Skip to content

Commit f04650b

Browse files
arsenmarsenm
authored
LoongArch: Add test for llvm.exp10 intrinsic (#148606)
1 parent efa5063 commit f04650b

File tree

1 file changed

+362
-0
lines changed

1 file changed

+362
-0
lines changed

‎llvm/test/CodeGen/LoongArch/llvm.exp10.ll

Lines changed: 362 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,362 @@
1+
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
2+
; RUN: llc -mtriple=loongarch32 -mattr=+d < %s | FileCheck -check-prefix=LA32 %s
3+
; RUN: llc -mtriple=loongarch64 -mattr=+d < %s | FileCheck -check-prefix=LA64 %s
4+
5+
define half @exp10_f16(half %x) #0 {
6+
; LA32-LABEL: exp10_f16:
7+
; LA32: # %bb.0:
8+
; LA32-NEXT: addi.w $sp, $sp, -16
9+
; LA32-NEXT: st.w $ra, $sp, 12 # 4-byte Folded Spill
10+
; LA32-NEXT: bl __extendhfsf2
11+
; LA32-NEXT: bl exp10f
12+
; LA32-NEXT: bl __truncsfhf2
13+
; LA32-NEXT: movfr2gr.s $a0, $fa0
14+
; LA32-NEXT: lu12i.w $a1, -16
15+
; LA32-NEXT: or $a0, $a0, $a1
16+
; LA32-NEXT: movgr2fr.w $fa0, $a0
17+
; LA32-NEXT: ld.w $ra, $sp, 12 # 4-byte Folded Reload
18+
; LA32-NEXT: addi.w $sp, $sp, 16
19+
; LA32-NEXT: ret
20+
;
21+
; LA64-LABEL: exp10_f16:
22+
; LA64: # %bb.0:
23+
; LA64-NEXT: addi.d $sp, $sp, -16
24+
; LA64-NEXT: st.d $ra, $sp, 8 # 8-byte Folded Spill
25+
; LA64-NEXT: pcaddu18i $ra, %call36(__extendhfsf2)
26+
; LA64-NEXT: jirl $ra, $ra, 0
27+
; LA64-NEXT: pcaddu18i $ra, %call36(exp10f)
28+
; LA64-NEXT: jirl $ra, $ra, 0
29+
; LA64-NEXT: pcaddu18i $ra, %call36(__truncsfhf2)
30+
; LA64-NEXT: jirl $ra, $ra, 0
31+
; LA64-NEXT: movfr2gr.s $a0, $fa0
32+
; LA64-NEXT: lu12i.w $a1, -16
33+
; LA64-NEXT: or $a0, $a0, $a1
34+
; LA64-NEXT: movgr2fr.w $fa0, $a0
35+
; LA64-NEXT: ld.d $ra, $sp, 8 # 8-byte Folded Reload
36+
; LA64-NEXT: addi.d $sp, $sp, 16
37+
; LA64-NEXT: ret
38+
%r = call half @llvm.exp10.f16(half %x)
39+
ret half %r
40+
}
41+
42+
define <2 x half> @exp10_v2f16(<2 x half> %x) #0 {
43+
; LA32-LABEL: exp10_v2f16:
44+
; LA32: # %bb.0:
45+
; LA32-NEXT: addi.w $sp, $sp, -16
46+
; LA32-NEXT: st.w $ra, $sp, 12 # 4-byte Folded Spill
47+
; LA32-NEXT: st.w $fp, $sp, 8 # 4-byte Folded Spill
48+
; LA32-NEXT: fst.d $fs0, $sp, 0 # 8-byte Folded Spill
49+
; LA32-NEXT: movgr2fr.w $fs0, $a1
50+
; LA32-NEXT: movgr2fr.w $fa0, $a0
51+
; LA32-NEXT: bl __extendhfsf2
52+
; LA32-NEXT: bl exp10f
53+
; LA32-NEXT: bl __truncsfhf2
54+
; LA32-NEXT: movfr2gr.s $fp, $fa0
55+
; LA32-NEXT: fmov.s $fa0, $fs0
56+
; LA32-NEXT: bl __extendhfsf2
57+
; LA32-NEXT: bl exp10f
58+
; LA32-NEXT: bl __truncsfhf2
59+
; LA32-NEXT: movfr2gr.s $a1, $fa0
60+
; LA32-NEXT: move $a0, $fp
61+
; LA32-NEXT: fld.d $fs0, $sp, 0 # 8-byte Folded Reload
62+
; LA32-NEXT: ld.w $fp, $sp, 8 # 4-byte Folded Reload
63+
; LA32-NEXT: ld.w $ra, $sp, 12 # 4-byte Folded Reload
64+
; LA32-NEXT: addi.w $sp, $sp, 16
65+
; LA32-NEXT: ret
66+
;
67+
; LA64-LABEL: exp10_v2f16:
68+
; LA64: # %bb.0:
69+
; LA64-NEXT: addi.d $sp, $sp, -32
70+
; LA64-NEXT: st.d $ra, $sp, 24 # 8-byte Folded Spill
71+
; LA64-NEXT: st.d $fp, $sp, 16 # 8-byte Folded Spill
72+
; LA64-NEXT: move $fp, $a0
73+
; LA64-NEXT: movgr2fr.w $fa0, $a1
74+
; LA64-NEXT: pcaddu18i $ra, %call36(__extendhfsf2)
75+
; LA64-NEXT: jirl $ra, $ra, 0
76+
; LA64-NEXT: pcaddu18i $ra, %call36(exp10f)
77+
; LA64-NEXT: jirl $ra, $ra, 0
78+
; LA64-NEXT: pcaddu18i $ra, %call36(__truncsfhf2)
79+
; LA64-NEXT: jirl $ra, $ra, 0
80+
; LA64-NEXT: movfr2gr.s $a0, $fa0
81+
; LA64-NEXT: st.h $a0, $sp, 2
82+
; LA64-NEXT: movgr2fr.w $fa0, $fp
83+
; LA64-NEXT: pcaddu18i $ra, %call36(__extendhfsf2)
84+
; LA64-NEXT: jirl $ra, $ra, 0
85+
; LA64-NEXT: pcaddu18i $ra, %call36(exp10f)
86+
; LA64-NEXT: jirl $ra, $ra, 0
87+
; LA64-NEXT: pcaddu18i $ra, %call36(__truncsfhf2)
88+
; LA64-NEXT: jirl $ra, $ra, 0
89+
; LA64-NEXT: movfr2gr.s $a0, $fa0
90+
; LA64-NEXT: st.h $a0, $sp, 0
91+
; LA64-NEXT: vld $vr0, $sp, 0
92+
; LA64-NEXT: vpickve2gr.h $a0, $vr0, 0
93+
; LA64-NEXT: vpickve2gr.h $a1, $vr0, 1
94+
; LA64-NEXT: ld.d $fp, $sp, 16 # 8-byte Folded Reload
95+
; LA64-NEXT: ld.d $ra, $sp, 24 # 8-byte Folded Reload
96+
; LA64-NEXT: addi.d $sp, $sp, 32
97+
; LA64-NEXT: ret
98+
%r = call <2 x half> @llvm.exp10.v2f16(<2 x half> %x)
99+
ret <2 x half> %r
100+
}
101+
102+
define float @exp10_f32(float %x) #0 {
103+
; LA32-LABEL: exp10_f32:
104+
; LA32: # %bb.0:
105+
; LA32-NEXT: b exp10f
106+
;
107+
; LA64-LABEL: exp10_f32:
108+
; LA64: # %bb.0:
109+
; LA64-NEXT: pcaddu18i $t8, %call36(exp10f)
110+
; LA64-NEXT: jr $t8
111+
%r = call float @llvm.exp10.f32(float %x)
112+
ret float %r
113+
}
114+
115+
define <2 x float> @exp10_v2f32(<2 x float> %x) #0 {
116+
; LA32-LABEL: exp10_v2f32:
117+
; LA32: # %bb.0:
118+
; LA32-NEXT: addi.w $sp, $sp, -32
119+
; LA32-NEXT: st.w $ra, $sp, 28 # 4-byte Folded Spill
120+
; LA32-NEXT: fst.d $fs0, $sp, 16 # 8-byte Folded Spill
121+
; LA32-NEXT: fst.d $fs1, $sp, 8 # 8-byte Folded Spill
122+
; LA32-NEXT: fmov.s $fs0, $fa1
123+
; LA32-NEXT: bl exp10f
124+
; LA32-NEXT: fmov.s $fs1, $fa0
125+
; LA32-NEXT: fmov.s $fa0, $fs0
126+
; LA32-NEXT: bl exp10f
127+
; LA32-NEXT: fmov.s $fa1, $fa0
128+
; LA32-NEXT: fmov.s $fa0, $fs1
129+
; LA32-NEXT: fld.d $fs1, $sp, 8 # 8-byte Folded Reload
130+
; LA32-NEXT: fld.d $fs0, $sp, 16 # 8-byte Folded Reload
131+
; LA32-NEXT: ld.w $ra, $sp, 28 # 4-byte Folded Reload
132+
; LA32-NEXT: addi.w $sp, $sp, 32
133+
; LA32-NEXT: ret
134+
;
135+
; LA64-LABEL: exp10_v2f32:
136+
; LA64: # %bb.0:
137+
; LA64-NEXT: addi.d $sp, $sp, -48
138+
; LA64-NEXT: st.d $ra, $sp, 40 # 8-byte Folded Spill
139+
; LA64-NEXT: vst $vr0, $sp, 0 # 16-byte Folded Spill
140+
; LA64-NEXT: vreplvei.w $vr0, $vr0, 0
141+
; LA64-NEXT: # kill: def $f0 killed $f0 killed $vr0
142+
; LA64-NEXT: pcaddu18i $ra, %call36(exp10f)
143+
; LA64-NEXT: jirl $ra, $ra, 0
144+
; LA64-NEXT: # kill: def $f0 killed $f0 def $vr0
145+
; LA64-NEXT: vst $vr0, $sp, 16 # 16-byte Folded Spill
146+
; LA64-NEXT: vld $vr0, $sp, 0 # 16-byte Folded Reload
147+
; LA64-NEXT: vreplvei.w $vr0, $vr0, 1
148+
; LA64-NEXT: # kill: def $f0 killed $f0 killed $vr0
149+
; LA64-NEXT: pcaddu18i $ra, %call36(exp10f)
150+
; LA64-NEXT: jirl $ra, $ra, 0
151+
; LA64-NEXT: # kill: def $f0 killed $f0 def $vr0
152+
; LA64-NEXT: vld $vr1, $sp, 16 # 16-byte Folded Reload
153+
; LA64-NEXT: vpackev.w $vr0, $vr0, $vr1
154+
; LA64-NEXT: ld.d $ra, $sp, 40 # 8-byte Folded Reload
155+
; LA64-NEXT: addi.d $sp, $sp, 48
156+
; LA64-NEXT: ret
157+
%r = call <2 x float> @llvm.exp10.v2f32(<2 x float> %x)
158+
ret <2 x float> %r
159+
}
160+
161+
define double @exp10_f64(double %x) #0 {
162+
; LA32-LABEL: exp10_f64:
163+
; LA32: # %bb.0:
164+
; LA32-NEXT: b exp10
165+
;
166+
; LA64-LABEL: exp10_f64:
167+
; LA64: # %bb.0:
168+
; LA64-NEXT: pcaddu18i $t8, %call36(exp10)
169+
; LA64-NEXT: jr $t8
170+
%r = call double @llvm.exp10.f64(double %x)
171+
ret double %r
172+
}
173+
174+
define <2 x double> @exp10_v2f64(<2 x double> %x) #0 {
175+
; LA32-LABEL: exp10_v2f64:
176+
; LA32: # %bb.0:
177+
; LA32-NEXT: addi.w $sp, $sp, -32
178+
; LA32-NEXT: st.w $ra, $sp, 28 # 4-byte Folded Spill
179+
; LA32-NEXT: fst.d $fs0, $sp, 16 # 8-byte Folded Spill
180+
; LA32-NEXT: fst.d $fs1, $sp, 8 # 8-byte Folded Spill
181+
; LA32-NEXT: fmov.d $fs0, $fa1
182+
; LA32-NEXT: bl exp10
183+
; LA32-NEXT: fmov.d $fs1, $fa0
184+
; LA32-NEXT: fmov.d $fa0, $fs0
185+
; LA32-NEXT: bl exp10
186+
; LA32-NEXT: fmov.d $fa1, $fa0
187+
; LA32-NEXT: fmov.d $fa0, $fs1
188+
; LA32-NEXT: fld.d $fs1, $sp, 8 # 8-byte Folded Reload
189+
; LA32-NEXT: fld.d $fs0, $sp, 16 # 8-byte Folded Reload
190+
; LA32-NEXT: ld.w $ra, $sp, 28 # 4-byte Folded Reload
191+
; LA32-NEXT: addi.w $sp, $sp, 32
192+
; LA32-NEXT: ret
193+
;
194+
; LA64-LABEL: exp10_v2f64:
195+
; LA64: # %bb.0:
196+
; LA64-NEXT: addi.d $sp, $sp, -48
197+
; LA64-NEXT: st.d $ra, $sp, 40 # 8-byte Folded Spill
198+
; LA64-NEXT: vst $vr0, $sp, 0 # 16-byte Folded Spill
199+
; LA64-NEXT: vreplvei.d $vr0, $vr0, 0
200+
; LA64-NEXT: # kill: def $f0_64 killed $f0_64 killed $vr0
201+
; LA64-NEXT: pcaddu18i $ra, %call36(exp10)
202+
; LA64-NEXT: jirl $ra, $ra, 0
203+
; LA64-NEXT: movfr2gr.d $a0, $fa0
204+
; LA64-NEXT: vinsgr2vr.d $vr0, $a0, 0
205+
; LA64-NEXT: vst $vr0, $sp, 16 # 16-byte Folded Spill
206+
; LA64-NEXT: vld $vr0, $sp, 0 # 16-byte Folded Reload
207+
; LA64-NEXT: vreplvei.d $vr0, $vr0, 1
208+
; LA64-NEXT: # kill: def $f0_64 killed $f0_64 killed $vr0
209+
; LA64-NEXT: pcaddu18i $ra, %call36(exp10)
210+
; LA64-NEXT: jirl $ra, $ra, 0
211+
; LA64-NEXT: movfr2gr.d $a0, $fa0
212+
; LA64-NEXT: vld $vr0, $sp, 16 # 16-byte Folded Reload
213+
; LA64-NEXT: vinsgr2vr.d $vr0, $a0, 1
214+
; LA64-NEXT: ld.d $ra, $sp, 40 # 8-byte Folded Reload
215+
; LA64-NEXT: addi.d $sp, $sp, 48
216+
; LA64-NEXT: ret
217+
%r = call <2 x double> @llvm.exp10.v2f64(<2 x double> %x)
218+
ret <2 x double> %r
219+
}
220+
221+
define fp128 @exp10_f128(fp128 %x) #0 {
222+
; LA32-LABEL: exp10_f128:
223+
; LA32: # %bb.0:
224+
; LA32-NEXT: addi.w $sp, $sp, -48
225+
; LA32-NEXT: st.w $ra, $sp, 44 # 4-byte Folded Spill
226+
; LA32-NEXT: st.w $fp, $sp, 40 # 4-byte Folded Spill
227+
; LA32-NEXT: ld.w $a2, $a1, 0
228+
; LA32-NEXT: ld.w $a3, $a1, 4
229+
; LA32-NEXT: ld.w $a4, $a1, 8
230+
; LA32-NEXT: ld.w $a1, $a1, 12
231+
; LA32-NEXT: move $fp, $a0
232+
; LA32-NEXT: st.w $a1, $sp, 20
233+
; LA32-NEXT: st.w $a4, $sp, 16
234+
; LA32-NEXT: st.w $a3, $sp, 12
235+
; LA32-NEXT: addi.w $a0, $sp, 24
236+
; LA32-NEXT: addi.w $a1, $sp, 8
237+
; LA32-NEXT: st.w $a2, $sp, 8
238+
; LA32-NEXT: bl exp10l
239+
; LA32-NEXT: ld.w $a0, $sp, 36
240+
; LA32-NEXT: ld.w $a1, $sp, 32
241+
; LA32-NEXT: ld.w $a2, $sp, 28
242+
; LA32-NEXT: ld.w $a3, $sp, 24
243+
; LA32-NEXT: st.w $a0, $fp, 12
244+
; LA32-NEXT: st.w $a1, $fp, 8
245+
; LA32-NEXT: st.w $a2, $fp, 4
246+
; LA32-NEXT: st.w $a3, $fp, 0
247+
; LA32-NEXT: ld.w $fp, $sp, 40 # 4-byte Folded Reload
248+
; LA32-NEXT: ld.w $ra, $sp, 44 # 4-byte Folded Reload
249+
; LA32-NEXT: addi.w $sp, $sp, 48
250+
; LA32-NEXT: ret
251+
;
252+
; LA64-LABEL: exp10_f128:
253+
; LA64: # %bb.0:
254+
; LA64-NEXT: addi.d $sp, $sp, -16
255+
; LA64-NEXT: st.d $ra, $sp, 8 # 8-byte Folded Spill
256+
; LA64-NEXT: pcaddu18i $ra, %call36(exp10l)
257+
; LA64-NEXT: jirl $ra, $ra, 0
258+
; LA64-NEXT: ld.d $ra, $sp, 8 # 8-byte Folded Reload
259+
; LA64-NEXT: addi.d $sp, $sp, 16
260+
; LA64-NEXT: ret
261+
%r = call fp128 @llvm.exp10.f128(fp128 %x)
262+
ret fp128 %r
263+
}
264+
265+
define <2 x fp128> @exp10_v2f128(<2 x fp128> %x) #0 {
266+
; LA32-LABEL: exp10_v2f128:
267+
; LA32: # %bb.0:
268+
; LA32-NEXT: addi.w $sp, $sp, -96
269+
; LA32-NEXT: st.w $ra, $sp, 92 # 4-byte Folded Spill
270+
; LA32-NEXT: st.w $fp, $sp, 88 # 4-byte Folded Spill
271+
; LA32-NEXT: st.w $s0, $sp, 84 # 4-byte Folded Spill
272+
; LA32-NEXT: st.w $s1, $sp, 80 # 4-byte Folded Spill
273+
; LA32-NEXT: st.w $s2, $sp, 76 # 4-byte Folded Spill
274+
; LA32-NEXT: st.w $s3, $sp, 72 # 4-byte Folded Spill
275+
; LA32-NEXT: ld.w $s0, $a1, 16
276+
; LA32-NEXT: ld.w $s1, $a1, 20
277+
; LA32-NEXT: ld.w $s2, $a1, 24
278+
; LA32-NEXT: ld.w $s3, $a1, 28
279+
; LA32-NEXT: ld.w $a2, $a1, 0
280+
; LA32-NEXT: ld.w $a3, $a1, 4
281+
; LA32-NEXT: ld.w $a4, $a1, 8
282+
; LA32-NEXT: ld.w $a1, $a1, 12
283+
; LA32-NEXT: move $fp, $a0
284+
; LA32-NEXT: st.w $a1, $sp, 20
285+
; LA32-NEXT: st.w $a4, $sp, 16
286+
; LA32-NEXT: st.w $a3, $sp, 12
287+
; LA32-NEXT: addi.w $a0, $sp, 24
288+
; LA32-NEXT: addi.w $a1, $sp, 8
289+
; LA32-NEXT: st.w $a2, $sp, 8
290+
; LA32-NEXT: bl exp10l
291+
; LA32-NEXT: st.w $s3, $sp, 52
292+
; LA32-NEXT: st.w $s2, $sp, 48
293+
; LA32-NEXT: st.w $s1, $sp, 44
294+
; LA32-NEXT: addi.w $a0, $sp, 56
295+
; LA32-NEXT: addi.w $a1, $sp, 40
296+
; LA32-NEXT: st.w $s0, $sp, 40
297+
; LA32-NEXT: bl exp10l
298+
; LA32-NEXT: ld.w $a0, $sp, 24
299+
; LA32-NEXT: ld.w $a1, $sp, 28
300+
; LA32-NEXT: ld.w $a2, $sp, 32
301+
; LA32-NEXT: ld.w $a3, $sp, 36
302+
; LA32-NEXT: ld.w $a4, $sp, 68
303+
; LA32-NEXT: ld.w $a5, $sp, 64
304+
; LA32-NEXT: ld.w $a6, $sp, 60
305+
; LA32-NEXT: ld.w $a7, $sp, 56
306+
; LA32-NEXT: st.w $a4, $fp, 28
307+
; LA32-NEXT: st.w $a5, $fp, 24
308+
; LA32-NEXT: st.w $a6, $fp, 20
309+
; LA32-NEXT: st.w $a7, $fp, 16
310+
; LA32-NEXT: st.w $a3, $fp, 12
311+
; LA32-NEXT: st.w $a2, $fp, 8
312+
; LA32-NEXT: st.w $a1, $fp, 4
313+
; LA32-NEXT: st.w $a0, $fp, 0
314+
; LA32-NEXT: ld.w $s3, $sp, 72 # 4-byte Folded Reload
315+
; LA32-NEXT: ld.w $s2, $sp, 76 # 4-byte Folded Reload
316+
; LA32-NEXT: ld.w $s1, $sp, 80 # 4-byte Folded Reload
317+
; LA32-NEXT: ld.w $s0, $sp, 84 # 4-byte Folded Reload
318+
; LA32-NEXT: ld.w $fp, $sp, 88 # 4-byte Folded Reload
319+
; LA32-NEXT: ld.w $ra, $sp, 92 # 4-byte Folded Reload
320+
; LA32-NEXT: addi.w $sp, $sp, 96
321+
; LA32-NEXT: ret
322+
;
323+
; LA64-LABEL: exp10_v2f128:
324+
; LA64: # %bb.0:
325+
; LA64-NEXT: addi.d $sp, $sp, -48
326+
; LA64-NEXT: st.d $ra, $sp, 40 # 8-byte Folded Spill
327+
; LA64-NEXT: st.d $fp, $sp, 32 # 8-byte Folded Spill
328+
; LA64-NEXT: st.d $s0, $sp, 24 # 8-byte Folded Spill
329+
; LA64-NEXT: st.d $s1, $sp, 16 # 8-byte Folded Spill
330+
; LA64-NEXT: st.d $s2, $sp, 8 # 8-byte Folded Spill
331+
; LA64-NEXT: st.d $s3, $sp, 0 # 8-byte Folded Spill
332+
; LA64-NEXT: ld.d $fp, $a1, 16
333+
; LA64-NEXT: ld.d $s0, $a1, 24
334+
; LA64-NEXT: ld.d $a2, $a1, 0
335+
; LA64-NEXT: ld.d $a1, $a1, 8
336+
; LA64-NEXT: move $s1, $a0
337+
; LA64-NEXT: move $a0, $a2
338+
; LA64-NEXT: pcaddu18i $ra, %call36(exp10l)
339+
; LA64-NEXT: jirl $ra, $ra, 0
340+
; LA64-NEXT: move $s2, $a0
341+
; LA64-NEXT: move $s3, $a1
342+
; LA64-NEXT: move $a0, $fp
343+
; LA64-NEXT: move $a1, $s0
344+
; LA64-NEXT: pcaddu18i $ra, %call36(exp10l)
345+
; LA64-NEXT: jirl $ra, $ra, 0
346+
; LA64-NEXT: st.d $a1, $s1, 24
347+
; LA64-NEXT: st.d $a0, $s1, 16
348+
; LA64-NEXT: st.d $s3, $s1, 8
349+
; LA64-NEXT: st.d $s2, $s1, 0
350+
; LA64-NEXT: ld.d $s3, $sp, 0 # 8-byte Folded Reload
351+
; LA64-NEXT: ld.d $s2, $sp, 8 # 8-byte Folded Reload
352+
; LA64-NEXT: ld.d $s1, $sp, 16 # 8-byte Folded Reload
353+
; LA64-NEXT: ld.d $s0, $sp, 24 # 8-byte Folded Reload
354+
; LA64-NEXT: ld.d $fp, $sp, 32 # 8-byte Folded Reload
355+
; LA64-NEXT: ld.d $ra, $sp, 40 # 8-byte Folded Reload
356+
; LA64-NEXT: addi.d $sp, $sp, 48
357+
; LA64-NEXT: ret
358+
%r = call <2 x fp128> @llvm.exp10.v2f128(<2 x fp128> %x)
359+
ret <2 x fp128> %r
360+
}
361+
362+
attributes #0 = { nounwind }

0 commit comments

Comments
 (0)