Skip to content

Resolve test failure with XMVector3ProjectStream on ARM-NEON #363

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.

Sign up for GitHub

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

Merged
Chuck Walbourn (walbourn) merged 1 commit into from
Nov 4, 2026
Merged

Resolve test failure with XMVector3ProjectStream on ARM-NEON #363

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 3 additions & 3 deletions .github/workflows/arm64test.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -37,13 +37,13 @@ jobs:
fail-fast: false

matrix:
build_type: [arm64-Release]
build_type: [arm64-Debug, arm64-Release]

steps:
- uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
- uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1

- name: Clone test repository
uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
with:
repository: walbourn/directxmathtest
path: Tests
Expand Down
36 changes: 18 additions & 18 deletions Inc/DirectXMathVector.inl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10834,40 +10834,40 @@ _Use_decl_annotations_ inline XMFLOAT3* XM_CALLCONV XMVector3ProjectStream(XMFLO
pInputVector += sizeof(XMFLOAT3) * 4;

float32x2_t r3 = vget_low_f32(Transform.r[3]);
float32x2_t r = vget_low_f32(Transform.r[0]);
XMVECTOR vResult0 = vmlaq_lane_f32(vdupq_lane_f32(r3, 0), V.val[0], r, 0); // Ax+M
XMVECTOR vResult1 = vmlaq_lane_f32(vdupq_lane_f32(r3, 1), V.val[0], r, 1); // Bx+N
float32x2_t r = vget_low_f32(Transform.r[2]);
XMVECTOR vResult0 = vmlaq_lane_f32(vdupq_lane_f32(r3, 0), V.val[2], r, 0); // Iz+M
XMVECTOR vResult1 = vmlaq_lane_f32(vdupq_lane_f32(r3, 1), V.val[2], r, 1); // Jz+N

XM_PREFETCH(pInputVector);

r3 = vget_high_f32(Transform.r[3]);
r = vget_high_f32(Transform.r[0]);
XMVECTOR vResult2 = vmlaq_lane_f32(vdupq_lane_f32(r3, 0), V.val[0], r, 0); // Cx+O
XMVECTOR W = vmlaq_lane_f32(vdupq_lane_f32(r3, 1), V.val[0], r, 1); // Dx+P
r = vget_high_f32(Transform.r[2]);
XMVECTOR vResult2 = vmlaq_lane_f32(vdupq_lane_f32(r3, 0), V.val[2], r, 0); // Kz+O
XMVECTOR W = vmlaq_lane_f32(vdupq_lane_f32(r3, 1), V.val[2], r, 1); // Lz+P

XM_PREFETCH(pInputVector + XM_CACHE_LINE_SIZE);

r = vget_low_f32(Transform.r[1]);
vResult0 = vmlaq_lane_f32(vResult0, V.val[1], r, 0); // Ax+Ey+M
vResult1 = vmlaq_lane_f32(vResult1, V.val[1], r, 1); // Bx+Fy+N
vResult0 = vmlaq_lane_f32(vResult0, V.val[1], r, 0); // Ey+Iz+M
vResult1 = vmlaq_lane_f32(vResult1, V.val[1], r, 1); // Fy+Jz+N

XM_PREFETCH(pInputVector + (XM_CACHE_LINE_SIZE * 2));

r = vget_high_f32(Transform.r[1]);
vResult2 = vmlaq_lane_f32(vResult2, V.val[1], r, 0); // Cx+Gy+O
W = vmlaq_lane_f32(W, V.val[1], r, 1); // Dx+Hy+P
vResult2 = vmlaq_lane_f32(vResult2, V.val[1], r, 0); // Gy+Kz+O
W = vmlaq_lane_f32(W, V.val[1], r, 1); // Hy+Lz+P

XM_PREFETCH(pInputVector + (XM_CACHE_LINE_SIZE * 3));

r = vget_low_f32(Transform.r[2]);
vResult0 = vmlaq_lane_f32(vResult0, V.val[2], r, 0); // Ax+Ey+Iz+M
vResult1 = vmlaq_lane_f32(vResult1, V.val[2], r, 1); // Bx+Fy+Jz+N
r = vget_low_f32(Transform.r[0]);
vResult0 = vmlaq_lane_f32(vResult0, V.val[0], r, 0); // Ax+Ey+Iz+M
vResult1 = vmlaq_lane_f32(vResult1, V.val[0], r, 1); // Bx+Fy+Jz+N

XM_PREFETCH(pInputVector + (XM_CACHE_LINE_SIZE * 4));

r = vget_high_f32(Transform.r[2]);
vResult2 = vmlaq_lane_f32(vResult2, V.val[2], r, 0); // Cx+Gy+Kz+O
W = vmlaq_lane_f32(W, V.val[2], r, 1); // Dx+Hy+Lz+P
r = vget_high_f32(Transform.r[0]);
vResult2 = vmlaq_lane_f32(vResult2, V.val[0], r, 0); // Cx+Gy+Kz+O
W = vmlaq_lane_f32(W, V.val[0], r, 1); // Dx+Hy+Lz+P

XM_PREFETCH(pInputVector + (XM_CACHE_LINE_SIZE * 5));

Expand Down Expand Up @@ -10912,9 +10912,9 @@ _Use_decl_annotations_ inline XMFLOAT3* XM_CALLCONV XMVector3ProjectStream(XMFLO
float32x2_t VH = vld1_lane_f32(reinterpret_cast<const float*>(pInputVector) + 2, zero, 0);
pInputVector += InputStride;

XMVECTOR vResult = vmlaq_lane_f32(Transform.r[3], Transform.r[0], VL, 0); // X
XMVECTOR vResult = vmlaq_lane_f32(Transform.r[3], Transform.r[2], VH, 0); // Z
vResult = vmlaq_lane_f32(vResult, Transform.r[1], VL, 1); // Y
vResult = vmlaq_lane_f32(vResult, Transform.r[2], VH, 0); // Z
vResult = vmlaq_lane_f32(vResult, Transform.r[0], VL, 0); // X

VH = vget_high_f32(vResult);
XMVECTOR W = vdupq_lane_f32(VH, 1);
Expand Down
Loading