A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 15, 17:15 UTC
[MISC] Refine no available block debug msg (#15076)
Signed-off-by: Yi Liu <yiliu4@habana.ai> Signed-off-by: yiliu30 <yi4.liu@intel.com> Co-authored-by: Yi Liu <yiliu4@habana.ai>
Yi Liu committed
Mar 25, 2025 at 00:01 UTC
9cc645141da1210d88bdf8aa5a7f0c725862b1fd
1 file changed
+10
-4
vllm/distributed/device_communicators/shm_broadcast.py
+10
-4
@@ -357,8 +357,11 @@ class MessageQueue:
357
# if we wait for a long time, log a message
358
if (time.monotonic() - start_time
359
> VLLM_RINGBUFFER_WARNING_INTERVAL * n_warning):
360
- logger.debug("No available block found in %s second. ",
361
- VLLM_RINGBUFFER_WARNING_INTERVAL)
360
+ logger.debug(
361
+ ("No available shared memory broadcast block found"
362
+ " in %s second."),
363
+ VLLM_RINGBUFFER_WARNING_INTERVAL,
364
+ )
365
n_warning += 1
366
367
# if we time out, raise an exception
@@ -415,8 +418,11 @@ class MessageQueue:
418
# if we wait for a long time, log a message
419
if (time.monotonic() - start_time
420
> VLLM_RINGBUFFER_WARNING_INTERVAL * n_warning):
418
- logger.debug("No available block found in %s second. ",
419
- VLLM_RINGBUFFER_WARNING_INTERVAL)
421
+ logger.debug(
422
+ ("No available shared memory broadcast block found"
423
+ "in %s second."),
424
+ VLLM_RINGBUFFER_WARNING_INTERVAL,
425
+ )
426
n_warning += 1
427
428
# if we time out, raise an exception