C)[None, :]
    )
    idx_n = idx_n[:, None]
    idx_c = idx_y_c[None, :] + group * GROUP_OUT_C
    idx_h = idx_y_h[:, None]
    idx_w = idx_y_w[:, None]

    # inductor generates a suffix
    {{store_output(("idx_n", "idx_c", "idx_h", "idx_w"), "acc", "mask")}}
)namegridsourcea  
        idx_x_d = d - PADDING_D + idx_y_d * STRIDE_D
        idx_x_h = i - PADDING_H + idx_y_h * STRIDE_H
        idx_x_w = j - PADDING_W + idx_y_w * STRIDE_W
        idx_x_c = tl.arange(0, BLOCK_K) + k

        x_ptrs = x_base + (
            (idx_x_d * stride_xd)[:, None]
            + (idx_x_h * stride_xh)[:, None]
            + (idx_x_w * stride_xw)[:, None]
            + (idx_x_c * stride_xc)[None, :]
        )
        mask_x = (
            (idx_n < BATCH)[:, None]
            & (idx_x_d >= 0)[:, None]
            & (idx_x_d < IN_D)[:, None]
            & (idx_x_h >= 0)[:, None]
            & (idx_x_h < IN_H)[:, None]
            & (idx_x_w >= 0)[:, None]
            & (idx_x_w < IN_W)[:, None]
            & (idx_x_c < GROUP_IN_C)[None, :]
        )
        matrix_x = tl.load(x_ptrs, mask=mask_x, other=0.0)

        w_ptrs = w_base + (
            (idx_x_c * stride_wc_in)[:, None] +
            (d * stride_wd) + (i * stride_wh) + (j * stride_ww)
        )
        mask_w = (idx_x_c[:, None] < GROUP_IN_C) & (idx_y_c[None, :] < GROUP_OUT_C)
        matrix_w = tl.load(w_ptrs, mask=mask_w, other=0.0)
        acc += tl.dot(matrix_x, matrix_w, allow_tf32=ALLOW_TF32)
convolution3daH  
{{def_kernel("X", "W")}}
    # Tensor dimensions
    BATCH = {{size("X", 0)}}
    IN_C = {{size("X", 1)}}
    IN_D = {{size("X", 2)}}
    IN_H = {{size("X", 3)}}
    IN_W = {{size("X", 4)}}
    OUT_C = {{size(None, 1)}}
    OUT_D = {{size(None, 2)}}
    OUT_H = {{size(None, 3)}}
    OUT_W = {{size(None, 4)}}

    # Strides:
    stride_xn = {{stride("X", 0)}}
    stride_xc = {{stride("X", 1)}}
    stride_xd = {{stride("X", 2)}}
    stride_xh = {{stride("X", 3)}}
    stride_xw = {{stride("X", 4)}}
    stride_wc_out = {{stride("W", 0)}}
    stride_wc_in = {{stride("W", 1)}}
    stride_wd = {{stride("W", 2)}}
    stride_wh = {{stride("W", 3)}}
    stride_ww = {{stride("W", 4)}}

    ndhw = tl.program_id(0) * BLOCK_M + tl.arange(0, BLOCK_M)
    idx_y_w = ndhw % OUT_W
    ndh = ndhw // OUT_W
    idx_y_h = ndh % OUT_H
    nd = ndh // OUT_H
    idx_y_d = nd % OUT_D
    idx_n = nd // OUT_D
    idx_y_c = tl.program_id(1) * BLOCK_N + tl.arange(0, BLOCK_N)

{% if GROUPS == 1 %}
    group = 0
    GROUP_IN_C = IN_C
    GROUP_OUT_C = OUT_C
{% else %}
    group = tl.program_id(2)
    GROUP_IN_C = IN_C // GROUPS
    GROUP_OUT_C = OUT_C // GROUPS
{% endif %}

    x_base = X + (group * stride_xc * GROUP_IN_C + idx_n * stride_xn)[:, None]
    w_base = (
        W + (group * stride_wc_out * GROUP_OUT_C + idx_y_c * stride_wc_out)[None, :]
    )

    acc = tl.zeros((BLOCK_M, BLOCK_N), dtype=tl.float32)

{% if UNROLL %}
{% for d in range(KERNEL_D) %}
{% for i in range(KERNEL_H) %}
{% for j in range(KERNEL_W) %}
    d = {{d}}
    i = {{i}}
    j = {{j}}
    for k in range(0, GROUP_IN_C, BLOCK_K):
        aF  
{% endfor %}
{% endfor %}
{% endfor %}
{% else %}
    # Could be simplified, but slightly slower:
    # for d in range(KERNEL_D):
    #   for i in range(KERNEL_H):
    #     for j in range(KERNEL_W):
    #         for k in range(0, GROUP_IN_C, BLOCK_K):
    BLOCK_K_COUNT = (GROUP_IN_C + BLOCK_K - 1) // BLOCK_K
    for dijk in range(KERNEL_D * KERNEL_H * KERNEL_W * BLOCK_K_COUNT):
        k = (dijk % BLOCK_K_COUNT) * BLOCK_K
        dij = dijk // BLOCK_K_COUNT
        j = dij % KERNEL_W
        di = dij // KERNEL_W
        i = di % KERNEL_H
        d = di // KERNEL_H
        a  
{% endif %}

    mask = (
        (idx_n < BATCH)[:, None]
        & (idx_y_d < OUT_D)[:, None]
        & (idx_y_h < OUT_H)[:, None]
        & (idx_y_w < OUT_W)[:, None]
        & (idx_y_c < GROUP_OUT_C)[None, :]
    )
    idx_n = idx_n[:, None]
    idx_c = idx_y_c[None, :] + group * GROUP_OUT_C
    idx_d = idx_y_d[:, None]
    idx_h = idx_y_h[:, None]
    idx_w = idx_y_w[:, None]

    # inductor generates a suffix
    {{store_output(("idx_n", "idx_c", "idx_d", "idx_h", "idx_w"), "acc", "mask")}}
zat::convolutionF)has_out_variantop_overloadc          
         [         R                  " [         R                  " US5      S5      n[         R                  " U R                  SSSS5      UR                  SS5      UR                  SSSS5      S9$ )Nr   r      r   )out)torchsqueezematmulpermute)xr'   r?   s      r*   conv1x1_via_mmrE   U  s]    emmAr*B/A<<			!Q1qyyACKK1a4K r,   c                  R    \ rS rSr% S\S'   S\S'   S\S'   S\S'   S\S'   S	\S
'   Srg)ConvLayoutParamsi_  tuple[int, ...]stridepaddingdilationbool
transposedoutput_paddingintgroupsr#   N)__name__
__module____qualname____firstlineno____annotations____static_attributes__r#   r,   r*   rG   rG   _  s%    ##Kr,   rG   c	                p   [         R                  R                     [        R                  R
                  R                  [        R                  " U SS9[        R                  " USS9[        R                  " USS9[         R                  R                  R                  U5      [         R                  R                  R                  U5      [         R                  R                  R                  U5      U[         R                  R                  R                  U5      U5	      n	[        R                  " U	R                  5       5      n
[        R                  " U	R                  5       5      nSSS5        [        R                  " U R                  5       U R!                  5       W
U5      $ ! , (       d  f       NC= f)z)Determine output layout for a convolutionT)guard_shapeN)r   graph	fake_moder@   opsatenconvolutionr
   ir_node_to_tensorsizevars
size_hintsconvert_shape_to_inductorsizerI   FixedLayoutget_device_or_error	get_dtype)rD   weightbiasrI   rJ   rK   rM   rN   rP   outputsizess              r*   conv_layoutrj   h  s)    
		++  5  T:  48GG''/GG''0GG''1GG''7

 ,,V[[];--fmmo> 
 >>			  
	s   EF''
F5c                    [        [        [        U 5      5      5      nUR                  SUR	                  S5      5        U$ )Nr   r=   )listreversedrangeinsertpop)rankorders     r*   channels_last_orderrs     s0    %+&'E	LLEIIbM"Lr,   c                   [        UR                  5       5      n[        US-
  5       H  n[        [        R
                     " USS9nM!     [        [        R                     " USS/5      n[        R                  R                  U [        U5      5      n [        [        U5      5      nUR                  UR                  S5      5        [        [        R                     " X5      n U R                  5       Gt pg[        [        R                     " U [        U5      U/5      n Uc  [        [        R                      " X5      nO[        [        R"                     " X U5      n[        [        R                     " U/ UQSP5      n[        [        U5      5      n	U	R%                  SU	R                  S5      5        [        [        R                     " X5      $ )Nr   r=   dimr   r   )lenget_sizern   Lr\   rA   rC   r
   ExternKernelrequire_stride_orderrs   rl   appendrp   reshaper   mmaddmmro   )
rD   rf   rg   rq   _	x_permuteri   in_chanresultresult_permutes
             r*   convert_1x1_conv_to_mmr     sR   v !D4!8_4<<R0 t||_VaV,F
,,Q0CD0IJAU4[!IY]]1%&	$,,%AjjlOU	$,,M%0':;A|477A&4::t/t||_V\u\b\2F%+&N!^//34T\\?622r,   c	                  ^ ^^^ [        U5      n[        U5      n[        U5      n[        U5      n[        U[        5      (       d)  [        R                  R
                  R                  U5      n[        U[        5      (       d   e[        [        R                  R
                  R                  U5      5      n[        [        R                  R
                  R                  U5      5      nUUUUUUS.m[        R                  " T 5      n	[        T R                  5       5      [        TR                  5       5      S-
  :X  aU  [        [        R                     " [        [        [        R                      " T S/T R                  5       Q5      TU40 TD6SS9$ [        R                  R
                  R                  TR                  5       5      tpn[        T R                  5       5      S:X  a  [        U5      S:X  a  U	S:X  a  TR#                  SU-   SU-   SU-   SU-   S	.5        [        [        R$                     " T S
S9m [        [        R$                     " TS
S9m[        [        R                     " [        T TU40 TD6S
S9$ [        U5      m['        UT5      n['        UT5      n['        UT5      n['        UT5      nUUUU 4S jn[(        R*                  =(       d    [(        R,                  n[(        R.                  (       d  U(       a  U" 5       (       a  [1        U5      (       a  [1        U5      (       a  [3        U5      (       a  [1        U5      (       ap  U(       di  [3        U5      (       aY  US:X  aS  [        R                  R
                  R5                  [7        T R                  5       5      S5      (       a  [9        T TU5      $ Ubf  U	S:w  a`  [        T TS 40 TD6n[        [        R:                     " U[        [        R<                     " X/R                  5       S   /TS/-  -   5      5      $ T R?                  5         TR?                  5         [        R                  R@                  (       av  TS
:X  ap  [        R                  =RB                  S-  sl!        [        RD                  RG                  T 5      m [        RD                  RG                  T5      m[I        T TS 40 TD6nO[I        T TS 40 TD6n[        RJ                  " [        R                  R
                  RM                  URN                  5      5      n[        RD                  RQ                  T U5      m [        RD                  RQ                  TU5      m/ SQnUc  T T/nS TS'   URS                  SS5        O\T TU/nUR?                  5         URU                  5         [        R                  R
                  R                  UR                  5       5        / n[V        RX                  RZ                  R]                  S5      (       a  [^        R`                  " UUU40 TD6/n[V        RX                  RZ                  R]                  S5      (       Ga  [c        U5      (       Gat  [1        U5      (       Gac  U(       Gd[  [3        U5      (       GaJ  [        R                  R
                  Re                  UT R                  5       S   5      (       Ga	  [1        U5      (       aK  [1        U5      (       a;  [3        U5      (       a+  US:X  a%  URg                  [h        Ra                  UU5      5        [        Rj                  Rm                  U	5      nU" [7        T R                  5       S   /T R                  5       S
S  Q5      U
U40 [o        U	[p        5      D6 GHC  nTS
:X  a  [r        Rt                  " U4T T4UUS   US   US   US   US   US   U[1        U5      [V        Rv                  Rx                  Rz                  UR|                  UR~                  S.UR                  D6  M  TS:X  d  M  [        Rt                  " U40 ST T4_SU_SUS   _SUS   _SUS
   _SUS   _SUS   _SUS
   _SUS   _SUS   _SUS
   _SU_S[1        U5      _S[V        Rv                  Rx                  Rz                  _SUR|                  _S UR~                  _UR                  D6  GMF     [        U5      (       a.  [        R                  " UUT T4Ub  U4O	[        5       -   UUUUTS!9  [        S"UUU5      $ )#N)rI   rJ   rK   rM   rN   rP   r   r   ru   r>   xpu)r   )r   )rI   rJ   rK   rN   r   c                    > [         R                  R                  (       a  TS:X  a  g[        TTS 40 TD6n [        R
                  " [         R                  R                  R                  U R                  5      5      nU[        R                  :H  $ )Nr   T)
r   rY   
layout_optrj   r
   get_stride_orderr_   r`   rI   NHWC_STRIDE_ORDER)layoutreq_stride_orderkwargsndimrf   rD   s     r*   channels_last_conv'convolution.<locals>.channels_last_conv  sl    77$!)Q77..GG''6
  2#7#777r,   cpurg   ATENTRITON)input_nodesr   KERNEL_HKERNEL_WSTRIDE_HSTRIDE_W	PADDING_H	PADDING_Wr"   UNROLL
ALLOW_TF32
num_stages	num_warpsr   r   KERNEL_Dr   r   STRIDE_Dr   r   	PADDING_Dr   r   r"   r   r   r   r   )r   rI   rJ   rK   rP   n_spatial_dimensionsr]   )Ftuple
isinstancerO   r   rY   r_   evaluate_static_shapeevaluate_static_shapesr
   get_device_typerw   rx   ry   r\   rA   r]   expandupdate	unsqueezer   r	   max_autotunemax_autotune_gemmconv_1x1_as_mmr   r   statically_known_gtr   r   addviewrealizer   num_channels_last_convrz   require_channels_lastrj   r   r`   rI   r{   ro   freeze_layoutr@   	_inductorutils_use_conv_autotune_backendaten_convolutionbindr   statically_known_equalsr|   aten_conv1x1_via_mmchoicesget_conv_configsr   r4   conv2d_templatemaybe_append_choicebackendscudnn
allow_tf32r   r   r   conv3d_templater   r   add_ck_conv_choicesr   )rD   rf   rg   rI   rJ   rK   rM   rN   rP   device_typeout_chanr   kernel_shaper   autotuning_gemmr   r   r   ordered_kwargs_for_cpp_kernelargsr   conv_configscfgr   r   s   ``                     @@r*   r]   r]     s    6]FGnGXH>*Nfc""!!77?fc"""" 177##::6BCFAGG$$;;GDEG  ( F $$Q'K
1::<C 12Q66$++q1*<qzz|*<=vtVvV
 	

 ()ww'7'7'N'N($H 1::<A#l"3q"8[E=Q-'> 8O"&"7		
 dnnaQ'4>>"6q164262
 	

 |D&$'F7D)GHd+H!.$7N8 8 ))EV-E-EO 
		?7I7K7KL!!FOOWH^$$aKGG00qzz|1LaPP%a66K50Q77{AdiiL(9!(<'=s
'JK
 	
 IIK
NN
 	wwdai	&&!+&OO11!4 66v>Q77Q77..GG''6
 OO004DE55f>NO%! |6{v%,,Q764 	//@G77??!!- 	
 	88BB''H^$$GG44Wajjl1oNN L!!!!!NN.33D&AByy11+>1::<?>QZZ\!"-=>?
 {,CD	
C qy33!"F!)!_)!_#AY#AY%aj%aj! #<0$~~33>>"~~!mm!" jj#& 33!"F " *!_	
 *!_ *!_ $AY $AY $AY &aj &aj &aj "  #<0!"  %~~33>>#$  #~~%& "mmjj)7
b F## 44F$2BwP!%		
 %]GT6JJr,   c                     [        XX#XEXgU5	      $ N)r]   )rD   rf   rg   rI   rJ   rK   rM   rN   rP   	benchmarkdeterministiccudnn_enabledr   s                r*   _convolutionr     s      	4JPV r,   c                    U R                   [        R                  R                  R                  R
                  :X  d   e[        R                  R                  (       a  X4$ [        U /UQ70 UD6$ r   )
targetr@   r[   r\   r]   defaultr   rY   r   r   )fx_noder   r   s      r*   constrain_conv_to_fx_stridesr     sR    >>UYY^^77?????ww|&w@@@@r,   )rD   r   rf   r   rg   Optional[TensorBox]rI   Sequence[int]rJ   rH   rK   rH   rM   rL   rN   rH   rP   rO   returnz	ir.Layout)rD   r   rf   r   rg   r   rI   r   rJ   r   rK   r   rM   rL   rN   r   rP   rO   )=
__future__r   loggingtypingr   r   r   r@   -torch._inductor.codegen.rocm.ck_conv_templater    r	   r
   loweringr   r   r   ry   r   select_algorithmr   r   r   r   r   r   r   r   r   r   r   virtualizedr   	mm_commonr   collections.abcr   r   	getLoggerrQ   logr[   r\   r+   r/   r4   LOOP_BODY_2Dr   LOOP_BODY_3Dr   r]   r   r   rE   r   rG   rj   rs   r   r   r   r#   r,   r*   <module>r      s   "  5 5  R      ' (! yy~~    8
 !		3h i4jkCH IDJKUYvB !		;x y<z{O` aPbccgR &	  ((	  )> y       	 
       $     F3. 4##$oKoKoK oK 	oK
 oK oK oK "oK oK %oKd 4$$% &(A d&&(D Er,   