(NV_ENC_ERR_INVALID_PARAM): EncodeAPI Internal Error

[BumbleLord]

Since starting streaming today: I get intermittent crashes in HiTMAN WOA, all relating to the same error:

"[obs-nvenc] d3d11_encode: nv.nvEncMapInputResource(enc->session, &map) failed: 8 (NV_ENC_ERR_INVALID_PARAM): EncodeAPI Internal Error."

I normally stream at 5500 kbps CBR with NVENC, and it was fine the last time I streamed a few days ago. I always run OBS in Admin mode, and am on version 31.0.0 of OBS and latest GPU drivers. total display res between two monitors is 3840x1080. I monitored temps and power usage on the GPU while recording, right before the most recent crash, and didn't notice any spikes in temp or power usage.

System Specs:

• ASRock B460 Steel Legend ATX LGA1200 Motherboard
• Intel i7-10700
• Thermalright Peerless Assassin 120 SE
• NVIDIA RTX 4070 SUPER
• G.Skill Aegis 32 GB (4 x 8 GB) DDR4-3000 CL16
• SanDisk SSD PLUS 2 TB
• Western Digital WD Black 4 TB
• Cooler Master MasterBox NR600
• Corsair TX650M Gold 650 W
• Asus VG278Q 27.0" 1920 x 1080 144 Hz Monitor
• KYY Portable Monitor 15.6inch 1080P FHD

Things I've tried:

• Checking for OBS and GPU Driver updates
• Checking permissions for web-based sources and setting them all to "Read Access to User Information Scenes etc."
• Turning Multi-Pass Mode to Full from Quarter Res
• Turning off scaling down to 720p
• Resetting to a brand new OBS Profile with NVENC
• Recording and Streaming the game
  

 

[BumbleLord]

Still getting the same crashes as before, with no change when recording in 720p or 1080p. Here's what I've done since the last post:

• Reseated GPU/Checked power cable connection was secure.
• Updated OBS to 31.0.1
• Reinstalled fresh most up-to-date GPU drivers with DDU (fresh install to just the Driver and NVIDIA Broadcast, skipped App)
• Removed scenes for games I regularly play but are hidden when not being played
• Tried 566.14 versions of the driver (both Studio and GR)

Last thing I can think to try is a new PSU, even the one in there is only about 3 years old and should be enough for the system. It was also kept dust free with a filtered intake and pretty regular cleaning.

If this doesn't work, I'm returning both the PSU and GPU and going back to AMD I guess...FML

 

[koala]

17:35:11.302: Device Remove/Reset!  Rebuilding all assets...
[...]
17:35:12.440: Failed to rebuild shared texture: 0x80070057

That's the sign of a graphics driver crash. This could happen with faulty/failing GPU hardware. You already refreshed your GPU driver, so it's probably the hardware. Check overheating, revert overclocking, check power supply.

 

[BumbleLord]

I monitored the temps and power usage during the last crash, temps barely crept over 50. The only OC is whatever is built into the card by the manufacturer (ASUS). This card is brand new, just put it into the system 10 days ago.

Is there any way other than testing with a new PSU to rule out mobo issues?