_gpu_dxg_r5a_sem.nx

_gpu_dxg_r5a_sem.nx -- SOVEREIGN-GPU R5a: make the real RTX 5080 EXECUTE a Nishi-emitted pushbuffer. HARDWARE-RUNG-UP, NISHI-ECOSYSTEM-ONLY: Nishi EMITS the raw NVIDIA host/FIFO semaphore-release pushbuffer ITSELF (the GPU's native command format is the last-mile hardware protocol, learned from the open spec but emitted sovereignly -- NO libcuda/Mesa/NVIDIA software linked or called). The GPU writes a known payload to a GPU VA backed by our EXISTINGHEAP system memory; we read it back CPU-side = the GPU ran OUR commands. Pushbuffer (GF100 method header = (mode<<29)|(count<<16)|(subch<<13)|(method>>2)); host-method semaphore release (NVC56F, stable Ampere->Blackwell): SEM_ADDR_LO@0x5c, SEM_ADDR_HI@0x60, SEM_PAYLOAD_LO@0x64, SEM_PAYLOAD_HI@0x68, SEM_EXECUTE@0x6c (OPERATION_RELEASE=1, PAYLOAD_SIZE_32BIT=0 -> value 0x1). 6 dwords: [0]=0x20050017 (increasing,count=5,subch0,method 0x5c>>2=0x17) [1]=addr_lo [2]=addr_hi [3]=payload [4]=0 [5]=0x1 SUBMITCOMMAND = 0xC180470F (size 384, size-swept): command_buffer@0=pushbuffer GPU VA, command_length@8, broadcast_context_count@0x18=1, broadcast_context[0]@0x20=context. RISK: a malformed submit may TDR (WSL recovers). license_tier: ORIGINAL