Switching From An RTX 4090 To iGPU Via Display Cable Allegedly Helped A User Gain 2.5GB VRAM, Allowing Him To Double The Context Window In Qwen3.8-27B To 132K

Switching From An RTX 4090 To iGPU Via Display Cable Allegedly Helped A User Gain 2.5GB VRAM, Allowing Him To Double The Context Window In Qwen3.8-27B To 132K

The RTX 4090 with its 24GB VRAM is capable of handling mid-sized LLMs like Qwen3.8-27B, but that also depends on what level of quantization you’re running, not to mention that you’re going to have to make adjustments and other optimizations, such as tweaking the Context Window to stay within that video memory limit. However, one user has apparently discovered that running an external display through the iGPU instead of the RTX 4090’s HDMI or Dis…