Ampere, Hopper, Blackwell now produce AMP/HOP/BWL suffixes (like ADA)
so RTX cards across generations are distinguishable: RTX6000ADA vs
RTX6000BWL. LOVELACE remains a skip token as it duplicates ADA info.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
RTX 6000 ADA and A6000 are distinct cards — RTX_4000_ADA_SFF now
produces RTX4000ADA instead of RTX, avoiding visual ambiguity with
the segment separator (10xRTX4000ADA vs 10xRTX-1x…).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
MB_ lots (e.g. MB_INTEL_..._GPU8) are incorrectly categorized as GPU
in the pricelist. Two fixes:
- Skip MB_ lots in buildGPUSegment regardless of pricelist category
- Add INTEL to vendor token skip list in parseGPUModel (was missing,
unlike AMD/NV/NVIDIA which were already skipped)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
