Gaming tests w/kernel 7.3

General system Gaming, sound and video
1

Running xanmod-edge, on version 7.3.1 (as of 2026-11-01), and dmemcg-booster.

First tested game: Titan Quest 2

This game consistently uses 5-6 GB of NVRAM, which pushes by RTX 4060 (w/ 8 GB) to the brink. The only way I could play before was to restart my PC, load basically nothing but Steam.

Today, I ran a fully-loaded system with TQ2. Not only was NVRAM never taxed (hit 7.3 GB briefly in one of the most graphically-intense areas, the Temple of the Moirai), there was absolutely no stuttering: never dropped below my FPS max limit of 90.

Next test will be Cities: Skylines, which has been a system killer unless I disabled almost all mods.

2

Still waiting for 7.3 on the stock and/or Zen kernels. Just had an update, but not to 7.3 yet…

7.2.11.arch1-1 -> 7.2.14.arch1-1 and 7.2.11.zen1-1 -> 7.2.14.zen1-1

3

Hmmm…it hasn’t even hit core-testing yet.

4

Isn’t that mesa only (AMD/Intel)?

5

According to this Reddit post (and subsequent Nvidia driver link), it’s now supported. How well it works is another story.

6

My initial testing of Cities: Skylines was successful. Game never got above 7.2 GB of NVRAM, although I did notice a minor spike of ZRAM usage (<10%). And this was with several mods, including Move It, which was notorious for killing my machine.

7

Moved to 7.3 Xanmod via Chaotic AUR.

─❯ uname -r
7.3.0-xanmod1-1-edge-x64v3
─❯

My GPU is a ZOTAC Nvidia GeForce RTX 3060 Ti (8GB GDDR6 VRAM). My only issues were occasional stuttering in Metro Exodus and X4: Foundations the past several weeks. The past 90 minutes or so playing Metro Exodus have been stutter-free. Next weekend (this week at work will be a long week with little time to play games) I’ll test further. But so far, so good!

8

Chaotic must be behind. I’m on the following (from this afternoon):

linux-xanmod-edge-linux-bin-x64v3 7.3.1-1
linux-xanmod-lts-linux-bin-x64v3 6.18.60-1
9

I didn’t even think to check for a bin version in the AUR.

10

After going through a crapload of kernel compiles during my time with Gentoo, I basically refuse to do that anymore.

11

It feels like kernel upgrades (e.g., from 7.2 to 7.3) are arriving later and later.

12

Try a bore or pds kernel. They are famous for better gaming performance.

13

Another update. Played another 2+ hour session, went right to the NVRAM max, never slowed down in the slightest. Still got some of these messages:

Nov 02 22:14:49 kernel: NVRM: intermapRegisterDmaMapping: Failed to insert new mapping node for range 0xDF7E020000-0xDF7E0AFFFF!
Nov 02 22:14:49 kernel: NVRM: GPU0 nvAssertFailedNoLog: Assertion failed: NV_OK == status @ gpu_vaspace.c:4586
Nov 02 22:14:49 kernel: NVRM: GPU0 nvAssertFailedNoLog: Assertion failed: NV_OK == status @ gpu_vaspace.c:2080
Nov 02 22:14:49 kernel: NVRM: nvCheckFailedNoLog: Check failed: 0 == (pMapNode->gpuMask & gpuMask) @ gpu_vaspace.c:4503
Nov 02 22:14:49 kernel: NVRM: nvCheckOkFailedNoLog: Check failed: Invalid argument to call [NV_ERR_INVALID_ARGUMENT] (0x0000001F) returned from _gvaspaceMappingInsert(pGVAS, pGpu, pVASBlock, vaLo, vaHi, flags) @ gpu_vaspace.c:2036
Nov 02 22:14:49 kernel: NVRM: nvCheckFailedNoLog: Check failed: NV_OK == status @ virt_mem_allocator_gm107.c:2563
Nov 02 22:14:49 kernel: NVRM: dmaAllocMapping_GM107: can't update VA space for mapping @vaddr=0x216a00000
Nov 02 22:14:49 kernel: NVRM: intermapRegisterDmaMapping: Failed to insert new mapping node for range 0xDF7E020000-0xDF7E0AFFFF!
Nov 02 22:14:49 kernel: NVRM: GPU0 nvAssertFailedNoLog: Assertion failed: NV_OK == status @ gpu_vaspace.c:4586
Nov 02 22:14:49 kernel: NVRM: GPU0 nvAssertFailedNoLog: Assertion failed: NV_OK == status @ gpu_vaspace.c:2080
Nov 02 22:14:51 kernel: NVRM: intermapRegisterDmaMapping: Failed to insert new mapping node for range 0xDF7E020000-0xDF7E0AFFFF!
Nov 02 22:14:51 kernel: NVRM: GPU0 nvAssertFailedNoLog: Assertion failed: NV_OK == status @ gpu_vaspace.c:4586
Nov 02 22:14:51 kernel: NVRM: GPU0 nvAssertFailedNoLog: Assertion failed: NV_OK == status @ gpu_vaspace.c:2080
Nov 02 22:15:01 kernel: NVRM: nvCheckFailedNoLog: Check failed: 0 == (pMapNode->gpuMask & gpuMask) @ gpu_vaspace.c:4503
Nov 02 22:15:01 kernel: NVRM: nvCheckOkFailedNoLog: Check failed: Invalid argument to call [NV_ERR_INVALID_ARGUMENT] (0x0000001F) returned from _gvaspaceMappingInsert(pGVAS, pGpu, pVASBlock, vaLo, vaHi, flags) @ gpu_vaspace.c:2036
Nov 02 22:15:01 kernel: NVRM: nvCheckFailedNoLog: Check failed: NV_OK == status @ virt_mem_allocator_gm107.c:2563
Nov 02 22:15:01 kernel: NVRM: dmaAllocMapping_GM107: can't update VA space for mapping @vaddr=0x1f4600000
Nov 02 22:15:03 kernel: NVRM: intermapRegisterDmaMapping: Failed to insert new mapping node for range 0xDF7E020000-0xDF7E0AFFFF!
Nov 02 22:15:03 kernel: NVRM: GPU0 nvAssertFailedNoLog: Assertion failed: NV_OK == status @ gpu_vaspace.c:4586
Nov 02 22:15:03 kernel: NVRM: GPU0 nvAssertFailedNoLog: Assertion failed: NV_OK == status @ gpu_vaspace.c:2080

These would have been extremely noticeable before, even to the point of the game freezing until I could free up some NVRAM from other processes.