Functions View for Build Insights in Visual Studio 2022 17.8 Eve Silfanus November 17th, 20230 2 Introduction We are excited to unveil a new feature in Build Insights for Visual Studio: Functions View! This feature is available in Visual Studio 2022 version 17.8. Functions View offers essential insights into functions and forceinlines within your codebases. Download Visual Studio 2022 17.8 We extend our sincere thanks thanks to the developer community, especially our game studio partners, for actively providing feedback. Your contributions are invaluable in shaping this new feature. For more details about Build Insights and to explore other features like Included Files and Include Tree Views, please visit our initial announcement blogpost. Code Generation Insights with Functions View Functions View is a powerful tool that displays the impact of each function on the total build time by analyzing code generation times and forceinlines. Forceinlines, commonly used to boost runtime efficiency, can also influence build times. The following sample code is based on a public code sample by Aras Pranckevičius. We will use it to show you how you can optimize your builds with Functions View. To setup your project, create a C++ Console application and copy the following sample code: #include struct float4 { __m128 val; float4() { val = _mm_setzero_ps(); } float4(float x) { val = _mm_set1_ps(x); } float4(float x, float y) { val = _mm_set_ps(y, x, y, x); } float4(float x, float y, float z) { val = _mm_set_ps(0.f, z, y, x); } float4(float x, float y, float z, float w) { val = _mm_set_ps(w, z, y, x); } float4(__m128 v) { val = v; } }; static __forceinline float4 operator+(const float4& a, const float4& b) { return float4(_mm_add_ps(a.val, b.val)); } static __forceinline float4 operator-(const float4& a, const float4& b) { return float4(_mm_sub_ps(a.val, b.val)); } static __forceinline float4 operator*(const float4& a, const float4& b) { return float4(_mm_mul_ps(a.val, b.val)); } static __forceinline float4 operator/(const float4& a, const float4& b) { return float4(_mm_div_ps(a.val, b.val)); } static __forceinline float4 csum(const float4& p) { __m128 r = _mm_add_ps(p.val, _mm_castsi128_ps(_mm_shuffle_epi32(_mm_castps_si128(p.val), _MM_SHUFFLE(0, 3, 2, 1)))); return _mm_add_ps(r, _mm_castsi128_ps(_mm_shuffle_epi32(_mm_castps_si128(r), _MM_SHUFFLE(1, 0, 3, 2)))); } static __forceinline float4 dot(const float4& p0, const float4& p1) { return csum(p0 * p1); } static __forceinline float4 dot(const float4& p) { return dot(p, p); } static __forceinline float4 rsqrt(const float4& x) { #define C0 9.999998e-01f #define C1 3.0000002e+00f #define C2 .5f #define C3 340282346638528859811704183484516925440.f __m128 e = _mm_mul_ps(_mm_rsqrt_ps((__m128) x.val), _mm_set_ps(C0, C0, C0, C0)); e = _mm_min_ps(e, _mm_set_ps(C3, C3, C3, C3)); return _mm_mul_ps(_mm_mul_ps(e, _mm_set_ps(C2, C2, C2, C2)), _mm_sub_ps(_mm_set_ps(C1, C1, C1, C1), _mm_mul_ps(_mm_mul_ps(x.val, e), e))); } static __forceinline float4 normalize(const float4& v) { return v * rsqrt(dot(v)); } static __forceinline float4 ident() { return float4(0.f, 0.f, 0.f, 1.f); } static __forceinline float4 sampleFun1(const float4& x, const float4& y) { return csum(x) / x + y; } static __forceinline float4 sampleFun2(const float4& q1, const float4& q2) { return sampleFun1(q1 * q2, q2 – q1) * (q1 + q2); } static float4 sampleFun3(const float4& pq, const float4& mask) { const float c8 = 0.923879532511287f; const float s8 = 0.38268343236509f; const float g = 5.82842712474619f; float4 ch = float4(2) * (normalize(pq) – normalize(mask)); float4 sh = pq * normalize(ch); float4 r = ((g * sh * sh – ch * ch) + sh / float4(s8, s8, s8, c8)) * mask; return normalize(r); } struct matrix […]
