diff --git a/include/dwarfs/worker_group.h b/include/dwarfs/worker_group.h
index ec13e261..4dd35fae 100644
--- a/include/dwarfs/worker_group.h
+++ b/include/dwarfs/worker_group.h
@@ -72,6 +72,7 @@ class worker_group {
   bool add_job(job_t&& job) { return impl_->add_job(std::move(job)); }
   size_t size() const { return impl_->size(); }
   size_t queue_size() const { return impl_->queue_size(); }
+  double get_cpu_time() const { return impl_->get_cpu_time(); }
 
   template <typename T>
   bool add_job(std::packaged_task<T()>&& task) {
@@ -88,6 +89,7 @@ class worker_group {
     virtual bool add_job(job_t&& job) = 0;
     virtual size_t size() const = 0;
     virtual size_t queue_size() const = 0;
+    virtual double get_cpu_time() const = 0;
   };
 
  private:
diff --git a/src/dwarfs/scanner.cpp b/src/dwarfs/scanner.cpp
index ca2762f7..aebbc157 100644
--- a/src/dwarfs/scanner.cpp
+++ b/src/dwarfs/scanner.cpp
@@ -613,6 +613,8 @@ void scanner_<LoggerPolicy>::scan(filesystem_writer& fsw,
   LOG_INFO << "waiting for background scanners...";
   wg_.wait();
 
+  LOG_INFO << "scanning CPU time: " << time_with_unit(wg_.get_cpu_time());
+
   LOG_INFO << "finalizing file inodes...";
   uint32_t first_device_inode = first_file_inode;
   fs.finalize(first_device_inode);
@@ -682,6 +684,9 @@ void scanner_<LoggerPolicy>::scan(filesystem_writer& fsw,
 
   blockify.wait();
 
+  LOG_INFO << "segmenting/blockifying CPU time: "
+           << time_with_unit(blockify.get_cpu_time());
+
   bm.finish_blocks();
   wg_.wait();
 
diff --git a/src/dwarfs/worker_group.cpp b/src/dwarfs/worker_group.cpp
index bb8f8990..03b7d4de 100644
--- a/src/dwarfs/worker_group.cpp
+++ b/src/dwarfs/worker_group.cpp
@@ -22,26 +22,44 @@
 #include <atomic>
 #include <condition_variable>
 #include <cstdint>
+#include <cstring>
+#include <ctime>
 #include <mutex>
 #include <queue>
 #include <string>
 #include <thread>
+#include <type_traits>
 #include <vector>
 
 #include <sys/resource.h>
 #include <sys/time.h>
-#include <time.h>
 #include <unistd.h>
 
+#include <pthread.h>
+
 #include <folly/Conv.h>
 #include <folly/system/ThreadName.h>
 
 #include "dwarfs/error.h"
 #include "dwarfs/semaphore.h"
+#include "dwarfs/util.h"
 #include "dwarfs/worker_group.h"
 
 namespace dwarfs {
 
+namespace {
+
+pthread_t std_to_pthread_id(std::thread::id tid) {
+  static_assert(std::is_same_v<pthread_t, std::thread::native_handle_type>);
+  static_assert(sizeof(std::thread::id) ==
+                sizeof(std::thread::native_handle_type));
+  pthread_t id{0};
+  std::memcpy(&id, &tid, sizeof(id));
+  return id;
+}
+
+} // namespace
+
 template <typename Policy>
 class basic_worker_group final : public worker_group::impl, private Policy {
  public:
@@ -55,6 +73,7 @@ class basic_worker_group final : public worker_group::impl, private Policy {
     if (num_workers < 1) {
       DWARFS_THROW(runtime_error, "invalid number of worker threads");
     }
+
     if (!group_name) {
       group_name = "worker";
     }
@@ -153,6 +172,22 @@ class basic_worker_group final : public worker_group::impl, private Policy {
     return jobs_.size();
   }
 
+  double get_cpu_time() const override {
+    std::lock_guard lock(mx_);
+    double t = 0.0;
+
+    for (auto const& w : workers_) {
+      ::clockid_t cid;
+      struct ::timespec ts;
+      if (::pthread_getcpuclockid(std_to_pthread_id(w.get_id()), &cid) == 0 &&
+          ::clock_gettime(cid, &ts) == 0) {
+        t += ts.tv_sec + 1e-9 * ts.tv_nsec;
+      }
+    }
+
+    return t;
+  }
+
  private:
   using jobs_t = std::queue<worker_group::job_t>;
 
diff --git a/src/mkdwarfs.cpp b/src/mkdwarfs.cpp
index 1909cd03..308edca6 100644
--- a/src/mkdwarfs.cpp
+++ b/src/mkdwarfs.cpp
@@ -865,6 +865,9 @@ int mkdwarfs(int argc, char** argv) {
     return 1;
   }
 
+  LOG_INFO << "compression CPU time: "
+           << time_with_unit(wg_compress.get_cpu_time());
+
   ofs.close();
 
   if (ofs.bad()) {