From c4422e567811051e40469551c58a8078158a6656 Mon Sep 17 00:00:00 2001
From: Nicolas Patry <patry.nicolas@protonmail.com>
Date: Fri, 18 Aug 2023 19:28:56 +0200
Subject: [PATCH] Adding small benchmark script. (#881)

# What does this PR do?

<!--
Congratulations! You've made it this far! You're not quite done yet
though.

Once merged, your PR is going to appear in the release notes with the
title you set, so make sure it's a great title that fully reflects the
extent of your awesome contribution.

Then, please replace this with a description of the change and which
issue is fixed (if applicable). Please also include relevant motivation
and context. List any dependencies (if any) that are required for this
change.

Once you're done, someone will review your PR shortly (see the section
"Who can review?" below to tag some potential reviewers). They may
suggest changes to make the code even better. If no one reviewed your PR
after a week has passed, don't hesitate to post a new comment
@-mentioning the same persons---sometimes notifications get lost.
-->

<!-- Remove if not applicable -->

Fixes # (issue)


## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the
other checks if that's the case).
- [ ] Did you read the [contributor
guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#start-contributing-pull-requests),
      Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the
[forum](https://discuss.huggingface.co/)? Please add a link
      to it if that's the case.
- [ ] Did you make sure to update the documentation with your changes?
Here are the
[documentation
guidelines](https://github.com/huggingface/transformers/tree/main/docs),
and
[here are tips on formatting
docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] Did you write any new necessary tests?


## Who can review?

Anyone in the community is free to review the PR once the tests have
passed. Feel free to tag
members/contributors who may be interested in your PR.

<!-- Your PR will be replied to more quickly if you can figure out the
right person to tag with @


@OlivierDehaene OR @Narsil

 -->
---
 load_tests/common.js | 64 ++++++++++++++++++++++++++++++++++++++++++++
 load_tests/tgi.js    | 17 ++++++++++++
 load_tests/vllm.js   | 17 ++++++++++++
 3 files changed, 98 insertions(+)
 create mode 100644 load_tests/common.js
 create mode 100644 load_tests/tgi.js
 create mode 100644 load_tests/vllm.js

diff --git a/load_tests/common.js b/load_tests/common.js
new file mode 100644
index 00000000..be812e9b
--- /dev/null
+++ b/load_tests/common.js
@@ -0,0 +1,64 @@
+import { check, randomSeed } from 'k6';
+import http from 'k6/http';
+import { Trend, Counter } from 'k6/metrics';
+import { randomItem } from 'https://jslib.k6.io/k6-utils/1.2.0/index.js';
+
+const seed = 0;
+
+const host = __ENV.HOST || '127.0.0.1:8000';
+const timePerToken = new Trend('time_per_token', true);
+const throughput = new Counter('tokens_per_s');
+
+randomSeed(seed);
+// const shareGPT = JSON.parse(open("ShareGPT_V3_unfiltered_cleaned_split.json"))
+const shareGPT = JSON.parse(open("small.json"))
+
+
+export function get_options(reference_latency_ms){
+    return {
+        thresholds: {
+            http_req_failed: ['rate==0'],
+            time_per_token: [{
+                threshold: `p(50)<${3 * reference_latency_ms}`,
+                abortOnFail: true,
+                delayAbortEval: '10s'
+            }],
+        },
+        scenarios: {
+            load_test: {
+                executor: 'constant-arrival-rate',
+                duration: '60s',
+                preAllocatedVUs: 100,
+                rate: 10,
+                timeUnit: '1s',
+            },
+        },
+    };
+}
+
+
+export function run(host, generate_payload, max_new_tokens) {
+    const headers = {'Content-Type': 'application/json'};
+    const query = randomItem(shareGPT);
+    const payload = JSON.stringify(generate_payload(query));
+    const res = http.post(`http://${host}/generate`, payload, {
+        headers,
+    });
+    if(res.status >= 400 && res.status < 500){
+        return;
+    }
+
+    check(res, {
+        'Post status is 200': (r) => res.status === 200,
+    });
+    const n_tokens = max_new_tokens;
+    const timings = res.timings.duration;
+
+    if (res.status === 200) {
+        const latency_ms_per_token = timings / n_tokens;
+        timePerToken.add(latency_ms_per_token);
+        const latency_in_s = latency_ms_per_token / 1000;
+        const individual_throughput = 1 / latency_in_s;
+        throughput.add(individual_throughput);
+    }
+}
diff --git a/load_tests/tgi.js b/load_tests/tgi.js
new file mode 100644
index 00000000..93a0e278
--- /dev/null
+++ b/load_tests/tgi.js
@@ -0,0 +1,17 @@
+import { get_options, run } from "./common.js";
+ 
+const reference_latency_ms = 30;
+const host = __ENV.HOST || '127.0.0.1:8000';
+const max_new_tokens = 50;
+
+
+function generate_payload(gpt){
+    const input = gpt["conversations"][0]["value"];
+    return {"inputs": input, "parameters": {"max_new_tokens": max_new_tokens, "temperature" : 0.5}}
+}
+
+export const options = get_options(reference_latency_ms);
+
+export default function(){
+    run(host, generate_payload, max_new_tokens);
+}
diff --git a/load_tests/vllm.js b/load_tests/vllm.js
new file mode 100644
index 00000000..fcb38262
--- /dev/null
+++ b/load_tests/vllm.js
@@ -0,0 +1,17 @@
+import { get_options, run } from "./common.js";
+ 
+const reference_latency_ms = 22;
+const host = __ENV.HOST || '127.0.0.1:8000';
+const max_new_tokens = 50;
+
+
+function generate_payload(gpt){
+    const input = gpt["conversations"][0]["value"];
+    return {"prompt": input, "temperature": 0.5, "ignore_eos": true}
+}
+
+export const options = get_options(reference_latency_ms);
+
+export default function(){
+    run(host, generate_payload, max_new_tokens);
+}