SkyPilot: Run AI and batch jobs on any infra (Kubernetes or 12+ clouds). Get unified execution, cost savings, and high GPU availability via a simple interface.
This PR adds hardcoded TPU V6e price to our catalog. Currently, TPU V6e is not in GPC SKUs, so we use the pricing in GCP TPU Pricing first.
There are some missing zones in the pricing page and I set their price to 0.0 so user can launch on those regions.
The hardcode catalog is generated by this script:
zones = ['us-east5-b', 'us-east5-c', 'us-south1-a', 'europe-west4-a', 'us-central1-c', 'us-south1-c', 'us-central1-b', 'asia-northeast1-b', 'us-east1-d', 'us-central2-b']
acc_nums = [1, 4, 8, 16, 32, 64, 128, 256]
region2price = {
'us-east1': 2.7,
'us-east5': 2.7,
'europe-west4': 2.97,
'asia-northeast1': 3.24,
}
no_pricing_info_region = set()
for n in acc_nums:
acc_name = f'tpu-v6e-{n}'
for z in zones:
r = '-'.join(z.split('-')[:-1])
if r not in region2price:
no_pricing_info_region.add(r)
continue
p = region2price[r] * n
print(f' ,{acc_name},1,,,{acc_name},{p},,{r},{z}')
print(no_pricing_info_region)
Tested (run the relevant ones):
[x] Code formatting: bash format.sh
[x] Any manual or new tests for this PR (please specify below)
This PR adds hardcoded TPU V6e price to our catalog. Currently, TPU V6e is not in GPC SKUs, so we use the pricing in GCP TPU Pricing first.
There are some missing zones in the pricing page and I set their price to 0.0 so user can launch on those regions.
The hardcode catalog is generated by this script:
Tested (run the relevant ones):
bash format.sh
python -m sky.clouds.service_catalog.data_fetchers.fetch_gcp --single-threaded --all-regions
pytest tests/test_smoke.py
pytest tests/test_smoke.py::test_fill_in_the_name
conda deactivate; bash -i tests/backward_compatibility_tests.sh