diff --git a/services/inputflinger/dispatcher/Android.bp b/services/inputflinger/dispatcher/Android.bp
index 1a0ec48..8b2b843 100644
--- a/services/inputflinger/dispatcher/Android.bp
+++ b/services/inputflinger/dispatcher/Android.bp
@@ -46,6 +46,7 @@
         "InputState.cpp",
         "InputTarget.cpp",
         "LatencyAggregator.cpp",
+        "LatencyAggregatorWithHistograms.cpp",
         "LatencyTracker.cpp",
         "Monitor.cpp",
         "TouchedWindow.cpp",
diff --git a/services/inputflinger/dispatcher/InputDispatcher.cpp b/services/inputflinger/dispatcher/InputDispatcher.cpp
index 250e72c..da3ac20 100644
--- a/services/inputflinger/dispatcher/InputDispatcher.cpp
+++ b/services/inputflinger/dispatcher/InputDispatcher.cpp
@@ -155,6 +155,10 @@
 // Number of recent events to keep for debugging purposes.
 constexpr size_t RECENT_QUEUE_MAX_SIZE = 10;
 
+// Interval at which we should push the atom gathering input event latencies in
+// LatencyAggregatorWithHistograms
+constexpr nsecs_t LATENCY_STATISTICS_PUSH_INTERVAL = 6 * 3600 * 1000000000LL; // 6 hours
+
 // Event log tags. See EventLogTags.logtags for reference.
 constexpr int LOGTAG_INPUT_INTERACTION = 62000;
 constexpr int LOGTAG_INPUT_FOCUS = 62001;
@@ -944,8 +948,13 @@
         mFocusedDisplayId(ui::LogicalDisplayId::DEFAULT),
         mWindowTokenWithPointerCapture(nullptr),
         mAwaitedApplicationDisplayId(ui::LogicalDisplayId::INVALID),
-        mLatencyAggregator(),
-        mLatencyTracker(&mLatencyAggregator) {
+        mInputEventTimelineProcessor(
+                input_flags::enable_per_device_input_latency_metrics()
+                        ? std::move(std::unique_ptr<InputEventTimelineProcessor>(
+                                  new LatencyAggregatorWithHistograms()))
+                        : std::move(std::unique_ptr<InputEventTimelineProcessor>(
+                                  new LatencyAggregator()))),
+        mLatencyTracker(*mInputEventTimelineProcessor) {
     mLooper = sp<Looper>::make(false);
     mReporter = createInputReporter();
 
@@ -1017,6 +1026,11 @@
         const nsecs_t nextAnrCheck = processAnrsLocked();
         nextWakeupTime = std::min(nextWakeupTime, nextAnrCheck);
 
+        if (input_flags::enable_per_device_input_latency_metrics()) {
+            const nsecs_t nextStatisticsPush = processLatencyStatisticsLocked();
+            nextWakeupTime = std::min(nextWakeupTime, nextStatisticsPush);
+        }
+
         // We are about to enter an infinitely long sleep, because we have no commands or
         // pending or queued events
         if (nextWakeupTime == LLONG_MAX) {
@@ -1097,6 +1111,21 @@
     return LLONG_MIN;
 }
 
+/**
+ * Check if enough time has passed since the last latency statistics push.
+ * Return the time at which we should wake up next.
+ */
+nsecs_t InputDispatcher::processLatencyStatisticsLocked() {
+    const nsecs_t currentTime = now();
+    // Log the atom recording latency statistics if more than 6 hours passed from the last
+    // push
+    if (currentTime - mLastStatisticPushTime >= LATENCY_STATISTICS_PUSH_INTERVAL) {
+        mInputEventTimelineProcessor->pushLatencyStatistics();
+        mLastStatisticPushTime = currentTime;
+    }
+    return mLastStatisticPushTime + LATENCY_STATISTICS_PUSH_INTERVAL;
+}
+
 std::chrono::nanoseconds InputDispatcher::getDispatchingTimeoutLocked(
         const std::shared_ptr<Connection>& connection) {
     if (connection->monitor) {
@@ -6055,7 +6084,7 @@
     dump += StringPrintf(INDENT2 "KeyRepeatTimeout: %" PRId64 "ms\n",
                          ns2ms(mConfig.keyRepeatTimeout));
     dump += mLatencyTracker.dump(INDENT2);
-    dump += mLatencyAggregator.dump(INDENT2);
+    dump += mInputEventTimelineProcessor->dump(INDENT2);
     dump += INDENT "InputTracer: ";
     dump += mTracer == nullptr ? "Disabled" : "Enabled";
 }
diff --git a/services/inputflinger/dispatcher/InputDispatcher.h b/services/inputflinger/dispatcher/InputDispatcher.h
index 87dfd1d..7060efa 100644
--- a/services/inputflinger/dispatcher/InputDispatcher.h
+++ b/services/inputflinger/dispatcher/InputDispatcher.h
@@ -28,6 +28,7 @@
 #include "InputTarget.h"
 #include "InputThread.h"
 #include "LatencyAggregator.h"
+#include "LatencyAggregatorWithHistograms.h"
 #include "LatencyTracker.h"
 #include "Monitor.h"
 #include "TouchState.h"
@@ -326,6 +327,7 @@
     std::chrono::nanoseconds mMonitorDispatchingTimeout GUARDED_BY(mLock);
 
     nsecs_t processAnrsLocked() REQUIRES(mLock);
+    nsecs_t processLatencyStatisticsLocked() REQUIRES(mLock);
     std::chrono::nanoseconds getDispatchingTimeoutLocked(
             const std::shared_ptr<Connection>& connection) REQUIRES(mLock);
 
@@ -697,7 +699,8 @@
                                          DeviceId deviceId) const REQUIRES(mLock);
 
     // Statistics gathering.
-    LatencyAggregator mLatencyAggregator GUARDED_BY(mLock);
+    nsecs_t mLastStatisticPushTime = 0;
+    std::unique_ptr<InputEventTimelineProcessor> mInputEventTimelineProcessor GUARDED_BY(mLock);
     LatencyTracker mLatencyTracker GUARDED_BY(mLock);
     void traceInboundQueueLengthLocked() REQUIRES(mLock);
     void traceOutboundQueueLength(const Connection& connection);
diff --git a/services/inputflinger/dispatcher/InputEventTimeline.h b/services/inputflinger/dispatcher/InputEventTimeline.h
index 951fcc8..4552708 100644
--- a/services/inputflinger/dispatcher/InputEventTimeline.h
+++ b/services/inputflinger/dispatcher/InputEventTimeline.h
@@ -121,13 +121,21 @@
 class InputEventTimelineProcessor {
 protected:
     InputEventTimelineProcessor() {}
-    virtual ~InputEventTimelineProcessor() {}
 
 public:
+    virtual ~InputEventTimelineProcessor() {}
+
     /**
      * Process the provided timeline
      */
     virtual void processTimeline(const InputEventTimeline& timeline) = 0;
+
+    /**
+     * Trigger a push for the input event latency statistics
+     */
+    virtual void pushLatencyStatistics() = 0;
+
+    virtual std::string dump(const char* prefix) const = 0;
 };
 
 } // namespace inputdispatcher
diff --git a/services/inputflinger/dispatcher/LatencyAggregator.cpp b/services/inputflinger/dispatcher/LatencyAggregator.cpp
index 4ddd2e9..d0e9d7c 100644
--- a/services/inputflinger/dispatcher/LatencyAggregator.cpp
+++ b/services/inputflinger/dispatcher/LatencyAggregator.cpp
@@ -28,6 +28,8 @@
 using dist_proc::aggregation::KllQuantile;
 using std::chrono_literals::operator""ms;
 
+namespace {
+
 // Convert the provided nanoseconds into hundreds of microseconds.
 // Use hundreds of microseconds (as opposed to microseconds) to preserve space.
 static inline int64_t ns2hus(nsecs_t nanos) {
@@ -74,6 +76,8 @@
     return std::chrono::milliseconds(std::stoi(millis));
 }
 
+} // namespace
+
 namespace android::inputdispatcher {
 
 /**
@@ -125,6 +129,9 @@
     processSlowEvent(timeline);
 }
 
+// This version of LatencyAggregator doesn't push any atoms
+void LatencyAggregator::pushLatencyStatistics() {}
+
 void LatencyAggregator::processStatistics(const InputEventTimeline& timeline) {
     std::scoped_lock lock(mLock);
     // Before we do any processing, check that we have not yet exceeded MAX_SIZE
diff --git a/services/inputflinger/dispatcher/LatencyAggregator.h b/services/inputflinger/dispatcher/LatencyAggregator.h
index d6d1686..468add1 100644
--- a/services/inputflinger/dispatcher/LatencyAggregator.h
+++ b/services/inputflinger/dispatcher/LatencyAggregator.h
@@ -57,6 +57,8 @@
      */
     void processTimeline(const InputEventTimeline& timeline) override;
 
+    void pushLatencyStatistics() override;
+
     std::string dump(const char* prefix) const;
 
     ~LatencyAggregator();
diff --git a/services/inputflinger/dispatcher/LatencyAggregatorWithHistograms.cpp b/services/inputflinger/dispatcher/LatencyAggregatorWithHistograms.cpp
new file mode 100644
index 0000000..881a96b
--- /dev/null
+++ b/services/inputflinger/dispatcher/LatencyAggregatorWithHistograms.cpp
@@ -0,0 +1,345 @@
+/*
+ * Copyright 2024 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#define LOG_TAG "LatencyAggregatorWithHistograms"
+#include "../InputDeviceMetricsSource.h"
+#include "InputDispatcher.h"
+
+#include <inttypes.h>
+#include <log/log_event_list.h>
+#include <statslog.h>
+
+#include <android-base/logging.h>
+#include <android-base/stringprintf.h>
+#include <input/Input.h>
+#include <log/log.h>
+#include <server_configurable_flags/get_flags.h>
+
+using android::base::StringPrintf;
+using std::chrono_literals::operator""ms;
+
+namespace {
+
+// Convert the provided nanoseconds into hundreds of microseconds.
+// Use hundreds of microseconds (as opposed to microseconds) to preserve space.
+static inline int64_t ns2hus(nsecs_t nanos) {
+    return ns2us(nanos) / 100;
+}
+
+// Category (=namespace) name for the input settings that are applied at boot time
+static const char* INPUT_NATIVE_BOOT = "input_native_boot";
+// Feature flag name for the threshold of end-to-end touch latency that would trigger
+// SlowEventReported atom to be pushed
+static const char* SLOW_EVENT_MIN_REPORTING_LATENCY_MILLIS =
+        "slow_event_min_reporting_latency_millis";
+// Feature flag name for the minimum delay before reporting a slow event after having just reported
+// a slow event. This helps limit the amount of data sent to the server
+static const char* SLOW_EVENT_MIN_REPORTING_INTERVAL_MILLIS =
+        "slow_event_min_reporting_interval_millis";
+
+// If an event has end-to-end latency > 200 ms, it will get reported as a slow event.
+std::chrono::milliseconds DEFAULT_SLOW_EVENT_MIN_REPORTING_LATENCY = 200ms;
+// If we receive two slow events less than 1 min apart, we will only report 1 of them.
+std::chrono::milliseconds DEFAULT_SLOW_EVENT_MIN_REPORTING_INTERVAL = 60000ms;
+
+static std::chrono::milliseconds getSlowEventMinReportingLatency() {
+    std::string millis = server_configurable_flags::
+            GetServerConfigurableFlag(INPUT_NATIVE_BOOT, SLOW_EVENT_MIN_REPORTING_LATENCY_MILLIS,
+                                      std::to_string(
+                                              DEFAULT_SLOW_EVENT_MIN_REPORTING_LATENCY.count()));
+    return std::chrono::milliseconds(std::stoi(millis));
+}
+
+static std::chrono::milliseconds getSlowEventMinReportingInterval() {
+    std::string millis = server_configurable_flags::
+            GetServerConfigurableFlag(INPUT_NATIVE_BOOT, SLOW_EVENT_MIN_REPORTING_INTERVAL_MILLIS,
+                                      std::to_string(
+                                              DEFAULT_SLOW_EVENT_MIN_REPORTING_INTERVAL.count()));
+    return std::chrono::milliseconds(std::stoi(millis));
+}
+
+} // namespace
+
+namespace android::inputdispatcher {
+
+int32_t LatencyStageIndexToAtomEnum(LatencyStageIndex latencyStageIndex) {
+    switch (latencyStageIndex) {
+        case LatencyStageIndex::EVENT_TO_READ:
+            return util::INPUT_EVENT_LATENCY_REPORTED__LATENCY_STAGE__EVENT_TO_READ;
+        case LatencyStageIndex::READ_TO_DELIVER:
+            return util::INPUT_EVENT_LATENCY_REPORTED__LATENCY_STAGE__READ_TO_DELIVER;
+        case LatencyStageIndex::DELIVER_TO_CONSUME:
+            return util::INPUT_EVENT_LATENCY_REPORTED__LATENCY_STAGE__DELIVER_TO_CONSUME;
+        case LatencyStageIndex::CONSUME_TO_FINISH:
+            return util::INPUT_EVENT_LATENCY_REPORTED__LATENCY_STAGE__CONSUME_TO_FINISH;
+        case LatencyStageIndex::CONSUME_TO_GPU_COMPLETE:
+            return util::INPUT_EVENT_LATENCY_REPORTED__LATENCY_STAGE__CONSUME_TO_GPU_COMPLETE;
+        case LatencyStageIndex::GPU_COMPLETE_TO_PRESENT:
+            return util::INPUT_EVENT_LATENCY_REPORTED__LATENCY_STAGE__GPU_COMPLETE_TO_PRESENT;
+        case LatencyStageIndex::END_TO_END:
+            return util::INPUT_EVENT_LATENCY_REPORTED__LATENCY_STAGE__END_TO_END;
+        default:
+            return util::INPUT_EVENT_LATENCY_REPORTED__LATENCY_STAGE__UNKNOWN_LATENCY_STAGE;
+    }
+}
+
+int32_t InputEventTypeEnumToAtomEnum(InputEventActionType inputEventActionType) {
+    switch (inputEventActionType) {
+        case InputEventActionType::UNKNOWN_INPUT_EVENT:
+            return util::INPUT_EVENT_LATENCY_REPORTED__INPUT_EVENT_TYPE__UNKNOWN_INPUT_EVENT;
+        case InputEventActionType::MOTION_ACTION_DOWN:
+            return util::INPUT_EVENT_LATENCY_REPORTED__INPUT_EVENT_TYPE__MOTION_ACTION_DOWN;
+        case InputEventActionType::MOTION_ACTION_MOVE:
+            return util::INPUT_EVENT_LATENCY_REPORTED__INPUT_EVENT_TYPE__MOTION_ACTION_MOVE;
+        case InputEventActionType::MOTION_ACTION_UP:
+            return util::INPUT_EVENT_LATENCY_REPORTED__INPUT_EVENT_TYPE__MOTION_ACTION_UP;
+        case InputEventActionType::MOTION_ACTION_HOVER_MOVE:
+            return util::INPUT_EVENT_LATENCY_REPORTED__INPUT_EVENT_TYPE__MOTION_ACTION_HOVER_MOVE;
+        case InputEventActionType::MOTION_ACTION_SCROLL:
+            return util::INPUT_EVENT_LATENCY_REPORTED__INPUT_EVENT_TYPE__MOTION_ACTION_SCROLL;
+        case InputEventActionType::KEY:
+            return util::INPUT_EVENT_LATENCY_REPORTED__INPUT_EVENT_TYPE__KEY;
+    }
+}
+
+void LatencyAggregatorWithHistograms::processTimeline(const InputEventTimeline& timeline) {
+    processStatistics(timeline);
+    processSlowEvent(timeline);
+}
+
+void LatencyAggregatorWithHistograms::addSampleToHistogram(
+        const InputEventLatencyIdentifier& identifier, LatencyStageIndex latencyStageIndex,
+        nsecs_t latency) {
+    // Only record positive values for the statistics
+    if (latency > 0) {
+        auto it = mHistograms.find(identifier);
+        if (it != mHistograms.end()) {
+            it->second[static_cast<size_t>(latencyStageIndex)].addSample(ns2hus(latency));
+        }
+    }
+}
+
+void LatencyAggregatorWithHistograms::processStatistics(const InputEventTimeline& timeline) {
+    // Only gather data for Down, Move and Up motion events and Key events
+    if (!(timeline.inputEventActionType == InputEventActionType::MOTION_ACTION_DOWN ||
+          timeline.inputEventActionType == InputEventActionType::MOTION_ACTION_MOVE ||
+          timeline.inputEventActionType == InputEventActionType::MOTION_ACTION_UP ||
+          timeline.inputEventActionType == InputEventActionType::KEY))
+        return;
+
+    // Don't collect data for unidentified devices. This situation can occur for the first few input
+    // events produced when an input device is first connected
+    if (timeline.vendorId == 0xFFFF && timeline.productId == 0xFFFF) return;
+
+    InputEventLatencyIdentifier identifier = {timeline.vendorId, timeline.productId,
+                                              timeline.sources, timeline.inputEventActionType};
+    // Check if there's a value in mHistograms map associated to identifier.
+    // If not, add an array with 7 empty histograms as an entry
+    if (mHistograms.count(identifier) == 0) {
+        if (static_cast<int32_t>(timeline.inputEventActionType) - 1 < 0) {
+            LOG(FATAL) << "Action index is smaller than 0. Action type: "
+                       << ftl::enum_string(timeline.inputEventActionType);
+            return;
+        }
+        size_t actionIndex =
+                static_cast<size_t>(static_cast<int32_t>(timeline.inputEventActionType) - 1);
+        if (actionIndex >= NUM_INPUT_EVENT_TYPES) {
+            LOG(FATAL) << "Action index greater than the number of input event types. Action Type: "
+                       << ftl::enum_string(timeline.inputEventActionType)
+                       << "; Action Type Index: " << actionIndex;
+            return;
+        }
+
+        std::array<Histogram, 7> histograms =
+                {Histogram(allBinSizes[binSizesMappings[0][actionIndex]]),
+                 Histogram(allBinSizes[binSizesMappings[1][actionIndex]]),
+                 Histogram(allBinSizes[binSizesMappings[2][actionIndex]]),
+                 Histogram(allBinSizes[binSizesMappings[3][actionIndex]]),
+                 Histogram(allBinSizes[binSizesMappings[4][actionIndex]]),
+                 Histogram(allBinSizes[binSizesMappings[5][actionIndex]]),
+                 Histogram(allBinSizes[binSizesMappings[6][actionIndex]])};
+        mHistograms.insert({identifier, histograms});
+    }
+
+    // Process common ones first
+    const nsecs_t eventToRead = timeline.readTime - timeline.eventTime;
+    addSampleToHistogram(identifier, LatencyStageIndex::EVENT_TO_READ, eventToRead);
+
+    // Now process per-connection ones
+    for (const auto& [connectionToken, connectionTimeline] : timeline.connectionTimelines) {
+        if (!connectionTimeline.isComplete()) {
+            continue;
+        }
+        const nsecs_t readToDeliver = connectionTimeline.deliveryTime - timeline.readTime;
+        const nsecs_t deliverToConsume =
+                connectionTimeline.consumeTime - connectionTimeline.deliveryTime;
+        const nsecs_t consumeToFinish =
+                connectionTimeline.finishTime - connectionTimeline.consumeTime;
+        const nsecs_t gpuCompletedTime =
+                connectionTimeline.graphicsTimeline[GraphicsTimeline::GPU_COMPLETED_TIME];
+        const nsecs_t presentTime =
+                connectionTimeline.graphicsTimeline[GraphicsTimeline::PRESENT_TIME];
+        const nsecs_t consumeToGpuComplete = gpuCompletedTime - connectionTimeline.consumeTime;
+        const nsecs_t gpuCompleteToPresent = presentTime - gpuCompletedTime;
+        const nsecs_t endToEnd = presentTime - timeline.eventTime;
+
+        addSampleToHistogram(identifier, LatencyStageIndex::READ_TO_DELIVER, readToDeliver);
+        addSampleToHistogram(identifier, LatencyStageIndex::DELIVER_TO_CONSUME, deliverToConsume);
+        addSampleToHistogram(identifier, LatencyStageIndex::CONSUME_TO_FINISH, consumeToFinish);
+        addSampleToHistogram(identifier, LatencyStageIndex::CONSUME_TO_GPU_COMPLETE,
+                             consumeToGpuComplete);
+        addSampleToHistogram(identifier, LatencyStageIndex::GPU_COMPLETE_TO_PRESENT,
+                             gpuCompleteToPresent);
+        addSampleToHistogram(identifier, LatencyStageIndex::END_TO_END, endToEnd);
+    }
+}
+
+void LatencyAggregatorWithHistograms::pushLatencyStatistics() {
+    for (auto& [id, histograms] : mHistograms) {
+        auto [vendorId, productId, sources, action] = id;
+        for (size_t latencyStageIndex = static_cast<size_t>(LatencyStageIndex::EVENT_TO_READ);
+             latencyStageIndex < static_cast<size_t>(LatencyStageIndex::SIZE);
+             ++latencyStageIndex) {
+            // Convert sources set to vector for atom logging:
+            std::vector<int32_t> sourcesVector = {};
+            for (auto& elem : sources) {
+                sourcesVector.push_back(static_cast<int32_t>(elem));
+            }
+
+            // convert histogram bin counts array to vector for atom logging:
+            std::array arr = histograms[latencyStageIndex].getBinCounts();
+            std::vector<int32_t> binCountsVector(arr.begin(), arr.end());
+
+            if (static_cast<int32_t>(action) - 1 < 0) {
+                ALOGW("Action index is smaller than 0. Action type: %s",
+                      ftl::enum_string(action).c_str());
+                continue;
+            }
+            size_t actionIndex = static_cast<size_t>(static_cast<int32_t>(action) - 1);
+            if (actionIndex >= NUM_INPUT_EVENT_TYPES) {
+                ALOGW("Action index greater than the number of input event types. Action Type: %s; "
+                      "Action Type Index: %zu",
+                      ftl::enum_string(action).c_str(), actionIndex);
+                continue;
+            }
+
+            stats_write(android::util::INPUT_EVENT_LATENCY_REPORTED, vendorId, productId,
+                        sourcesVector, InputEventTypeEnumToAtomEnum(action),
+                        LatencyStageIndexToAtomEnum(
+                                static_cast<LatencyStageIndex>(latencyStageIndex)),
+                        histogramVersions[latencyStageIndex][actionIndex], binCountsVector);
+        }
+    }
+    mHistograms.clear();
+}
+
+// TODO (b/270049345): For now, we just copied the code from LatencyAggregator to populate the old
+// atom, but eventually we should migrate this to use the new SlowEventReported atom
+void LatencyAggregatorWithHistograms::processSlowEvent(const InputEventTimeline& timeline) {
+    static const std::chrono::duration sSlowEventThreshold = getSlowEventMinReportingLatency();
+    static const std::chrono::duration sSlowEventReportingInterval =
+            getSlowEventMinReportingInterval();
+    for (const auto& [token, connectionTimeline] : timeline.connectionTimelines) {
+        if (!connectionTimeline.isComplete()) {
+            continue;
+        }
+        mNumEventsSinceLastSlowEventReport++;
+        const nsecs_t presentTime =
+                connectionTimeline.graphicsTimeline[GraphicsTimeline::PRESENT_TIME];
+        const std::chrono::nanoseconds endToEndLatency =
+                std::chrono::nanoseconds(presentTime - timeline.eventTime);
+        if (endToEndLatency < sSlowEventThreshold) {
+            continue;
+        }
+        // This is a slow event. Before we report it, check if we are reporting too often
+        const std::chrono::duration elapsedSinceLastReport =
+                std::chrono::nanoseconds(timeline.eventTime - mLastSlowEventTime);
+        if (elapsedSinceLastReport < sSlowEventReportingInterval) {
+            mNumSkippedSlowEvents++;
+            continue;
+        }
+
+        const nsecs_t eventToRead = timeline.readTime - timeline.eventTime;
+        const nsecs_t readToDeliver = connectionTimeline.deliveryTime - timeline.readTime;
+        const nsecs_t deliverToConsume =
+                connectionTimeline.consumeTime - connectionTimeline.deliveryTime;
+        const nsecs_t consumeToFinish =
+                connectionTimeline.finishTime - connectionTimeline.consumeTime;
+        const nsecs_t gpuCompletedTime =
+                connectionTimeline.graphicsTimeline[GraphicsTimeline::GPU_COMPLETED_TIME];
+        const nsecs_t consumeToGpuComplete = gpuCompletedTime - connectionTimeline.consumeTime;
+        const nsecs_t gpuCompleteToPresent = presentTime - gpuCompletedTime;
+
+        android::util::stats_write(android::util::SLOW_INPUT_EVENT_REPORTED,
+                                   timeline.inputEventActionType ==
+                                           InputEventActionType::MOTION_ACTION_DOWN,
+                                   static_cast<int32_t>(ns2us(eventToRead)),
+                                   static_cast<int32_t>(ns2us(readToDeliver)),
+                                   static_cast<int32_t>(ns2us(deliverToConsume)),
+                                   static_cast<int32_t>(ns2us(consumeToFinish)),
+                                   static_cast<int32_t>(ns2us(consumeToGpuComplete)),
+                                   static_cast<int32_t>(ns2us(gpuCompleteToPresent)),
+                                   static_cast<int32_t>(ns2us(endToEndLatency.count())),
+                                   static_cast<int32_t>(mNumEventsSinceLastSlowEventReport),
+                                   static_cast<int32_t>(mNumSkippedSlowEvents));
+        mNumEventsSinceLastSlowEventReport = 0;
+        mNumSkippedSlowEvents = 0;
+        mLastSlowEventTime = timeline.readTime;
+    }
+}
+
+std::string LatencyAggregatorWithHistograms::dump(const char* prefix) const {
+    std::string statisticsStr = StringPrintf("%s Histograms:\n", prefix);
+    for (const auto& [id, histograms] : mHistograms) {
+        auto [vendorId, productId, sources, action] = id;
+
+        std::string identifierStr =
+                StringPrintf("%s  Identifier: vendor %d, product %d, sources: {", prefix, vendorId,
+                             productId);
+        bool firstSource = true;
+        for (const auto& source : sources) {
+            if (!firstSource) {
+                identifierStr += ", ";
+            }
+            identifierStr += StringPrintf("%d", static_cast<int32_t>(source));
+            firstSource = false;
+        }
+        identifierStr += StringPrintf("}, action: %d\n", static_cast<int32_t>(action));
+
+        std::string histogramsStr;
+        for (size_t stageIndex = 0; stageIndex < static_cast<size_t>(LatencyStageIndex::SIZE);
+             stageIndex++) {
+            const auto& histogram = histograms[stageIndex];
+            const std::array<int, NUM_BINS>& binCounts = histogram.getBinCounts();
+
+            histogramsStr += StringPrintf("%s   %zu: ", prefix, stageIndex);
+            histogramsStr += StringPrintf("%d", binCounts[0]);
+            for (size_t bin = 1; bin < NUM_BINS; bin++) {
+                histogramsStr += StringPrintf(", %d", binCounts[bin]);
+            }
+            histogramsStr += StringPrintf("\n");
+        }
+        statisticsStr += identifierStr + histogramsStr;
+    }
+
+    return StringPrintf("%sLatencyAggregatorWithHistograms:\n", prefix) + statisticsStr +
+            StringPrintf("%s  mLastSlowEventTime=%" PRId64 "\n", prefix, mLastSlowEventTime) +
+            StringPrintf("%s  mNumEventsSinceLastSlowEventReport = %zu\n", prefix,
+                         mNumEventsSinceLastSlowEventReport) +
+            StringPrintf("%s  mNumSkippedSlowEvents = %zu\n", prefix, mNumSkippedSlowEvents);
+}
+
+} // namespace android::inputdispatcher
diff --git a/services/inputflinger/dispatcher/LatencyAggregatorWithHistograms.h b/services/inputflinger/dispatcher/LatencyAggregatorWithHistograms.h
new file mode 100644
index 0000000..2ceb0e7
--- /dev/null
+++ b/services/inputflinger/dispatcher/LatencyAggregatorWithHistograms.h
@@ -0,0 +1,164 @@
+/*
+ * Copyright 2024 The Android Open Source Project
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#pragma once
+
+#include <android-base/thread_annotations.h>
+#include <utils/Timers.h>
+
+#include "InputEventTimeline.h"
+
+namespace android::inputdispatcher {
+
+static constexpr size_t NUM_BINS = 20;
+static constexpr size_t NUM_INPUT_EVENT_TYPES = 6;
+
+enum class LatencyStageIndex : size_t {
+    EVENT_TO_READ = 0,
+    READ_TO_DELIVER = 1,
+    DELIVER_TO_CONSUME = 2,
+    CONSUME_TO_FINISH = 3,
+    CONSUME_TO_GPU_COMPLETE = 4,
+    GPU_COMPLETE_TO_PRESENT = 5,
+    END_TO_END = 6,
+    SIZE = 7, // must be last
+};
+
+// Let's create a full timeline here:
+// eventTime
+// readTime
+// <---- after this point, the data becomes per-connection
+// deliveryTime // time at which the event was sent to the receiver
+// consumeTime  // time at which the receiver read the event
+// finishTime   // time at which the dispatcher reads the response from the receiver that the event
+// was processed
+// GraphicsTimeline::GPU_COMPLETED_TIME
+// GraphicsTimeline::PRESENT_TIME
+
+/**
+ * Keep histograms with latencies of the provided events
+ */
+class LatencyAggregatorWithHistograms final : public InputEventTimelineProcessor {
+public:
+    /**
+     * Record a complete event timeline
+     */
+    void processTimeline(const InputEventTimeline& timeline) override;
+
+    void pushLatencyStatistics() override;
+
+    std::string dump(const char* prefix) const override;
+
+private:
+    // ---------- Slow event handling ----------
+    void processSlowEvent(const InputEventTimeline& timeline);
+    nsecs_t mLastSlowEventTime = 0;
+    // How many slow events have been skipped due to rate limiting
+    size_t mNumSkippedSlowEvents = 0;
+    // How many events have been received since the last time we reported a slow event
+    size_t mNumEventsSinceLastSlowEventReport = 0;
+
+    // ---------- Statistics handling ----------
+    /**
+     * Data structure to gather time samples into NUM_BINS buckets
+     */
+    class Histogram {
+    public:
+        Histogram(const std::array<int, NUM_BINS - 1>& binSizes) : mBinSizes(binSizes) {
+            mBinCounts.fill(0);
+        }
+
+        // Increments binCounts of the appropriate bin when adding a new sample
+        void addSample(int64_t sample) {
+            size_t binIndex = getSampleBinIndex(sample);
+            mBinCounts[binIndex]++;
+        }
+
+        const std::array<int32_t, NUM_BINS>& getBinCounts() const { return mBinCounts; }
+
+    private:
+        // reference to an array that represents the range of values each bin holds.
+        // in bin i+1 live samples such that *mBinSizes[i] <= sample < *mBinSizes[i+1]
+        const std::array<int, NUM_BINS - 1>& mBinSizes;
+        std::array<int32_t, NUM_BINS>
+                mBinCounts; // the number of samples that currently live in each bin
+
+        size_t getSampleBinIndex(int64_t sample) {
+            auto it = std::upper_bound(mBinSizes.begin(), mBinSizes.end(), sample);
+            return std::distance(mBinSizes.begin(), it);
+        }
+    };
+
+    void processStatistics(const InputEventTimeline& timeline);
+
+    // Identifier for the an input event. If two input events have the same identifiers we
+    // want to use the same histograms to count the latency samples
+    using InputEventLatencyIdentifier =
+            std::tuple<uint16_t /*vendorId*/, uint16_t /*productId*/,
+                       const std::set<InputDeviceUsageSource> /*sources*/,
+                       InputEventActionType /*inputEventActionType*/>;
+
+    // Maps an input event identifier to an array of 7 histograms, one for each latency
+    // stage. It is cleared after an atom push
+    std::map<InputEventLatencyIdentifier, std::array<Histogram, 7>> mHistograms;
+
+    void addSampleToHistogram(const InputEventLatencyIdentifier& identifier,
+                              LatencyStageIndex latencyStageIndex, nsecs_t time);
+
+    // Stores all possible arrays of bin sizes. The order in the vector does not matter, as long
+    // as binSizesMappings points to the right index
+    static constexpr std::array<std::array<int, NUM_BINS - 1>, 6> allBinSizes = {
+            {{10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100},
+             {1, 2, 3, 4, 5, 6, 8, 10, 12, 14, 16, 18, 20, 22, 24, 26, 28, 30, 32},
+             {15, 30, 45, 60, 75, 90, 105, 120, 135, 150, 165, 180, 195, 210, 225, 240, 255, 270,
+              285},
+             {40, 80, 120, 160, 200, 240, 280, 320, 360, 400, 440, 480, 520, 560, 600, 640, 680,
+              720, 760},
+             {20, 40, 60, 80, 100, 120, 140, 160, 180, 200, 220, 240, 260, 280, 300, 320, 340, 360,
+              380},
+             {200, 300, 400, 500, 600, 700, 800, 900, 1000, 1100, 1200, 1300, 1400, 1500, 1600,
+              1700, 1800, 1900, 2000}}};
+
+    // Stores indexes in allBinSizes to use with each {LatencyStage, InputEventType} pair.
+    // Bin sizes for a certain latencyStage and inputEventType are at:
+    // *(allBinSizes[binSizesMappings[latencyStageIndex][inputEventTypeIndex]])
+    // inputEventTypeIndex is the int value of InputEventActionType enum decreased by 1 since we
+    // don't want to record latencies for unknown events.
+    // e.g. MOTION_ACTION_DOWN is 0, MOTION_ACTION_MOVE is 1...
+    static constexpr std::array<std::array<int8_t, NUM_INPUT_EVENT_TYPES>,
+                                static_cast<size_t>(LatencyStageIndex::SIZE)>
+            binSizesMappings = {{{0, 0, 0, 0, 0, 0},
+                                 {1, 1, 1, 1, 1, 1},
+                                 {1, 1, 1, 1, 1, 1},
+                                 {2, 2, 2, 2, 2, 2},
+                                 {3, 3, 3, 3, 3, 3},
+                                 {4, 4, 4, 4, 4, 4},
+                                 {5, 5, 5, 5, 5, 5}}};
+
+    // Similar to binSizesMappings, but holds the index of the array of bin ranges to use on the
+    // server. The index gets pushed with the atom within the histogram_version field.
+    static constexpr std::array<std::array<int8_t, NUM_INPUT_EVENT_TYPES>,
+                                static_cast<size_t>(LatencyStageIndex::SIZE)>
+            histogramVersions = {{{0, 0, 0, 0, 0, 0},
+                                  {1, 1, 1, 1, 1, 1},
+                                  {1, 1, 1, 1, 1, 1},
+                                  {2, 2, 2, 2, 2, 2},
+                                  {3, 3, 3, 3, 3, 3},
+                                  {4, 4, 4, 4, 4, 4},
+                                  {5, 5, 5, 5, 5, 5}}};
+};
+
+} // namespace android::inputdispatcher
diff --git a/services/inputflinger/dispatcher/LatencyTracker.cpp b/services/inputflinger/dispatcher/LatencyTracker.cpp
index 69024b3..d203fba 100644
--- a/services/inputflinger/dispatcher/LatencyTracker.cpp
+++ b/services/inputflinger/dispatcher/LatencyTracker.cpp
@@ -62,10 +62,8 @@
     }
 }
 
-LatencyTracker::LatencyTracker(InputEventTimelineProcessor* processor)
-      : mTimelineProcessor(processor) {
-    LOG_ALWAYS_FATAL_IF(processor == nullptr);
-}
+LatencyTracker::LatencyTracker(InputEventTimelineProcessor& processor)
+      : mTimelineProcessor(&processor) {}
 
 void LatencyTracker::trackListener(int32_t inputEventId, nsecs_t eventTime, nsecs_t readTime,
                                    DeviceId deviceId,
diff --git a/services/inputflinger/dispatcher/LatencyTracker.h b/services/inputflinger/dispatcher/LatencyTracker.h
index b4053ba..80d3f2f 100644
--- a/services/inputflinger/dispatcher/LatencyTracker.h
+++ b/services/inputflinger/dispatcher/LatencyTracker.h
@@ -42,7 +42,7 @@
      * Create a LatencyTracker.
      * param reportingFunction: the function that will be called in order to report full latency.
      */
-    LatencyTracker(InputEventTimelineProcessor* processor);
+    LatencyTracker(InputEventTimelineProcessor& processor);
     /**
      * Start keeping track of an event identified by inputEventId. This must be called first.
      * If duplicate events are encountered (events that have the same eventId), none of them will be
diff --git a/services/inputflinger/tests/LatencyTracker_test.cpp b/services/inputflinger/tests/LatencyTracker_test.cpp
index 0f92833..3f14c23 100644
--- a/services/inputflinger/tests/LatencyTracker_test.cpp
+++ b/services/inputflinger/tests/LatencyTracker_test.cpp
@@ -87,7 +87,7 @@
         connection1 = sp<BBinder>::make();
         connection2 = sp<BBinder>::make();
 
-        mTracker = std::make_unique<LatencyTracker>(this);
+        mTracker = std::make_unique<LatencyTracker>(*this);
         setDefaultInputDeviceInfo(*mTracker);
     }
     void TearDown() override {}
@@ -106,6 +106,8 @@
     void processTimeline(const InputEventTimeline& timeline) override {
         mReceivedTimelines.push_back(timeline);
     }
+    void pushLatencyStatistics() override {}
+    std::string dump(const char* prefix) const { return ""; };
     std::deque<InputEventTimeline> mReceivedTimelines;
 };
 
diff --git a/services/inputflinger/tests/fuzzers/LatencyTrackerFuzzer.cpp b/services/inputflinger/tests/fuzzers/LatencyTrackerFuzzer.cpp
index 695eb3c..908fa40 100644
--- a/services/inputflinger/tests/fuzzers/LatencyTrackerFuzzer.cpp
+++ b/services/inputflinger/tests/fuzzers/LatencyTrackerFuzzer.cpp
@@ -38,6 +38,10 @@
             connectionTimeline.isComplete();
         }
     };
+
+    void pushLatencyStatistics() override {}
+
+    std::string dump(const char* prefix) const { return ""; };
 };
 
 static sp<IBinder> getConnectionToken(FuzzedDataProvider& fdp,
@@ -53,7 +57,7 @@
     FuzzedDataProvider fdp(data, size);
 
     EmptyProcessor emptyProcessor;
-    LatencyTracker tracker(&emptyProcessor);
+    LatencyTracker tracker(emptyProcessor);
 
     // Make some pre-defined tokens to ensure that some timelines are complete.
     std::array<sp<IBinder> /*token*/, 10> predefinedTokens;