MagicBeanDev/Hookshot

Archived

Watch

Fork 0

PERFORMANCE IS HORRENDOUS ON A 5070 and a 5700x #844

Closed

opened 2026-11-08 22:58:15 +00:00 by jobedthedope · 2 comments

jobedthedope commented

2026-11-08 22:58:15 +00:00

(Migrated from github.com)

Hookshot version

v2.0.5

What kind of problem

Low fps or stutter

What happened

why is the fps so low? this version currently has no bugs but the fps is so low, lower than the emulator even

Where in the game

kokiri forest

How often

Every time

Graphics card and driver version

rtx 5070 latest drivers

CPU, RAM and operating system

5700x 16gb ram win 11

Changes from the defaults

No response

Other software hooked into the game

No, nothing hooked in

hookshot_log.txt

Screenshots or video

Before you send it

I'm on the newest release, or the problem is still there on it.
I searched the open issues for this problem.
I'm not attaching game files, keys, hookshot.exe, the .bin caches or saves.

### Hookshot version v2.0.5 ### What kind of problem Low fps or stutter ### What happened why is the fps so low? this version currently has no bugs but the fps is so low, lower than the emulator even <img width="2560" height="1440" alt="Image" src="https://github.com/user-attachments/assets/1a2dbc4a-2b17-436d-acb9-75843b880306" /> ### Where in the game kokiri forest ### How often Every time ### Graphics card and driver version rtx 5070 latest drivers ### CPU, RAM and operating system 5700x 16gb ram win 11 ### Changes from the defaults _No response_ ### Other software hooked into the game No, nothing hooked in ### hookshot_log.txt  ### Screenshots or video <img width="2560" height="1440" alt="Image" src="https://github.com/user-attachments/assets/e174b3be-d3e7-416b-bbdd-1598facf38db" /> ### Before you send it - [x] I'm on the newest release, or the problem is still there on it. - [x] I searched the open issues for this problem. - [x] I'm not attaching game files, keys, hookshot.exe, the .bin caches or saves.

IgorAntunes66 commented

2026-11-08 23:17:30 +00:00

(Migrated from github.com)

RTX 5070: persistent gameplay stutter after cache warm-up, irregular frame times and occasional long stalls

Summary

I experience persistent stuttering while walking around Kokiri Forest and moving the camera. Frame generation improves the displayed FPS but does not eliminate the interruptions.

Repeated traversal after cache warm-up still produces uneven frame delivery. Some samples contain no queued shader jobs or reported shader/pipeline waits, so initial shader compilation may not explain all of the stuttering.

Hardware and environment

OS: Windows 11
CPU: Intel Core i7-12700K
GPU: NVIDIA GeForce RTX 5070, 12 GB
RAM: 32 GB DDR5
NVIDIA driver: 621.01 at the time of the measurements
Installation: Hookshot Builder
Game language: Portuguese (Brazil)
Vulkan 1.4; mesh shaders, host memory import and pipeline libraries detected
Controller: DualSense, recognized through SDL

Version clarification

The initial screenshot identifies v2.0.2, source 998ef8f9. The installation was subsequently updated during troubleshooting.

The latest local log identifies:

[hos host] hookshot v2.0.5 (source cd9cebb6)

The earlier measurements below were collected before this latest log. They should not all be interpreted as a controlled v2.0.5 reproduction.

Steps used to reproduce

Load a save in Kokiri Forest.
Walk around the village and rotate the camera.
Repeat the same route within the same session to reduce first-visit compilation effects.
Compare frame generation off versus 2x, and frame pacing on versus off.

Expected behavior

More consistent frame delivery when revisiting an already explored area, or diagnostic information that distinguishes compilation stalls from other runtime stalls.

Observations from earlier troubleshooting sessions

Gameplay summaries commonly reported approximately 40–50 FPS, with some lighter views approaching 60 FPS.
With frame generation enabled at 2x, the displayed counter reached approximately 91 FPS, but visible stuttering remained.
Presentation interval summaries recorded frames longer than 40 ms.
One summary reported:

GPU time 15.0 ms a frame (the longest submission 518.1 ms)

In other stuttering sections, the summaries reported:

0.00 waits for shaders or pipelines (0.0 ms)
0 jobs queued

GPU frame times were often approximately 10–14 ms while overall frame delivery was slower.
Some samples reported zero failed graphics programs.
Internal vcores diagnostics included occasional long waits. I do not know whether these waits caused the visible interruptions.

These are sampled and aggregate diagnostics, not a timestamped frame-by-frame trace.

Tests already attempted

Test | Observation -- | -- Frame generation disabled | Uneven frame delivery remained Frame pacing enabled/disabled | Disabling it did not eliminate the stalls Repeated traversal with an increasingly warm cache | Compilation activity decreased, but stuttering remained Reduced overall GPU memory occupancy | Stalls were not eliminated

Total GPU memory use measured through nvidia-smi decreased from approximately 11.0–11.3 GiB to approximately 9.6–9.7 GiB.

These are adapter-wide measurements, not Hookshot-only allocations. The settings changed during testing, so this was not a controlled comparison of quality presets.

Latest v2.0.5 log — separate findings

The latest log reports:

3150 pipelines (320708 ms)
53617 programs (97 failed)
translation 453166 ms in all

Its initial program-cache message reports:

0 program keys kept, 0 KB

A pipeline cache was present at startup.

The latest tail shows approximately 60 FPS, but with only 2–211 draws and 14 dispatches per frame. This is not comparable to the earlier outdoor gameplay samples with roughly 1,700–3,000 draws. I am therefore not treating those 60 FPS summaries as evidence that outdoor stuttering is fixed.

Latest saved configuration

This configuration was read after the latest session and does not necessarily describe every earlier measurement:

quality_preset=0
fullscreen=1
gpu_device=1
frame_generation=2
frame_pacing=1
resolution=1
upscaling=0
sharpness=0
handheld_profile=0
ultrawide=0
robust_buffers=1
fps_counter=1
game_speed=1

Related reports

#388: cold-cache stuttering and flickering on an RTX 5070 Ti
#124: RTX 5070 stuttering in new areas

I cannot confirm whether this is the same underlying issue. Please let me know if these findings belong in an existing thread.

Diagnostic questions

Which diagnostics would help distinguish runtime synchronization/resource-loading stalls from shader or pipeline compilation?
What is needed to investigate the isolated 518 ms GPU submission?
Should the 97 failed graphics programs in the v2.0.5 log be investigated separately?

I can provide a fresh, short gameplay reproduction with fixed settings and a sanitized log. No game files, keys, generated executables or caches are included in this report.

</html> </html>

<html> <body> <html><head></head><body>RTX 5070: persistent gameplay stutter after cache warm-up, irregular frame times and occasional long stalls<h2>Summary</h2>I experience persistent stuttering while walking around Kokiri Forest and moving the camera. Frame generation improves the displayed FPS but does not eliminate the interruptions.Repeated traversal after cache warm-up still produces uneven frame delivery. Some samples contain no queued shader jobs or reported shader/pipeline waits, so initial shader compilation may not explain all of the stuttering.<h2>Hardware and environment</h2><ul><li>OS: Windows 11</li><li>CPU: Intel Core i7-12700K</li><li>GPU: NVIDIA GeForce RTX 5070, 12 GB</li><li>RAM: 32 GB DDR5</li><li>NVIDIA driver: 621.01 at the time of the measurements</li><li>Installation: Hookshot Builder</li><li>Game language: Portuguese (Brazil)</li><li>Vulkan 1.4; mesh shaders, host memory import and pipeline libraries detected</li><li>Controller: DualSense, recognized through SDL</li></ul><h2>Version clarification</h2>The initial screenshot identifies v2.0.2, source <code>998ef8f9</code>. The installation was subsequently updated during troubleshooting.The latest local log identifies:<pre><code>[hos host] hookshot v2.0.5 (source cd9cebb6)</code></pre>The earlier measurements below were collected before this latest log. They should not all be interpreted as a controlled v2.0.5 reproduction.<h2>Steps used to reproduce</h2><ol start="1"><li>Load a save in Kokiri Forest.</li><li>Walk around the village and rotate the camera.</li><li>Repeat the same route within the same session to reduce first-visit compilation effects.</li><li>Compare frame generation off versus 2x, and frame pacing on versus off.</li></ol><h2>Expected behavior</h2>More consistent frame delivery when revisiting an already explored area, or diagnostic information that distinguishes compilation stalls from other runtime stalls.<h2>Observations from earlier troubleshooting sessions</h2><ul><li>Gameplay summaries commonly reported approximately 40–50 FPS, with some lighter views approaching 60 FPS.</li><li>With frame generation enabled at 2x, the displayed counter reached approximately 91 FPS, but visible stuttering remained.</li><li>Presentation interval summaries recorded frames longer than 40 ms.</li><li>One summary reported:</li></ul><pre><code>GPU time 15.0 ms a frame (the longest submission 518.1 ms)</code></pre><ul><li>In other stuttering sections, the summaries reported:</li></ul><pre><code>0.00 waits for shaders or pipelines (0.0 ms) 0 jobs queued</code></pre><ul><li>GPU frame times were often approximately 10–14 ms while overall frame delivery was slower.</li><li>Some samples reported zero failed graphics programs.</li><li>Internal <code>vcores</code> diagnostics included occasional long waits. I do not know whether these waits caused the visible interruptions.</li></ul>These are sampled and aggregate diagnostics, not a timestamped frame-by-frame trace.<h2>Tests already attempted</h2> Test | Observation -- | -- Frame generation disabled | Uneven frame delivery remained Frame pacing enabled/disabled | Disabling it did not eliminate the stalls Repeated traversal with an increasingly warm cache | Compilation activity decreased, but stuttering remained Reduced overall GPU memory occupancy | Stalls were not eliminated Total GPU memory use measured through <code>nvidia-smi</code> decreased from approximately 11.0–11.3 GiB to approximately 9.6–9.7 GiB.These are adapter-wide measurements, not Hookshot-only allocations. The settings changed during testing, so this was not a controlled comparison of quality presets.<h2>Latest v2.0.5 log — separate findings</h2>The latest log reports:<pre><code>3150 pipelines (320708 ms) 53617 programs (97 failed) translation 453166 ms in all</code></pre>Its initial program-cache message reports:<pre><code>0 program keys kept, 0 KB</code></pre>A pipeline cache was present at startup.The latest tail shows approximately 60 FPS, but with only 2–211 draws and 14 dispatches per frame. This is not comparable to the earlier outdoor gameplay samples with roughly 1,700–3,000 draws. I am therefore not treating those 60 FPS summaries as evidence that outdoor stuttering is fixed.<h2>Latest saved configuration</h2>This configuration was read after the latest session and does not necessarily describe every earlier measurement:<pre><code>quality_preset=0 fullscreen=1 gpu_device=1 frame_generation=2 frame_pacing=1 resolution=1 upscaling=0 sharpness=0 handheld_profile=0 ultrawide=0 robust_buffers=1 fps_counter=1 game_speed=1</code></pre><h2>Related reports</h2><ul><li>#388: cold-cache stuttering and flickering on an RTX 5070 Ti</li><li>#124: RTX 5070 stuttering in new areas</li></ul>I cannot confirm whether this is the same underlying issue. Please let me know if these findings belong in an existing thread.<h2>Diagnostic questions</h2><ul><li>Which diagnostics would help distinguish runtime synchronization/resource-loading stalls from shader or pipeline compilation?</li><li>What is needed to investigate the isolated 518 ms GPU submission?</li><li>Should the 97 failed graphics programs in the v2.0.5 log be investigated separately?</li></ul>I can provide a fresh, short gameplay reproduction with fixed settings and a sanitized log. No game files, keys, generated executables or caches are included in this report.</body></html> </body> </html>

NotRexed commented

2026-11-09 05:59:05 +00:00

(Migrated from github.com)

Tracked in #995, follow there. That issue collects every report of low FPS while the GPU is not fully used, and shader stutter, and this one is linked from it with its GPU, version and log details. If it still happens on v2.0.6 (https://github.com/NotRexed/Hookshot/releases/tag/v2.0.6; Update next to Play in the launcher), please add a comment there with your hookshot_log.txt.

Commenting is not possible because the repository is archived.