[Fix, Test] (DirectVulkan, ShaderTranspiler, TraceReplay): repair iterationRP's missing reduction barrier

Program 203 reuses prefixSumCache for a second subgroup reduction before every workgroup invocation has consumed the first result. Add a fingerprint-gated SPIR-V pass that inserts the missing Workgroup acquire-release barrier while preserving native subgroup operations.

Keep the repair opt-in behind MOBILEGL_ITERATIONRP_FIX_BARRIER, cover insertion, pass-through, and idempotence, and enable it together with the existing iterationRP subgroup repairs for the matching Linux and Android CI retraces.
This commit is contained in:
2026-08-19 23:42:17 -04:00
parent 5bd8ef01e5
commit c09045fe59
20 changed files with 583 additions and 2 deletions
@@ -164,6 +164,21 @@ bool LoadMobileGL(const Request& request, std::string& error) {
} else {
unsetenv("MOBILEGL_COHERENT_AS_FLUSH");
}
if (request.fixIterationRPSubgroupScratch) {
setenv("MOBILEGL_FIX_ITERATIONRP_SUBGROUP_SCRATCH", "1", 1);
} else {
unsetenv("MOBILEGL_FIX_ITERATIONRP_SUBGROUP_SCRATCH");
}
if (request.deriveNumSubgroups) {
setenv("MOBILEGL_DERIVE_NUM_SUBGROUPS", "1", 1);
} else {
unsetenv("MOBILEGL_DERIVE_NUM_SUBGROUPS");
}
if (request.iterationRPFixBarrier) {
setenv("MOBILEGL_ITERATIONRP_FIX_BARRIER", "1", 1);
} else {
unsetenv("MOBILEGL_ITERATIONRP_FIX_BARRIER");
}
if (request.fboAttachmentDumps.empty()) {
unsetenv("MOBILEGL_TRACE_DUMP_FBO_ATTACHMENTS");
} else {
@@ -818,6 +833,10 @@ bool WriteResultJson(const Request& request, const Result& result) {
<< (request.avoidAngleLlvmpipeSamplerMipmapMinFilter ? "true" : "false") << ",\n";
file << " \"avoidAngleLlvmpipeExplicitLodBias\": "
<< (request.avoidAngleLlvmpipeExplicitLodBias ? "true" : "false") << ",\n";
file << " \"fixIterationRPSubgroupScratch\": " << (request.fixIterationRPSubgroupScratch ? "true" : "false")
<< ",\n";
file << " \"deriveNumSubgroups\": " << (request.deriveNumSubgroups ? "true" : "false") << ",\n";
file << " \"iterationRPFixBarrier\": " << (request.iterationRPFixBarrier ? "true" : "false") << ",\n";
file << " \"holdMs\": " << request.holdMs << ",\n";
file << " \"mismatchPixels\": " << result.mismatchPixels << "\n";
file << "}\n";
@@ -44,6 +44,9 @@ struct Request {
bool avoidAngleLlvmpipeSamplerMipmapMinFilter = false;
bool avoidAngleLlvmpipeExplicitLodBias = false;
bool coherentAsFlush = false;
bool fixIterationRPSubgroupScratch = false;
bool deriveNumSubgroups = false;
bool iterationRPFixBarrier = false;
int holdMs = 0;
};
@@ -122,6 +122,9 @@ Java_top_mobilegl_plugin_trace_TraceReplayActivity_nativeRunTraceReplay(JNIEnv*
jboolean avoidAngleLlvmpipeSamplerMipmapMinFilter,
jboolean avoidAngleLlvmpipeExplicitLodBias,
jboolean coherentAsFlush,
jboolean fixIterationRPSubgroupScratch,
jboolean deriveNumSubgroups,
jboolean iterationRPFixBarrier,
jstring texture2dDumps) {
mobilegl_trace::Request request;
request.tracePath = ToString(env, tracePath);
@@ -150,6 +153,9 @@ Java_top_mobilegl_plugin_trace_TraceReplayActivity_nativeRunTraceReplay(JNIEnv*
avoidAngleLlvmpipeSamplerMipmapMinFilter == JNI_TRUE;
request.avoidAngleLlvmpipeExplicitLodBias = avoidAngleLlvmpipeExplicitLodBias == JNI_TRUE;
request.coherentAsFlush = coherentAsFlush == JNI_TRUE;
request.fixIterationRPSubgroupScratch = fixIterationRPSubgroupScratch == JNI_TRUE;
request.deriveNumSubgroups = deriveNumSubgroups == JNI_TRUE;
request.iterationRPFixBarrier = iterationRPFixBarrier == JNI_TRUE;
ScopedTraceReplayState replayState;
mobilegl_trace_set_requested_size(request.width, request.height);