The juicy part starts at min 1.10:
[Video] Nvidia FAILURE Laboratory: testing chips down to the atom
- XMIT
- [ XMIT ]
- Location: Austin, TX area
- Main keyboard: XMIT Hall Effect
- Main mouse: CST L-Trac Trackball
- Favorite switch: XMIT 60g Tactile Hall Effect
- DT Pro Member: 0093
To me, NVIDIA's emulation facilities are rather more interesting. Testing chips mechanically is one thing. Testing chips logically, and simulating a chip that isn't even made yet, is something entirely different.
https://blogs.nvidia.com/blog/2011/05/1 ... ation-lab/
https://blogs.nvidia.com/blog/2011/05/1 ... ation-lab/
- derzemel
- Location: Bucharest, Romania
- Main keyboard: FC660C, SSK, TX-1800 Nixie
- Main mouse: Mionix Naos 7000
- Favorite switch: Alps SKCL/SKCM tactile
It actually is the best way to simulate a new design. But simulating a chip with 8 billion or more transistors is really remarkable. No wonder those machines are so big and cost so much.XMIT wrote: To me, NVIDIA's emulation facilities are rather more interesting. Testing chips mechanically is one thing. Testing chips logically, and simulating a chip that isn't even made yet, is something entirely different.
https://blogs.nvidia.com/blog/2011/05/1 ... ation-lab/
This giant I/O plug is fascinating


Anyway, some level of emulation is necessary when designing new electronics parts, especially very complex ones.