Dbgpu

Includes the fix from https://github.com/painebenjamin/dbgpu/issues/2, whereby GPU's that listed their memory capacity in megabytes in base-2 (e.g. 1024 MB) would incorrectly report much smaller memory sizes. This fix only affects older model GPUs, modern GPUs are not affected.

Includes the (theoretical) performance of the new 50-series.

Also includes changes from 1, including TPU ID and URL in the output. Example:

sh
dbgpu lookup 3090ti --fuzzy


╒══════════════════════════════════════════════════════════════════════════════════════════════╕
│ GeForce RTX 3090 Ti │
├──────────────────────────────────────────────────────────────────────────────────────────────┤
│ GPU Name | GA102 │
│ Manufacturer | NVIDIA │
│ Architecture | Ampere │
│ Foundry | Samsung │
│ Process Size | 8 nm │
│ Transistor Count | 28.3 billion │
│ Transistor Density | 45.1 million/mm² │
│ Die Size | 628 mm² │
│ Chip Package | BGA-3328 │
│ Release Date | 2022-01-27 │
│ Generation | GeForce 30 │
│ Bus Interface | PCIe 4.0 x16 │
│ Base Clock | 1,560 MHz │
│ Boost Clock | 1,860 MHz │
│ Memory Clock | 1,313 MHz │
│ Memory Size | 24.0 GB │
│ Memory Type | GDDR6X │
│ Memory Bus | 384 bit │
│ Memory Bandwidth | 1,010.0 GB/s │
│ Shading Units | 10,752 │
│ Texture Mapping Units | 336 │
│ Render Output Processors | 112 │
│ Streaming Multiprocessors | 84 │
│ Tensor Cores | 336 │
│ Ray Tracing Cores | 84 │
│ L1 Cache | 128.0 KB │
│ L2 Cache | 6.0 MB │
│ Thermal Design Power | 450 W │
│ Board Length | 336 mm │
│ Board Width | 140 mm │
│ Board Slot Width | Triple-slot │
│ Suggested PSU | 850 W │
│ Power Connectors | 1x 16-pin │
│ Display Connectors | 1x HDMI 2.1, 3x DisplayPort 1.4a │