Blur Busters Forums

Sandy wrote: ↑

26 Jun 2024, 14:34

Yes, I also found the hit registration problem! As far as simply solving input delay is concerned, many people don’t know that disabling NVIDIA thread optimization (globally disabling) and restarting the computer can eliminate input delay. Now my input delay problem is solved. Next, I will continue to study the hit registration problem. I suspect the problem is still in the NVIDIA control panel.

howiec wrote: ↑

26 Jun 2024, 14:45

Sandy wrote: ↑

26 Jun 2024, 14:34

Yes, I also found the hit registration problem! As far as simply solving input delay is concerned, many people don’t know that disabling NVIDIA thread optimization (globally disabling) and restarting the computer can eliminate input delay. Now my input delay problem is solved. Next, I will continue to study the hit registration problem. I suspect the problem is still in the NVIDIA control panel.

Unfortunately, the underlying inconsistency is not a netcode/network/hit-reg issue (although those effects obviously can make things feel worse).

What I'm describing is the fact that BOTH the arc-distance AND rate of angle change are heavily affected by any system performance or workload changes even if you send the exact same CPI counts per period to the game.

This problem is not the same as network/netcode related calculations that determine when and where your shots are fired from vs objects/targets positions in the game state vs when and where your target is from their respective (client) end.

howiec wrote: ↑

26 Jun 2024, 14:45

Sandy wrote: ↑

26 Jun 2024, 14:34

Yes, I also found the hit registration problem! As far as simply solving input delay is concerned, many people don’t know that disabling NVIDIA thread optimization (globally disabling) and restarting the computer can eliminate input delay. Now my input delay problem is solved. Next, I will continue to study the hit registration problem. I suspect the problem is still in the NVIDIA control panel.

Unfortunately, the underlying inconsistency is not a netcode/network/hit-reg issue (although those effects obviously can make things feel worse).

What I'm describing is the fact that BOTH the aim arc-distance AND rate of angle change are heavily affected by any system performance or workload changes even if you send the exact same CPI counts per period to the game.

This problem is not the same as network/netcode related calculations that determine when and where your shots are fired from vs objects/targets positions in the game state vs when and where your target is from their respective (client) end.

Sandy wrote: ↑

27 Jun 2024, 11:32

Maybe try setting CPU C1E C3 C6 in the motherboard BIOS or completely disabling all power saving modes?

Sandy wrote: ↑

27 Jun 2024, 11:32

howiec wrote: ↑

26 Jun 2024, 14:45

Sandy wrote: ↑

26 Jun 2024, 14:34

Yes, I also found the hit registration problem! As far as simply solving input delay is concerned, many people don’t know that disabling NVIDIA thread optimization (globally disabling) and restarting the computer can eliminate input delay. Now my input delay problem is solved. Next, I will continue to study the hit registration problem. I suspect the problem is still in the NVIDIA control panel.

Unfortunately, the underlying inconsistency is not a netcode/network/hit-reg issue (although those effects obviously can make things feel worse).

What I'm describing is the fact that BOTH the aim arc-distance AND rate of angle change are heavily affected by any system performance or workload changes even if you send the exact same CPI counts per period to the game.

This problem is not the same as network/netcode related calculations that determine when and where your shots are fired from vs objects/targets positions in the game state vs when and where your target is from their respective (client) end.

Maybe try setting CPU C1E C3 C6 in the motherboard BIOS or completely disabling all power saving modes?

howiec wrote: ↑

27 Jun 2024, 13:54

Sandy wrote: ↑

27 Jun 2024, 11:32

howiec wrote: ↑

26 Jun 2024, 14:45

Sandy wrote: ↑

26 Jun 2024, 14:34

Yes, I also found the hit registration problem! As far as simply solving input delay is concerned, many people don’t know that disabling NVIDIA thread optimization (globally disabling) and restarting the computer can eliminate input delay. Now my input delay problem is solved. Next, I will continue to study the hit registration problem. I suspect the problem is still in the NVIDIA control panel.

Unfortunately, the underlying inconsistency is not a netcode/network/hit-reg issue (although those effects obviously can make things feel worse).

What I'm describing is the fact that BOTH the aim arc-distance AND rate of angle change are heavily affected by any system performance or workload changes even if you send the exact same CPI counts per period to the game.

This problem is not the same as network/netcode related calculations that determine when and where your shots are fired from vs objects/targets positions in the game state vs when and where your target is from their respective (client) end.

Maybe try setting CPU C1E C3 C6 in the motherboard BIOS or completely disabling all power saving modes?

Disabling C-states / power-saving features are certainly recommended and are amongst the most obvious performance optimizations.

I have a pretty well-optimized PC across the board.

However, I've settled for not disabling every single Windows service possible due to the underlying fact that Apex itself places different workloads due to differing resource requirements between maps, etc. such that you would have to have different settings to compensate per map, and Apex / Steam client updates, etc. also make a significant difference.

Not to mention that you can't defer Windows, antivirus, and application updates forever (unless maybe if you dual boot) and you can't precisely control Windows thread, processes, handles, affinity, etc.

In other words, you could have the highest performing PC that is 100% identical in both HW and SW settings but even then, you will have different effective aim/sens per map. Sacrificing PC functionality or going thru massive inconveniences just to try and maintain consistency isn't worth it when you can only achieve a relatively short period of consistency with Apex before anything screws it up.

Sandy wrote: ↑

28 Jun 2024, 07:41

I just tried disabling all C-states and various combinations and my situation got worse. Now I reset the BIOS and disabled NVIDIA thread optimization and my input lag is gone again, but I still have the hit registration issue. That is to say, NVIDIA thread optimization is directly related to the mouse to screen delay. This setting has been repeatedly verified. I tried Win10 yesterday and now I switched back to Win7.

• Assuming you've increased the performance of your PC, such as by using higher clocks or lowering RAM latency, or disabling C-states, etc., hit-reg shouldn't be negatively affected (generally speaking) because that is more directly related to netcode, connection quality, and server/network quality.
• Threaded optimization obviously can impact parallelism and latency/consistency depending on various details, so I doubt anyone's arguing against it having a real effect depending on the game.
• Ultimately, the configuration (HW & SW incl. settings/tweaks) that feels or works best for you may be different for someone else.
• I can achieve very low input delay for Apex in my system but that's not the issue.
  The problem is consistency in Apex. It is abundantly clear that Apex processes/handles aim input, angle adjustments, etc. vs rendering in a pathetically poor way. This is easily tested and reproducible across users.
  So even if you had a supercomputer and negliglbe input delay, and perfect network and server performance, the underlying way Apex is coded for input/aim handling would still result in the same aforementioned problems.

howiec wrote: ↑

01 Jul 2024, 10:39

Sandy wrote: ↑

28 Jun 2024, 07:41

I just tried disabling all C-states and various combinations and my situation got worse. Now I reset the BIOS and disabled NVIDIA thread optimization and my input lag is gone again, but I still have the hit registration issue. That is to say, NVIDIA thread optimization is directly related to the mouse to screen delay. This setting has been repeatedly verified. I tried Win10 yesterday and now I switched back to Win7.

A few things to note:

• Assuming you've increased the performance of your PC, such as by using higher clocks or lowering RAM latency, or disabling C-states, etc., hit-reg shouldn't be negatively affected (generally speaking) because that is more directly related to netcode, connection quality, and server/network quality.
• Threaded optimization obviously can impact parallelism and latency/consistency depending on various details, so I doubt anyone's arguing against it having a real effect depending on the game.
• Ultimately, the configuration (HW & SW incl. settings/tweaks) that feels or works best for you may be different for someone else.
• I can achieve very low input delay for Apex in my system but that's not the issue.
  The problem is consistency in Apex. It is abundantly clear that Apex processes/handles aim input, angle adjustments, etc. vs rendering in a pathetically poor way. This is easily tested and reproducible across users.
  So even if you had a supercomputer and negliglbe input delay, and perfect network and server performance, the underlying way Apex is coded for input/aim handling would still result in the same aforementioned problems.

Sandy wrote: ↑

01 Jul 2024, 11:04

I tested Cs1.6 and there was still a hit registration problem in single-player games without an Internet connection. It felt like your screen happened ahead of time, but the real event was delayed. In short, you need to know that everything behind this points to "computing" and NVIDIA controls too much of Windows computing. I was testing different versions of drivers today and found that the input delay between different versions of drivers is very different.

Sandy wrote: ↑

01 Jul 2024, 11:04

howiec wrote: ↑

01 Jul 2024, 10:39

Sandy wrote: ↑

28 Jun 2024, 07:41

I just tried disabling all C-states and various combinations and my situation got worse. Now I reset the BIOS and disabled NVIDIA thread optimization and my input lag is gone again, but I still have the hit registration issue. That is to say, NVIDIA thread optimization is directly related to the mouse to screen delay. This setting has been repeatedly verified. I tried Win10 yesterday and now I switched back to Win7.

A few things to note:

• Assuming you've increased the performance of your PC, such as by using higher clocks or lowering RAM latency, or disabling C-states, etc., hit-reg shouldn't be negatively affected (generally speaking) because that is more directly related to netcode, connection quality, and server/network quality.
• Threaded optimization obviously can impact parallelism and latency/consistency depending on various details, so I doubt anyone's arguing against it having a real effect depending on the game.
• Ultimately, the configuration (HW & SW incl. settings/tweaks) that feels or works best for you may be different for someone else.
• I can achieve very low input delay for Apex in my system but that's not the issue.
  The problem is consistency in Apex. It is abundantly clear that Apex processes/handles aim input, angle adjustments, etc. vs rendering in a pathetically poor way. This is easily tested and reproducible across users.
  So even if you had a supercomputer and negliglbe input delay, and perfect network and server performance, the underlying way Apex is coded for input/aim handling would still result in the same aforementioned problems.

I tested Cs1.6 and there was still a hit registration problem in single-player games without an Internet connection. It felt like your screen happened ahead of time, but the real event was delayed. In short, you need to know that everything behind this points to "computing" and NVIDIA controls too much of Windows computing. I was testing different versions of drivers today and found that the input delay between different versions of drivers is very different. This surprised me. I never touched the driver before.