low latency mode [ON] work. oscilloscope measurement

Everything about latency. Tips, testing methods, mouse lag, display lag, game engine lag, network lag, whole input lag chain, VSYNC OFF vs VSYNC ON, and more! Input Lag Articles on Blur Busters.
Post Reply
blackstorm82
Posts: 23
Joined: 01 Jun 2020, 20:23

low latency mode [ON] work. oscilloscope measurement

Post by blackstorm82 » 02 Aug 2021, 18:19

Currently, there is data measured by a user with an oscilloscope.

his argument

NVIDIA LLM [ON] MODE
"Works regardless of GPU load rate" said

He capped 60 FPS with a 360HZ monitor and tested the LLM ON and OFF.

His test environment is full of loopholes.
Are you sure that LLM ON mode will work even if it's not 99% GPU?

His argument seems like NVIDIA is saying that as long as you turn on Reflex, it will always work.
What do you think?

He's also the one who claims that the GSYNC DISPLAY module has a GPU scaling feature.
He's a big believer in the mechanical figures of his oscilloscope.

His argument is the same as saying that if you set the Maximum pre-rendered frames setting to 1-4 in the past, it will work unconditionally regardless of the GPU load rate.

If his assertion is correct, he is renewing the claims of intellectuals around the world.
Does LLM mode and max prerender really work without regard to graphics card load rate?

I trusted and learned from Blubuster's material. I learned that llm only works in gpu bound situations.

Please help me to convince me. Blubuster officials or members

User avatar
jorimt
Posts: 2481
Joined: 04 Nov 2016, 10:44
Location: USA

Re: low latency mode [ON] work. oscilloscope measurement

Post by jorimt » 03 Aug 2021, 08:43

blackstorm82 wrote:
02 Aug 2021, 18:19
Are you sure that LLM ON mode will work even if it's not 99% GPU?
In supported games, LLM will reduce the maximum number of pre-rendered frames that can be in the render queue whether the GPU is at 99% or not.

However, when the GPU isn't at 99%, the render queue may already be filled at or below what LLM is set it to anyway, even if LLM was off, which means you won't notice an input lag difference until the GPU is at 99% and LLM is on.
(jorimt: /jor-uhm-tee/)
Author: Blur Busters "G-SYNC 101" Series

Displays: ASUS PG27AQN, LG 48CX VR: Beyond, Quest 3, Reverb G2, Index OS: Windows 11 Pro Case: Fractal Design Torrent PSU: Seasonic PRIME TX-1000 MB: ASUS Z790 Hero CPU: Intel i9-13900k w/Noctua NH-U12A GPU: GIGABYTE RTX 4090 GAMING OC RAM: 32GB G.SKILL Trident Z5 DDR5 6400MHz CL32 SSDs: 2TB WD_BLACK SN850 (OS), 4TB WD_BLACK SN850X (Games) Keyboards: Wooting 60HE, Logitech G915 TKL Mice: Razer Viper Mini SE, Razer Viper 8kHz Sound: Creative Sound Blaster Katana V2 (speakers/amp/DAC), AFUL Performer 8 (IEMs)

blackstorm82
Posts: 23
Joined: 01 Jun 2020, 20:23

Re: low latency mode [ON] work. oscilloscope measurement

Post by blackstorm82 » 03 Aug 2021, 09:40

jorimt wrote:
03 Aug 2021, 08:43
blackstorm82 wrote:
02 Aug 2021, 18:19
Are you sure that LLM ON mode will work even if it's not 99% GPU?
In supported games, LLM will reduce the maximum number of pre-rendered frames that can be in the render queue whether the GPU is at 99% or not.

However, when the GPU isn't at 99%, the render queue may already be filled at or below what LLM is set it to anyway, even if LLM was off, which means you won't notice an input lag difference until the GPU is at 99% and LLM is on.
Thank you. I'm sorry. Let me ask you one more time.

The engine of a particular game is really strange. If the rendering queue is created as "1-3" OR "2-3"
It's a game with figures that are hard to judge accurately.

Should I say that setting the LLM to "ON" and reducing it to "1" works for input delays?
(graphic card load rate is not 99%) = Not GPUBOUND

It is confusing because many people feel that the input delay has been accelerated by changing LLM to "ON" in the environment other than GPU-BOUND.
I think they are mistaken.
Last edited by blackstorm82 on 03 Aug 2021, 09:55, edited 4 times in total.

pox02
Posts: 259
Joined: 28 Sep 2018, 06:04

Re: low latency mode [ON] work. oscilloscope measurement

Post by pox02 » 03 Aug 2021, 09:45

blackstorm82 wrote:
03 Aug 2021, 09:40
jorimt wrote:
03 Aug 2021, 08:43
blackstorm82 wrote:
02 Aug 2021, 18:19
Are you sure that LLM ON mode will work even if it's not 99% GPU?
In supported games, LLM will reduce the maximum number of pre-rendered frames that can be in the render queue whether the GPU is at 99% or not.

However, when the GPU isn't at 99%, the render queue may already be filled at or below what LLM is set it to anyway, even if LLM was off, which means you won't notice an input lag difference until the GPU is at 99% and LLM is on.
Thank you.

The engine of a particular game is really strange. If the rendering queue is created as "1-3" OR "2-3"

Should I say that setting the LLM to "ON" and reducing it to "1" works for input delays?
(graphic card load rate is not 99%) = Not GPUBOUND
Try this

LLM to ultra frame limiter v3 3 below ur hz gsync on vsync off enjoy the lowest latency :)
monitors xg258q aw2518hf 27GK750F-B pg248q xg240r lg w2363d-pf xb270hu XL2546 XL2546K NXG252R

User avatar
jorimt
Posts: 2481
Joined: 04 Nov 2016, 10:44
Location: USA

Re: low latency mode [ON] work. oscilloscope measurement

Post by jorimt » 03 Aug 2021, 10:38

blackstorm82 wrote:
03 Aug 2021, 09:40
The engine of a particular game is really strange. If the rendering queue is created as "1-3" OR "2-3"
It's a game with figures that are hard to judge accurately.

Should I say that setting the LLM to "ON" and reducing it to "1" works for input delays?
(graphic card load rate is not 99%) = Not GPUBOUND
Again, all LLM does is set the maximum amount of pre-rendered frames able to render at a time. It does not set a static limit.

Pre-rendered frames aren't constantly 1 frame or 2 frames or 3 frames. Sometimes the render queue is filled with 1 frame, sometimes with 2, sometimes with 3, sometimes with 0, and so on. LLM only determines how many can be queued in it at once.

If the render queue is set to 3 pre-rendered frames with LLM off and the GPU is below 99%, the render queue can contain between 0 and 3 pre-rendered frame at any given time. It's only when the GPU is at or above 99% that the render queue typically increases to 3 in that case, as the CPU is then forced to wait for the GPU to finish, so the CPU creates more pre-rendered frames in advance while it waits.

However, if the render queue is set to 3 pre-rendered frames, you set LLM to on, and the render queue can now only have 1 pre-rendered frame, if the GPU doesn't reach 99%, the game probably would have only had 1 or less pre-rendered frames in the queue without LLM anyway, since the GPU is now able to keep up with the CPU.

Example...

Render queue set to 3 + LLM off + GPU usage below 99%:
[1 can be full or empty > 2 can be full or empty > 3 can be full or empty]

Render queue set to 3 + LLM off + GPU usage above 99%:
[1 is full > 2 is full > 3 is full]

Render queue set to 3 + LLM on + GPU usage below 99%:
[1 can be full or empty] (with LLM on, render queue is now limited to 1, so there is only 1 slot to be filled)

Render queue set to 3 + LLM on + GPU usage above 99%:
[1 is full] (with LLM on, render queue is now limited to 1, so there is only 1 slot to be filled)

It varies, but that's the idea.
(jorimt: /jor-uhm-tee/)
Author: Blur Busters "G-SYNC 101" Series

Displays: ASUS PG27AQN, LG 48CX VR: Beyond, Quest 3, Reverb G2, Index OS: Windows 11 Pro Case: Fractal Design Torrent PSU: Seasonic PRIME TX-1000 MB: ASUS Z790 Hero CPU: Intel i9-13900k w/Noctua NH-U12A GPU: GIGABYTE RTX 4090 GAMING OC RAM: 32GB G.SKILL Trident Z5 DDR5 6400MHz CL32 SSDs: 2TB WD_BLACK SN850 (OS), 4TB WD_BLACK SN850X (Games) Keyboards: Wooting 60HE, Logitech G915 TKL Mice: Razer Viper Mini SE, Razer Viper 8kHz Sound: Creative Sound Blaster Katana V2 (speakers/amp/DAC), AFUL Performer 8 (IEMs)

Post Reply