1target datalayout = "e-m:e-i64:64-i128:128-n32:64-S128"
2target triple = "aarch64-linux-android"
3
4%struct.rs_allocation = type { i64*, i64*, i64*, i64* }
5
6declare i8* @rsOffset(%struct.rs_allocation* nocapture readonly %a, i32 %sizeOf, i32 %x, i32 %y, i32 %z)
7declare i8* @rsOffsetNs(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z)
8
9; The loads and stores in this file are annotated with RenderScript-specific
10; information for the type based alias analysis, such that the TBAA analysis
11; understands that loads and stores from two allocations with different types
12; can never access the same memory element. This is different from C, where
13; a char or uchar load/store is special as it can alias with about everything.
14;
15; The TBAA tree in this file has the the node "RenderScript Distinct TBAA" as
16; its root.
17; This means all loads/stores that share this common root can be proven to not
18; alias. However, the alias analysis still has to assume MayAlias between
19; memory accesses in this file and memory accesses annotated with the C/C++
20; TBAA metadata.
21; A node named "RenderScript TBAA" wraps our distinct TBAA root node.
22; If we can ensure that all accesses to elements loaded from RenderScript
23; allocations are either annotated with the RenderScript TBAA information or
24; not annotated at all, but never annotated with the C/C++ metadata, we
25; can add the "RenderScript TBAA" tree under the C/C++ TBAA tree. This enables
26; TBAA to prove that an access to data from the RenderScript allocation
27; does not alias with a load/store accessing something not part of a RenderScript
28; allocation.
29; We do this by swapping the second operand of "RenderScript TBAA" with the node
30; for "Simple C/C++ TBAA", thus connecting these TBAA groups. The other root
31; node (with no children) can then safely be dropped from the analysis.
32
33!13 = !{!"RenderScript Distinct TBAA"}
34!14 = !{!"RenderScript TBAA", !13}
35!15 = !{!"allocation", !14}
36
37!21 = !{!"char", !15}
38define void @rsSetElementAtImpl_char(%struct.rs_allocation* nocapture readonly %a, i8 signext %val, i32 %x, i32 %y, i32 %z) #1 {
39  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 1, i32 %x, i32 %y, i32 %z) #2
40  store i8 %val, i8* %1, align 1, !tbaa !21
41  ret void
42}
43
44define signext i8 @rsGetElementAtImpl_char(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
45  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 1, i32 %x, i32 %y, i32 %z) #2
46  %2 = load i8, i8* %1, align 1, !tbaa !21
47  ret i8 %2
48}
49
50!22 = !{!"char2", !15}
51define void @rsSetElementAtImpl_char2(%struct.rs_allocation* nocapture readonly %a, i16 %val, i32 %x, i32 %y, i32 %z) #1 {
52  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 2, i32 %x, i32 %y, i32 %z) #2
53  %2 = bitcast i8* %1 to <2 x i8>*
54  %3 = bitcast i16 %val to <2 x i8>
55  store <2 x i8> %3, <2 x i8>* %2, align 2, !tbaa !22
56  ret void
57}
58
59define <2 x i8> @rsGetElementAtImpl_char2(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
60  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 2, i32 %x, i32 %y, i32 %z) #2
61  %2 = bitcast i8* %1 to <2 x i8>*
62  %3 = load <2 x i8>, <2 x i8>* %2, align 2, !tbaa !22
63  ret <2 x i8> %3
64}
65
66!23 = !{!"char3", !15}
67define void @rsSetElementAtImpl_char3(%struct.rs_allocation* nocapture readonly %a, i32 %val, i32 %x, i32 %y, i32 %z) #1 {
68  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 4, i32 %x, i32 %y, i32 %z) #2
69  %2 = bitcast i32 %val to <4 x i8>
70  %3 = shufflevector <4 x i8> %2, <4 x i8> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 undef>
71  %4 = bitcast i8* %1 to <4 x i8>*
72  store <4 x i8> %3, <4 x i8>* %4, align 4, !tbaa !23
73  ret void
74}
75
76define <3 x i8> @rsGetElementAtImpl_char3(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
77  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 4, i32 %x, i32 %y, i32 %z) #2
78  %2 = bitcast i8* %1 to <4 x i8>*
79  %3 = load <4 x i8>, <4 x i8>* %2, align 4, !tbaa !23
80  %4 = shufflevector <4 x i8> %3, <4 x i8> undef, <3 x i32> <i32 0, i32 1, i32 2>
81  ret <3 x i8> %4
82}
83
84!24 = !{!"char4", !15}
85define void @rsSetElementAtImpl_char4(%struct.rs_allocation* nocapture readonly %a, i32 %val, i32 %x, i32 %y, i32 %z) #1 {
86  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 4, i32 %x, i32 %y, i32 %z) #2
87  %2 = bitcast i8* %1 to <4 x i8>*
88  %3 = bitcast i32 %val to <4 x i8>
89  store <4 x i8> %3, <4 x i8>* %2, align 4, !tbaa !24
90  ret void
91}
92
93define <4 x i8> @rsGetElementAtImpl_char4(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
94  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 4, i32 %x, i32 %y, i32 %z) #2
95  %2 = bitcast i8* %1 to <4 x i8>*
96  %3 = load <4 x i8>, <4 x i8>* %2, align 4, !tbaa !24
97  ret <4 x i8> %3
98}
99
100!25 = !{!"uchar", !15}
101define void @rsSetElementAtImpl_uchar(%struct.rs_allocation* nocapture readonly %a, i8 zeroext %val, i32 %x, i32 %y, i32 %z) #1 {
102  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 1, i32 %x, i32 %y, i32 %z) #2
103  store i8 %val, i8* %1, align 1, !tbaa !25
104  ret void
105}
106
107define zeroext i8 @rsGetElementAtImpl_uchar(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
108  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 1, i32 %x, i32 %y, i32 %z) #2
109  %2 = load i8, i8* %1, align 1, !tbaa !25
110  ret i8 %2
111}
112
113!26 = !{!"uchar2", !15}
114define void @rsSetElementAtImpl_uchar2(%struct.rs_allocation* nocapture readonly %a, i16 %val, i32 %x, i32 %y, i32 %z) #1 {
115  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 2, i32 %x, i32 %y, i32 %z) #2
116  %2 = bitcast i8* %1 to <2 x i8>*
117  %3 = bitcast i16 %val to <2 x i8>
118  store <2 x i8> %3, <2 x i8>* %2, align 2, !tbaa !26
119  ret void
120}
121
122define <2 x i8> @rsGetElementAtImpl_uchar2(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
123  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 2, i32 %x, i32 %y, i32 %z) #2
124  %2 = bitcast i8* %1 to <2 x i8>*
125  %3 = load <2 x i8>, <2 x i8>* %2, align 2, !tbaa !26
126  ret <2 x i8> %3
127}
128
129!27 = !{!"uchar3", !15}
130define void @rsSetElementAtImpl_uchar3(%struct.rs_allocation* nocapture readonly %a, i32 %val, i32 %x, i32 %y, i32 %z) #1 {
131  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 4, i32 %x, i32 %y, i32 %z) #2
132  %2 = bitcast i32 %val to <4 x i8>
133  %3 = shufflevector <4 x i8> %2, <4 x i8> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 undef>
134  %4 = bitcast i8* %1 to <4 x i8>*
135  store <4 x i8> %3, <4 x i8>* %4, align 4, !tbaa !27
136  ret void
137}
138
139define <3 x i8> @rsGetElementAtImpl_uchar3(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
140  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 4, i32 %x, i32 %y, i32 %z) #2
141  %2 = bitcast i8* %1 to <4 x i8>*
142  %3 = load <4 x i8>, <4 x i8>* %2, align 4, !tbaa !27
143  %4 = shufflevector <4 x i8> %3, <4 x i8> undef, <3 x i32> <i32 0, i32 1, i32 2>
144  ret <3 x i8> %4
145}
146
147!28 = !{!"uchar4", !15}
148define void @rsSetElementAtImpl_uchar4(%struct.rs_allocation* nocapture readonly %a, i32 %val, i32 %x, i32 %y, i32 %z) #1 {
149  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 4, i32 %x, i32 %y, i32 %z) #2
150  %2 = bitcast i8* %1 to <4 x i8>*
151  %3 = bitcast i32 %val to <4 x i8>
152  store <4 x i8> %3, <4 x i8>* %2, align 4, !tbaa !28
153  ret void
154}
155
156define <4 x i8> @rsGetElementAtImpl_uchar4(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
157  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 4, i32 %x, i32 %y, i32 %z) #2
158  %2 = bitcast i8* %1 to <4 x i8>*
159  %3 = load <4 x i8>, <4 x i8>* %2, align 4, !tbaa !28
160  ret <4 x i8> %3
161}
162
163!29 = !{!"short", !15}
164define void @rsSetElementAtImpl_short(%struct.rs_allocation* nocapture readonly %a, i16 signext %val, i32 %x, i32 %y, i32 %z) #1 {
165  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 2, i32 %x, i32 %y, i32 %z) #2
166  %2 = bitcast i8* %1 to i16*
167  store i16 %val, i16* %2, align 2, !tbaa !29
168  ret void
169}
170
171define signext i16 @rsGetElementAtImpl_short(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
172  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 2, i32 %x, i32 %y, i32 %z) #2
173  %2 = bitcast i8* %1 to i16*
174  %3 = load i16, i16* %2, align 2, !tbaa !29
175  ret i16 %3
176}
177
178!30 = !{!"short2", !15}
179define void @rsSetElementAtImpl_short2(%struct.rs_allocation* nocapture readonly %a, i32 %val, i32 %x, i32 %y, i32 %z) #1 {
180  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 4, i32 %x, i32 %y, i32 %z) #2
181  %2 = bitcast i8* %1 to <2 x i16>*
182  %3 = bitcast i32 %val to <2 x i16>
183  store <2 x i16> %3, <2 x i16>* %2, align 4, !tbaa !30
184  ret void
185}
186
187define <2 x i16> @rsGetElementAtImpl_short2(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
188  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 4, i32 %x, i32 %y, i32 %z) #2
189  %2 = bitcast i8* %1 to <2 x i16>*
190  %3 = load <2 x i16>, <2 x i16>* %2, align 4, !tbaa !30
191  ret <2 x i16> %3
192}
193
194!31 = !{!"short3", !15}
195define void @rsSetElementAtImpl_short3(%struct.rs_allocation* nocapture readonly %a, <2 x i32> %val, i32 %x, i32 %y, i32 %z) #1 {
196  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 8, i32 %x, i32 %y, i32 %z) #2
197  %2 = bitcast <2 x i32> %val to <4 x i16>
198  %3 = shufflevector <4 x i16> %2, <4 x i16> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 undef>
199  %4 = bitcast i8* %1 to <4 x i16>*
200  store <4 x i16> %3, <4 x i16>* %4, align 8, !tbaa !31
201  ret void
202}
203
204define <3 x i16> @rsGetElementAtImpl_short3(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
205  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 8, i32 %x, i32 %y, i32 %z) #2
206  %2 = bitcast i8* %1 to <4 x i16>*
207  %3 = load <4 x i16>, <4 x i16>* %2, align 8, !tbaa !31
208  %4 = shufflevector <4 x i16> %3, <4 x i16> undef, <3 x i32> <i32 0, i32 1, i32 2>
209  ret <3 x i16> %4
210}
211
212!32 = !{!"short4", !15}
213define void @rsSetElementAtImpl_short4(%struct.rs_allocation* nocapture readonly %a, <4 x i16> %val, i32 %x, i32 %y, i32 %z) #1 {
214  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 8, i32 %x, i32 %y, i32 %z) #2
215  %2 = bitcast i8* %1 to <4 x i16>*
216  store <4 x i16> %val, <4 x i16>* %2, align 8, !tbaa !32
217  ret void
218}
219
220define <4 x i16> @rsGetElementAtImpl_short4(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
221  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 8, i32 %x, i32 %y, i32 %z) #2
222  %2 = bitcast i8* %1 to <4 x i16>*
223  %3 = load <4 x i16>, <4 x i16>* %2, align 8, !tbaa !32
224  ret <4 x i16> %3
225}
226
227!33 = !{!"ushort", !15}
228define void @rsSetElementAtImpl_ushort(%struct.rs_allocation* nocapture readonly %a, i16 zeroext %val, i32 %x, i32 %y, i32 %z) #1 {
229  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 2, i32 %x, i32 %y, i32 %z) #2
230  %2 = bitcast i8* %1 to i16*
231  store i16 %val, i16* %2, align 2, !tbaa !33
232  ret void
233}
234
235define zeroext i16 @rsGetElementAtImpl_ushort(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
236  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 2, i32 %x, i32 %y, i32 %z) #2
237  %2 = bitcast i8* %1 to i16*
238  %3 = load i16, i16* %2, align 2, !tbaa !33
239  ret i16 %3
240}
241
242!34 = !{!"ushort2", !15}
243define void @rsSetElementAtImpl_ushort2(%struct.rs_allocation* nocapture readonly %a, i32 %val, i32 %x, i32 %y, i32 %z) #1 {
244  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 4, i32 %x, i32 %y, i32 %z) #2
245  %2 = bitcast i8* %1 to <2 x i16>*
246  %3 = bitcast i32 %val to <2 x i16>
247  store <2 x i16> %3, <2 x i16>* %2, align 4, !tbaa !34
248  ret void
249}
250
251define <2 x i16> @rsGetElementAtImpl_ushort2(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
252  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 4, i32 %x, i32 %y, i32 %z) #2
253  %2 = bitcast i8* %1 to <2 x i16>*
254  %3 = load <2 x i16>, <2 x i16>* %2, align 4, !tbaa !34
255  ret <2 x i16> %3
256}
257
258!35 = !{!"ushort3", !15}
259define void @rsSetElementAtImpl_ushort3(%struct.rs_allocation* nocapture readonly %a, <2 x i32> %val, i32 %x, i32 %y, i32 %z) #1 {
260  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 8, i32 %x, i32 %y, i32 %z) #2
261  %2 = bitcast <2 x i32> %val to <4 x i16>
262  %3 = shufflevector <4 x i16> %2, <4 x i16> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 undef>
263  %4 = bitcast i8* %1 to <4 x i16>*
264  store <4 x i16> %3, <4 x i16>* %4, align 8, !tbaa !35
265  ret void
266}
267
268define <3 x i16> @rsGetElementAtImpl_ushort3(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
269  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 8, i32 %x, i32 %y, i32 %z) #2
270  %2 = bitcast i8* %1 to <4 x i16>*
271  %3 = load <4 x i16>, <4 x i16>* %2, align 8, !tbaa !35
272  %4 = shufflevector <4 x i16> %3, <4 x i16> undef, <3 x i32> <i32 0, i32 1, i32 2>
273  ret <3 x i16> %4
274}
275
276!36 = !{!"ushort4", !15}
277define void @rsSetElementAtImpl_ushort4(%struct.rs_allocation* nocapture readonly %a, <4 x i16> %val, i32 %x, i32 %y, i32 %z) #1 {
278  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 8, i32 %x, i32 %y, i32 %z) #2
279  %2 = bitcast i8* %1 to <4 x i16>*
280  store <4 x i16> %val, <4 x i16>* %2, align 8, !tbaa !36
281  ret void
282}
283
284define <4 x i16> @rsGetElementAtImpl_ushort4(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
285  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 8, i32 %x, i32 %y, i32 %z) #2
286  %2 = bitcast i8* %1 to <4 x i16>*
287  %3 = load <4 x i16>, <4 x i16>* %2, align 8, !tbaa !36
288  ret <4 x i16> %3
289}
290
291!37 = !{!"int", !15}
292define void @rsSetElementAtImpl_int(%struct.rs_allocation* nocapture readonly %a, i32 %val, i32 %x, i32 %y, i32 %z) #1 {
293  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 4, i32 %x, i32 %y, i32 %z) #2
294  %2 = bitcast i8* %1 to i32*
295  store i32 %val, i32* %2, align 4, !tbaa !37
296  ret void
297}
298
299define i32 @rsGetElementAtImpl_int(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
300  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 4, i32 %x, i32 %y, i32 %z) #2
301  %2 = bitcast i8* %1 to i32*
302  %3 = load i32, i32* %2, align 4, !tbaa !37
303  ret i32 %3
304}
305
306!38 = !{!"int2", !15}
307define void @rsSetElementAtImpl_int2(%struct.rs_allocation* nocapture readonly %a, <2 x i32> %val, i32 %x, i32 %y, i32 %z) #1 {
308  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 8, i32 %x, i32 %y, i32 %z) #2
309  %2 = bitcast i8* %1 to <2 x i32>*
310  store <2 x i32> %val, <2 x i32>* %2, align 8, !tbaa !38
311  ret void
312}
313
314define <2 x i32> @rsGetElementAtImpl_int2(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
315  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 8, i32 %x, i32 %y, i32 %z) #2
316  %2 = bitcast i8* %1 to <2 x i32>*
317  %3 = load <2 x i32>, <2 x i32>* %2, align 8, !tbaa !38
318  ret <2 x i32> %3
319}
320
321!39 = !{!"int3", !15}
322define void @rsSetElementAtImpl_int3(%struct.rs_allocation* nocapture readonly %a, <4 x i32> %val, i32 %x, i32 %y, i32 %z) #1 {
323  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 16, i32 %x, i32 %y, i32 %z) #2
324  %2 = shufflevector <4 x i32> %val, <4 x i32> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 undef>
325  %3 = bitcast i8* %1 to <4 x i32>*
326  store <4 x i32> %2, <4 x i32>* %3, align 16, !tbaa !39
327  ret void
328}
329
330define <3 x i32> @rsGetElementAtImpl_int3(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
331  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 16, i32 %x, i32 %y, i32 %z) #2
332  %2 = bitcast i8* %1 to <4 x i32>*
333  %3 = load <4 x i32>, <4 x i32>* %2, align 8, !tbaa !39
334  %4 = shufflevector <4 x i32> %3, <4 x i32> undef, <3 x i32> <i32 0, i32 1, i32 2>
335  ret <3 x i32> %4
336}
337
338!40 = !{!"int4", !15}
339define void @rsSetElementAtImpl_int4(%struct.rs_allocation* nocapture readonly %a, <4 x i32> %val, i32 %x, i32 %y, i32 %z) #1 {
340  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 16, i32 %x, i32 %y, i32 %z) #2
341  %2 = bitcast i8* %1 to <4 x i32>*
342  store <4 x i32> %val, <4 x i32>* %2, align 16, !tbaa !40
343  ret void
344}
345
346define <4 x i32> @rsGetElementAtImpl_int4(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
347  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 16, i32 %x, i32 %y, i32 %z) #2
348  %2 = bitcast i8* %1 to <4 x i32>*
349  %3 = load <4 x i32>, <4 x i32>* %2, align 16, !tbaa !40
350  ret <4 x i32> %3
351}
352
353!41 = !{!"uint", !15}
354define void @rsSetElementAtImpl_uint(%struct.rs_allocation* nocapture readonly %a, i32 %val, i32 %x, i32 %y, i32 %z) #1 {
355  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 4, i32 %x, i32 %y, i32 %z) #2
356  %2 = bitcast i8* %1 to i32*
357  store i32 %val, i32* %2, align 4, !tbaa !41
358  ret void
359}
360
361define i32 @rsGetElementAtImpl_uint(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
362  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 4, i32 %x, i32 %y, i32 %z) #2
363  %2 = bitcast i8* %1 to i32*
364  %3 = load i32, i32* %2, align 4, !tbaa !41
365  ret i32 %3
366}
367
368!42 = !{!"uint2", !15}
369define void @rsSetElementAtImpl_uint2(%struct.rs_allocation* nocapture readonly %a, <2 x i32> %val, i32 %x, i32 %y, i32 %z) #1 {
370  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 8, i32 %x, i32 %y, i32 %z) #2
371  %2 = bitcast i8* %1 to <2 x i32>*
372  store <2 x i32> %val, <2 x i32>* %2, align 8, !tbaa !42
373  ret void
374}
375
376define <2 x i32> @rsGetElementAtImpl_uint2(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
377  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 8, i32 %x, i32 %y, i32 %z) #2
378  %2 = bitcast i8* %1 to <2 x i32>*
379  %3 = load <2 x i32>, <2 x i32>* %2, align 8, !tbaa !42
380  ret <2 x i32> %3
381}
382
383!43 = !{!"uint3", !15}
384define void @rsSetElementAtImpl_uint3(%struct.rs_allocation* nocapture readonly %a, <4 x i32> %val, i32 %x, i32 %y, i32 %z) #1 {
385  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 16, i32 %x, i32 %y, i32 %z) #2
386  %2 = shufflevector <4 x i32> %val, <4 x i32> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 undef>
387  %3 = bitcast i8* %1 to <4 x i32>*
388  store <4 x i32> %2, <4 x i32>* %3, align 16, !tbaa !43
389  ret void
390}
391
392define <3 x i32> @rsGetElementAtImpl_uint3(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
393  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 16, i32 %x, i32 %y, i32 %z) #2
394  %2 = bitcast i8* %1 to <4 x i32>*
395  %3 = load <4 x i32>, <4 x i32>* %2, align 8, !tbaa !43
396  %4 = shufflevector <4 x i32> %3, <4 x i32> undef, <3 x i32> <i32 0, i32 1, i32 2>
397  ret <3 x i32> %4
398}
399
400!44 = !{!"uint4", !15}
401define void @rsSetElementAtImpl_uint4(%struct.rs_allocation* nocapture readonly %a, <4 x i32> %val, i32 %x, i32 %y, i32 %z) #1 {
402  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 16, i32 %x, i32 %y, i32 %z) #2
403  %2 = bitcast i8* %1 to <4 x i32>*
404  store <4 x i32> %val, <4 x i32>* %2, align 16, !tbaa !44
405  ret void
406}
407
408define <4 x i32> @rsGetElementAtImpl_uint4(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
409  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 16, i32 %x, i32 %y, i32 %z) #2
410  %2 = bitcast i8* %1 to <4 x i32>*
411  %3 = load <4 x i32>, <4 x i32>* %2, align 16, !tbaa !44
412  ret <4 x i32> %3
413}
414
415!45 = !{!"long", !15}
416define void @rsSetElementAtImpl_long(%struct.rs_allocation* nocapture readonly %a, i64 %val, i32 %x, i32 %y, i32 %z) #1 {
417  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 8, i32 %x, i32 %y, i32 %z) #2
418  %2 = bitcast i8* %1 to i64*
419  store i64 %val, i64* %2, align 8, !tbaa !45
420  ret void
421}
422
423define i64 @rsGetElementAtImpl_long(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
424  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 8, i32 %x, i32 %y, i32 %z) #2
425  %2 = bitcast i8* %1 to i64*
426  %3 = load i64, i64* %2, align 8, !tbaa !45
427  ret i64 %3
428}
429
430!46 = !{!"long2", !15}
431define void @rsSetElementAtImpl_long2(%struct.rs_allocation* nocapture readonly %a, <2 x i64> %val, i32 %x, i32 %y, i32 %z) #1 {
432  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 16, i32 %x, i32 %y, i32 %z) #2
433  %2 = bitcast i8* %1 to <2 x i64>*
434  store <2 x i64> %val, <2 x i64>* %2, align 16, !tbaa !46
435  ret void
436}
437
438define <2 x i64> @rsGetElementAtImpl_long2(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
439  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 16, i32 %x, i32 %y, i32 %z) #2
440  %2 = bitcast i8* %1 to <2 x i64>*
441  %3 = load <2 x i64>, <2 x i64>* %2, align 16, !tbaa !46
442  ret <2 x i64> %3
443}
444
445!47 = !{!"long3", !15}
446define void @rsSetElementAtImpl_long3(%struct.rs_allocation* nocapture readonly %a, <3 x i64>* %val, i32 %x, i32 %y, i32 %z) #1 {
447  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 32, i32 %x, i32 %y, i32 %z) #2
448  %2 = load <3 x i64>, <3 x i64>* %val
449  %3 = shufflevector <3 x i64> %2, <3 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 undef>
450  %4 = bitcast i8* %1 to <4 x i64>*
451  store <4 x i64> %3, <4 x i64>* %4, align 32, !tbaa !47
452  ret void
453}
454
455define void @rsGetElementAtImpl_long3(<3 x i64>* noalias nocapture sret %agg.result, %struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #1 {
456  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 32, i32 %x, i32 %y, i32 %z) #2
457  %2 = bitcast i8* %1 to <4 x i64>*
458  %3 = load <4 x i64>, <4 x i64>* %2, align 32
459  %4 = bitcast <3 x i64>* %agg.result to <4 x i64>*
460  store <4 x i64> %3, <4 x i64>* %4, align 32, !tbaa !47
461  ret void
462}
463
464!48 = !{!"long4", !15}
465define void @rsSetElementAtImpl_long4(%struct.rs_allocation* nocapture readonly %a, <4 x i64>* %val, i32 %x, i32 %y, i32 %z) #1 {
466  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 32, i32 %x, i32 %y, i32 %z) #2
467  %2 = load <4 x i64>, <4 x i64>* %val
468  %3 = bitcast i8* %1 to <4 x i64>*
469  store <4 x i64> %2, <4 x i64>* %3, align 32, !tbaa !48
470  ret void
471}
472
473define void @rsGetElementAtImpl_long4(<4 x i64>* noalias nocapture sret %agg.result, %struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #1 {
474  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 32, i32 %x, i32 %y, i32 %z) #2
475  %2 = bitcast i8* %1 to <4 x i64>*
476  %3 = load <4 x i64>, <4 x i64>* %2, align 32, !tbaa !15
477  store <4 x i64> %3, <4 x i64>* %agg.result, align 32, !tbaa !48
478  ret void
479}
480
481!49 = !{!"ulong", !15}
482define void @rsSetElementAtImpl_ulong(%struct.rs_allocation* nocapture readonly %a, i64 %val, i32 %x, i32 %y, i32 %z) #1 {
483  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 8, i32 %x, i32 %y, i32 %z) #2
484  %2 = bitcast i8* %1 to i64*
485  store i64 %val, i64* %2, align 8, !tbaa !49
486  ret void
487}
488
489define i64 @rsGetElementAtImpl_ulong(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
490  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 8, i32 %x, i32 %y, i32 %z) #2
491  %2 = bitcast i8* %1 to i64*
492  %3 = load i64, i64* %2, align 8, !tbaa !49
493  ret i64 %3
494}
495
496!50 = !{!"ulong2", !15}
497define void @rsSetElementAtImpl_ulong2(%struct.rs_allocation* nocapture readonly %a, <2 x i64> %val, i32 %x, i32 %y, i32 %z) #1 {
498  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 16, i32 %x, i32 %y, i32 %z) #2
499  %2 = bitcast i8* %1 to <2 x i64>*
500  store <2 x i64> %val, <2 x i64>* %2, align 16, !tbaa !50
501  ret void
502}
503
504define <2 x i64> @rsGetElementAtImpl_ulong2(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
505  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 16, i32 %x, i32 %y, i32 %z) #2
506  %2 = bitcast i8* %1 to <2 x i64>*
507  %3 = load <2 x i64>, <2 x i64>* %2, align 16, !tbaa !50
508  ret <2 x i64> %3
509}
510
511!51 = !{!"ulong3", !15}
512define void @rsSetElementAtImpl_ulong3(%struct.rs_allocation* nocapture readonly %a, <3 x i64>* %val, i32 %x, i32 %y, i32 %z) #1 {
513  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 32, i32 %x, i32 %y, i32 %z) #2
514  %2 = load <3 x i64>, <3 x i64>* %val
515  %3 = shufflevector <3 x i64> %2, <3 x i64> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 undef>
516  %4 = bitcast i8* %1 to <4 x i64>*
517  store <4 x i64> %3, <4 x i64>* %4, align 32, !tbaa !51
518  ret void
519}
520
521define void @rsGetElementAtImpl_ulong3(<3 x i64>* noalias nocapture sret %agg.result, %struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #1 {
522  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 32, i32 %x, i32 %y, i32 %z) #2
523  %2 = bitcast i8* %1 to <4 x i64>*
524  %3 = load <4 x i64>, <4 x i64>* %2, align 32
525  %4 = bitcast <3 x i64>* %agg.result to <4 x i64>*
526  store <4 x i64> %3, <4 x i64>* %4, align 32, !tbaa !51
527  ret void
528}
529
530!52 = !{!"ulong4", !15}
531define void @rsSetElementAtImpl_ulong4(%struct.rs_allocation* nocapture readonly %a, <4 x i64>* %val, i32 %x, i32 %y, i32 %z) #1 {
532  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 32, i32 %x, i32 %y, i32 %z) #2
533  %2 = load <4 x i64>, <4 x i64>* %val
534  %3 = bitcast i8* %1 to <4 x i64>*
535  store <4 x i64> %2, <4 x i64>* %3, align 32, !tbaa !52
536  ret void
537}
538
539define void @rsGetElementAtImpl_ulong4(<4 x i64>* noalias nocapture sret %agg.result, %struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #1 {
540  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 32, i32 %x, i32 %y, i32 %z) #2
541  %2 = bitcast i8* %1 to <4 x i64>*
542  %3 = load <4 x i64>, <4 x i64>* %2, align 32, !tbaa !15
543  store <4 x i64> %3, <4 x i64>* %agg.result, align 32, !tbaa !52
544  ret void
545}
546
547!53 = !{!"float", !15}
548define void @rsSetElementAtImpl_float(%struct.rs_allocation* nocapture readonly %a, float %val, i32 %x, i32 %y, i32 %z) #1 {
549  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 4, i32 %x, i32 %y, i32 %z) #2
550  %2 = bitcast i8* %1 to float*
551  store float %val, float* %2, align 4, !tbaa !53
552  ret void
553}
554
555define float @rsGetElementAtImpl_float(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
556  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 4, i32 %x, i32 %y, i32 %z) #2
557  %2 = bitcast i8* %1 to float*
558  %3 = load float, float* %2, align 4, !tbaa !53
559  ret float %3
560}
561
562!54 = !{!"float2", !15}
563define void @rsSetElementAtImpl_float2(%struct.rs_allocation* nocapture readonly %a, <2 x float> %val, i32 %x, i32 %y, i32 %z) #1 {
564  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 8, i32 %x, i32 %y, i32 %z) #2
565  %2 = bitcast i8* %1 to <2 x float>*
566  store <2 x float> %val, <2 x float>* %2, align 8, !tbaa !54
567  ret void
568}
569
570define <2 x float> @rsGetElementAtImpl_float2(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
571  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 8, i32 %x, i32 %y, i32 %z) #2
572  %2 = bitcast i8* %1 to <2 x float>*
573  %3 = load <2 x float>, <2 x float>* %2, align 8, !tbaa !54
574  ret <2 x float> %3
575}
576
577!55 = !{!"float3", !15}
578define void @rsSetElementAtImpl_float3(%struct.rs_allocation* nocapture readonly %a, <4 x i32> %val, i32 %x, i32 %y, i32 %z) #1 {
579  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 16, i32 %x, i32 %y, i32 %z) #2
580  %2 = bitcast <4 x i32> %val to <4 x float>
581  %3 = shufflevector <4 x float> %2, <4 x float> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 undef>
582  %4 = bitcast i8* %1 to <4 x float>*
583  store <4 x float> %3, <4 x float>* %4, align 16, !tbaa !55
584  ret void
585}
586
587define <3 x float> @rsGetElementAtImpl_float3(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
588  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 16, i32 %x, i32 %y, i32 %z) #2
589  %2 = bitcast i8* %1 to <4 x float>*
590  %3 = load <4 x float>, <4 x float>* %2, align 8, !tbaa !55
591  %4 = shufflevector <4 x float> %3, <4 x float> undef, <3 x i32> <i32 0, i32 1, i32 2>
592  ret <3 x float> %4
593}
594
595!56 = !{!"float4", !15}
596define void @rsSetElementAtImpl_float4(%struct.rs_allocation* nocapture readonly %a, <4 x float> %val, i32 %x, i32 %y, i32 %z) #1 {
597  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 16, i32 %x, i32 %y, i32 %z) #2
598  %2 = bitcast i8* %1 to <4 x float>*
599  store <4 x float> %val, <4 x float>* %2, align 16, !tbaa !56
600  ret void
601}
602
603define <4 x float> @rsGetElementAtImpl_float4(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
604  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 16, i32 %x, i32 %y, i32 %z) #2
605  %2 = bitcast i8* %1 to <4 x float>*
606  %3 = load <4 x float>, <4 x float>* %2, align 16, !tbaa !56
607  ret <4 x float> %3
608}
609
610!57 = !{!"double", !15}
611define void @rsSetElementAtImpl_double(%struct.rs_allocation* nocapture readonly %a, double %val, i32 %x, i32 %y, i32 %z) #1 {
612  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 8, i32 %x, i32 %y, i32 %z) #2
613  %2 = bitcast i8* %1 to double*
614  store double %val, double* %2, align 8, !tbaa !57
615  ret void
616}
617
618define double @rsGetElementAtImpl_double(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
619  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 8, i32 %x, i32 %y, i32 %z) #2
620  %2 = bitcast i8* %1 to double*
621  %3 = load double, double* %2, align 8, !tbaa !57
622  ret double %3
623}
624
625!58 = !{!"double2", !15}
626define void @rsSetElementAtImpl_double2(%struct.rs_allocation* nocapture readonly %a, <2 x double> %val, i32 %x, i32 %y, i32 %z) #1 {
627  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 16, i32 %x, i32 %y, i32 %z) #2
628  %2 = bitcast i8* %1 to <2 x double>*
629  store <2 x double> %val, <2 x double>* %2, align 16, !tbaa !58
630  ret void
631}
632
633define <2 x double> @rsGetElementAtImpl_double2(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
634  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 16, i32 %x, i32 %y, i32 %z) #2
635  %2 = bitcast i8* %1 to <2 x double>*
636  %3 = load <2 x double>, <2 x double>* %2, align 16, !tbaa !58
637  ret <2 x double> %3
638}
639
640!59 = !{!"double3", !15}
641define void @rsSetElementAtImpl_double3(%struct.rs_allocation* nocapture readonly %a, <3 x double>* %val, i32 %x, i32 %y, i32 %z) #1 {
642  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 32, i32 %x, i32 %y, i32 %z) #2
643  %2 = load <3 x double>, <3 x double>* %val
644  %3 = shufflevector <3 x double> %2, <3 x double> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 undef>
645  %4 = bitcast i8* %1 to <4 x double>*
646  store <4 x double> %3, <4 x double>* %4, align 32, !tbaa !59
647  ret void
648}
649
650
651define void @rsGetElementAtImpl_double3(<3 x double>* noalias nocapture sret %agg.result, %struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #1 {
652  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 32, i32 %x, i32 %y, i32 %z) #2
653  %2 = bitcast i8* %1 to <4 x double>*
654  %3 = load <4 x double>, <4 x double>* %2, align 32
655  %4 = bitcast <3 x double>* %agg.result to <4 x double>*
656  store <4 x double> %3, <4 x double>* %4, align 32, !tbaa !59
657  ret void
658}
659
660!60 = !{!"double4", !15}
661define void @rsSetElementAtImpl_double4(%struct.rs_allocation* nocapture readonly %a, <4 x double>* %val, i32 %x, i32 %y, i32 %z) #1 {
662  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 32, i32 %x, i32 %y, i32 %z) #2
663  %2 = load <4 x double>, <4 x double>* %val
664  %3 = bitcast i8* %1 to <4 x double>*
665  store <4 x double> %2, <4 x double>* %3, align 32, !tbaa !60
666  ret void
667}
668define void @rsGetElementAtImpl_double4(<4 x double>* noalias nocapture sret %agg.result, %struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #1 {
669  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a, i32 32, i32 %x, i32 %y, i32 %z) #2
670  %2 = bitcast i8* %1 to <4 x double>*
671  %3 = load <4 x double>, <4 x double>* %2, align 32, !tbaa !15
672  store <4 x double> %3, <4 x double>* %agg.result, align 32, !tbaa !60
673  ret void
674}
675
676!61 = !{!"half", !15}
677define void @rsSetElementAtImpl_half(%struct.rs_allocation* nocapture readonly %a.coerce, half %val, i32 %x, i32 %y, i32 %z) #1 {
678  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a.coerce, i32 2, i32 %x, i32 %y, i32 %z) #2
679  %2 = bitcast i8* %1 to half*
680  store half %val, half* %2, align 2, !tbaa !61
681  ret void
682}
683
684define half @rsGetElementAtImpl_half(%struct.rs_allocation* nocapture readonly %a.coerce, i32 %x, i32 %y, i32 %z) #0 {
685  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a.coerce, i32 2, i32 %x, i32 %y, i32 %z) #2
686  %2 = bitcast i8* %1 to half*
687  %3 = load half, half* %2, align 2, !tbaa !61
688  ret half %3
689}
690
691!62 = !{!"half2", !15}
692define void @rsSetElementAtImpl_half2(%struct.rs_allocation* nocapture readonly %a.coerce, i32 %val, i32 %x, i32 %y, i32 %z) #1 {
693  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a.coerce, i32 4, i32 %x, i32 %y, i32 %z) #2
694  %2 = bitcast i8* %1 to <2 x half>*
695  %3 = bitcast i32 %val to <2 x half>
696  store <2 x half> %3, <2 x half>* %2, align 4, !tbaa !62
697  ret void
698}
699
700define <2 x half> @rsGetElementAtImpl_half2(%struct.rs_allocation* nocapture readonly %a.coerce, i32 %x, i32 %y, i32 %z) #0 {
701  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a.coerce, i32 4, i32 %x, i32 %y, i32 %z) #2
702  %2 = bitcast i8* %1 to <2 x half>*
703  %3 = load <2 x half>, <2 x half>* %2, align 4, !tbaa !62
704  ret <2 x half> %3
705}
706
707!63 = !{!"half3", !15}
708define void @rsSetElementAtImpl_half3(%struct.rs_allocation* nocapture readonly %a.coerce, <2 x i32> %val, i32 %x, i32 %y, i32 %z) #1 {
709  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a.coerce, i32 8, i32 %x, i32 %y, i32 %z) #2
710  %2 = bitcast <2 x i32> %val to <4 x half>
711  %3 = shufflevector <4 x half> %2, <4 x half> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 undef>
712  %4 = bitcast i8* %1 to <4 x half>*
713  store <4 x half> %3, <4 x half>* %4, align 8, !tbaa !63
714  ret void
715}
716
717define <3 x half> @rsGetElementAtImpl_half3(%struct.rs_allocation* nocapture readonly %a.coerce, i32 %x, i32 %y, i32 %z) #1 {
718  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a.coerce, i32 8, i32 %x, i32 %y, i32 %z) #2
719  %2 = bitcast i8* %1 to <4 x half>*
720  %3 = load <4 x half>, <4 x half>* %2, align 8, !tbaa !63
721  %4 = shufflevector <4 x half> %3, <4 x half> undef, <3 x i32> <i32 0, i32 1, i32 2>
722  ret <3 x half> %4
723}
724
725!64 = !{!"half4", !15}
726define void @rsSetElementAtImpl_half4(%struct.rs_allocation* nocapture readonly %a.coerce, <4 x half> %val, i32 %x, i32 %y, i32 %z) #1 {
727  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a.coerce, i32 8, i32 %x, i32 %y, i32 %z) #2
728  %2 = bitcast i8* %1 to <4 x half>*
729  store <4 x half> %val, <4 x half>* %2, align 8, !tbaa !64
730  ret void
731}
732
733define <4 x half> @rsGetElementAtImpl_half4(%struct.rs_allocation* nocapture readonly %a.coerce, i32 %x, i32 %y, i32 %z) #0 {
734  %1 = tail call i8* @rsOffset(%struct.rs_allocation* %a.coerce, i32 8, i32 %x, i32 %y, i32 %z) #2
735  %2 = bitcast i8* %1 to <4 x half>*
736  %3 = load <4 x half>, <4 x half>* %2, align 8, !tbaa !64
737  ret <4 x half> %3
738}
739
740
741define void @__rsAllocationVLoadXImpl_long4(<4 x i64>* noalias nocapture sret %agg.result, %struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #1 {
742  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
743  %2 = bitcast i8* %1 to <4 x i64>*
744  %3 = load <4 x i64>, <4 x i64>* %2, align 8
745  store <4 x i64> %3, <4 x i64>* %agg.result
746  ret void
747}
748define void @__rsAllocationVLoadXImpl_long3(<3 x i64>* noalias nocapture sret %agg.result, %struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #1 {
749  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
750  %2 = bitcast i8* %1 to <3 x i64>*
751  %3 = load <3 x i64>, <3 x i64>* %2, align 8
752  store <3 x i64> %3, <3 x i64>* %agg.result
753  ret void
754}
755define <2 x i64> @__rsAllocationVLoadXImpl_long2(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
756  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
757  %2 = bitcast i8* %1 to <2 x i64>*
758  %3 = load <2 x i64>, <2 x i64>* %2, align 8
759  ret <2 x i64> %3
760}
761
762define void @__rsAllocationVLoadXImpl_ulong4(<4 x i64>* noalias nocapture sret %agg.result, %struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #1 {
763  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
764  %2 = bitcast i8* %1 to <4 x i64>*
765  %3 = load <4 x i64>, <4 x i64>* %2, align 8
766  store <4 x i64> %3, <4 x i64>* %agg.result
767  ret void
768}
769define void @__rsAllocationVLoadXImpl_ulong3(<3 x i64>* noalias nocapture sret %agg.result, %struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #1 {
770  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
771  %2 = bitcast i8* %1 to <3 x i64>*
772  %3 = load <3 x i64>, <3 x i64>* %2, align 8
773  store <3 x i64> %3, <3 x i64>* %agg.result
774  ret void
775}
776define <2 x i64> @__rsAllocationVLoadXImpl_ulong2(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
777  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
778  %2 = bitcast i8* %1 to <2 x i64>*
779  %3 = load <2 x i64>, <2 x i64>* %2, align 8
780  ret <2 x i64> %3
781}
782
783define <4 x i32> @__rsAllocationVLoadXImpl_int4(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
784  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
785  %2 = bitcast i8* %1 to <4 x i32>*
786  %3 = load <4 x i32>, <4 x i32>* %2, align 4
787  ret <4 x i32> %3
788}
789define <3 x i32> @__rsAllocationVLoadXImpl_int3(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
790  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
791  %2 = bitcast i8* %1 to <3 x i32>*
792  %3 = load <3 x i32>, <3 x i32>* %2, align 4
793  ret <3 x i32> %3
794}
795define <2 x i32> @__rsAllocationVLoadXImpl_int2(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
796  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
797  %2 = bitcast i8* %1 to <2 x i32>*
798  %3 = load <2 x i32>, <2 x i32>* %2, align 4
799  ret <2 x i32> %3
800}
801
802define <4 x i32> @__rsAllocationVLoadXImpl_uint4(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
803  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
804  %2 = bitcast i8* %1 to <4 x i32>*
805  %3 = load <4 x i32>, <4 x i32>* %2, align 4
806  ret <4 x i32> %3
807}
808define <3 x i32> @__rsAllocationVLoadXImpl_uint3(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
809  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
810  %2 = bitcast i8* %1 to <3 x i32>*
811  %3 = load <3 x i32>, <3 x i32>* %2, align 4
812  ret <3 x i32> %3
813}
814define <2 x i32> @__rsAllocationVLoadXImpl_uint2(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
815  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
816  %2 = bitcast i8* %1 to <2 x i32>*
817  %3 = load <2 x i32>, <2 x i32>* %2, align 4
818  ret <2 x i32> %3
819}
820
821define <4 x i16> @__rsAllocationVLoadXImpl_short4(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
822  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
823  %2 = bitcast i8* %1 to <4 x i16>*
824  %3 = load <4 x i16>, <4 x i16>* %2, align 2
825  ret <4 x i16> %3
826}
827define <3 x i16> @__rsAllocationVLoadXImpl_short3(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
828  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
829  %2 = bitcast i8* %1 to <3 x i16>*
830  %3 = load <3 x i16>, <3 x i16>* %2, align 2
831  ret <3 x i16> %3
832}
833define <2 x i16> @__rsAllocationVLoadXImpl_short2(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
834  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
835  %2 = bitcast i8* %1 to <2 x i16>*
836  %3 = load <2 x i16>, <2 x i16>* %2, align 2
837  ret <2 x i16> %3
838}
839
840define <4 x i16> @__rsAllocationVLoadXImpl_ushort4(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
841  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
842  %2 = bitcast i8* %1 to <4 x i16>*
843  %3 = load <4 x i16>, <4 x i16>* %2, align 2
844  ret <4 x i16> %3
845}
846define <3 x i16> @__rsAllocationVLoadXImpl_ushort3(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
847  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
848  %2 = bitcast i8* %1 to <3 x i16>*
849  %3 = load <3 x i16>, <3 x i16>* %2, align 2
850  ret <3 x i16> %3
851}
852define <2 x i16> @__rsAllocationVLoadXImpl_ushort2(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
853  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
854  %2 = bitcast i8* %1 to <2 x i16>*
855  %3 = load <2 x i16>, <2 x i16>* %2, align 2
856  ret <2 x i16> %3
857}
858
859define <4 x i8> @__rsAllocationVLoadXImpl_char4(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
860  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
861  %2 = bitcast i8* %1 to <4 x i8>*
862  %3 = load <4 x i8>, <4 x i8>* %2, align 1
863  ret <4 x i8> %3
864}
865define <3 x i8> @__rsAllocationVLoadXImpl_char3(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
866  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
867  %2 = bitcast i8* %1 to <3 x i8>*
868  %3 = load <3 x i8>, <3 x i8>* %2, align 1
869  ret <3 x i8> %3
870}
871define <2 x i8> @__rsAllocationVLoadXImpl_char2(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
872  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
873  %2 = bitcast i8* %1 to <2 x i8>*
874  %3 = load <2 x i8>, <2 x i8>* %2, align 1
875  ret <2 x i8> %3
876}
877
878define <4 x i8> @__rsAllocationVLoadXImpl_uchar4(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
879  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
880  %2 = bitcast i8* %1 to <4 x i8>*
881  %3 = load <4 x i8>, <4 x i8>* %2, align 1
882  ret <4 x i8> %3
883}
884define <3 x i8> @__rsAllocationVLoadXImpl_uchar3(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
885  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
886  %2 = bitcast i8* %1 to <3 x i8>*
887  %3 = load <3 x i8>, <3 x i8>* %2, align 1
888  ret <3 x i8> %3
889}
890define <2 x i8> @__rsAllocationVLoadXImpl_uchar2(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
891  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
892  %2 = bitcast i8* %1 to <2 x i8>*
893  %3 = load <2 x i8>, <2 x i8>* %2, align 1
894  ret <2 x i8> %3
895}
896
897define <4 x float> @__rsAllocationVLoadXImpl_float4(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
898  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
899  %2 = bitcast i8* %1 to <4 x float>*
900  %3 = load <4 x float>, <4 x float>* %2, align 4
901  ret <4 x float> %3
902}
903define <3 x float> @__rsAllocationVLoadXImpl_float3(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
904  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
905  %2 = bitcast i8* %1 to <3 x float>*
906  %3 = load <3 x float>, <3 x float>* %2, align 4
907  ret <3 x float> %3
908}
909define <2 x float> @__rsAllocationVLoadXImpl_float2(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
910  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
911  %2 = bitcast i8* %1 to <2 x float>*
912  %3 = load <2 x float>, <2 x float>* %2, align 4
913  ret <2 x float> %3
914}
915
916define void @__rsAllocationVLoadXImpl_double4(<4 x double>* noalias nocapture sret %agg.result, %struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #1 {
917  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
918  %2 = bitcast i8* %1 to <4 x double>*
919  %3 = load <4 x double>, <4 x double>* %2, align 8
920  store <4 x double> %3, <4 x double>* %agg.result
921  ret void
922}
923define void @__rsAllocationVLoadXImpl_double3(<3 x double>* noalias nocapture sret %agg.result, %struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #1 {
924  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
925  %2 = bitcast i8* %1 to <3 x double>*
926  %3 = load <3 x double>, <3 x double>* %2, align 8
927  store <3 x double> %3, <3 x double>* %agg.result
928  ret void
929}
930define <2 x double> @__rsAllocationVLoadXImpl_double2(%struct.rs_allocation* nocapture readonly %a, i32 %x, i32 %y, i32 %z) #0 {
931  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
932  %2 = bitcast i8* %1 to <2 x double>*
933  %3 = load <2 x double>, <2 x double>* %2, align 8
934  ret <2 x double> %3
935}
936
937
938define void @__rsAllocationVStoreXImpl_long4(%struct.rs_allocation* nocapture readonly %a, <4 x i64>* %val, i32 %x, i32 %y, i32 %z) #1 {
939  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
940  %2 = load <4 x i64>, <4 x i64>* %val
941  %3 = bitcast i8* %1 to <4 x i64>*
942  store <4 x i64> %2, <4 x i64>* %3, align 8
943  ret void
944}
945define void @__rsAllocationVStoreXImpl_long3(%struct.rs_allocation* nocapture readonly %a, <3 x i64>* %val, i32 %x, i32 %y, i32 %z) #1 {
946  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
947  %2 = load <3 x i64>, <3 x i64>* %val
948  %3 = bitcast i8* %1 to <3 x i64>*
949  store <3 x i64> %2, <3 x i64>* %3, align 8
950  ret void
951}
952define void @__rsAllocationVStoreXImpl_long2(%struct.rs_allocation* nocapture readonly %a, <2 x i64> %val, i32 %x, i32 %y, i32 %z) #1 {
953  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
954  %2 = bitcast i8* %1 to <2 x i64>*
955  store <2 x i64> %val, <2 x i64>* %2, align 8
956  ret void
957}
958
959define void @__rsAllocationVStoreXImpl_ulong4(%struct.rs_allocation* nocapture readonly %a, <4 x i64>* %val, i32 %x, i32 %y, i32 %z) #1 {
960  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
961  %2 = load <4 x i64>, <4 x i64>* %val
962  %3 = bitcast i8* %1 to <4 x i64>*
963  store <4 x i64> %2, <4 x i64>* %3, align 8
964  ret void
965}
966define void @__rsAllocationVStoreXImpl_ulong3(%struct.rs_allocation* nocapture readonly %a, <3 x i64>* %val, i32 %x, i32 %y, i32 %z) #1 {
967  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
968  %2 = load <3 x i64>, <3 x i64>* %val
969  %3 = bitcast i8* %1 to <3 x i64>*
970  store <3 x i64> %2, <3 x i64>* %3, align 8
971  ret void
972}
973define void @__rsAllocationVStoreXImpl_ulong2(%struct.rs_allocation* nocapture readonly %a, <2 x i64> %val, i32 %x, i32 %y, i32 %z) #1 {
974  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
975  %2 = bitcast i8* %1 to <2 x i64>*
976  store <2 x i64> %val, <2 x i64>* %2, align 8
977  ret void
978}
979
980define void @__rsAllocationVStoreXImpl_int4(%struct.rs_allocation* nocapture readonly %a, <4 x i32> %val, i32 %x, i32 %y, i32 %z) #1 {
981  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
982  %2 = bitcast i8* %1 to <4 x i32>*
983  store <4 x i32> %val, <4 x i32>* %2, align 4
984  ret void
985}
986define void @__rsAllocationVStoreXImpl_int3(%struct.rs_allocation* nocapture readonly %a, <4 x i32> %val, i32 %x, i32 %y, i32 %z) #1 {
987  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
988  %2 = bitcast i8* %1 to <3 x i32>*
989  %3 = shufflevector <4 x i32> %val, <4 x i32> undef, <3 x i32> <i32 0, i32 1, i32 2>
990  store <3 x i32> %3, <3 x i32>* %2, align 4
991  ret void
992}
993define void @__rsAllocationVStoreXImpl_int2(%struct.rs_allocation* nocapture readonly %a, <2 x i32> %val, i32 %x, i32 %y, i32 %z) #1 {
994  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
995  %2 = bitcast i8* %1 to <2 x i32>*
996  store <2 x i32> %val, <2 x i32>* %2, align 4
997  ret void
998}
999
1000define void @__rsAllocationVStoreXImpl_uint4(%struct.rs_allocation* nocapture readonly %a, <4 x i32> %val, i32 %x, i32 %y, i32 %z) #1 {
1001  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
1002  %2 = bitcast i8* %1 to <4 x i32>*
1003  store <4 x i32> %val, <4 x i32>* %2, align 4
1004  ret void
1005}
1006define void @__rsAllocationVStoreXImpl_uint3(%struct.rs_allocation* nocapture readonly %a, <4 x i32> %val, i32 %x, i32 %y, i32 %z) #1 {
1007  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
1008  %2 = bitcast i8* %1 to <3 x i32>*
1009  %3 = shufflevector <4 x i32> %val, <4 x i32> undef, <3 x i32> <i32 0, i32 1, i32 2>
1010  store <3 x i32> %3, <3 x i32>* %2, align 4
1011  ret void
1012}
1013define void @__rsAllocationVStoreXImpl_uint2(%struct.rs_allocation* nocapture readonly %a, <2 x i32> %val, i32 %x, i32 %y, i32 %z) #1 {
1014  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
1015  %2 = bitcast i8* %1 to <2 x i32>*
1016  store <2 x i32> %val, <2 x i32>* %2, align 4
1017  ret void
1018}
1019
1020define void @__rsAllocationVStoreXImpl_short4(%struct.rs_allocation* nocapture readonly %a, <4 x i16> %val, i32 %x, i32 %y, i32 %z) #1 {
1021  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
1022  %2 = bitcast i8* %1 to <4 x i16>*
1023  store <4 x i16> %val, <4 x i16>* %2, align 2
1024  ret void
1025}
1026define void @__rsAllocationVStoreXImpl_short3(%struct.rs_allocation* nocapture readonly %a, <2 x i32> %val, i32 %x, i32 %y, i32 %z) #1 {
1027  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
1028  %2 = bitcast i8* %1 to <3 x i16>*
1029  %3 = bitcast <2 x i32> %val to <4 x i16>
1030  %4 = shufflevector <4 x i16> %3, <4 x i16> undef, <3 x i32> <i32 0, i32 1, i32 2>
1031  store <3 x i16> %4, <3 x i16>* %2, align 2
1032  ret void
1033}
1034define void @__rsAllocationVStoreXImpl_short2(%struct.rs_allocation* nocapture readonly %a, i32 %val, i32 %x, i32 %y, i32 %z) #1 {
1035  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
1036  %2 = bitcast i8* %1 to <2 x i16>*
1037  %3 = bitcast i32 %val to <2 x i16>
1038  store <2 x i16> %3, <2 x i16>* %2, align 2
1039  ret void
1040}
1041
1042define void @__rsAllocationVStoreXImpl_ushort4(%struct.rs_allocation* nocapture readonly %a, <4 x i16> %val, i32 %x, i32 %y, i32 %z) #1 {
1043  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
1044  %2 = bitcast i8* %1 to <4 x i16>*
1045  store <4 x i16> %val, <4 x i16>* %2, align 2
1046  ret void
1047}
1048define void @__rsAllocationVStoreXImpl_ushort3(%struct.rs_allocation* nocapture readonly %a, <2 x i32> %val, i32 %x, i32 %y, i32 %z) #1 {
1049  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
1050  %2 = bitcast i8* %1 to <3 x i16>*
1051  %3 = bitcast <2 x i32> %val to <4 x i16>
1052  %4 = shufflevector <4 x i16> %3, <4 x i16> undef, <3 x i32> <i32 0, i32 1, i32 2>
1053  store <3 x i16> %4, <3 x i16>* %2, align 2
1054  ret void
1055}
1056define void @__rsAllocationVStoreXImpl_ushort2(%struct.rs_allocation* nocapture readonly %a, i32 %val, i32 %x, i32 %y, i32 %z) #1 {
1057  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
1058  %2 = bitcast i8* %1 to <2 x i16>*
1059  %3 = bitcast i32 %val to <2 x i16>
1060  store <2 x i16> %3, <2 x i16>* %2, align 2
1061  ret void
1062}
1063
1064define void @__rsAllocationVStoreXImpl_char4(%struct.rs_allocation* nocapture readonly %a, i32 %val, i32 %x, i32 %y, i32 %z) #1 {
1065  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
1066  %2 = bitcast i8* %1 to <4 x i8>*
1067  %3 = bitcast i32 %val to <4 x i8>
1068  store <4 x i8> %3, <4 x i8>* %2, align 1
1069  ret void
1070}
1071define void @__rsAllocationVStoreXImpl_char3(%struct.rs_allocation* nocapture readonly %a, i32 %val, i32 %x, i32 %y, i32 %z) #1 {
1072  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
1073  %2 = bitcast i8* %1 to <3 x i8>*
1074  %3 = bitcast i32 %val to <4 x i8>
1075  %4 = shufflevector <4 x i8> %3, <4 x i8> undef, <3 x i32> <i32 0, i32 1, i32 2>
1076  store <3 x i8> %4, <3 x i8>* %2, align 1
1077  ret void
1078}
1079define void @__rsAllocationVStoreXImpl_char2(%struct.rs_allocation* nocapture readonly %a, i16 %val, i32 %x, i32 %y, i32 %z) #1 {
1080  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
1081  %2 = bitcast i8* %1 to <2 x i8>*
1082  %3 = bitcast i16 %val to <2 x i8>
1083  store <2 x i8> %3, <2 x i8>* %2, align 8
1084  ret void
1085}
1086
1087define void @__rsAllocationVStoreXImpl_uchar4(%struct.rs_allocation* nocapture readonly %a, i32 %val, i32 %x, i32 %y, i32 %z) #1 {
1088  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
1089  %2 = bitcast i8* %1 to <4 x i8>*
1090  %3 = bitcast i32 %val to <4 x i8>
1091  store <4 x i8> %3, <4 x i8>* %2, align 1
1092  ret void
1093}
1094define void @__rsAllocationVStoreXImpl_uchar3(%struct.rs_allocation* nocapture readonly %a, i32 %val, i32 %x, i32 %y, i32 %z) #1 {
1095  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
1096  %2 = bitcast i8* %1 to <3 x i8>*
1097  %3 = bitcast i32 %val to <4 x i8>
1098  %4 = shufflevector <4 x i8> %3, <4 x i8> undef, <3 x i32> <i32 0, i32 1, i32 2>
1099  store <3 x i8> %4, <3 x i8>* %2, align 1
1100  ret void
1101}
1102define void @__rsAllocationVStoreXImpl_uchar2(%struct.rs_allocation* nocapture readonly %a, i16 %val, i32 %x, i32 %y, i32 %z) #1 {
1103  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
1104  %2 = bitcast i8* %1 to <2 x i8>*
1105  %3 = bitcast i16 %val to <2 x i8>
1106  store <2 x i8> %3, <2 x i8>* %2, align 8
1107  ret void
1108}
1109
1110define void @__rsAllocationVStoreXImpl_float4(%struct.rs_allocation* nocapture readonly %a, <4 x float> %val, i32 %x, i32 %y, i32 %z) #1 {
1111  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
1112  %2 = bitcast i8* %1 to <4 x float>*
1113  store <4 x float> %val, <4 x float>* %2, align 4
1114  ret void
1115}
1116define void @__rsAllocationVStoreXImpl_float3(%struct.rs_allocation* nocapture readonly %a, <4 x i32> %val, i32 %x, i32 %y, i32 %z) #1 {
1117  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
1118  %2 = bitcast i8* %1 to <3 x float>*
1119  %3 = bitcast <4 x i32> %val to <4 x float>
1120  %4 = shufflevector <4 x float> %3, <4 x float> undef, <3 x i32> <i32 0, i32 1, i32 2>
1121  store <3 x float> %4, <3 x float>* %2, align 4
1122  ret void
1123}
1124define void @__rsAllocationVStoreXImpl_float2(%struct.rs_allocation* nocapture readonly %a, <2 x float> %val, i32 %x, i32 %y, i32 %z) #1 {
1125  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
1126  %2 = bitcast i8* %1 to <2 x float>*
1127  store <2 x float> %val, <2 x float>* %2, align 4
1128  ret void
1129}
1130
1131define void @__rsAllocationVStoreXImpl_double4(%struct.rs_allocation* nocapture readonly %a, <4 x double>* %val, i32 %x, i32 %y, i32 %z) #1 {
1132  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
1133  %2 = load <4 x double>, <4 x double>* %val
1134  %3 = bitcast i8* %1 to <4 x double>*
1135  store <4 x double> %2, <4 x double>* %3, align 8
1136  ret void
1137}
1138define void @__rsAllocationVStoreXImpl_double3(%struct.rs_allocation* nocapture readonly %a, <3 x double>* %val, i32 %x, i32 %y, i32 %z) #1 {
1139  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
1140  %2 = load <3 x double>, <3 x double>* %val
1141  %3 = bitcast i8* %1 to <3 x double>*
1142  store <3 x double> %2, <3 x double>* %3, align 8
1143  ret void
1144}
1145define void @__rsAllocationVStoreXImpl_double2(%struct.rs_allocation* nocapture readonly %a, <2 x double> %val, i32 %x, i32 %y, i32 %z) #1 {
1146  %1 = tail call i8* @rsOffsetNs(%struct.rs_allocation* %a, i32 %x, i32 %y, i32 %z) #2
1147  %2 = bitcast i8* %1 to <2 x double>*
1148  store <2 x double> %val, <2 x double>* %2, align 8
1149  ret void
1150}
1151
1152
1153attributes #0 = { nounwind readonly "less-precise-fpmad"="false" "no-frame-pointer-elim"="true" "no-frame-pointer-elim-non-leaf"="true" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "unsafe-fp-math"="false" "use-soft-float"="false" }
1154attributes #1 = { nounwind "less-precise-fpmad"="false" "no-frame-pointer-elim"="true" "no-frame-pointer-elim-non-leaf"="true" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "unsafe-fp-math"="false" "use-soft-float"="false" }
1155attributes #2 = { nobuiltin }
1156
1157