Statements (24)
| Predicate | Object |
|---|---|
| gptkbp:instanceOf |
gptkb:technology
|
| gptkbp:allows |
twice as many operations per clock at lower precision
|
| gptkbp:category |
GPU feature
|
| gptkbp:competitor |
gptkb:NVIDIA_Tensor_Cores
|
| gptkbp:developedBy |
gptkb:AMD
|
| gptkbp:enables |
FP16 operations
INT8 operations mixed precision computing |
| gptkbp:improves |
performance in compatible workloads
|
| gptkbp:introducedIn |
2017
|
| gptkbp:marketedAs |
gptkb:AMD
|
| gptkbp:relatedTo |
gptkb:Vulkan_API
gptkb:DirectX_12 |
| gptkbp:supportedBy |
gptkb:AMD_Radeon_RX_5000_series
gptkb:AMD_Radeon_VII gptkb:AMD_Radeon_RX_6000_series gptkb:AMD_Radeon_RX_Vega gptkb:AMD_Vega_architecture |
| gptkbp:usedFor |
gptkb:machine_learning
gaming |
| gptkbp:usedIn |
graphics processing
|
| gptkbp:bfsParent |
gptkb:Vega_architecture
|
| gptkbp:bfsLayer |
6
|
| https://www.w3.org/2000/01/rdf-schema#label |
Rapid Packed Math
|