NVIDIA Kepler architecture
E758261
NVIDIA Kepler architecture is a GPU microarchitecture designed to deliver high parallel computing performance and improved energy efficiency for graphics and high-performance computing workloads.
Observed surface forms (1)
| Surface form | Occurrences |
|---|---|
| Nvidia Kepler | 1 |
Statements (49)
| Predicate | Object |
|---|---|
| instanceOf | GPU microarchitecture ⓘ |
| architectureGeneration | NVIDIA 3rd-generation unified shader architecture NERFINISHED ⓘ |
| codename | Kepler NERFINISHED ⓘ |
| designGoal |
enhance GPU computing for HPC
ⓘ
improve compute density ⓘ increase performance per watt ⓘ |
| developer | NVIDIA NERFINISHED ⓘ |
| family |
NVIDIA GeForce
NERFINISHED
ⓘ
NVIDIA Quadro NERFINISHED ⓘ NVIDIA Tesla NERFINISHED ⓘ |
| firstLaunchYear | 2012 ⓘ |
| keyFeature |
ECC memory support on selected models
ⓘ
GPU Boost technology NERFINISHED ⓘ Hyper-Q NERFINISHED ⓘ PCI Express 3.0 support ⓘ SMX streaming multiprocessor design NERFINISHED ⓘ dynamic parallelism ⓘ enhanced warp schedulers ⓘ high parallel computing performance ⓘ improved energy efficiency ⓘ improved instruction throughput ⓘ improved power management ⓘ support for CUDA programming model ⓘ support for DirectX 11 ⓘ support for OpenGL 4.x ⓘ |
| marketSegment |
consumer graphics
ⓘ
data center computing ⓘ professional visualization ⓘ |
| notableImprovementOver |
higher shader count compared to Fermi
ⓘ
reduced power consumption compared to Fermi ⓘ |
| predecessor | NVIDIA Fermi architecture NERFINISHED ⓘ |
| processNode | 28 nm ⓘ |
| successor | NVIDIA Maxwell architecture NERFINISHED ⓘ |
| supportsAPI |
CUDA 5.x generation
ⓘ
DirectCompute NERFINISHED ⓘ OpenCL NERFINISHED ⓘ |
| supportsFeature | double-precision floating point on selected models ⓘ |
| targetApplication |
GPGPU
ⓘ
graphics processing ⓘ high-performance computing ⓘ |
| usedInProduct |
GeForce GTX 660 Ti
NERFINISHED
ⓘ
GeForce GTX 670 NERFINISHED ⓘ GeForce GTX 680 NERFINISHED ⓘ GeForce GTX 690 NERFINISHED ⓘ Quadro K5000 NERFINISHED ⓘ Quadro K6000 NERFINISHED ⓘ Tesla K20 NERFINISHED ⓘ Tesla K20X NERFINISHED ⓘ Tesla K40 NERFINISHED ⓘ |
Referenced by (2)
Full triples — surface form annotated when it differs from this entity's canonical label.
subject surface form:
Nvidia Maxwell
this entity surface form:
Nvidia Kepler