github-starred/ollama
2
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2026-07-11 11:07:38 -05:00
Code Issues 16.9k Packages Projects Releases 122 Wiki Activity

[GH-ISSUE #11060] New Ollama Engine Causes Severe Performance Degradation (10x Slower) #53809

New Issue
Closed
opened 2026-04-29 04:47:33 -05:00 by GiteaMirror · 9 comments
GiteaMirror commented 2026-04-29 04:47:33 -05:00
Owner
Copy Link

Originally created by @AmesianX on GitHub (Jun 13, 2025).
Original GitHub issue: https://github.com/ollama/ollama/issues/11060

Originally assigned to: @jmorganca on GitHub.

What is the issue?

<html>

📝 Technical Reporting

Hello Ollama Team,

I'm using Ollama for local LLM inference, and I've encountered a critical performance regression in the latest version of the engine.
By setting the environment variable -e OLLAMA_NEW_ENGINE=false, the inference speed of Qwen3 increases dramatically, while using the new engine causes a significant slowdown.

🔧 Environment Info

  • GPU: NVIDIA RTX 3090 Ti (1x)
  • OLLAMA Version: v0.9 (new engine) and v0.7 (old engine)
  • Model: Qwen3 (or similar large LLM)
  • Test Method: ollama pull qwen3:30b-a3b + Open WebUI

📌 Problem Description

  • With the new engine:

    • Inference speed is 10x slower (from 100 to 12–60 tokens/second).
    • Response latency is long and unstable (takes several seconds before the response starts).
    • GPU utilization is inconsistent, even on a single RTX 3090 Ti.

  • With the old engine:

    • High-speed inference (over 100 tokens/second).
    • Stable and fast response time (response starts immediately after request).
    • Efficient GPU usage.

🧠 Possible Cause

  • The new engine may have changed how the model is loaded or cached.
  • There could be issues with GPU memory management or CUDA interface changes.
  • Possible regression in dynamic batching or caching strategies.

📌 Request

  • Please investigate the performance degradation in the new engine.
  • Provide a fix or performance improvement if possible.
  • Maintain the option to use the old engine (-e OLLAMA_NEW_ENGINE=false is currently working).
  • Add official notes or explanations about performance changes in future releases.

📌 Thank You

Thank you for your time and effort in maintaining such a great tool.
I look forward to your response.

</html>

Relevant log output

OS

No response

GPU

No response

CPU

No response

Ollama version

No response

Originally created by @AmesianX on GitHub (Jun 13, 2025). Original GitHub issue: https://github.com/ollama/ollama/issues/11060 Originally assigned to: @jmorganca on GitHub. ### What is the issue? <html> <body> <!--StartFragment--><hr class="border-gray-100 dark:border-gray-850" style="box-sizing: border-box; border-width: 1px 0px 0px; border-style: solid; border-color: rgb(236, 236, 236); border-image: initial; margin: 3em 0px; padding: 0px; height: 0px; color: rgb(78, 78, 78); margin-block: 16px; font-family: -apple-system, BlinkMacSystemFont, Inter, Vazirmatn, ui-sans-serif, system-ui, &quot;Segoe UI&quot;, Roboto, Ubuntu, Cantarell, &quot;Noto Sans&quot;, sans-serif, &quot;Helvetica Neue&quot;, Arial, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;, &quot;Segoe UI Symbol&quot;, &quot;Noto Color Emoji&quot;; font-size: 16px; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; font-weight: 400; letter-spacing: normal; orphans: 2; text-align: start; text-indent: 0px; text-transform: none; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; white-space: pre-line; background-color: rgb(255, 255, 255); text-decoration-thickness: initial; text-decoration-style: initial; text-decoration-color: initial;"><h2 dir="auto" style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px 0px 1em; padding: 0px; font-size: 1.5em; font-weight: 600; color: oklch(0.21 0.034 264.665); line-height: 1.33333; margin-block: 4px; --tw-font-weight: 600; font-family: -apple-system, BlinkMacSystemFont, Inter, Vazirmatn, ui-sans-serif, system-ui, &quot;Segoe UI&quot;, Roboto, Ubuntu, Cantarell, &quot;Noto Sans&quot;, sans-serif, &quot;Helvetica Neue&quot;, Arial, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;, &quot;Segoe UI Symbol&quot;, &quot;Noto Color Emoji&quot;; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; letter-spacing: normal; orphans: 2; text-align: start; text-indent: 0px; text-transform: none; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; white-space: pre-line; background-color: rgb(255, 255, 255); text-decoration-thickness: initial; text-decoration-style: initial; text-decoration-color: initial;">📝 Technical Reporting</h2><blockquote dir="auto" style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px 0px 1.6em; padding: 0px; color: oklch(0.21 0.034 264.665); border-inline-start: 2px solid rgb(236, 236, 236); quotes: &quot;“&quot; &quot;”&quot; &quot;‘&quot; &quot;’&quot;; padding-inline-start: 1em; font-style: normal; font-weight: 400; margin-block: 0px; --tw-font-weight: 400; font-family: -apple-system, BlinkMacSystemFont, Inter, Vazirmatn, ui-sans-serif, system-ui, &quot;Segoe UI&quot;, Roboto, Ubuntu, Cantarell, &quot;Noto Sans&quot;, sans-serif, &quot;Helvetica Neue&quot;, Arial, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;, &quot;Segoe UI Symbol&quot;, &quot;Noto Color Emoji&quot;; font-size: 16px; font-variant-ligatures: normal; font-variant-caps: normal; letter-spacing: normal; orphans: 2; text-align: start; text-indent: 0px; text-transform: none; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; white-space: pre-line; background-color: rgb(255, 255, 255); text-decoration-thickness: initial; text-decoration-style: initial; text-decoration-color: initial;"><p dir="auto" style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 1.25em 0px; padding: 0px; margin-block: 0px;">Hello Ollama Team,</p><div class="my-2" style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px; padding: 0px; margin-block: 8px;"></div><p dir="auto" style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 1.25em 0px; padding: 0px; margin-block: 0px;">I'm using Ollama for local LLM inference, and I've encountered a <strong style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px; padding: 0px; font-weight: 600; color: inherit;">critical performance regression</strong> in the latest version of the engine.<br style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px; padding: 0px;">By setting the environment variable <code class="codespan cursor-pointer" style="box-sizing: border-box; border: 0px solid rgb(235, 87, 87); margin: 0px; padding: 3px 8px; font-family: ui-monospace, SFMono-Regular, Menlo, Monaco, Consolas, &quot;Liberation Mono&quot;, &quot;Courier New&quot;, monospace; font-feature-settings: normal; font-variation-settings: normal; font-size: 0.8em; cursor: pointer; width: auto; overflow-x: auto; color: rgb(235, 87, 87); margin-inline: 2px; border-radius: 6px; background-color: rgb(236, 236, 236); font-weight: 600;">-e OLLAMA_NEW_ENGINE=false</code>, the inference speed of <strong style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px; padding: 0px; font-weight: 600; color: inherit;">Qwen3</strong> increases dramatically, while using the new engine causes a <strong style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px; padding: 0px; font-weight: 600; color: inherit;">significant slowdown</strong>. </p></blockquote><hr class="border-gray-100 dark:border-gray-850" style="box-sizing: border-box; border-width: 1px 0px 0px; border-style: solid; border-color: rgb(236, 236, 236); border-image: initial; margin: 3em 0px; padding: 0px; height: 0px; color: rgb(78, 78, 78); margin-block: 16px; font-family: -apple-system, BlinkMacSystemFont, Inter, Vazirmatn, ui-sans-serif, system-ui, &quot;Segoe UI&quot;, Roboto, Ubuntu, Cantarell, &quot;Noto Sans&quot;, sans-serif, &quot;Helvetica Neue&quot;, Arial, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;, &quot;Segoe UI Symbol&quot;, &quot;Noto Color Emoji&quot;; font-size: 16px; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; font-weight: 400; letter-spacing: normal; orphans: 2; text-align: start; text-indent: 0px; text-transform: none; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; white-space: pre-line; background-color: rgb(255, 255, 255); text-decoration-thickness: initial; text-decoration-style: initial; text-decoration-color: initial;"><h3 dir="auto" style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px 0px 0.6em; padding: 0px; font-size: 1.25em; font-weight: 600; color: oklch(0.21 0.034 264.665); line-height: 1.6; margin-block: 4px; --tw-font-weight: 600; font-family: -apple-system, BlinkMacSystemFont, Inter, Vazirmatn, ui-sans-serif, system-ui, &quot;Segoe UI&quot;, Roboto, Ubuntu, Cantarell, &quot;Noto Sans&quot;, sans-serif, &quot;Helvetica Neue&quot;, Arial, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;, &quot;Segoe UI Symbol&quot;, &quot;Noto Color Emoji&quot;; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; letter-spacing: normal; orphans: 2; text-align: start; text-indent: 0px; text-transform: none; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; white-space: pre-line; background-color: rgb(255, 255, 255); text-decoration-thickness: initial; text-decoration-style: initial; text-decoration-color: initial;">🔧 Environment Info</h3><ul dir="auto" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0px 0px 1.25em; padding: 0px; list-style: disc; padding-inline-start: 1.625em; margin-block: 0px; color: rgb(78, 78, 78); font-family: -apple-system, BlinkMacSystemFont, Inter, Vazirmatn, ui-sans-serif, system-ui, &quot;Segoe UI&quot;, Roboto, Ubuntu, Cantarell, &quot;Noto Sans&quot;, sans-serif, &quot;Helvetica Neue&quot;, Arial, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;, &quot;Segoe UI Symbol&quot;, &quot;Noto Color Emoji&quot;; font-size: 16px; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; font-weight: 400; letter-spacing: normal; orphans: 2; text-align: start; text-indent: 0px; text-transform: none; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; white-space: pre-line; background-color: rgb(255, 255, 255); text-decoration-thickness: initial; text-decoration-style: initial; text-decoration-color: initial;"><li class="text-start" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0.5em 0px; padding: 0px; text-align: start; padding-inline-start: 0.375em; margin-block: 0px;"><strong style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px; padding: 0px; font-weight: 600; color: oklch(0.21 0.034 264.665);">GPU</strong>: NVIDIA RTX 3090 Ti (1x)</li><li class="text-start" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0.5em 0px; padding: 0px; text-align: start; padding-inline-start: 0.375em; margin-block: 0px;"><strong style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px; padding: 0px; font-weight: 600; color: oklch(0.21 0.034 264.665);">OLLAMA Version</strong>: v0.9 (new engine) and v0.7 (old engine)</li><li class="text-start" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0.5em 0px; padding: 0px; text-align: start; padding-inline-start: 0.375em; margin-block: 0px;"><strong style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px; padding: 0px; font-weight: 600; color: oklch(0.21 0.034 264.665);">Model</strong>: Qwen3 (or similar large LLM)</li><li class="text-start" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0.5em 0px; padding: 0px; text-align: start; padding-inline-start: 0.375em; margin-block: 0px;"><strong style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px; padding: 0px; font-weight: 600; color: oklch(0.21 0.034 264.665);">Test Method</strong>: <code class="codespan cursor-pointer" style="box-sizing: border-box; border: 0px solid rgb(235, 87, 87); margin: 0px; padding: 3px 8px; font-family: ui-monospace, SFMono-Regular, Menlo, Monaco, Consolas, &quot;Liberation Mono&quot;, &quot;Courier New&quot;, monospace; font-feature-settings: normal; font-variation-settings: normal; font-size: 0.8em; cursor: pointer; width: auto; overflow-x: auto; color: rgb(235, 87, 87); margin-inline: 2px; border-radius: 6px; background-color: rgb(236, 236, 236); font-weight: 600;">ollama pull qwen3:30b-a3b + Open WebUI </code></li></ul><div class="my-2" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0px; padding: 0px; margin-block: 8px; color: rgb(78, 78, 78); font-family: -apple-system, BlinkMacSystemFont, Inter, Vazirmatn, ui-sans-serif, system-ui, &quot;Segoe UI&quot;, Roboto, Ubuntu, Cantarell, &quot;Noto Sans&quot;, sans-serif, &quot;Helvetica Neue&quot;, Arial, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;, &quot;Segoe UI Symbol&quot;, &quot;Noto Color Emoji&quot;; font-size: 16px; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; font-weight: 400; letter-spacing: normal; orphans: 2; text-align: start; text-indent: 0px; text-transform: none; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; white-space: pre-line; background-color: rgb(255, 255, 255); text-decoration-thickness: initial; text-decoration-style: initial; text-decoration-color: initial;"></div><hr class="border-gray-100 dark:border-gray-850" style="box-sizing: border-box; border-width: 1px 0px 0px; border-style: solid; border-color: rgb(236, 236, 236); border-image: initial; margin: 3em 0px; padding: 0px; height: 0px; color: rgb(78, 78, 78); margin-block: 16px; font-family: -apple-system, BlinkMacSystemFont, Inter, Vazirmatn, ui-sans-serif, system-ui, &quot;Segoe UI&quot;, Roboto, Ubuntu, Cantarell, &quot;Noto Sans&quot;, sans-serif, &quot;Helvetica Neue&quot;, Arial, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;, &quot;Segoe UI Symbol&quot;, &quot;Noto Color Emoji&quot;; font-size: 16px; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; font-weight: 400; letter-spacing: normal; orphans: 2; text-align: start; text-indent: 0px; text-transform: none; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; white-space: pre-line; background-color: rgb(255, 255, 255); text-decoration-thickness: initial; text-decoration-style: initial; text-decoration-color: initial;"><div class="absolute top-1 right-1.5 z-20 invisible group-hover:visible" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0px; padding: 0px; visibility: hidden; position: absolute; top: 4px; right: 6px; z-index: 20;"><div class="flex" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0px; padding: 0px; display: flex;"><button class="p-1 rounded-lg bg-transparent transition" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0px; padding: 4px; font: inherit; letter-spacing: inherit; color: inherit; opacity: 1; background-color: rgba(0, 0, 0, 0); border-radius: 8px; appearance: button; cursor: pointer; transition-property: color, background-color, border-color, outline-color, text-decoration-color, fill, stroke, --tw-gradient-from, --tw-gradient-via, --tw-gradient-to, opacity, box-shadow, transform, translate, scale, rotate, filter, -webkit-backdrop-filter, backdrop-filter; transition-timing-function: cubic-bezier(0.4, 0, 0.2, 1); transition-duration: 0.15s;"><svg xmlns="http://www.w3.org/2000/svg" fill="none" viewBox="0 0 24 24" stroke-width="1.5" stroke="currentColor" class=" size-3.5"><path stroke-linecap="round" stroke-linejoin="round" d="M3 16.5v2.25A2.25 2.25 0 0 0 5.25 21h13.5A2.25 2.25 0 0 0 21 18.75V16.5M16.5 12 12 16.5m0 0L7.5 12m4.5 4.5V3"></path></svg></button></div></div></div><hr class="border-gray-100 dark:border-gray-850" style="box-sizing: border-box; border-width: 1px 0px 0px; border-style: solid; border-color: rgb(236, 236, 236); border-image: initial; margin: 3em 0px; padding: 0px; height: 0px; color: rgb(78, 78, 78); margin-block: 16px; font-family: -apple-system, BlinkMacSystemFont, Inter, Vazirmatn, ui-sans-serif, system-ui, &quot;Segoe UI&quot;, Roboto, Ubuntu, Cantarell, &quot;Noto Sans&quot;, sans-serif, &quot;Helvetica Neue&quot;, Arial, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;, &quot;Segoe UI Symbol&quot;, &quot;Noto Color Emoji&quot;; font-size: 16px; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; font-weight: 400; letter-spacing: normal; orphans: 2; text-align: start; text-indent: 0px; text-transform: none; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; white-space: pre-line; background-color: rgb(255, 255, 255); text-decoration-thickness: initial; text-decoration-style: initial; text-decoration-color: initial;"><h3 dir="auto" style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px 0px 0.6em; padding: 0px; font-size: 1.25em; font-weight: 600; color: oklch(0.21 0.034 264.665); line-height: 1.6; margin-block: 4px; --tw-font-weight: 600; font-family: -apple-system, BlinkMacSystemFont, Inter, Vazirmatn, ui-sans-serif, system-ui, &quot;Segoe UI&quot;, Roboto, Ubuntu, Cantarell, &quot;Noto Sans&quot;, sans-serif, &quot;Helvetica Neue&quot;, Arial, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;, &quot;Segoe UI Symbol&quot;, &quot;Noto Color Emoji&quot;; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; letter-spacing: normal; orphans: 2; text-align: start; text-indent: 0px; text-transform: none; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; white-space: pre-line; background-color: rgb(255, 255, 255); text-decoration-thickness: initial; text-decoration-style: initial; text-decoration-color: initial;">📌 Problem Description</h3><ul dir="auto" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0px 0px 1.25em; padding: 0px; list-style: disc; padding-inline-start: 1.625em; margin-block: 0px; color: rgb(78, 78, 78); font-family: -apple-system, BlinkMacSystemFont, Inter, Vazirmatn, ui-sans-serif, system-ui, &quot;Segoe UI&quot;, Roboto, Ubuntu, Cantarell, &quot;Noto Sans&quot;, sans-serif, &quot;Helvetica Neue&quot;, Arial, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;, &quot;Segoe UI Symbol&quot;, &quot;Noto Color Emoji&quot;; font-size: 16px; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; font-weight: 400; letter-spacing: normal; orphans: 2; text-align: start; text-indent: 0px; text-transform: none; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; white-space: pre-line; background-color: rgb(255, 255, 255); text-decoration-thickness: initial; text-decoration-style: initial; text-decoration-color: initial;"><li class="text-start" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0.5em 0px; padding: 0px; text-align: start; padding-inline-start: 0.375em; margin-block: 0px;"><p style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 1.25em 0px; padding: 0px; display: inline; margin-block: 0px;"><strong style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px; padding: 0px; font-weight: 600; color: oklch(0.21 0.034 264.665);">With the new engine</strong>:</p><ul dir="auto" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0.75em 0px; padding: 0px; list-style: disc; padding-inline-start: 1.625em; margin-block: 0px;"><li class="text-start" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0.5em 0px; padding: 0px; text-align: start; padding-inline-start: 0.375em; margin-block: 0px;">Inference speed is <strong style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px; padding: 0px; font-weight: 600; color: oklch(0.21 0.034 264.665);">10x slower</strong> (from 100 to 12–60 tokens/second).</li><li class="text-start" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0.5em 0px; padding: 0px; text-align: start; padding-inline-start: 0.375em; margin-block: 0px;"><strong style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px; padding: 0px; font-weight: 600; color: oklch(0.21 0.034 264.665);">Response latency is long and unstable</strong> (takes several seconds before the response starts).</li><li class="text-start" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0.5em 0px; padding: 0px; text-align: start; padding-inline-start: 0.375em; margin-block: 0px;"><strong style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px; padding: 0px; font-weight: 600; color: oklch(0.21 0.034 264.665);">GPU utilization is inconsistent</strong>, even on a single RTX 3090 Ti.</li></ul></li><li class="text-start" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0.5em 0px; padding: 0px; text-align: start; padding-inline-start: 0.375em; margin-block: 0px;"><p style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 1.25em 0px; padding: 0px; display: inline; margin-block: 0px;"><strong style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px; padding: 0px; font-weight: 600; color: oklch(0.21 0.034 264.665);">With the old engine</strong>:</p><ul dir="auto" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0.75em 0px; padding: 0px; list-style: disc; padding-inline-start: 1.625em; margin-block: 0px;"><li class="text-start" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0.5em 0px; padding: 0px; text-align: start; padding-inline-start: 0.375em; margin-block: 0px;"><strong style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px; padding: 0px; font-weight: 600; color: oklch(0.21 0.034 264.665);">High-speed inference</strong> (over 100 tokens/second).</li><li class="text-start" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0.5em 0px; padding: 0px; text-align: start; padding-inline-start: 0.375em; margin-block: 0px;"><strong style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px; padding: 0px; font-weight: 600; color: oklch(0.21 0.034 264.665);">Stable and fast response time</strong> (response starts immediately after request).</li><li class="text-start" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0.5em 0px; padding: 0px; text-align: start; padding-inline-start: 0.375em; margin-block: 0px;"><strong style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px; padding: 0px; font-weight: 600; color: oklch(0.21 0.034 264.665);">Efficient GPU usage</strong>.</li></ul></li></ul><div class="my-2" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0px; padding: 0px; margin-block: 8px; color: rgb(78, 78, 78); font-family: -apple-system, BlinkMacSystemFont, Inter, Vazirmatn, ui-sans-serif, system-ui, &quot;Segoe UI&quot;, Roboto, Ubuntu, Cantarell, &quot;Noto Sans&quot;, sans-serif, &quot;Helvetica Neue&quot;, Arial, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;, &quot;Segoe UI Symbol&quot;, &quot;Noto Color Emoji&quot;; font-size: 16px; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; font-weight: 400; letter-spacing: normal; orphans: 2; text-align: start; text-indent: 0px; text-transform: none; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; white-space: pre-line; background-color: rgb(255, 255, 255); text-decoration-thickness: initial; text-decoration-style: initial; text-decoration-color: initial;"></div><hr class="border-gray-100 dark:border-gray-850" style="box-sizing: border-box; border-width: 1px 0px 0px; border-style: solid; border-color: rgb(236, 236, 236); border-image: initial; margin: 3em 0px; padding: 0px; height: 0px; color: rgb(78, 78, 78); margin-block: 16px; font-family: -apple-system, BlinkMacSystemFont, Inter, Vazirmatn, ui-sans-serif, system-ui, &quot;Segoe UI&quot;, Roboto, Ubuntu, Cantarell, &quot;Noto Sans&quot;, sans-serif, &quot;Helvetica Neue&quot;, Arial, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;, &quot;Segoe UI Symbol&quot;, &quot;Noto Color Emoji&quot;; font-size: 16px; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; font-weight: 400; letter-spacing: normal; orphans: 2; text-align: start; text-indent: 0px; text-transform: none; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; white-space: pre-line; background-color: rgb(255, 255, 255); text-decoration-thickness: initial; text-decoration-style: initial; text-decoration-color: initial;"><h3 dir="auto" style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px 0px 0.6em; padding: 0px; font-size: 1.25em; font-weight: 600; color: oklch(0.21 0.034 264.665); line-height: 1.6; margin-block: 4px; --tw-font-weight: 600; font-family: -apple-system, BlinkMacSystemFont, Inter, Vazirmatn, ui-sans-serif, system-ui, &quot;Segoe UI&quot;, Roboto, Ubuntu, Cantarell, &quot;Noto Sans&quot;, sans-serif, &quot;Helvetica Neue&quot;, Arial, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;, &quot;Segoe UI Symbol&quot;, &quot;Noto Color Emoji&quot;; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; letter-spacing: normal; orphans: 2; text-align: start; text-indent: 0px; text-transform: none; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; white-space: pre-line; background-color: rgb(255, 255, 255); text-decoration-thickness: initial; text-decoration-style: initial; text-decoration-color: initial;">🧠 Possible Cause</h3><ul dir="auto" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0px 0px 1.25em; padding: 0px; list-style: disc; padding-inline-start: 1.625em; margin-block: 0px; color: rgb(78, 78, 78); font-family: -apple-system, BlinkMacSystemFont, Inter, Vazirmatn, ui-sans-serif, system-ui, &quot;Segoe UI&quot;, Roboto, Ubuntu, Cantarell, &quot;Noto Sans&quot;, sans-serif, &quot;Helvetica Neue&quot;, Arial, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;, &quot;Segoe UI Symbol&quot;, &quot;Noto Color Emoji&quot;; font-size: 16px; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; font-weight: 400; letter-spacing: normal; orphans: 2; text-align: start; text-indent: 0px; text-transform: none; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; white-space: pre-line; background-color: rgb(255, 255, 255); text-decoration-thickness: initial; text-decoration-style: initial; text-decoration-color: initial;"><li class="text-start" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0.5em 0px; padding: 0px; text-align: start; padding-inline-start: 0.375em; margin-block: 0px;">The new engine may have <strong style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px; padding: 0px; font-weight: 600; color: oklch(0.21 0.034 264.665);">changed how the model is loaded or cached</strong>.</li><li class="text-start" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0.5em 0px; padding: 0px; text-align: start; padding-inline-start: 0.375em; margin-block: 0px;">There could be <strong style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px; padding: 0px; font-weight: 600; color: oklch(0.21 0.034 264.665);">issues with GPU memory management</strong> or <strong style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px; padding: 0px; font-weight: 600; color: oklch(0.21 0.034 264.665);">CUDA interface changes</strong>.</li><li class="text-start" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0.5em 0px; padding: 0px; text-align: start; padding-inline-start: 0.375em; margin-block: 0px;">Possible <strong style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px; padding: 0px; font-weight: 600; color: oklch(0.21 0.034 264.665);">regression in dynamic batching or caching strategies</strong>.</li></ul><div class="my-2" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0px; padding: 0px; margin-block: 8px; color: rgb(78, 78, 78); font-family: -apple-system, BlinkMacSystemFont, Inter, Vazirmatn, ui-sans-serif, system-ui, &quot;Segoe UI&quot;, Roboto, Ubuntu, Cantarell, &quot;Noto Sans&quot;, sans-serif, &quot;Helvetica Neue&quot;, Arial, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;, &quot;Segoe UI Symbol&quot;, &quot;Noto Color Emoji&quot;; font-size: 16px; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; font-weight: 400; letter-spacing: normal; orphans: 2; text-align: start; text-indent: 0px; text-transform: none; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; white-space: pre-line; background-color: rgb(255, 255, 255); text-decoration-thickness: initial; text-decoration-style: initial; text-decoration-color: initial;"></div><hr class="border-gray-100 dark:border-gray-850" style="box-sizing: border-box; border-width: 1px 0px 0px; border-style: solid; border-color: rgb(236, 236, 236); border-image: initial; margin: 3em 0px; padding: 0px; height: 0px; color: rgb(78, 78, 78); margin-block: 16px; font-family: -apple-system, BlinkMacSystemFont, Inter, Vazirmatn, ui-sans-serif, system-ui, &quot;Segoe UI&quot;, Roboto, Ubuntu, Cantarell, &quot;Noto Sans&quot;, sans-serif, &quot;Helvetica Neue&quot;, Arial, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;, &quot;Segoe UI Symbol&quot;, &quot;Noto Color Emoji&quot;; font-size: 16px; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; font-weight: 400; letter-spacing: normal; orphans: 2; text-align: start; text-indent: 0px; text-transform: none; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; white-space: pre-line; background-color: rgb(255, 255, 255); text-decoration-thickness: initial; text-decoration-style: initial; text-decoration-color: initial;"><h3 dir="auto" style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px 0px 0.6em; padding: 0px; font-size: 1.25em; font-weight: 600; color: oklch(0.21 0.034 264.665); line-height: 1.6; margin-block: 4px; --tw-font-weight: 600; font-family: -apple-system, BlinkMacSystemFont, Inter, Vazirmatn, ui-sans-serif, system-ui, &quot;Segoe UI&quot;, Roboto, Ubuntu, Cantarell, &quot;Noto Sans&quot;, sans-serif, &quot;Helvetica Neue&quot;, Arial, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;, &quot;Segoe UI Symbol&quot;, &quot;Noto Color Emoji&quot;; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; letter-spacing: normal; orphans: 2; text-align: start; text-indent: 0px; text-transform: none; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; white-space: pre-line; background-color: rgb(255, 255, 255); text-decoration-thickness: initial; text-decoration-style: initial; text-decoration-color: initial;">📌 Request</h3><ul dir="auto" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0px 0px 1.25em; padding: 0px; list-style: disc; padding-inline-start: 1.625em; margin-block: 0px; color: rgb(78, 78, 78); font-family: -apple-system, BlinkMacSystemFont, Inter, Vazirmatn, ui-sans-serif, system-ui, &quot;Segoe UI&quot;, Roboto, Ubuntu, Cantarell, &quot;Noto Sans&quot;, sans-serif, &quot;Helvetica Neue&quot;, Arial, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;, &quot;Segoe UI Symbol&quot;, &quot;Noto Color Emoji&quot;; font-size: 16px; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; font-weight: 400; letter-spacing: normal; orphans: 2; text-align: start; text-indent: 0px; text-transform: none; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; white-space: pre-line; background-color: rgb(255, 255, 255); text-decoration-thickness: initial; text-decoration-style: initial; text-decoration-color: initial;"><li class="text-start" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0.5em 0px; padding: 0px; text-align: start; padding-inline-start: 0.375em; margin-block: 0px;">Please <strong style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px; padding: 0px; font-weight: 600; color: oklch(0.21 0.034 264.665);">investigate the performance degradation</strong> in the new engine.</li><li class="text-start" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0.5em 0px; padding: 0px; text-align: start; padding-inline-start: 0.375em; margin-block: 0px;">Provide <strong style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px; padding: 0px; font-weight: 600; color: oklch(0.21 0.034 264.665);">a fix or performance improvement</strong> if possible.</li><li class="text-start" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0.5em 0px; padding: 0px; text-align: start; padding-inline-start: 0.375em; margin-block: 0px;">Maintain the <strong style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px; padding: 0px; font-weight: 600; color: oklch(0.21 0.034 264.665);">option to use the old engine</strong> (<code class="codespan cursor-pointer" style="box-sizing: border-box; border: 0px solid rgb(235, 87, 87); margin: 0px; padding: 3px 8px; font-family: ui-monospace, SFMono-Regular, Menlo, Monaco, Consolas, &quot;Liberation Mono&quot;, &quot;Courier New&quot;, monospace; font-feature-settings: normal; font-variation-settings: normal; font-size: 0.8em; cursor: pointer; width: auto; overflow-x: auto; color: rgb(235, 87, 87); margin-inline: 2px; border-radius: 6px; background-color: rgb(236, 236, 236); font-weight: 600;">-e OLLAMA_NEW_ENGINE=false</code> is currently working).</li><li class="text-start" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0.5em 0px; padding: 0px; text-align: start; padding-inline-start: 0.375em; margin-block: 0px;">Add <strong style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px; padding: 0px; font-weight: 600; color: oklch(0.21 0.034 264.665);">official notes or explanations</strong> about performance changes in future releases.</li></ul><div class="my-2" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0px; padding: 0px; margin-block: 8px; color: rgb(78, 78, 78); font-family: -apple-system, BlinkMacSystemFont, Inter, Vazirmatn, ui-sans-serif, system-ui, &quot;Segoe UI&quot;, Roboto, Ubuntu, Cantarell, &quot;Noto Sans&quot;, sans-serif, &quot;Helvetica Neue&quot;, Arial, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;, &quot;Segoe UI Symbol&quot;, &quot;Noto Color Emoji&quot;; font-size: 16px; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; font-weight: 400; letter-spacing: normal; orphans: 2; text-align: start; text-indent: 0px; text-transform: none; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; white-space: pre-line; background-color: rgb(255, 255, 255); text-decoration-thickness: initial; text-decoration-style: initial; text-decoration-color: initial;"></div><hr class="border-gray-100 dark:border-gray-850" style="box-sizing: border-box; border-width: 1px 0px 0px; border-style: solid; border-color: rgb(236, 236, 236); border-image: initial; margin: 3em 0px; padding: 0px; height: 0px; color: rgb(78, 78, 78); margin-block: 16px; font-family: -apple-system, BlinkMacSystemFont, Inter, Vazirmatn, ui-sans-serif, system-ui, &quot;Segoe UI&quot;, Roboto, Ubuntu, Cantarell, &quot;Noto Sans&quot;, sans-serif, &quot;Helvetica Neue&quot;, Arial, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;, &quot;Segoe UI Symbol&quot;, &quot;Noto Color Emoji&quot;; font-size: 16px; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; font-weight: 400; letter-spacing: normal; orphans: 2; text-align: start; text-indent: 0px; text-transform: none; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; white-space: pre-line; background-color: rgb(255, 255, 255); text-decoration-thickness: initial; text-decoration-style: initial; text-decoration-color: initial;"><h3 dir="auto" style="box-sizing: border-box; border: 0px solid oklch(0.21 0.034 264.665); margin: 0px 0px 0.6em; padding: 0px; font-size: 1.25em; font-weight: 600; color: oklch(0.21 0.034 264.665); line-height: 1.6; margin-block: 4px; --tw-font-weight: 600; font-family: -apple-system, BlinkMacSystemFont, Inter, Vazirmatn, ui-sans-serif, system-ui, &quot;Segoe UI&quot;, Roboto, Ubuntu, Cantarell, &quot;Noto Sans&quot;, sans-serif, &quot;Helvetica Neue&quot;, Arial, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;, &quot;Segoe UI Symbol&quot;, &quot;Noto Color Emoji&quot;; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; letter-spacing: normal; orphans: 2; text-align: start; text-indent: 0px; text-transform: none; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; white-space: pre-line; background-color: rgb(255, 255, 255); text-decoration-thickness: initial; text-decoration-style: initial; text-decoration-color: initial;">📌 Thank You</h3><p dir="auto" style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0px 0px 1.25em; padding: 0px; margin-block: 0px; color: rgb(78, 78, 78); font-family: -apple-system, BlinkMacSystemFont, Inter, Vazirmatn, ui-sans-serif, system-ui, &quot;Segoe UI&quot;, Roboto, Ubuntu, Cantarell, &quot;Noto Sans&quot;, sans-serif, &quot;Helvetica Neue&quot;, Arial, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;, &quot;Segoe UI Symbol&quot;, &quot;Noto Color Emoji&quot;; font-size: 16px; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; font-weight: 400; letter-spacing: normal; orphans: 2; text-align: start; text-indent: 0px; text-transform: none; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; white-space: pre-line; background-color: rgb(255, 255, 255); text-decoration-thickness: initial; text-decoration-style: initial; text-decoration-color: initial;">Thank you for your time and effort in maintaining such a great tool.<br style="box-sizing: border-box; border: 0px solid rgb(78, 78, 78); margin: 0px; padding: 0px;">I look forward to your response.</p><!--EndFragment--> </body> </html> ### Relevant log output ```shell ``` ### OS _No response_ ### GPU _No response_ ### CPU _No response_ ### Ollama version _No response_
GiteaMirror added the bugneeds more info labels 2026-04-29 04:47:33 -05:00
GiteaMirror commented 2026-04-29 04:47:35 -05:00
Author
Owner
Copy Link

@jmorganca commented on GitHub (Jun 13, 2025):

Hi there, which version of Ollama is this? In a more recent version this performance issue was fixed.

<!-- gh-comment-id:2968703374 --> @jmorganca commented on GitHub (Jun 13, 2025): Hi there, which version of Ollama is this? In a more recent version this performance issue was fixed.
GiteaMirror commented 2026-04-29 04:47:36 -05:00
Author
Owner
Copy Link

@jmorganca commented on GitHub (Jun 13, 2025):

Oops, sorry. Just saw the version you had.

<!-- gh-comment-id:2968704577 --> @jmorganca commented on GitHub (Jun 13, 2025): Oops, sorry. Just saw the version you had.
GiteaMirror commented 2026-04-29 04:47:39 -05:00
Author
Owner
Copy Link

@AmesianX commented on GitHub (Jun 13, 2025):

Hi there, which version of Ollama is this? In a more recent version this performance issue was fixed.

I always remove the "ollama/ollama:latest" docker image completely using "docker rmi ollama/ollama:latest" before running Ollama with Docker.
If you have fixed any issues in the latest version of Ollama, you must definitely replace the Docker image with the latest version.
Otherwise, many people will experience inconvenience.

<!-- gh-comment-id:2968734384 --> @AmesianX commented on GitHub (Jun 13, 2025): > Hi there, which version of Ollama is this? In a more recent version this performance issue was fixed. I always remove the "ollama/ollama:latest" docker image completely using "docker rmi ollama/ollama:latest" before running Ollama with Docker. If you have fixed any issues in the latest version of Ollama, you must definitely replace the Docker image with the latest version. Otherwise, many people will experience inconvenience.
GiteaMirror commented 2026-04-29 04:47:40 -05:00
Author
Owner
Copy Link

@apunkt commented on GitHub (Jun 13, 2025):

Tested v0.9.0 native install in Ubuntu 24.04 with qwen3:30b-a3b on Tesla M40:

old-engine: 18.15 token/s
new-engine: 18.65 token/s

<!-- gh-comment-id:2969571221 --> @apunkt commented on GitHub (Jun 13, 2025): Tested v0.9.0 native install in Ubuntu 24.04 with qwen3:30b-a3b on Tesla M40: old-engine: 18.15 token/s new-engine: 18.65 token/s
GiteaMirror commented 2026-04-29 04:47:41 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Jun 13, 2025):

Server logs will aid in debugging.

<!-- gh-comment-id:2969770616 --> @rick-github commented on GitHub (Jun 13, 2025): [Server logs](https://github.com/ollama/ollama/blob/main/docs/troubleshooting.md#how-to-troubleshoot-issues) will aid in debugging.
GiteaMirror commented 2026-04-29 04:47:41 -05:00
Author
Owner
Copy Link

@smileyboy2019 commented on GitHub (Jun 14, 2025):

0.9 在liunx 速度确实慢，并且导致以前的一些模型无法运行。

<!-- gh-comment-id:2972203166 --> @smileyboy2019 commented on GitHub (Jun 14, 2025): 0.9 在liunx 速度确实慢，并且导致以前的一些模型无法运行。
GiteaMirror commented 2026-04-29 04:47:42 -05:00
Author
Owner
Copy Link

@Rokazas commented on GitHub (Jun 16, 2025):

same here. The last good version for me is 0.6.8. Newer versions do not use GPU, just CPU for me. previous was sharing the tasks.

<!-- gh-comment-id:2976848042 --> @Rokazas commented on GitHub (Jun 16, 2025): same here. The last good version for me is 0.6.8. Newer versions do not use GPU, just CPU for me. previous was sharing the tasks.
GiteaMirror commented 2026-04-29 04:47:42 -05:00
Author
Owner
Copy Link

@rick-github commented on GitHub (Jun 16, 2025):

Server logs will aid in debugging.

<!-- gh-comment-id:2976885712 --> @rick-github commented on GitHub (Jun 16, 2025): [Server logs](https://github.com/ollama/ollama/blob/main/docs/troubleshooting.md#how-to-troubleshoot-issues) will aid in debugging.
GiteaMirror commented 2026-04-29 04:47:43 -05:00
Author
Owner
Copy Link

@fyun89 commented on GitHub (Jun 17, 2025):

@AmesianX how are you including this environment variable? I tried to look for this but couldnt find it.

I normally use the background service (systctrl) on my Ubuntu machine. So I thought having the variable in ~/.bashrc would do the trick but I did not see any boost in performance.

I also tried using the "-e" flag at the end of "ollama serve" but it didnt recognize that.

-Edit-
I also tried updating the ollama.service config file but I still don't see the mentioned recovery of speed.

<!-- gh-comment-id:2979058681 --> @fyun89 commented on GitHub (Jun 17, 2025): @AmesianX how are you including this environment variable? I tried to look for this but couldnt find it. I normally use the background service (systctrl) on my Ubuntu machine. So I thought having the variable in ~/.bashrc would do the trick but I did not see any boost in performance. I also tried using the "-e" flag at the end of "ollama serve" but it didnt recognize that. -Edit- I also tried updating the ollama.service config file but I still don't see the mentioned recovery of speed.
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
amd
api
app
bug
build
cli
cloud
compatibility
context-length
create
docker
documentation
embeddings
feature request
feedback wanted
good first issue
gpt-oss
gpu
harmony
help wanted
image
install
intel
js
launch
linux
macos
memory
mlx
model
needs more info
networking
nvidia
ollama.com
performance
pull-request
Mirrored from GitHub Pull Request
 python
question
registry
rendering
thinking
tools
top
vulkan
windows
wsl
No Label bug needs more info
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/ollama#53809
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?