Loading files
This file is pinned to the selected release. Download the original to open it in your design tool.
a480b9028This file is pinned to the selected release. Download the original to open it in your design tool.
// Copyright 2020 ETH Zurich and University of Bologna.
// Licensed under the Apache License, Version 2.0, see LICENSE for details.
// SPDX-License-Identifier: Apache-2.0
#include <snrt.h>
int main() {
int errs = 46;
if (snrt_is_compute_core()) {
uint32_t i_a = 0x4048F5C3; // 3.14 0
uint32_t i_an = 0xC048F5C3; // -3.14
uint32_t i_b = 0x3FCF1AA0; // 1.618 2
uint32_t i_bn = 0xBFCF1AA0; // -1.618
uint32_t i_c = 0x4018FFEB; // 2.39062
uint32_t i_cn = 0xC018FFEB; // -2.39062
uint32_t i_d = 0x3E801FFB; // 0.250244 6
uint32_t i_dn = 0xBE801FFB; // -0.250244
uint32_t i_e = 0x3F000000; // 0.5
uint32_t i_en = 0xBF000000; // -0.5
uint32_t i_f = 0x42C83F36; // 100.123456789 10
uint32_t i_fn = 0xC2C83F36; // -100.123456789
int res0 = 0;
uint32_t res1 = 0;
uint32_t res2 = 0;
uint32_t res3 = 0;
uint32_t res4 = 0;
asm volatile(
"fmv.s.x ft0, %0\n"
"fmv.s.x ft1, %1\n"
"fmv.s.x ft2, %2\n"
"fmv.s.x ft3, %3\n"
"vfcpka.b.s ft4, ft0, ft2\n"
"vfcpkb.b.s ft4, ft0, ft2\n"
"vfcpkc.b.s ft4, ft0, ft2\n"
"vfcpkd.b.s ft4, ft0, ft2\n" // ft4 =
// {3.14, 1.618, 3.14, 1.618, 3.14, 1.618,
// 3.14, 1.618}
"vfcpka.b.s ft5, ft1, ft3\n"
"vfcpkb.b.s ft5, ft1, ft3\n"
"vfcpkc.b.s ft5, ft1, ft3\n"
"vfcpkd.b.s ft5, ft1, ft3\n" // ft5 = {-3.14, -1.618,-3.14, -1.618,
// -3.14, -1.618, -3.14, -1.618}
"vfcpka.b.s ft6, ft0, ft3\n"
"vfcpkb.b.s ft6, ft0, ft3\n"
"vfcpkc.b.s ft6, ft0, ft3\n"
"vfcpkd.b.s ft6, ft0, ft3\n" // ft6 = {3.14, -1.618, 3.14,
// -1.618, 3.14, -1.618, 3.14, -1.618}
"vfcpka.b.s ft7, ft1, ft2\n"
"vfcpkb.b.s ft7, ft1, ft2\n"
"vfcpkc.b.s ft7, ft1, ft2\n"
"vfcpkd.b.s ft7, ft1, ft2\n" // ft7 = {-3.14, 1.618, -3.14, 1.618,
// -3.14, 1.618, -3.14, 1.618}
: "+r"(i_a), "+r"(i_an), "+r"(i_b), "+r"(i_bn));
// VFSGNJ
asm volatile(
"vfsgnj.b ft0, ft4, ft4\n"
"vfeq.b %0, ft4, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
asm volatile(
"vfsgnj.b ft0, ft4, ft5\n"
"vfeq.b %0, ft5, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
asm volatile(
"vfsgnj.b ft0, ft5, ft6\n"
"vfeq.b %0, ft6, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
asm volatile(
"vfsgnj.b ft0, ft5, ft7\n"
"vfeq.b %0, ft7, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
// VFSGNJ.R
asm volatile(
"vfsgnj.r.b ft0, ft4, ft4\n"
"vfeq.b %0, ft4, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
asm volatile(
"vfsgnj.r.b ft0, ft4, ft5\n"
"vfeq.b %0, ft5, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
asm volatile(
"vfsgnj.r.b ft0, ft5, ft6\n"
"vfeq.b %0, ft4, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
asm volatile(
"vfsgnj.r.b ft0, ft5, ft7\n"
"vfeq.b %0, ft5, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
// VFSGNJN
asm volatile(
"vfsgnjn.b ft0, ft4, ft4\n"
"vfeq.b %0, ft5, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
asm volatile(
"vfsgnjn.b ft0, ft4, ft5\n"
"vfeq.b %0, ft4, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
asm volatile(
"vfsgnjn.b ft0, ft5, ft6\n"
"vfeq.b %0, ft7, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
asm volatile(
"vfsgnjn.b ft0, ft5, ft7\n"
"vfeq.b %0, ft6, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
// VFSGNJN.R
asm volatile(
"vfsgnjn.r.b ft0, ft4, ft4\n"
"vfeq.b %0, ft5, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
asm volatile(
"vfsgnjn.r.b ft0, ft4, ft5\n"
"vfeq.b %0, ft4, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
asm volatile(
"vfsgnjn.r.b ft0, ft5, ft6\n"
"vfeq.b %0, ft5, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
asm volatile(
"vfsgnjn.r.b ft0, ft5, ft7\n"
"vfeq.b %0, ft4, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
// VFSGNJX
asm volatile(
"vfsgnjx.b ft0, ft4, ft4\n"
"vfeq.b %0, ft4, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
asm volatile(
"vfsgnjx.b ft0, ft4, ft5\n"
"vfeq.b %0, ft5, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
asm volatile(
"vfsgnjx.b ft0, ft5, ft6\n"
"vfeq.b %0, ft7, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
asm volatile(
"vfsgnjx.b ft0, ft5, ft7\n"
"vfeq.b %0, ft6, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
// VFSGNJX.R
asm volatile(
"vfsgnjx.r.b ft0, ft4, ft4\n"
"vfeq.b %0, ft4, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
asm volatile(
"vfsgnjx.r.b ft0, ft4, ft5\n"
"vfeq.b %0, ft5, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
asm volatile(
"vfsgnjx.r.b ft0, ft5, ft6\n"
"vfeq.b %0, ft5, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
asm volatile(
"vfsgnjx.r.b ft0, ft5, ft7\n"
"vfeq.b %0, ft4, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
// load new data
asm volatile(
"fmv.s.x ft0, %0\n" // 3.14
"fmv.s.x ft1, %1\n" // -1.618
"fmv.s.x ft2, %2\n" // 0.250244
"fmv.s.x ft3, %3\n" // 100.123456789
"vfcpka.b.s ft4, ft3, ft0\n"
"vfcpkb.b.s ft4, ft3, ft0\n"
"vfcpkc.b.s ft4, ft3, ft0\n"
"vfcpkd.b.s ft4, ft3, ft0\n" // ft4 = {100.123456789, 3.14,
// 100.123456789, 3.14,
// 100.123456789, 3.14,
// 100.123456789, 3.14}
"vfcpka.b.s ft5, ft2, ft1\n"
"vfcpkb.b.s ft5, ft2, ft1\n"
"vfcpkc.b.s ft5, ft2, ft1\n"
"vfcpkd.b.s ft5, ft2, ft1\n" // ft5 = {0.250244, -1.618, 0.250244,
// -1.618, 0.250244, -1.618, 0.250244,
// -1.618}
"vfcpka.b.s ft6, ft1, ft3\n"
"vfcpkb.b.s ft6, ft1, ft3\n"
"vfcpkc.b.s ft6, ft1, ft3\n"
"vfcpkd.b.s ft6, ft1, ft3\n" // ft6 = {-1.618, 100.123456789,
// -1.618, 100.123456789, -1.618,
// 100.123456789, -1.618,
// 100.123456789}
: "+r"(i_a), "+r"(i_bn), "+r"(i_d), "+r"(i_f));
// VFADD
// pack results
res1 = 0x42C00000;
res2 = 0x3FC00000;
res3 = 0xBFA00000;
res4 = 0x42C00000;
asm volatile(
"fmv.s.x ft0, %0\n"
"fmv.s.x ft1, %1\n"
"fmv.s.x ft2, %2\n"
"fmv.s.x ft3, %3\n"
// pack h values
"vfcpka.b.s ft7, ft0, ft1\n"
"vfcpkb.b.s ft7, ft0, ft1\n"
"vfcpkc.b.s ft7, ft0, ft1\n"
"vfcpkd.b.s ft7, ft0, ft1\n"
"vfcpka.b.s ft8, ft2, ft3\n"
"vfcpkb.b.s ft8, ft2, ft3\n"
"vfcpkc.b.s ft8, ft2, ft3\n"
"vfcpkd.b.s ft8, ft2, ft3\n"
: "+r"(res1), "+r"(res2), "+r"(res3), "+r"(res4));
asm volatile(
"vfadd.b ft0, ft4, ft5\n"
"vfeq.b %0, ft7, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
asm volatile(
"vfadd.b ft0, ft5, ft6\n"
"vfeq.b %0, ft8, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
// VFADD.R
// pack results
res1 = 0x42C00000;
res2 = 0x40400000;
res3 = 0xBFA00000;
res4 = 0xC0400000;
asm volatile(
"fmv.s.x ft0, %0\n"
"fmv.s.x ft1, %1\n"
"fmv.s.x ft2, %2\n"
"fmv.s.x ft3, %3\n"
// pack h values
"vfcpka.b.s ft7, ft0, ft1\n"
"vfcpkb.b.s ft7, ft0, ft1\n"
"vfcpkc.b.s ft7, ft0, ft1\n"
"vfcpkd.b.s ft7, ft0, ft1\n"
"vfcpka.b.s ft8, ft2, ft3\n"
"vfcpkb.b.s ft8, ft2, ft3\n"
"vfcpkc.b.s ft8, ft2, ft3\n"
"vfcpkd.b.s ft8, ft2, ft3\n"
: "+r"(res1), "+r"(res2), "+r"(res3), "+r"(res4));
asm volatile(
"vfadd.r.b ft0, ft4, ft5\n"
"vfeq.b %0, ft7, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
asm volatile(
"vfadd.r.b ft0, ft5, ft6\n"
"vfeq.b %0, ft8, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
// VFSUB
// pack results
res1 = 0x42C00000;
res2 = 0x40800000;
res3 = 0x3FE00000;
res4 = 0xC2C00000;
asm volatile(
"fmv.s.x ft0, %0\n"
"fmv.s.x ft1, %1\n"
"fmv.s.x ft2, %2\n"
"fmv.s.x ft3, %3\n"
// pack h values
"vfcpka.b.s ft7, ft0, ft1\n"
"vfcpkb.b.s ft7, ft0, ft1\n"
"vfcpkc.b.s ft7, ft0, ft1\n"
"vfcpkd.b.s ft7, ft0, ft1\n"
"vfcpka.b.s ft8, ft2, ft3\n"
"vfcpkb.b.s ft8, ft2, ft3\n"
"vfcpkc.b.s ft8, ft2, ft3\n"
"vfcpkd.b.s ft8, ft2, ft3\n"
: "+r"(res1), "+r"(res2), "+r"(res3), "+r"(res4));
asm volatile(
"vfsub.b ft0, ft4, ft5\n"
"vfeq.b %0, ft7, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
asm volatile(
"vfsub.b ft0, ft5, ft6\n"
"vfeq.b %0, ft8, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
// VFSUB.R
// pack results
res1 = 0x42C00000;
res2 = 0x40400000;
res3 = 0x3FE00000;
res4 = 0x00000000;
asm volatile(
"fmv.s.x ft0, %0\n"
"fmv.s.x ft1, %1\n"
"fmv.s.x ft2, %2\n"
"fmv.s.x ft3, %3\n"
// pack h values
"vfcpka.b.s ft7, ft0, ft1\n"
"vfcpkb.b.s ft7, ft0, ft1\n"
"vfcpkc.b.s ft7, ft0, ft1\n"
"vfcpkd.b.s ft7, ft0, ft1\n"
"vfcpka.b.s ft8, ft2, ft3\n"
"vfcpkb.b.s ft8, ft2, ft3\n"
"vfcpkc.b.s ft8, ft2, ft3\n"
"vfcpkd.b.s ft8, ft2, ft3\n"
: "+r"(res1), "+r"(res2), "+r"(res3), "+r"(res4));
asm volatile(
"vfsub.r.b ft0, ft4, ft5\n"
"vfeq.b %0, ft7, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
asm volatile(
"vfsub.r.b ft0, ft5, ft6\n"
"vfeq.b %0, ft8, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
// VFMUL
// pack results
res1 = 0x41C00000;
res2 = 0xC0800000;
res3 = 0xBEC00000;
res4 = 0xC3000000;
asm volatile(
"fmv.s.x ft0, %0\n"
"fmv.s.x ft1, %1\n"
"fmv.s.x ft2, %2\n"
"fmv.s.x ft3, %3\n"
// pack h values
"vfcpka.b.s ft7, ft0, ft1\n"
"vfcpkb.b.s ft7, ft0, ft1\n"
"vfcpkc.b.s ft7, ft0, ft1\n"
"vfcpkd.b.s ft7, ft0, ft1\n"
"vfcpka.b.s ft8, ft2, ft3\n"
"vfcpkb.b.s ft8, ft2, ft3\n"
"vfcpkc.b.s ft8, ft2, ft3\n"
"vfcpkd.b.s ft8, ft2, ft3\n"
: "+r"(res1), "+r"(res2), "+r"(res3), "+r"(res4));
asm volatile(
"vfmul.b ft0, ft4, ft5\n"
"vfeq.b %0, ft7, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
asm volatile(
"vfmul.b ft0, ft5, ft6\n"
"vfeq.b %0, ft8, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
// VFMUL.R
// pack results
res1 = 0x41C00000;
res2 = 0x3F400000;
res3 = 0xBEC00000;
res4 = 0x40000000;
asm volatile(
"fmv.s.x ft0, %0\n"
"fmv.s.x ft1, %1\n"
"fmv.s.x ft2, %2\n"
"fmv.s.x ft3, %3\n"
// pack h values
"vfcpka.b.s ft7, ft0, ft1\n"
"vfcpkb.b.s ft7, ft0, ft1\n"
"vfcpkc.b.s ft7, ft0, ft1\n"
"vfcpkd.b.s ft7, ft0, ft1\n"
"vfcpka.b.s ft8, ft2, ft3\n"
"vfcpkb.b.s ft8, ft2, ft3\n"
"vfcpkc.b.s ft8, ft2, ft3\n"
"vfcpkd.b.s ft8, ft2, ft3\n"
: "+r"(res1), "+r"(res2), "+r"(res3), "+r"(res4));
asm volatile(
"vfmul.r.b ft0, ft4, ft5\n"
"vfeq.b %0, ft7, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
asm volatile(
"vfmul.r.b ft0, ft5, ft6\n"
"vfeq.b %0, ft8, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
// VFMAC
// pack results
res1 = 0x41C00000;
res2 = 0xC0800000;
res3 = 0x41C00000;
res4 = 0xC3200000;
asm volatile(
"fmv.s.x ft0, %0\n"
"fmv.s.x ft1, %1\n"
"fmv.s.x ft2, %2\n"
"fmv.s.x ft3, %3\n"
// pack h values
"vfcpka.b.s ft7, ft0, ft1\n"
"vfcpkb.b.s ft7, ft0, ft1\n"
"vfcpkc.b.s ft7, ft0, ft1\n"
"vfcpkd.b.s ft7, ft0, ft1\n"
"vfcpka.b.s ft8, ft2, ft3\n"
"vfcpkb.b.s ft8, ft2, ft3\n"
"vfcpkc.b.s ft8, ft2, ft3\n"
"vfcpkd.b.s ft8, ft2, ft3\n"
// reset ft0
"fcvt.d.w ft0, zero\n"
: "+r"(res1), "+r"(res2), "+r"(res3), "+r"(res4));
asm volatile(
"vfmac.b ft0, ft4, ft5\n"
"vfeq.b %0, ft7, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
asm volatile(
"vfmac.b ft0, ft5, ft6\n"
"vfeq.b %0, ft8, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
// VFMAC.R
// pack results
res1 = 0x41C00000;
res2 = 0x3F400000;
res3 = 0x41C00000;
res4 = 0x40400000;
asm volatile(
"fmv.s.x ft0, %0\n"
"fmv.s.x ft1, %1\n"
"fmv.s.x ft2, %2\n"
"fmv.s.x ft3, %3\n"
// pack h values
"vfcpka.b.s ft7, ft0, ft1\n"
"vfcpkb.b.s ft7, ft0, ft1\n"
"vfcpkc.b.s ft7, ft0, ft1\n"
"vfcpkd.b.s ft7, ft0, ft1\n"
"vfcpka.b.s ft8, ft2, ft3\n"
"vfcpkb.b.s ft8, ft2, ft3\n"
"vfcpkc.b.s ft8, ft2, ft3\n"
"vfcpkd.b.s ft8, ft2, ft3\n"
// reset ft0
"fcvt.d.w ft0, zero\n"
: "+r"(res1), "+r"(res2), "+r"(res3), "+r"(res4));
asm volatile(
"vfmac.r.b ft0, ft4, ft5\n"
"vfeq.b %0, ft7, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
asm volatile(
"vfmac.r.b ft0, ft5, ft6\n"
"vfeq.b %0, ft8, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
// VFMRE
// pack results
res1 = 0xC1C00000;
res2 = 0x40800000;
res3 = 0xC1C00000;
res4 = 0x43200000;
asm volatile(
"fmv.s.x ft0, %0\n"
"fmv.s.x ft1, %1\n"
"fmv.s.x ft2, %2\n"
"fmv.s.x ft3, %3\n"
// pack h values
"vfcpka.b.s ft7, ft0, ft1\n"
"vfcpkb.b.s ft7, ft0, ft1\n"
"vfcpkc.b.s ft7, ft0, ft1\n"
"vfcpkd.b.s ft7, ft0, ft1\n"
"vfcpka.b.s ft8, ft2, ft3\n"
"vfcpkb.b.s ft8, ft2, ft3\n"
"vfcpkc.b.s ft8, ft2, ft3\n"
"vfcpkd.b.s ft8, ft2, ft3\n"
// reset ft0
"fcvt.d.w ft0, zero\n"
: "+r"(res1), "+r"(res2), "+r"(res3), "+r"(res4));
asm volatile(
"vfmre.b ft0, ft4, ft5\n"
"vfeq.b %0, ft7, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
asm volatile(
"vfmre.b ft0, ft5, ft6\n"
"vfeq.b %0, ft8, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
// pack results
res1 = 0x43000000;
res2 = 0xC3000000;
asm volatile(
"fmv.s.x ft1, %0\n"
"fmv.s.x ft2, %1\n"
// pack h values
"vfcpka.b.s ft7, ft1, ft2\n"
"vfcpkb.b.s ft7, ft1, ft2\n"
"vfcpkc.b.s ft7, ft1, ft2\n"
"vfcpkd.b.s ft7, ft1, ft2\n"
// do NOT reset ft0
: "+r"(res1), "+r"(res2));
asm volatile(
"vfmre.b ft0, ft4, ft6\n"
"vfeq.b %0, ft7, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
// VFMRE.R
// pack results
res1 = 0xC1C00000;
res2 = 0xBF400000;
res3 = 0xC1C00000;
res4 = 0xC0400000;
asm volatile(
"fmv.s.x ft0, %0\n"
"fmv.s.x ft1, %1\n"
"fmv.s.x ft2, %2\n"
"fmv.s.x ft3, %3\n"
// pack h values
"vfcpka.b.s ft7, ft0, ft1\n"
"vfcpkb.b.s ft7, ft0, ft1\n"
"vfcpkc.b.s ft7, ft0, ft1\n"
"vfcpkd.b.s ft7, ft0, ft1\n"
"vfcpka.b.s ft8, ft2, ft3\n"
"vfcpkb.b.s ft8, ft2, ft3\n"
"vfcpkc.b.s ft8, ft2, ft3\n"
"vfcpkd.b.s ft8, ft2, ft3\n"
// reset ft0
"fcvt.d.w ft0, zero\n"
: "+r"(res1), "+r"(res2), "+r"(res3), "+r"(res4));
asm volatile(
"vfmre.r.b ft0, ft4, ft5\n"
"vfeq.b %0, ft7, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
asm volatile(
"vfmre.r.b ft0, ft5, ft6\n"
"vfeq.b %0, ft8, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
// pack results
res1 = 0x43000000;
res2 = 0x3FC00000;
asm volatile(
"fmv.s.x ft1, %0\n"
"fmv.s.x ft2, %1\n"
// pack h values
"vfcpka.b.s ft7, ft1, ft2\n"
"vfcpkb.b.s ft7, ft1, ft2\n"
"vfcpkc.b.s ft7, ft1, ft2\n"
"vfcpkd.b.s ft7, ft1, ft2\n"
// do NOT reset ft0
: "+r"(res1), "+r"(res2));
asm volatile(
"vfmre.r.b ft0, ft4, ft6\n"
"vfeq.b %0, ft7, ft0\n"
: "+r"(res0));
errs -= (res0 == 0xff);
}
return errs;
}