Lines Matching refs:rdx
6 define fastcc float @reduction_cost_float(<4 x float> %rdx) {
7 …%rdx.shuf = shufflevector <4 x float> %rdx, <4 x float> undef, <4 x i32> <i32 2, i32 3, i32 undef,…
8 %bin.rdx = fadd <4 x float> %rdx, %rdx.shuf
9 …%rdx.shuf7 = shufflevector <4 x float> %bin.rdx, <4 x float> undef, <4 x i32> <i32 1, i32 undef, i…
10 %bin.rdx8 = fadd <4 x float> %bin.rdx, %rdx.shuf7
21 define fastcc i32 @reduction_cost_int(<8 x i32> %rdx) {
22 %rdx.shuf = shufflevector <8 x i32> %rdx, <8 x i32> undef,
25 %bin.rdx = add <8 x i32> %rdx, %rdx.shuf
26 %rdx.shuf.2 = shufflevector <8 x i32> %bin.rdx, <8 x i32> undef,
29 %bin.rdx.2 = add <8 x i32> %bin.rdx, %rdx.shuf.2
30 %rdx.shuf.3 = shufflevector <8 x i32> %bin.rdx.2, <8 x i32> undef,
33 %bin.rdx.3 = add <8 x i32> %bin.rdx.2, %rdx.shuf.3
38 %r = extractelement <8 x i32> %bin.rdx.3, i32 0
42 define fastcc float @pairwise_hadd(<4 x float> %rdx, float %f1) {
43 %rdx.shuf.0.0 = shufflevector <4 x float> %rdx, <4 x float> undef,
45 %rdx.shuf.0.1 = shufflevector <4 x float> %rdx, <4 x float> undef,
47 %bin.rdx.0 = fadd <4 x float> %rdx.shuf.0.0, %rdx.shuf.0.1
48 %rdx.shuf.1.0 = shufflevector <4 x float> %bin.rdx.0, <4 x float> undef,
50 %rdx.shuf.1.1 = shufflevector <4 x float> %bin.rdx.0, <4 x float> undef,
52 %bin.rdx.1 = fadd <4 x float> %rdx.shuf.1.0, %rdx.shuf.1.1
57 %r = extractelement <4 x float> %bin.rdx.1, i32 0
62 define fastcc float @pairwise_hadd_assoc(<4 x float> %rdx, float %f1) {
63 %rdx.shuf.0.0 = shufflevector <4 x float> %rdx, <4 x float> undef,
65 %rdx.shuf.0.1 = shufflevector <4 x float> %rdx, <4 x float> undef,
67 %bin.rdx.0 = fadd <4 x float> %rdx.shuf.0.1, %rdx.shuf.0.0
68 %rdx.shuf.1.0 = shufflevector <4 x float> %bin.rdx.0, <4 x float> undef,
70 %rdx.shuf.1.1 = shufflevector <4 x float> %bin.rdx.0, <4 x float> undef,
72 %bin.rdx.1 = fadd <4 x float> %rdx.shuf.1.0, %rdx.shuf.1.1
77 %r = extractelement <4 x float> %bin.rdx.1, i32 0
82 define fastcc float @pairwise_hadd_skip_first(<4 x float> %rdx, float %f1) {
83 %rdx.shuf.0.0 = shufflevector <4 x float> %rdx, <4 x float> undef,
85 %rdx.shuf.0.1 = shufflevector <4 x float> %rdx, <4 x float> undef,
87 %bin.rdx.0 = fadd <4 x float> %rdx.shuf.0.0, %rdx.shuf.0.1
88 %rdx.shuf.1.1 = shufflevector <4 x float> %bin.rdx.0, <4 x float> undef,
90 %bin.rdx.1 = fadd <4 x float> %bin.rdx.0, %rdx.shuf.1.1
95 %r = extractelement <4 x float> %bin.rdx.1, i32 0
100 define fastcc double @no_pairwise_reduction2double(<2 x double> %rdx, double %f1) {
101 %rdx.shuf = shufflevector <2 x double> %rdx, <2 x double> undef, <2 x i32> <i32 1, i32 undef>
102 %bin.rdx = fadd <2 x double> %rdx, %rdx.shuf
108 %r = extractelement <2 x double> %bin.rdx, i32 0
112 define fastcc float @no_pairwise_reduction4float(<4 x float> %rdx, float %f1) {
113 …%rdx.shuf = shufflevector <4 x float> %rdx, <4 x float> undef, <4 x i32> <i32 2, i32 3, i32 undef,…
114 %bin.rdx = fadd <4 x float> %rdx, %rdx.shuf
115 …%rdx.shuf7 = shufflevector <4 x float> %bin.rdx, <4 x float> undef, <4 x i32> <i32 1, i32 undef, i…
116 %bin.rdx8 = fadd <4 x float> %bin.rdx, %rdx.shuf7
126 define fastcc double @no_pairwise_reduction4double(<4 x double> %rdx, double %f1) {
127 …%rdx.shuf = shufflevector <4 x double> %rdx, <4 x double> undef, <4 x i32> <i32 2, i32 3, i32 unde…
128 %bin.rdx = fadd <4 x double> %rdx, %rdx.shuf
129 …%rdx.shuf7 = shufflevector <4 x double> %bin.rdx, <4 x double> undef, <4 x i32> <i32 1, i32 undef,…
130 %bin.rdx8 = fadd <4 x double> %bin.rdx, %rdx.shuf7
139 define fastcc float @no_pairwise_reduction8float(<8 x float> %rdx, float %f1) {
140 …%rdx.shuf3 = shufflevector <8 x float> %rdx, <8 x float> undef, <8 x i32> <i32 4, i32 5, i32 6, i3…
141 %bin.rdx4 = fadd <8 x float> %rdx, %rdx.shuf3
142 …%rdx.shuf = shufflevector <8 x float> %bin.rdx4, <8 x float> undef, <8 x i32> <i32 2, i32 3, i32 u…
143 %bin.rdx = fadd <8 x float> %bin.rdx4, %rdx.shuf
144 …%rdx.shuf7 = shufflevector <8 x float> %bin.rdx, <8 x float> undef, <8 x i32> <i32 1, i32 undef, i…
145 %bin.rdx8 = fadd <8 x float> %bin.rdx, %rdx.shuf7
154 define fastcc i64 @no_pairwise_reduction2i64(<2 x i64> %rdx, i64 %f1) {
155 %rdx.shuf = shufflevector <2 x i64> %rdx, <2 x i64> undef, <2 x i32> <i32 1, i32 undef>
156 %bin.rdx = add <2 x i64> %rdx, %rdx.shuf
162 %r = extractelement <2 x i64> %bin.rdx, i32 0
166 define fastcc i32 @no_pairwise_reduction4i32(<4 x i32> %rdx, i32 %f1) {
167 …%rdx.shuf = shufflevector <4 x i32> %rdx, <4 x i32> undef, <4 x i32> <i32 2, i32 3, i32 undef, i32…
168 %bin.rdx = add <4 x i32> %rdx, %rdx.shuf
169 …%rdx.shuf7 = shufflevector <4 x i32> %bin.rdx, <4 x i32> undef, <4 x i32> <i32 1, i32 undef, i32 u…
170 %bin.rdx8 = add <4 x i32> %bin.rdx, %rdx.shuf7
180 define fastcc i64 @no_pairwise_reduction4i64(<4 x i64> %rdx, i64 %f1) {
181 …%rdx.shuf = shufflevector <4 x i64> %rdx, <4 x i64> undef, <4 x i32> <i32 2, i32 3, i32 undef, i32…
182 %bin.rdx = add <4 x i64> %rdx, %rdx.shuf
183 …%rdx.shuf7 = shufflevector <4 x i64> %bin.rdx, <4 x i64> undef, <4 x i32> <i32 1, i32 undef, i32 u…
184 %bin.rdx8 = add <4 x i64> %bin.rdx, %rdx.shuf7
193 define fastcc i16 @no_pairwise_reduction8i16(<8 x i16> %rdx, i16 %f1) {
194 …%rdx.shuf3 = shufflevector <8 x i16> %rdx, <8 x i16> undef, <8 x i32> <i32 4, i32 5, i32 6, i32 7,…
195 %bin.rdx4 = add <8 x i16> %rdx, %rdx.shuf3
196 …%rdx.shuf = shufflevector <8 x i16> %bin.rdx4, <8 x i16> undef, <8 x i32> <i32 2, i32 3, i32 undef…
197 %bin.rdx = add <8 x i16> %bin.rdx4, %rdx.shuf
198 …%rdx.shuf7 = shufflevector <8 x i16> %bin.rdx, <8 x i16> undef, <8 x i32> <i32 1, i32 undef, i32 u…
199 %bin.rdx8 = add <8 x i16> %bin.rdx, %rdx.shuf7
209 define fastcc i32 @no_pairwise_reduction8i32(<8 x i32> %rdx, i32 %f1) {
210 …%rdx.shuf3 = shufflevector <8 x i32> %rdx, <8 x i32> undef, <8 x i32> <i32 4, i32 5, i32 6, i32 7,…
211 %bin.rdx4 = add <8 x i32> %rdx, %rdx.shuf3
212 …%rdx.shuf = shufflevector <8 x i32> %bin.rdx4, <8 x i32> undef, <8 x i32> <i32 2, i32 3, i32 undef…
213 %bin.rdx = add <8 x i32> %bin.rdx4, %rdx.shuf
214 …%rdx.shuf7 = shufflevector <8 x i32> %bin.rdx, <8 x i32> undef, <8 x i32> <i32 1, i32 undef, i32 u…
215 %bin.rdx8 = add <8 x i32> %bin.rdx, %rdx.shuf7
224 define fastcc double @pairwise_reduction2double(<2 x double> %rdx, double %f1) {
225 %rdx.shuf.1.0 = shufflevector <2 x double> %rdx, <2 x double> undef, <2 x i32> <i32 0, i32 undef>
226 %rdx.shuf.1.1 = shufflevector <2 x double> %rdx, <2 x double> undef, <2 x i32> <i32 1, i32 undef>
227 %bin.rdx8 = fadd <2 x double> %rdx.shuf.1.0, %rdx.shuf.1.1
237 define fastcc float @pairwise_reduction4float(<4 x float> %rdx, float %f1) {
238 …%rdx.shuf.0.0 = shufflevector <4 x float> %rdx, <4 x float> undef, <4 x i32> <i32 0, i32 2, i32 un…
239 …%rdx.shuf.0.1 = shufflevector <4 x float> %rdx, <4 x float> undef, <4 x i32> <i32 1, i32 3, i32 un…
240 %bin.rdx = fadd <4 x float> %rdx.shuf.0.0, %rdx.shuf.0.1
241 …%rdx.shuf.1.0 = shufflevector <4 x float> %bin.rdx, <4 x float> undef, <4 x i32> <i32 0, i32 undef…
242 …%rdx.shuf.1.1 = shufflevector <4 x float> %bin.rdx, <4 x float> undef, <4 x i32> <i32 1, i32 undef…
243 %bin.rdx8 = fadd <4 x float> %rdx.shuf.1.0, %rdx.shuf.1.1
253 define fastcc double @pairwise_reduction4double(<4 x double> %rdx, double %f1) {
254 …%rdx.shuf.0.0 = shufflevector <4 x double> %rdx, <4 x double> undef, <4 x i32> <i32 0, i32 2, i32 …
255 …%rdx.shuf.0.1 = shufflevector <4 x double> %rdx, <4 x double> undef, <4 x i32> <i32 1, i32 3, i32 …
256 %bin.rdx = fadd <4 x double> %rdx.shuf.0.0, %rdx.shuf.0.1
257 …%rdx.shuf.1.0 = shufflevector <4 x double> %bin.rdx, <4 x double> undef, <4 x i32> <i32 0, i32 und…
258 …%rdx.shuf.1.1 = shufflevector <4 x double> %bin.rdx, <4 x double> undef, <4 x i32> <i32 1, i32 und…
259 %bin.rdx8 = fadd <4 x double> %rdx.shuf.1.0, %rdx.shuf.1.1
268 define fastcc float @pairwise_reduction8float(<8 x float> %rdx, float %f1) {
269 …%rdx.shuf.0.0 = shufflevector <8 x float> %rdx, <8 x float> undef, <8 x i32> <i32 0, i32 2, i32 4,…
270 …%rdx.shuf.0.1 = shufflevector <8 x float> %rdx, <8 x float> undef, <8 x i32> <i32 1, i32 3, i32 5,…
271 %bin.rdx = fadd <8 x float> %rdx.shuf.0.0, %rdx.shuf.0.1
272 …%rdx.shuf.1.0 = shufflevector <8 x float> %bin.rdx, <8 x float> undef,<8 x i32> <i32 0, i32 2, i32…
273 …%rdx.shuf.1.1 = shufflevector <8 x float> %bin.rdx, <8 x float> undef,<8 x i32> <i32 1, i32 3, i32…
274 %bin.rdx8 = fadd <8 x float> %rdx.shuf.1.0, %rdx.shuf.1.1
275 …%rdx.shuf.2.0 = shufflevector <8 x float> %bin.rdx8, <8 x float> undef,<8 x i32> <i32 0, i32 undef…
276 …%rdx.shuf.2.1 = shufflevector <8 x float> %bin.rdx8, <8 x float> undef,<8 x i32> <i32 1, i32 undef…
277 %bin.rdx9 = fadd <8 x float> %rdx.shuf.2.0, %rdx.shuf.2.1
286 define fastcc i64 @pairwise_reduction2i64(<2 x i64> %rdx, i64 %f1) {
287 %rdx.shuf.1.0 = shufflevector <2 x i64> %rdx, <2 x i64> undef, <2 x i32> <i32 0, i32 undef>
288 %rdx.shuf.1.1 = shufflevector <2 x i64> %rdx, <2 x i64> undef, <2 x i32> <i32 1, i32 undef>
289 %bin.rdx8 = add <2 x i64> %rdx.shuf.1.0, %rdx.shuf.1.1
299 define fastcc i32 @pairwise_reduction4i32(<4 x i32> %rdx, i32 %f1) {
300 …%rdx.shuf.0.0 = shufflevector <4 x i32> %rdx, <4 x i32> undef, <4 x i32> <i32 0, i32 2, i32 undef,…
301 …%rdx.shuf.0.1 = shufflevector <4 x i32> %rdx, <4 x i32> undef, <4 x i32> <i32 1, i32 3, i32 undef,…
302 %bin.rdx = add <4 x i32> %rdx.shuf.0.0, %rdx.shuf.0.1
303 …%rdx.shuf.1.0 = shufflevector <4 x i32> %bin.rdx, <4 x i32> undef, <4 x i32> <i32 0, i32 undef, i3…
304 …%rdx.shuf.1.1 = shufflevector <4 x i32> %bin.rdx, <4 x i32> undef, <4 x i32> <i32 1, i32 undef, i3…
305 %bin.rdx8 = add <4 x i32> %rdx.shuf.1.0, %rdx.shuf.1.1
315 define fastcc i64 @pairwise_reduction4i64(<4 x i64> %rdx, i64 %f1) {
316 …%rdx.shuf.0.0 = shufflevector <4 x i64> %rdx, <4 x i64> undef, <4 x i32> <i32 0, i32 2, i32 undef,…
317 …%rdx.shuf.0.1 = shufflevector <4 x i64> %rdx, <4 x i64> undef, <4 x i32> <i32 1, i32 3, i32 undef,…
318 %bin.rdx = add <4 x i64> %rdx.shuf.0.0, %rdx.shuf.0.1
319 …%rdx.shuf.1.0 = shufflevector <4 x i64> %bin.rdx, <4 x i64> undef, <4 x i32> <i32 0, i32 undef, i3…
320 …%rdx.shuf.1.1 = shufflevector <4 x i64> %bin.rdx, <4 x i64> undef, <4 x i32> <i32 1, i32 undef, i3…
321 %bin.rdx8 = add <4 x i64> %rdx.shuf.1.0, %rdx.shuf.1.1
330 define fastcc i16 @pairwise_reduction8i16(<8 x i16> %rdx, i16 %f1) {
331 …%rdx.shuf.0.0 = shufflevector <8 x i16> %rdx, <8 x i16> undef, <8 x i32> <i32 0, i32 2, i32 4, i32…
332 …%rdx.shuf.0.1 = shufflevector <8 x i16> %rdx, <8 x i16> undef, <8 x i32> <i32 1, i32 3, i32 5, i32…
333 %bin.rdx = add <8 x i16> %rdx.shuf.0.0, %rdx.shuf.0.1
334 …%rdx.shuf.1.0 = shufflevector <8 x i16> %bin.rdx, <8 x i16> undef,<8 x i32> <i32 0, i32 2, i32 und…
335 …%rdx.shuf.1.1 = shufflevector <8 x i16> %bin.rdx, <8 x i16> undef,<8 x i32> <i32 1, i32 3, i32 und…
336 %bin.rdx8 = add <8 x i16> %rdx.shuf.1.0, %rdx.shuf.1.1
337 …%rdx.shuf.2.0 = shufflevector <8 x i16> %bin.rdx8, <8 x i16> undef,<8 x i32> <i32 0, i32 undef, i3…
338 …%rdx.shuf.2.1 = shufflevector <8 x i16> %bin.rdx8, <8 x i16> undef,<8 x i32> <i32 1, i32 undef, i3…
339 %bin.rdx9 = add <8 x i16> %rdx.shuf.2.0, %rdx.shuf.2.1
349 define fastcc i32 @pairwise_reduction8i32(<8 x i32> %rdx, i32 %f1) {
350 …%rdx.shuf.0.0 = shufflevector <8 x i32> %rdx, <8 x i32> undef, <8 x i32> <i32 0, i32 2, i32 4, i32…
351 …%rdx.shuf.0.1 = shufflevector <8 x i32> %rdx, <8 x i32> undef, <8 x i32> <i32 1, i32 3, i32 5, i32…
352 %bin.rdx = add <8 x i32> %rdx.shuf.0.0, %rdx.shuf.0.1
353 …%rdx.shuf.1.0 = shufflevector <8 x i32> %bin.rdx, <8 x i32> undef,<8 x i32> <i32 0, i32 2, i32 und…
354 …%rdx.shuf.1.1 = shufflevector <8 x i32> %bin.rdx, <8 x i32> undef,<8 x i32> <i32 1, i32 3, i32 und…
355 %bin.rdx8 = add <8 x i32> %rdx.shuf.1.0, %rdx.shuf.1.1
356 …%rdx.shuf.2.0 = shufflevector <8 x i32> %bin.rdx8, <8 x i32> undef,<8 x i32> <i32 0, i32 undef, i3…
357 …%rdx.shuf.2.1 = shufflevector <8 x i32> %bin.rdx8, <8 x i32> undef,<8 x i32> <i32 1, i32 undef, i3…
358 %bin.rdx9 = add <8 x i32> %rdx.shuf.2.0, %rdx.shuf.2.1