When installing binary tarballs, Spack has to download from its
binary mirrors.
Sometimes Spack has cache available for these mirrors.
That cache helps to order mirrors to increase the likelihood of
getting a direct hit.
However, currently, when Spack can't find a spec in any local cache
of mirrors, it's very dumb:
- A while ago it used to query each mirror to see if it had a spec,
and use that information to order the mirror again, only to go
about and do exactly a part of what it just did: fetch the spec
from that mirror confused
- Recently, it was changed to download a full index.json, which
can be multiple dozens of MBs of data and may take a minute to
process thanks to the blazing fast performance you get with
Python.
In a typical use case of concretizing with reuse, the full index.json
is already available, and it likely that the local cache gives a perfect
mirror ordering on install. (There's typically no need to update any
caches).
However, in the use case of Gitlab CI, the build jobs don't have cache,
and it would be smart to just do direct fetches instead of all the
redundant work of (1) and/or (2).
Also, direct fetches from mirrors will soon be fast enough to
prefer these direct fetches over the excruciating slowness of
index.json files.
* include Drishti
* fix syntax
* Update var/spack/repos/builtin/packages/drishti/package.py
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
* Update package.py
* Update var/spack/repos/builtin/packages/drishti/package.py
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
* py-sphinxcontrib-devhelp: add 1.0.2
* Update var/spack/repos/builtin/packages/py-sphinxcontrib-devhelp/package.py
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
* py-sphinxcontrib-applehelp: add 1.0.2
* Update var/spack/repos/builtin/packages/py-sphinxcontrib-applehelp/package.py
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
Co-authored-by: Adam J. Stewart <ajstewart426@gmail.com>
* ROCm 5.3.0 updates
* New patches for 5.3.0 on hip and hsakmt
* Adding additional build arguments in hip and llvm
* RVS updates for 5.3.0 release
* New patches and rocm-tensile, rocprofiler-dev, roctracer-dev recipe updates for 5.3.0
* Reverting OPENMP fix from rocm-tensile
* Removing the patch to compile without git and adding witout it
* Install library in to lib directory instead of lib64 across all platform
* Setting lib install directory to lib
* Disable gallivm coroutine for libllvm15
* Update llvm-amdgpu prefix path in hip-config.cmake.in
Removing libllvm15 from Mesa dependency removing
* hip-config.cmake.in update required from 5.2
* hip-config.cmake.in update required from 5.2 and above
* hip-config.cmake.in update required for all 5.2 release above
* Style check correction in hip update
* ginkgo: add missing include
* Patching hsa include path for rocm 5.3
* Restricting patch for llvm-15
* Style check error correction
* PIC flag required for the new test applications
* Passing -DCMAKE_POSITION_INDEPENDENT_CODE=ON in the cmake_args instead of setting -fPIC in CFLAGS
Co-authored-by: Cordell Bloor <Cordell.Bloor@amd.com>
gnulib/lib/malloca.c uses single value `static_assert()` only available in c-11
syntax. `gcc` seems to be fine, but `icc` needs extra flag.
Co-authored-by: Stephen Sachs <stesachs@amazon.com>
Since it's a header-only library there's nothing to build. However, the
default targets include tests and examples and there's no option to turn
them off during configuration time.
Per https://geant4.web.cern.ch/node/1837 the correct dependency for 10.6 is on `g4emlow@7.9.1`, not on both `g4emlow@7.9` and `g4emlow@7.9.1`.
This is a minor cosmetic fix. The concretization for 10.6 works just fine here. But this removes the duplicate entry.
This PR patches the f_check script to detect the ifort compiler and
ensure that F_COMPILER is iset to INTEL. This problem was introduced with
openblas-0.3.21. Without this patch, the value of F_COMPILER falls back
to G77 and icc rather than ifort is used for the linking stage. That
results in the openblas library missing libifcore, which in turn means
many Fotran programs can not be compiled with ifort.
Writing a long dependency like:
```python
depends_on(
"llvm"
"targets=amdgpu,bpf,nvptx,webassembly"
"version_suffix=jl +link_llvm_dylib ~internal_unwind"
)
```
when it should be formatted like this:
```python
depends_on(
"llvm"
" targets=amdgpu,bpf,nvptx,webassembly"
" version_suffix=jl +link_llvm_dylib ~internal_unwind"
)
```
can cause really subtle errors. Specifically, you'll get something like this in
the package sanity tests:
```
AttributeError: 'NoneType' object has no attribute 'rpartition'
```
because Spack happily constructs a class that has a dependency with name `None`.
We can catch this earlier by banning anonymous dependency specs directly in
`depends_on()`. This causes the package itself to fail to parse, and emits
a much better error message:
```
==> Error: Invalid dependency specification in package 'julia':
llvmtargets=amdgpu,bpf,nvptx,webassemblyversion_suffix=jl +link_llvm_dylib ~internal_unwind
```
* Only restrict CMake version in umpire when examples and rocm are enabled
* Add CMAKE_HIP_ARCHITECTURES to Umpire and lift cmake version restriction
Co-authored-by: Tom Scogland <scogland1@llnl.gov>
