-
Notifications
You must be signed in to change notification settings - Fork 4.2k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Riscv64 c906 d1 #3177
base: master
Are you sure you want to change the base?
Riscv64 c906 d1 #3177
Conversation
RVV_SPEC_0_7 update
replace vfmacc_vf with vfmacc_vv
Codecov Report
@@ Coverage Diff @@
## master #3177 +/- ##
==========================================
+ Coverage 90.34% 90.77% +0.43%
==========================================
Files 510 465 -45
Lines 136094 108462 -27632
==========================================
- Hits 122950 98458 -24492
+ Misses 13144 10004 -3140
Continue to review full report at Codecov.
|
replace vfmacc_vf with vfmacc_vv for better peformance
replace vfmacc_vf with vfmacc_vv for better performance
replace vfmacc_vf with vfmacc_vv for better performance
#if RVV_SPEC_0_7 | ||
vfloat16m1_t _v0 = vle16_v_f16m1(tmpptr, vl); | ||
vfloat16m1_t _val0 = vrgathervx_float16xm1(_v0, 0, vl); | ||
vfloat16m1_t _val1 = vrgathervx_float16xm1(_v0, 1, vl); | ||
vfloat16m1_t _val2 = vrgathervx_float16xm1(_v0, 2, vl); | ||
vfloat16m1_t _val3 = vrgathervx_float16xm1(_v0, 3, vl); | ||
vfloat16m1_t _val4 = vrgathervx_float16xm1(_v0, 4, vl); | ||
vfloat16m1_t _val5 = vrgathervx_float16xm1(_v0, 5, vl); | ||
vfloat16m1_t _val6 = vrgathervx_float16xm1(_v0, 6, vl); | ||
vfloat16m1_t _val7 = vrgathervx_float16xm1(_v0, 7, vl); | ||
tmpptr += 8; | ||
|
||
vfloat16m1_t _w0 = vle16_v_f16m1(kptr0, vl); | ||
_sum0 = vfmacc_vv_f16m1(_sum0, _val0, _w0, vl); | ||
_sum1 = vfmacc_vv_f16m1(_sum1, _val1, _w0, vl); | ||
_sum2 = vfmacc_vv_f16m1(_sum2, _val2, _w0, vl); | ||
_sum3 = vfmacc_vv_f16m1(_sum3, _val3, _w0, vl); | ||
_sum4 = vfmacc_vv_f16m1(_sum4, _val4, _w0, vl); | ||
_sum5 = vfmacc_vv_f16m1(_sum5, _val5, _w0, vl); | ||
_sum6 = vfmacc_vv_f16m1(_sum6, _val6, _w0, vl); | ||
_sum7 = vfmacc_vv_f16m1(_sum7, _val7, _w0, vl); | ||
|
||
kptr0 += packn; | ||
#else | ||
vfloat16m1_t _v0 = vle16_v_f16m1(tmpptr, vl); | ||
vfloat16m1_t _val0 = vrgather_vx_f16m1(_v0, 0, vl); | ||
vfloat16m1_t _val1 = vrgather_vx_f16m1(_v0, 1, vl); | ||
vfloat16m1_t _val2 = vrgather_vx_f16m1(_v0, 2, vl); | ||
vfloat16m1_t _val3 = vrgather_vx_f16m1(_v0, 3, vl); | ||
vfloat16m1_t _val4 = vrgather_vx_f16m1(_v0, 4, vl); | ||
vfloat16m1_t _val5 = vrgather_vx_f16m1(_v0, 5, vl); | ||
vfloat16m1_t _val6 = vrgather_vx_f16m1(_v0, 6, vl); | ||
vfloat16m1_t _val7 = vrgather_vx_f16m1(_v0, 7, vl); | ||
tmpptr += 8; | ||
|
||
vfloat16m1_t _w0 = vle16_v_f16m1(kptr0, vl); | ||
_sum0 = vfmacc_vf_f16m1(_sum0, val0, _w0, vl); | ||
_sum1 = vfmacc_vf_f16m1(_sum1, val1, _w0, vl); | ||
_sum2 = vfmacc_vf_f16m1(_sum2, val2, _w0, vl); | ||
_sum3 = vfmacc_vf_f16m1(_sum3, val3, _w0, vl); | ||
_sum4 = vfmacc_vf_f16m1(_sum4, val4, _w0, vl); | ||
_sum5 = vfmacc_vf_f16m1(_sum5, val5, _w0, vl); | ||
_sum6 = vfmacc_vf_f16m1(_sum6, val6, _w0, vl); | ||
_sum7 = vfmacc_vf_f16m1(_sum7, val7, _w0, vl); | ||
_sum0 = vfmacc_vv_f16m1(_sum0, _val0, _w0, vl); | ||
_sum1 = vfmacc_vv_f16m1(_sum1, _val1, _w0, vl); | ||
_sum2 = vfmacc_vv_f16m1(_sum2, _val2, _w0, vl); | ||
_sum3 = vfmacc_vv_f16m1(_sum3, _val3, _w0, vl); | ||
_sum4 = vfmacc_vv_f16m1(_sum4, _val4, _w0, vl); | ||
_sum5 = vfmacc_vv_f16m1(_sum5, _val5, _w0, vl); | ||
_sum6 = vfmacc_vv_f16m1(_sum6, _val6, _w0, vl); | ||
_sum7 = vfmacc_vv_f16m1(_sum7, _val7, _w0, vl); | ||
|
||
kptr0 += packn; | ||
#endif |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Stick to rvv-1.0 spec for intrinsic code, and define compatibility alias for rvv-0.7 in riscv_v_071_fix.h
No description provided.