TETensorRT · Beginners
Community homeHot discussions Latest posts Best of Knowledge base Resources About

TensorRT · Beginners · Those easily-missed type details in tensorrt-newbie

40
TEr/tensorrt-newbie·posted by alice_dev·6 hours agoTutorial

Those easily-missed type details in tensorrt-newbie

Short version: tensorrt-newbie needs almost no tuning at small and medium scale — the point where it starts to hurt is much further out than most people assume. Full measurements below.

We also fixed monitoring along the way: replaced average-based alerts with percentiles and split them per endpoint. False alerts dropped by about seventy percent and the on-call rotation visibly cheered up.

Worth noting: the official docs do cover this, just in a very inconspicuous spot. I only found it reading the source comments, where the author explains the reasoning — roughly "so that it degrades into predictable behaviour in extreme cases".

Jjvns.caExternal link · opens in a new tab
13 comments

13 comments

M
Ddev_zhou·28 minutes ago

We have run this in production for two years without hitting it. That said, we never reached this scale, so our experience is not really evidence here.

233
Hhuang_ke·28 minutes ago

Can you give a minimal reproduction? I ran it locally for ten minutes and could not reproduce on macOS with the latest version.

260
Ttang_haoOP·1 hour ago

Sharing our numbers, 8 cores 16GB, same scenario:

| Concurrency | P50 | P99 |
|---|---|---|
| 200 | 12ms | 88ms |
| 500 | 31ms | 340ms |

P99 clearly collapses at 500 concurrency, which lines up with your knee point.

34
Zzhu_zong·2 days ago

Agreeing with the above. One addition: with this option enabled the GC count in your metrics doubles, so adjust the alert threshold at the same time or it will keep firing.

375
Rrase·just nowedited

I just read the tensorrt-newbie source — the author actually explains the reasoning in a comment, roughly "so that it degrades into predictable behaviour in extreme cases".

219
Cchen_dev·yesterday

This matches what we see in production. We only hit it past 3k QPS; the earlier load tests showed nothing — the test traffic was too clean, with no long-tail requests.

464
Sslow_query·12 minutes agoLevel 6

Saved. I am reworking this area this week — this saves a lot of wrong turns.

111
Nnikic·3 minutes ago

There is actually a simpler fix that needs no architecture change: move this check up to the gateway and the problem disappears. The cost is one extra lookup at the gateway.

242
Kkite·yesterday

A question: what changes in a container with a 512Mi memory limit? That is how we run it in production.

202
Zzhu_zong·2 hours ago

Has anyone run a controlled experiment? I did, reducing it to a single variable, and the difference was 4% — within noise. So I suspect the main cause is something else.

103
Kkite·yesterday

This is not a tensorrt-newbie problem, it is a usage problem. The docs say this API is not thread-safe and you must lock around it yourself.

34
Oops_wang·12 minutes agoedited

Thanks for sharing real numbers — far more useful than the articles that only cover concepts.

19
Lli_ming·3 minutes agoedited

I see point 3 differently. The trade-off depends on your read/write ratio: read-heavy with little writing means caching actually widens the inconsistency window.

7

This is the post detail page /en/c/tensorrt-newbie/post/p6. Posts and comments are generated deterministically from a seeded PRNG, so the same post always renders the same content and the link can be shared, reloaded and indexed. In production this page reads MySQL for the post, Redis for hot-post caching, and fetches the whole comment tree in a single query on the path column.

See the database schema →