Skip to content

GPU: more constexpr cleanups to support Metal #15830

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
ktf merged 2 commits into from
Sep 22, 2026
+16 −16
Merged

GPU: more constexpr cleanups to support Metal #15830

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
1827d2d
TRD: more program-scope variables fixes
ktf Sep 22, 2026
59e0aa1
GPUTracking: drop static from the function-scope constants
ktf Sep 22, 2026
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
4 changes: 2 additions & 2 deletions Detectors/TRD/base/include/TRDBase/PadPlane.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -197,8 +197,8 @@ class PadPlane
GPUd() double getAnodeWireOffset() const { return mAnodeWireOffset; };

private:
static constexpr int MAXCOLS = 144;
static constexpr int MAXROWS = 16;
static GPUglobalconstexpr() int MAXCOLS = 144;
static GPUglobalconstexpr() int MAXROWS = 16;

int mLayer; // Layer number
int mStack; // Stack number
Expand Down
4 changes: 2 additions & 2 deletions GPU/GPUTracking/Merger/GPUTPCGMMerger.cxx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1480,8 +1480,8 @@ struct GPUTPCGMMerger_CompareClusterIds {

GPUd() void GPUTPCGMMerger::CollectMergedTracks(int32_t nBlocks, int32_t nThreads, int32_t iBlock, int32_t iThread)
{
static constexpr int32_t kMaxParts = 16;
static constexpr int32_t kMaxClusters = constants::MERGER_MAX_TRACK_CLUSTERS;
constexpr int32_t kMaxParts = 16;
constexpr int32_t kMaxClusters = constants::MERGER_MAX_TRACK_CLUSTERS;

GPUTPCGMSectorTrack* trackParts[kMaxParts];

Expand Down
12 changes: 6 additions & 6 deletions GPU/GPUTracking/Merger/GPUTPCGMTrackParam.cxx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -49,7 +49,7 @@ using namespace o2::tpc;

GPUd() bool GPUTPCGMTrackParam::Fit(GPUTPCGMMerger* GPUrestrict() merger, int32_t iTrk, GPUTPCGMMergedTrackHit* GPUrestrict() clusters, int32_t& GPUrestrict() N, int32_t& GPUrestrict() NTolerated, float& GPUrestrict() Alpha, int32_t attempt, float maxSinPhi, GPUTPCGMMergedTrack& GPUrestrict() track)
{
static constexpr float kDeg2Rad = M_PI / 180.f;
constexpr float kDeg2Rad = M_PI / 180.f;
CADEBUG(static constexpr float kSectAngle = 2 * M_PI / 18.f);

const GPUParam& GPUrestrict() param = merger->Param();
Expand Down Expand Up @@ -366,8 +366,8 @@ GPUd() bool GPUTPCGMTrackParam::Fit(GPUTPCGMMerger* GPUrestrict() merger, int32_

GPUdni() void GPUTPCGMTrackParam::MoveToReference(GPUTPCGMPropagator& prop, const GPUParam& param, float& Alpha)
{
static constexpr float kDeg2Rad = M_PI / 180.f;
static constexpr float kSectAngle = 2 * M_PI / 18.f;
constexpr float kDeg2Rad = M_PI / 180.f;
constexpr float kSectAngle = 2 * M_PI / 18.f;

if (param.rec.tpc.trackReferenceX <= 500) {
GPUTPCGMTrackParam save = *this;
Expand Down Expand Up @@ -575,7 +575,7 @@ GPUd() float GPUTPCGMTrackParam::AttachClusters(const GPUTPCGMMerger* GPUrestric

GPUd() bool GPUTPCGMTrackParam::AttachClustersPropagate(const GPUTPCGMMerger* GPUrestrict() Merger, int32_t sector, int32_t lastRow, int32_t toRow, int32_t iTrack, bool goodLeg, GPUTPCGMPropagator& GPUrestrict() prop, bool inFlyDirection, float maxSinPhi, bool dodEdx)
{
static constexpr float kSectAngle = 2 * M_PI / 18.f;
constexpr float kSectAngle = 2 * M_PI / 18.f;
if (Merger->Param().rec.tpc.disableRefitAttachment & 2) {
return dodEdx;
}
Expand Down Expand Up @@ -678,7 +678,7 @@ GPUdi() void GPUTPCGMTrackParam::AttachClustersLooperFollow(const GPUTPCGMMerger
float toX = mX;
bool inFlyDirection = (Merger->MergedTracks()[iTrack].Leg() & 1) ^ up;

static constexpr float kSectAngle = 2 * M_PI / 18.f;
constexpr float kSectAngle = 2 * M_PI / 18.f;
const GPUParam& GPUrestrict() param = Merger->Param();
bool right = (mP[2] < 0) ^ up;
const int32_t sectorSide = sector >= (int32_t)(GPUTPCGeometry::NSECTORS / 2) ? (GPUTPCGeometry::NSECTORS / 2) : 0;
Expand Down Expand Up @@ -744,7 +744,7 @@ GPUdi() void GPUTPCGMTrackParam::AttachClustersLooperFollow(const GPUTPCGMMerger

GPUdi() void GPUTPCGMTrackParam::AttachClustersLooper(const GPUTPCGMMerger* GPUrestrict() Merger, int32_t sector, int32_t iRow, int32_t iTrack, bool outwards, GPUTPCGMPropagator& GPUrestrict() prop)
{
static constexpr float kSectAngle = 2 * M_PI / 18.f;
constexpr float kSectAngle = 2 * M_PI / 18.f;
// Note that the coordinate system is rotated by 90 degree swapping X and Y!
float X = mP[2] > 0 ? mP[0] : -mP[0];
float Y = mP[2] > 0 ? -mX : mX;
Expand Down
4 changes: 2 additions & 2 deletions GPU/GPUTracking/Refit/GPUTrackingRefit.cxx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -398,8 +398,8 @@ GPUd() int32_t GPUTrackingRefit::RefitTrack(T& trkX, bool outward, bool resetCov
trk.NormalizeAlpha(alpha);
prop.SetAlpha(alpha);
} else if constexpr (std::is_same_v<S, TrackParCov>) {
static constexpr float kDeg2Rad = M_PI / 180.f;
static constexpr float kSectAngle = 2 * M_PI / 18.f;
constexpr float kDeg2Rad = M_PI / 180.f;
constexpr float kSectAngle = 2 * M_PI / 18.f;
if (mPparam->rec.tpc.trackReferenceX <= 500) {
if (prop->PropagateToXBxByBz(trk, mPparam->rec.tpc.trackReferenceX)) {
if (CAMath::Abs(trk.getY()) > trk.getX() * CAMath::Tan(kSectAngle / 2.f)) {
Expand Down
8 changes: 4 additions & 4 deletions GPU/GPUTracking/SectorTracker/GPUTPCNeighboursFinder.cxx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -73,11 +73,11 @@ GPUdii() void GPUTPCNeighboursFinder::Thread<0>(int32_t /*nBlocks*/, int32_t nTh
return;
}

static constexpr uint32_t UNROLL_GLOBAL = GPUCA_PAR_NEIGHBOURS_FINDER_UNROLL_GLOBAL > 1 ? GPUCA_PAR_NEIGHBOURS_FINDER_UNROLL_GLOBAL : 1;
constexpr uint32_t UNROLL_GLOBAL = GPUCA_PAR_NEIGHBOURS_FINDER_UNROLL_GLOBAL > 1 ? GPUCA_PAR_NEIGHBOURS_FINDER_UNROLL_GLOBAL : 1;
static_assert(constants::NEIGHBOURS_MAX_N % UNROLL_GLOBAL == 0);
static constexpr uint32_t MAX_SHARED = GPUCA_PAR_NEIGHBOURS_FINDER_MAX_NNEIGHUP;
static constexpr uint32_t MAX_GLOBAL = (MAX_SHARED < constants::NEIGHBOURS_MAX_N) ? (((constants::NEIGHBOURS_MAX_N - MAX_SHARED - 1) / UNROLL_GLOBAL + 1) * UNROLL_GLOBAL) : 0;
static constexpr uint32_t MAX_TOTAL = MAX_SHARED + MAX_GLOBAL;
constexpr uint32_t MAX_SHARED = GPUCA_PAR_NEIGHBOURS_FINDER_MAX_NNEIGHUP;
constexpr uint32_t MAX_GLOBAL = (MAX_SHARED < constants::NEIGHBOURS_MAX_N) ? (((constants::NEIGHBOURS_MAX_N - MAX_SHARED - 1) / UNROLL_GLOBAL + 1) * UNROLL_GLOBAL) : 0;
constexpr uint32_t MAX_TOTAL = MAX_SHARED + MAX_GLOBAL;

const float chi2Cut = 3.f * 3.f * 4 * (s.mUpDx * s.mUpDx + s.mDnDx * s.mDnDx);
// float chi2Cut = 3.f*3.f*(s.mUpDx*s.mUpDx + s.mDnDx*s.mDnDx ); //SG
Expand Down
Loading