Quote from: Worgarthe on December 14, 2025, 21:47:40I mean, again, and this is now a rhetorical question - why in the world would one pay +2980€ to go from 245HX + Blackwell 1000 to 245HX + Blackwell 5000 (that previously linked P16 Gen 3 configuration on Lenovo's site) when a 740-780€ 5070 Ti is simply going to destroy that Blackwell 5000? ¯\_(ツ)_/¯
Sure if they simply NEED to play Battlefield 6 on battery when they commute to work then ok, go crazy and get that Blackwell 5000, or they can save money and go with Blackwell 4000 (16 GB) for "just" +1420€ (and get even more obliterated by a 740-780€ 5070 Ti), but yeah...
FWIW there's a much bigger difference between the higher-end mobile GPU options for everything that isn't gaming, so it can absolutely make sense to spend more on P16 Gen 3 GPU upgrades even if the chassis is TGP‑limited.
RTX Pro 2000 / 5060 mobile – 1 NVENC encoder, 128‑bit bus, 8 GB VRAM
RTX Pro 3000 / 5070 Ti mobile – 1 NVENC encoder, 192‑bit bus, 12 GB VRAM
RTX Pro 4000 / 5080 mobile – 2 NVENC encoders, 256‑bit bus, 16 GB VRAM
RTX Pro 5000 / 5090 mobile – 3 NVENC encoders, 256‑bit bus, 24 GB VRAM
For video editing specifically, the RTX Pro 4000 mobile is the real workstation sweet spot: it delivers desktop 5080‑class encoding (100% more NVENC encoders vs the RTX Pro 3000 / 5070 Ti mobile) while staying relatively reasonable on cost and power.
The RTX Pro 5000 / 5090 mobile can then give you essentially "5090‑class" video encoding throughput, since it has the same three NVENC encoders as the desktop 5090, but for the big price jump over the RTX Pro 4000 mobile you're only getting 50% more NVENC encoders and there's zero further improvement to the memory bus.
All of these mobile parts use the same single NVDEC decoder, so you don't gain extra decode hardware by moving up the stack. On desktop, the 5080 and 5090 each have two NVDEC decoders, which can help with heavier multi‑track timeline video decode. The desktop 5070 Ti desktop by comparison only has a single NVENC encoder / NVDEC decoder so it's a bit limited as a workstation GPU.