Allow EQ to Use More Resources?

Is there a way to allow EQ to use more system resources? Maybe more VRAM? Could it mitigate lag, for example, where there are a lot of player models in the same place (PoK, Guild Lobby, Raids).

My System

• Intel i5-12600K
• Nvidia RTX 4060Ti 8GB
• 32GB DDR4 3600
• 2TB gen3 NVME
• Play EQ in a 1080p window on a 1440p desktop with shadows off but all modern player/mount/pet models on.

The problem: Major reduction in FPS when around too many different player models (textures) or in a very densely decorated guild hall (again, textures).
Testing

• Using MSI Afterburner
• Caveat: I'm running more than EQ including multiple tabs on Firefox including streaming movie.

Results

• Freshly loaded into the lobby
  • CPU 42C, 28.6w
  • Used System RAM 10,344 MB
  • GPU 45C, 11.5w
  • Used Video RAM 1,184 MB
  • 176 FPS
• Facing lobby pile with hidden pets, hidden mercs, and hidden players (only weapons visible)
  • CPU 42C, 31.4w
  • Used System RAM 10,220 MB
  • GPU 46C, 25.5w
  • Used Video RAM 1,214 MB
  • 90 FPS
• Facing lobby pile with player and pet models turned on
  • CPU 43C, 31.4w
  • Used System RAM 10,229 MB
  • GPU 43C, 25.7w
  • Used Video RAM 1,203 MB
  • 31 FPS

I recommend that players with modern GPU disable UseLitBatches in your eqclient.ini file.

UseLitBatches=FALSE



This is an old option that uses the CPU to sort graphics data into batches for the GPU. Modern GPUs don't necessarily benefit from this pre-processing and in many cases it lowers performance. By disabling UseLitBatches you save CPU processing time and allow the GPU to handle all the data instead. This is known to improve performance, on modern systems, while viewing large numbers of characters ("lag piles") and certain animated areas like the Kod`Taz graveyard.

I will try this tonight and report back.

Cairbrae are you aware that an indirect result of disabling LitBatches is a decrease in total clip plane? Although FPS improves, clip plane distance also decreases. Would it make sense that clip plane is being impacted related to this setting? I would want the development team to be aware so that LitBatches is never force disabled, thereby decreasing everyone's clip plane.

As an example on an RTX 4090: If one disables LitBatches then the clip plane in Greater Faydark decreases, and FPS increases (i.e., distance-oriented fog effect occurs closer to the player's POV). If one enables LitBatches, the clip plane in Greater Faydark will increase, and FPS will then perhaps relatedly decrease.

I will look for that as well.

Just tested clip plane relationship with LitBatches again. Seeing the same as ~1 year ago. I would estimate the clip plane horizon "fog of war" impact in Greater Faydark to be around -400 units in-game with LitBatches disabled.

That is: With LitBatches enabled, you gain 400 units of horizon visibility due to an extended clip plane (in Greater Faydark at least). With LitBatches disabled, you lose 400 units of horizon visibility.

Further, in comparing the Bristlebane POK lagpile with LitBatches enabled/disabled, I'm coming in at 29 FPS with the same field of view regardless of LitBatches' setting.

So I'd be interested to know how much of the FPS loss is actually related to CPU batching versus the indirect clip plane +/- relationship. Would also be interested to know if this clip plane LitBatches relationship is known by the devs and intended to exist.

Unfortunately, as it turns out UseLitBatches is already set to FALSE, so I turned it on for some testing.

• Freshly loaded into the lobby
  • CPU 37C, 24.9w
  • Used System RAM 9745 MB
  • GPU 43C, 21.1w
  • Used Video RAM 886 MB
  • 179 FPS
• Facing lobby pile with hidden pets, hidden mercs, and hidden players (only weapons visible)
  • CPU 44C, 21.1w
  • Used System RAM 9,754 MB
  • GPU 43C, 26.4w
  • Used Video RAM 895 MB
  • 90 FPS
• Facing lobby pile with player and pet models turned on
  • CPU 41C, 26.6w
  • Used System RAM 9,729 MB
  • GPU 44C, 21.0w
  • Used Video RAM 966 MB
  • 27 FPS

It would look as though (for my configuration), the setting provides a negligible performance difference in the Guild Lobby. I'll switch back to see if the change does anything beneficial.

No change. I even deleted eqclient.ini, loaded default and tested with both True/False values and both were within 3 FPS of each other.

Has greatly improved the guild lobby performance but didn't had an effect on non character lag like a not so crowded PoK so it works as described. Thank you for this note.

My concern is that the developers need to be aware of all unintended and unexpected relationships to LitBatches beyond just CPU-GPU batching (such as clip plane decreasing when LiBatches = disabled).

We would not want a forced change that impacts other universal/critical visual elements in the game.

I can confirm what Lodestar is writing. Just tested how the Far Clip Plane does work with UseLitBatches and found an old bug that is not related to it while doing so. It seems that the possible performance improvement by using UseLitBatches=FALSE is just the result of the reduced Far Clip Plane. I also tested Far Clip Plane 0 but the difference there is small.

Location: Greater Faydark when zoning in from PoK.
Far Clip Plane: 100

UseLitBatches=TRUE (standard)



UseLitBatches=FALSE

This was my estimation as well. Nothing is so significant to improving FPS in EQ as simply not being able to see things/looking away.

On the other hand, it feels like I'm leaving a LOT of performance headroom on the table with my CPU and GPU barely being utilized.

While the GPU does not get used at all or only for the LitBatches, there are new bottlenecks and other issues that can and do cause lag. The problem never was that unused ressources do not get used.

The game was fine and performed great on new CPUs before the work on the new UI engine and DirectX 11 got started and I do not get the point why all those troublesome changes got made as they hardly can improve something that was already about perfect.