Install this package:
emerge -a sci-libs/hipTensor
| Version | EAPI | Keywords | Slot |
|---|---|---|---|
| 10.0.0 | 8 | ~amd64 | 0/$(ver_cut 1-2) |
<pkgmetadata>
<maintainer type="person">
<email>iohann.s.titov@gmail.com</email>
<name>Ivan S. Titov</name>
</maintainer>
<longdescription lang="en">
hipTensor is AMD's tensor primitives library, covering contraction,
reduction and permutation of arbitrary-rank tensors through an
einsum-style descriptor API. It is the ROCm counterpart to NVIDIA's
cuTENSOR and is implemented on top of the contraction, reduction and
elementwise kernel instances that sci-libs/composable-kernel generates,
so it requires that package to be built with USE=hiptensor.
</longdescription>
<upstream>
<remote-id type="github">ROCm/rocm-libraries</remote-id>
</upstream>
</pkgmetadata>
Manage flags for this package:
euse -i <flag> -p sci-libs/hipTensor |
euse -E <flag> -p sci-libs/hipTensor |
euse -D <flag> -p sci-libs/hipTensor
| Flag | Description | 10.0.0 |
|---|---|---|
| "( | ⚠️ | ✓ |
| ( | ⚠️ | ✓ |
| ) | ⚠️ | ✓ |
| )" | ⚠️ | ✓ |
| gfx1100 | ⚠️ | ✓ |
| gfx1101 | ⚠️ | ✓ |
| gfx1102 | ⚠️ | ✓ |
| gfx1103 | ⚠️ | ✓ |
| gfx1150 | ⚠️ | ✓ |
| gfx1151 | ⚠️ | ✓ |
| gfx1152 | ⚠️ | ✓ |
| gfx1153 | ⚠️ | ✓ |
| gfx1200 | ⚠️ | ✓ |
| gfx1201 | ⚠️ | ✓ |
| gfx1250 | ⚠️ | ✓ |
| gfx908 | ⚠️ | ✓ |
| gfx90a | ⚠️ | ✓ |
| gfx942 | ⚠️ | ✓ |
| gfx950 | ⚠️ | ✓ |
| test | Build and run unit tests ⚠️ | ✓ |
| Type | File | Size | Versions |
|---|
| Type | File | Size |
|---|---|---|
| DIST | hiptensor-10.0.0.tar.gz | 586494 bytes |