diff --git a/Source/NNSupportFunctions/muriscv_nn_vec_mat_mult_t_s8.c b/Source/NNSupportFunctions/muriscv_nn_vec_mat_mult_t_s8.c
index 908aa9c4..ddd733a4 100644
--- a/Source/NNSupportFunctions/muriscv_nn_vec_mat_mult_t_s8.c
+++ b/Source/NNSupportFunctions/muriscv_nn_vec_mat_mult_t_s8.c
@@ -63,6 +63,7 @@ muriscv_nn_status muriscv_nn_vec_mat_mult_t_s8(const q7_t *lhs,
                                                const int32_t rhs_offset) // Currently Unused
 {
 #if defined(USE_VEXT)
+    if (address_offset == 1) { // TODO: remove this after #81 is fixed
 
     /* At some point in time there might be implementations with the Zvediv extension.
      * It might provide dot-product functions which would simplify this code here and
@@ -249,8 +250,134 @@ muriscv_nn_status muriscv_nn_vec_mat_mult_t_s8(const q7_t *lhs,
         *dst = (q7_t)res_scalar;
         dst += address_offset;
         rhs += rhs_cols;
+
+    }
+    } else { // TODO: remove this after #81 is fixed
+    /* Uses 5x loop unrolling in order to expose more ILP */
+    const int32_t row_loop_cnt = rhs_rows / 5;
+    for (int i_row_loop_cnt = 0; i_row_loop_cnt < row_loop_cnt; i_row_loop_cnt++)
+    {
+
+        const q7_t *lhs_ptr = lhs;
+        const q7_t *rhs_ptr_0 = &rhs[0];
+        const q7_t *rhs_ptr_1 = &rhs[rhs_cols];
+        const q7_t *rhs_ptr_2 = &rhs[rhs_cols * 2];
+        const q7_t *rhs_ptr_3 = &rhs[rhs_cols * 3];
+        const q7_t *rhs_ptr_4 = &rhs[rhs_cols * 4];
+
+        q31_t res00 = 0;
+        q31_t res01 = 0;
+        q31_t res02 = 0;
+        q31_t res03 = 0;
+        q31_t res04 = 0;
+        if (bias)
+        {
+            res00 = *bias++;
+            res01 = *bias++;
+            res02 = *bias++;
+            res03 = *bias++;
+            res04 = *bias++;
+        }
+        for (int32_t rhs_cols_idx = 0; rhs_cols_idx < rhs_cols; ++rhs_cols_idx)
+        {
+            const q31_t rhs_value0 = (int8_t)*rhs_ptr_0;
+            const q31_t rhs_value1 = (int8_t)*rhs_ptr_1;
+            const q31_t rhs_value2 = (int8_t)*rhs_ptr_2;
+            const q31_t rhs_value3 = (int8_t)*rhs_ptr_3;
+            const q31_t rhs_value4 = (int8_t)*rhs_ptr_4;
+            const q31_t lhs_value = (int8_t)*lhs_ptr + lhs_offset;
+
+            res00 += lhs_value * rhs_value0;
+            res01 += lhs_value * rhs_value1;
+            res02 += lhs_value * rhs_value2;
+            res03 += lhs_value * rhs_value3;
+            res04 += lhs_value * rhs_value4;
+
+            ++rhs_ptr_0;
+            ++rhs_ptr_1;
+            ++rhs_ptr_2;
+            ++rhs_ptr_3;
+            ++rhs_ptr_4;
+            ++lhs_ptr;
+        }
+        // Quantize down
+        res00 = muriscv_nn_requantize(res00, dst_multiplier, dst_shift);
+        res01 = muriscv_nn_requantize(res01, dst_multiplier, dst_shift);
+        res02 = muriscv_nn_requantize(res02, dst_multiplier, dst_shift);
+        res03 = muriscv_nn_requantize(res03, dst_multiplier, dst_shift);
+        res04 = muriscv_nn_requantize(res04, dst_multiplier, dst_shift);
+
+        // Add offset
+        res00 += dst_offset;
+        res01 += dst_offset;
+        res02 += dst_offset;
+        res03 += dst_offset;
+        res04 += dst_offset;
+
+        // Clamp the result
+        res00 = MAX(res00, activation_min);
+        res00 = MIN(res00, activation_max);
+        res01 = MAX(res01, activation_min);
+        res01 = MIN(res01, activation_max);
+        res02 = MAX(res02, activation_min);
+        res02 = MIN(res02, activation_max);
+        res03 = MAX(res03, activation_min);
+        res03 = MIN(res03, activation_max);
+        res04 = MAX(res04, activation_min);
+        res04 = MIN(res04, activation_max);
+
+        *dst = (q7_t)res00;
+        *(dst + address_offset) = (q7_t)res01;
+        *(dst + 2 * address_offset) = (q7_t)res02;
+        *(dst + 3 * address_offset) = (q7_t)res03;
+        *(dst + 4 * address_offset) = (q7_t)res04;
+        dst += 5 * address_offset;
+
+        rhs += 5 * rhs_cols;
+
+    }
+
+    /* Handle the leftover part from 5x loop unrolling */
+    const int loop_cnt = rhs_rows % 5;
+    for (int i_loop_cnt = 0; i_loop_cnt < loop_cnt; i_loop_cnt++)
+    {
+
+        const q7_t *lhs_ptr = &lhs[0];
+        const q7_t *rhs_ptr = &rhs[0];
+
+        q31_t res00 = 0;
+        if (bias)
+        {
+            res00 = *bias++;
+        }
+
+        for (int32_t rhs_cols_idx = 0; rhs_cols_idx < rhs_cols; ++rhs_cols_idx)
+        {
+            q31_t rhs_value0 = (int8_t)rhs_ptr[0];
+            q31_t lhs_value = (int8_t)lhs_ptr[0] + lhs_offset;
+
+            res00 += lhs_value * rhs_value0;
+
+            ++rhs_ptr;
+            ++lhs_ptr;
+        }
+
+        // Quantize down
+        res00 = muriscv_nn_requantize(res00, dst_multiplier, dst_shift);
+
+        // Add offset
+        res00 += dst_offset;
+
+        // Clamp the result
+        res00 = MAX(res00, activation_min);
+        res00 = MIN(res00, activation_max);
+
+        *dst = (int8_t)res00;
+        dst += address_offset;
+        rhs += rhs_cols;
         
     }
+    } // TODO: remove this after #81 is fixed
 
 #elif defined(USE_PEXT) /* defined(USE_VEXT) */