Repository navigation
Expand file tree
/
Copy pathvector_fma.S
More file actions
354 lines (345 loc) · 15.5 KB
/
Copy pathvector_fma.S
File metadata and controls
354 lines (345 loc) · 15.5 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
# Single-width vector fused multiply-add results and accrued flags.
.noshrink
# Enable FS and VS and select RNE.
.4byte 0x000062b7 # lui t0, 0x6
.4byte 0x60028293 # addi t0, t0, 0x600
.4byte 0x3002a073 # csrs mstatus, t0
.4byte 0x00305073 # csrwi fcsr, 0
.4byte 0xc1017657 # vsetivli a2, 2, e32, m1, tu, mu
.assert rd_addr == 12 "VL writes x12"
.assert rd_wdata == 0x2 "e32 VL is 2"
# Binary32 (1 + 2^-23) * (1 - 2^-23) - 1 has exact residual -2^-46.
.4byte 0x3f8002b7 # lui t0, 0x3f800
.4byte 0x00128293 # addi t0, t0, 1
.4byte 0x5e02c457 # vmv.v.x v8, t0
.4byte 0x3f800337 # lui t1, 0x3f800
.4byte 0xffe30313 # addi t1, t1, -2
.4byte 0x5e0344d7 # vmv.v.x v9, t1
.4byte 0xbf8003b7 # lui t2, 0xbf800
.4byte 0x5e03cb57 # vmv.v.x v22, t2
.4byte 0xb2941b57 # vfmacc.vv v22, v8, v9
.assert trap == 0 "binary32 fused accumulate retires"
.4byte 0x43602557 # vmv.x.s a0, v22
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0xffffffffa8800000 "binary32 fused residual is -2^-46"
.4byte 0x001025f3 # csrr a1, fflags
.assert rd_addr == 11 "fflags writes x11"
.assert rd_wdata == 0x0 "binary32 fused residual raises no flags"
# Binary32 destructive forms with vs1 or fs1 = 2, vs2 = 3, old vd = 4.
.4byte 0x400002b7 # lui t0, 0x40000
.4byte 0x40400337 # lui t1, 0x40400
.4byte 0x408003b7 # lui t2, 0x40800
.4byte 0xf0028053 # fmv.w.x ft0, t0
.4byte 0x5e02c457 # vmv.v.x v8, t0
.4byte 0x5e0344d7 # vmv.v.x v9, t1
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xb2941857 # vfmacc.vv v16, v8, v9
.assert trap == 0 "binary32 vfmacc.vv retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0x41200000 "binary32 vfmacc.vv yields 10"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xb6941857 # vfnmacc.vv v16, v8, v9
.assert trap == 0 "binary32 vfnmacc.vv retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0xffffffffc1200000 "binary32 vfnmacc.vv yields -10"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xba941857 # vfmsac.vv v16, v8, v9
.assert trap == 0 "binary32 vfmsac.vv retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0x40000000 "binary32 vfmsac.vv yields 2"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xbe941857 # vfnmsac.vv v16, v8, v9
.assert trap == 0 "binary32 vfnmsac.vv retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0xffffffffc0000000 "binary32 vfnmsac.vv yields -2"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xa2941857 # vfmadd.vv v16, v8, v9
.assert trap == 0 "binary32 vfmadd.vv retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0x41300000 "binary32 vfmadd.vv yields 11"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xa6941857 # vfnmadd.vv v16, v8, v9
.assert trap == 0 "binary32 vfnmadd.vv retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0xffffffffc1300000 "binary32 vfnmadd.vv yields -11"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xaa941857 # vfmsub.vv v16, v8, v9
.assert trap == 0 "binary32 vfmsub.vv retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0x40a00000 "binary32 vfmsub.vv yields 5"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xae941857 # vfnmsub.vv v16, v8, v9
.assert trap == 0 "binary32 vfnmsub.vv retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0xffffffffc0a00000 "binary32 vfnmsub.vv yields -5"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xb2905857 # vfmacc.vf v16, ft0, v9
.assert trap == 0 "binary32 vfmacc.vf retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0x41200000 "binary32 vfmacc.vf yields 10"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xb6905857 # vfnmacc.vf v16, ft0, v9
.assert trap == 0 "binary32 vfnmacc.vf retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0xffffffffc1200000 "binary32 vfnmacc.vf yields -10"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xba905857 # vfmsac.vf v16, ft0, v9
.assert trap == 0 "binary32 vfmsac.vf retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0x40000000 "binary32 vfmsac.vf yields 2"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xbe905857 # vfnmsac.vf v16, ft0, v9
.assert trap == 0 "binary32 vfnmsac.vf retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0xffffffffc0000000 "binary32 vfnmsac.vf yields -2"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xa2905857 # vfmadd.vf v16, ft0, v9
.assert trap == 0 "binary32 vfmadd.vf retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0x41300000 "binary32 vfmadd.vf yields 11"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xa6905857 # vfnmadd.vf v16, ft0, v9
.assert trap == 0 "binary32 vfnmadd.vf retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0xffffffffc1300000 "binary32 vfnmadd.vf yields -11"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xaa905857 # vfmsub.vf v16, ft0, v9
.assert trap == 0 "binary32 vfmsub.vf retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0x40a00000 "binary32 vfmsub.vf yields 5"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xae905857 # vfnmsub.vf v16, ft0, v9
.assert trap == 0 "binary32 vfnmsub.vf retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0xffffffffc0a00000 "binary32 vfnmsub.vf yields -5"
.4byte 0x001025f3 # csrr a1, fflags
.assert rd_addr == 11 "fflags writes x11"
.assert rd_wdata == 0x0 "exact binary32 forms raise no flags"
# Binary64 (1 + 2^-52) * (1 - 2^-52) - 1 has exact residual -2^-104.
.4byte 0xc1817657 # vsetivli a2, 2, e64, m1, tu, mu
.assert rd_addr == 12 "VL writes x12"
.assert rd_wdata == 0x2 "e64 VL is 2"
.4byte 0x3ff002b7 # lui t0, 0x3ff00
.4byte 0x02029293 # slli t0, t0, 32
.4byte 0x00128293 # addi t0, t0, 1
.4byte 0xf2028053 # fmv.d.x ft0, t0
.4byte 0x3ff00337 # lui t1, 0x3ff00
.4byte 0x02031313 # slli t1, t1, 32
.4byte 0xffe30313 # addi t1, t1, -2
.4byte 0x5e034bd7 # vmv.v.x v23, t1
.4byte 0xbff003b7 # lui t2, 0xbff00
.4byte 0x02039393 # slli t2, t2, 32
.4byte 0x5e03c557 # vmv.v.x v10, t2
.4byte 0xa2a05bd7 # vfmadd.vf v23, ft0, v10
.assert trap == 0 "binary64 fused multiply-add retires"
.4byte 0x43702557 # vmv.x.s a0, v23
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0xb970000000000000 "binary64 fused residual is -2^-104"
.4byte 0x001025f3 # csrr a1, fflags
.assert rd_addr == 11 "fflags writes x11"
.assert rd_wdata == 0x0 "binary64 fused residual raises no flags"
# Binary64 destructive forms with vs1 or fs1 = 2, vs2 = 3, old vd = 4.
.4byte 0x400002b7 # lui t0, 0x40000
.4byte 0x02029293 # slli t0, t0, 32
.4byte 0x40080337 # lui t1, 0x40080
.4byte 0x02031313 # slli t1, t1, 32
.4byte 0x401003b7 # lui t2, 0x40100
.4byte 0x02039393 # slli t2, t2, 32
.4byte 0xf2028053 # fmv.d.x ft0, t0
.4byte 0x5e02c457 # vmv.v.x v8, t0
.4byte 0x5e0344d7 # vmv.v.x v9, t1
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xb2941857 # vfmacc.vv v16, v8, v9
.assert trap == 0 "binary64 vfmacc.vv retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0x4024000000000000 "binary64 vfmacc.vv yields 10"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xb6941857 # vfnmacc.vv v16, v8, v9
.assert trap == 0 "binary64 vfnmacc.vv retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0xc024000000000000 "binary64 vfnmacc.vv yields -10"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xba941857 # vfmsac.vv v16, v8, v9
.assert trap == 0 "binary64 vfmsac.vv retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0x4000000000000000 "binary64 vfmsac.vv yields 2"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xbe941857 # vfnmsac.vv v16, v8, v9
.assert trap == 0 "binary64 vfnmsac.vv retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0xc000000000000000 "binary64 vfnmsac.vv yields -2"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xa2941857 # vfmadd.vv v16, v8, v9
.assert trap == 0 "binary64 vfmadd.vv retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0x4026000000000000 "binary64 vfmadd.vv yields 11"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xa6941857 # vfnmadd.vv v16, v8, v9
.assert trap == 0 "binary64 vfnmadd.vv retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0xc026000000000000 "binary64 vfnmadd.vv yields -11"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xaa941857 # vfmsub.vv v16, v8, v9
.assert trap == 0 "binary64 vfmsub.vv retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0x4014000000000000 "binary64 vfmsub.vv yields 5"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xae941857 # vfnmsub.vv v16, v8, v9
.assert trap == 0 "binary64 vfnmsub.vv retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0xc014000000000000 "binary64 vfnmsub.vv yields -5"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xb2905857 # vfmacc.vf v16, ft0, v9
.assert trap == 0 "binary64 vfmacc.vf retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0x4024000000000000 "binary64 vfmacc.vf yields 10"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xb6905857 # vfnmacc.vf v16, ft0, v9
.assert trap == 0 "binary64 vfnmacc.vf retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0xc024000000000000 "binary64 vfnmacc.vf yields -10"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xba905857 # vfmsac.vf v16, ft0, v9
.assert trap == 0 "binary64 vfmsac.vf retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0x4000000000000000 "binary64 vfmsac.vf yields 2"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xbe905857 # vfnmsac.vf v16, ft0, v9
.assert trap == 0 "binary64 vfnmsac.vf retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0xc000000000000000 "binary64 vfnmsac.vf yields -2"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xa2905857 # vfmadd.vf v16, ft0, v9
.assert trap == 0 "binary64 vfmadd.vf retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0x4026000000000000 "binary64 vfmadd.vf yields 11"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xa6905857 # vfnmadd.vf v16, ft0, v9
.assert trap == 0 "binary64 vfnmadd.vf retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0xc026000000000000 "binary64 vfnmadd.vf yields -11"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xaa905857 # vfmsub.vf v16, ft0, v9
.assert trap == 0 "binary64 vfmsub.vf retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0x4014000000000000 "binary64 vfmsub.vf yields 5"
.4byte 0x5e03c857 # vmv.v.x v16, t2
.4byte 0xae905857 # vfnmsub.vf v16, ft0, v9
.assert trap == 0 "binary64 vfnmsub.vf retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0xc014000000000000 "binary64 vfnmsub.vf yields -5"
.4byte 0x001025f3 # csrr a1, fflags
.assert rd_addr == 11 "fflags writes x11"
.assert rd_wdata == 0x0 "exact binary64 forms raise no flags"
# Lane zero computes 2 * 3 + 1. Lane one would compute 0 * Inf + 1.
.4byte 0xc1017657 # vsetivli a2, 2, e32, m1, tu, mu
.assert rd_addr == 12 "VL writes x12"
.assert rd_wdata == 0x2 "e32 VL is 2"
.4byte 0x5e003457 # vmv.v.i v8, 0
.4byte 0x400002b7 # lui t0, 0x40000
.4byte 0x4202e457 # vmv.s.x v8, t0
.4byte 0x7f800337 # lui t1, 0x7f800
.4byte 0x5e0344d7 # vmv.v.x v9, t1
.4byte 0x40400337 # lui t1, 0x40400
.4byte 0x420364d7 # vmv.s.x v9, t1
.4byte 0x3f8003b7 # lui t2, 0x3f800
.4byte 0x5e03c657 # vmv.v.x v12, t2
.4byte 0x00100293 # addi t0, zero, 1
.4byte 0x4202e057 # vmv.s.x v0, t0
# Use agnostic policies and observe only the enabled result and accrued flags.
.4byte 0xcd017657 # vsetivli a2, 2, e32, m1, ta, ma
.assert rd_addr == 12 "VL writes x12"
.assert rd_wdata == 0x2 "e32 VL is 2"
.4byte 0x00105073 # csrwi fflags, 0
.4byte 0xb0941657 # vfmacc.vv v12, v8, v9, v0.t
.assert trap == 0 "masked binary32 fused accumulate retires"
.4byte 0x42c02557 # vmv.x.s a0, v12
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0x40e00000 "enabled fused result is 7.0"
.4byte 0x001025f3 # csrr a1, fflags
.assert rd_addr == 11 "fflags writes x11"
.assert rd_wdata == 0x0 "masked 0 * Inf raises no flags"
# Restore the addend before enabling the formerly masked lane.
.4byte 0xc1017657 # vsetivli a2, 2, e32, m1, tu, mu
.assert rd_addr == 12 "VL writes x12"
.assert rd_wdata == 0x2 "e32 VL is 2"
.4byte 0x5e03c657 # vmv.v.x v12, t2
.4byte 0xb2941657 # vfmacc.vv v12, v8, v9
.assert trap == 0 "unmasked binary32 fused accumulate retires"
.4byte 0x001025f3 # csrr a1, fflags
.assert rd_addr == 11 "fflags writes x11"
.assert rd_wdata == 0x10 "active 0 * Inf raises NV"
.4byte 0xc100f657 # vsetivli a2, 1, e32, m1, tu, mu
.assert rd_addr == 12 "VL writes x12"
.assert rd_wdata == 0x1 "e32 VL is 1"
.4byte 0x3ec0bad7 # vslidedown.vi v21, v12, 1
.4byte 0x43502557 # vmv.x.s a0, v21
.assert rd_addr == 10 "vector element writes x10"
.assert rd_wdata == 0x7fc00000 "active invalid fused result is canonical NaN"
# The exact binary32 fused result 1 * 1 + 2^-24 lies halfway above 1.
.4byte 0xc1017657 # vsetivli a2, 2, e32, m1, tu, mu
.assert rd_addr == 12 "binary32 VL writes x12"
.assert rd_wdata == 0x2 "binary32 VL is two"
.4byte 0x00305073 # csrwi fcsr, 0
.4byte 0x3f8002b7 # lui t0, 0x3f800
.4byte 0x5e02c457 # vmv.v.x v8, t0
.4byte 0x5e02c4d7 # vmv.v.x v9, t0
.4byte 0x338002b7 # lui t0, 0x33800
.4byte 0x5e02c857 # vmv.v.x v16, t0
# RNE selects the even significand and raises inexact.
.4byte 0xb2941857 # vfmacc.vv v16, v8, v9
.assert trap == 0 "RNE fused accumulate retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "RNE fused result writes x10"
.assert rd_wdata == 0x3f800000 "RNE rounds the fused result to 1.0"
.4byte 0x001025f3 # csrr a1, fflags
.assert rd_addr == 11 "RNE fflags writes x11"
.assert rd_wdata == 0x1 "RNE fused result raises NX"
# RUP selects the next value above 1 and raises inexact.
.4byte 0x0021d073 # csrwi frm, 3
.4byte 0x00105073 # csrwi fflags, 0
.4byte 0x00202673 # csrr a2, frm
.assert rd_addr == 12 "frm writes x12"
.assert rd_wdata == 0x3 "frm selects RUP"
.4byte 0x5e02c857 # vmv.v.x v16, t0
.4byte 0xb2941857 # vfmacc.vv v16, v8, v9
.assert trap == 0 "RUP fused accumulate retires"
.4byte 0x43002557 # vmv.x.s a0, v16
.assert rd_addr == 10 "RUP fused result writes x10"
.assert rd_wdata == 0x3f800001 "RUP rounds the fused result above 1.0"
.4byte 0x001025f3 # csrr a1, fflags
.assert rd_addr == 11 "RUP fflags writes x11"
.assert rd_wdata == 0x1 "RUP fused result raises NX"
.4byte 0x00205073 # csrwi frm, 0