xKg95*dZddlmZddlmZddlZdZdZdZ Gdde Z Gdd e Z Gd d e Zd Zd ZdZdZddZedZdZej,ddfdZej,ddfdZdZdZdZd dZdZd!dZdZdZ dZ!y)"zz The Device Array API is not implemented in the simulator. This module provides stubs to allow tests to import correctly. )contextmanager) numpy_versionNzArray contains non-contiguous buffer and cannot be transferred as a single memory region. Please ensure contiguous buffer with numpy .ascontiguousarray()c"eZdZdZfdZxZS) FakeShapez The FakeShape class is used to provide a shape which does not allow negative indexing, similar to the shape in CUDA Python. (Numpy shape arrays allow negative indexing) cjt|tr|dkr tdtt||S)Nrztuple index out of range) isinstanceint IndexErrorsuperr __getitem__)selfk __class__s l/home/alanp/www/video.onchill/myenv/lib/python3.12/site-packages/numba/cuda/simulator/cudadrv/devicearray.pyr zFakeShape.__getitem__s1 a !a%78 8Y1!44)__name__ __module__ __qualname____doc__r __classcell__)rs@rrrs 55rrc@eZdZdZdZdZdZdZdZdZ dZ d Z y ) FakeWithinKernelCUDAArrayaG Created to emulate the behavior of arrays within kernels, where either array.item or array['item'] is valid (that is, give all structured arrays `numpy.recarray`-like semantics). This behaviour does not follow the semantics of Python and NumPy with non-jitted code, and will be deprecated and removed. cFt|tsJ||jd<y)N_item)r FakeCUDAArray__dict__r items r__init__z"FakeWithinKernelCUDAArray.__init__+s $ ...!% grc<t|tr t|S|SN)rrrrs r__wrap_if_fakez(FakeWithinKernelCUDAArray.__wrap_if_fake/s dM *,T2 2Krc|t|jjvr/|jt |jj|S|j|jj |Sr!)dirr_ary(_FakeWithinKernelCUDAArray__wrap_if_fakegetattrr )r attrnames r __getattr__z%FakeWithinKernelCUDAArray.__getattr__5sW s4::??+ +&&wtzz'IJ J&&tzz'='=h'GH Hrc<|jj||yr!r __setitem__)r nmvals r __setattr__z%FakeWithinKernelCUDAArray.__setattr__;s r3'rcV|j|jj|Sr!)r&rr )r idxs rr z%FakeWithinKernelCUDAArray.__getitem__>s"""4::#9#9##>??rc<|jj||yr!r+r r1r.s rr,z%FakeWithinKernelCUDAArray.__setitem__As sC(rc,t|jSr!)lenrr s r__len__z!FakeWithinKernelCUDAArray.__len__Ds4::rct||}d|jd}|rtfd|D|d<tfd|D}||i|S)NcRt|tr|jj}|Sr!)rrrr%objs r convert_fakesz@FakeWithinKernelCUDAArray.__array_ufunc__..convert_fakesQs#89iinnJroutc3.K|] }|ywr!).0or<s r z.Ys!@Cq-"2Cc3.K|] }|ywr!r?)r@ar<s rrBz.Zs4t!]1%trC)r'gettuple)r ufuncmethodargskwargscallr=r<s @r__array_ufunc__z)FakeWithinKernelCUDAArray.__array_ufunc__Gs]uf%  jj !!@C!@@F5M4t44T$V$$rN) rrrrrr&r)r/r r,r7rMr?rrrr"s1& I (@)%rrceZdZdZdZd&dZedZedZdZ d&dZ edZ d'd Z d Z d Zd(d Zd&dZedZdZdZdZdZdZdZdZdZdZdZdZdZdZdZdZ dZ!d Z"d!Z#d"Z$d#Z%d$Z&d&d%Z'y ))rzj Implements the interface of a DeviceArray/DeviceRecord, but mostly just wraps a NumPy array. Tc ||_||_yr!)r%streamr aryrPs rrzFakeCUDAArray.__init__fs  rc.|jjSr!r%nbytesr6s r alloc_sizezFakeCUDAArray.alloc_sizejsyyrc.|jjSr!rTr6s rrUzFakeCUDAArray.nbytesnsyyrcx t|j|}|S#t$r}d|z}t||d}~wwxYw)Nz#Wrapped array has no attribute '%s')r'r%AttributeError)r r(attremsgs rr)zFakeCUDAArray.__getattr__ssC -499h/DK -7(BC %1 , -s 949c.t|j|Sr!rr%)r rPs rbindzFakeCUDAArray.bind{sTYY//rc"|jSr!) transposer6s rTzFakeCUDAArray.T~s~~rNcVttj|j|S)N)axes)rnprar%)r rds rrazFakeCUDAArray.transposesR\\$))$?@@rc|jj|}t|tjtj fvr|St ||jS)NrP)r%r typerendarrayvoidrrP)r r1rets rr zFakeCUDAArray.__getitem__sEii##C( 9RZZ1 1J T[[9 9rc:|jj||Sr!)r%r,r3s rr,zFakeCUDAArray.__setitem__syy$$S#..rc| tj|j}n t||tj||j|Sr!)re empty_liker%check_array_compatibilitycopytorQs r copy_to_hostzFakeCUDAArray.copy_to_hosts: ;-- *C %dC 0 #tyy! rcXt|t|t|}}t|trt|t ||nAt j ||jdrdnddtdkrdnd}t ||t j|j|y) a Copy from the provided array into this array. This may be less forgiving than the CUDA Python implementation, which will copy data up to the length of the smallest of the two arrays, whereas this expects the size of the arrays to be equal. C_CONTIGUOUSCFTrFN)ordersubokcopy) sentry_contiguous array_corerrrorearrayflagsrrpr%)r rRrP self_coreary_cores rcopy_to_devicezFakeCUDAArray.copy_to_devices $(. 38 c= ) c " %i :xx&__^>rcLt|jj|i|Sr!)rr%reshapers rrzFakeCUDAArray.reshapes$.TYY..??@@rcLt|jj|i|Sr!)rr%viewrs rrzFakeCUDAArray.views!^TYY^^T,, 1101010001001 rrc|jr |js|Sg}|jD]#}|j|dk(rdn td%|t |S)aG Extract the repeated core of a broadcast array. Broadcast arrays are by definition non-contiguous due to repeated dimensions, i.e., dimensions with stride 0. In order to ascertain memory contiguity and copy the underlying data from such arrays, we must create a view without the repeated dimensions. rN)stridessizeappendslicerG)rR core_indexstrides rr|r|sS ;;chh J++v{!d < uZ !!rc|jj}tt|jt|j D]\}}|dkDs |dk7s||k7ry||z}y)z Returns True iff `ary` is C-style contiguous while ignoring broadcasted and 1-sized dimensions. As opposed to array_core(), it does not call require_context(), which can be quite expensive. rFT)dtypeitemsizezipreversedrr)rRrrrs r is_contiguousr s` 99  DXcii0(3;;2GH v 191v~ EMD I rcvt|}|jds|jdsttyy)Nrs F_CONTIGUOUS)r|r~ ValueErrorerrmsg_contiguous_buffer)rRcores rr{r{s7 c?D ::n %djj.H122/I %rc|j|j}}|j|jk7r%td|jd|j|j|jk7r%t d|jd|j|j |j k7r%t d|j d|j y)Nzincompatible dtype: z vs. zincompatible shape: zincompatible strides: )squeezer TypeErrorrrr)ary1ary2ary1sqary2sqs rroro!s\\^T\\^FF zzTZZTZZ12 2 ||v||#**djj23 3 ~~',, 67 7(rc tj|tdkrdndd}t|||jj dvrtj n |j}ttjtjt|j||j|j|jjt|S|j||y) NrvFTrzryMm)bufferrrr)rhrg)rer}rr{rcharint64rrirzr|rrrrhr)rRrPrzto buffer_dtypes r to_devicer.s ((3!.!7T Cc z#&99>>T#9rxxsyy  JJwwz#/44\Biiii   dS d"    #f-rc#Kdywr!r?)args rpinnedrAs  scRdD]}||vs|j|t|i|S)N)portablewc)pop device_array)rJrK unused_args r mapped_arrayrFs2(   JJz ")  ( ((rrtc4tj||||SN)rrrrxrerirrrrxs r pinned_arrayrM ::E7%u MMrc4tj||||Srrrs r managed_arrayrQrrcpd|vr|jdnd}ttj|i||S)NrPrrg)rrreri)rJrKrPs rrrUs4%-%7VZZ !QF T4V4V DDrc|jds|jds|jdkryt|jDcgc]}|}}|j ddgt |jz}|j j}|D]\}}|||<||j|z}t|Scc}w)z[ Given an array, compute strides for a new contiguous array of the same shape. rsrrNc |dS)Nrr?)xs rz0_contiguous_strides_like_array..ls!A$r)keyr) r~ndim enumeratersortr5rrrrG)rRr stridepermrri_perm_s r_contiguous_strides_like_arrayrZs yy CIIn$=Q( 46414J6OOO'cC $$G YY  F  #))F##  >7s B?cB|jdr|jdsyy)Nrrsrurt)r~)rRs r_order_like_arrayrws yy >)Brctt|}t|}t|j|j||SNr)rrrrr)rRrPrrxs rdevice_array_liker~4,S1G c "E ciisyy'# %%rctt|}t|}t|j|j||Sr)rrrrr)rRrrxs rpinned_array_likerrrct|tr|dfSt|tjs#tj|t dkrdndd}t |||dfS)NFrvTr)rrrerjr}rr)rRrPrzs r auto_devicer sZ#}%Ez c277 #hh '&0d S&$ ' --rct|ddS)z$Check if an object is a CUDA ndarrayrF)r'r:s ris_cuda_ndarrayr s 3*E 22rctfd}|dt|dt|dtj|dty)z,Verify the CUDA ndarray interface for an objct|s t|tt||st|d|y)Nz must be of type )hasattrrYrr')rZtypr;s r requires_attrz4verify_cuda_ndarray_interface..requires_attrs>sD! & &'#t,c2 D#!FG G3rrrrrN)require_cuda_ndarrayrGrerr )r;rs` rverify_cuda_ndarray_interfacers?H '5!)U#'288$&#rc0t|s tdy)z9Raises ValueError is is_cuda_ndarray(obj) evaluates Falsezrequire an cuda ndarray objectN)r rr:s rrrs 3 9:: r)rTNr)rT)"r contextlibrnumba.np.numpy_supportrnumpyre DeviceRecordfrom_record_likerrGrobjectrrr|rr{rorrrfloat64rrrrrrrr r rrr?rrrs&0 3 5 59%9%xX FX v"$  3 7.&  )!jj$cN "zz4sNE :%% .3  ;r