who is providing GPU based VM with hourly procing?

jpeng

Are there any LET providing this? thanks.

jmgcaguicla

I remember FluidStack doing hourly @lentro but I think they're bare metal instead of VMs.

Iirc Scaleway also has GPU instances

Out of alternatives, you could always go with Beff Jezos.

FlorinMarian

https://www.ovhcloud.com/en-ie/public-cloud/prices/#397

Shishov

https://selectel.ru/en/services/cloud/servers/gpu/

Erisa

Adding to the other excellent choices, here's another https://www.scaleway.com/en/gpu-instances/

lentro

@jmgcaguicla said:
I remember FluidStack doing hourly @lentro but I think they're bare metal instead of VMs.

Iirc Scaleway also has GPU instances

Out of alternatives, you could always go with Beff Jezos.

Thanks! I actually left Fluidstack and am building my own cloud GPU service for ML. It’ll be called TensorDock.

I’ll launch it in the far future but here is pricing if you’re interested:
https://tensordock-website.pages.dev/#pricing

Customize per resource what you need. I’ve got a few racks in Boston, Chicago (SBA Edge), and Las Vegas (Switch).

It’ll be cheaper than anything else on the market.

If you’re interested, PM me and I can get you set up on the private beta platform.

Billing is done with hourly pricing and then pro-rated to a per minute basis.

jmgcaguicla

@lentro said:

@jmgcaguicla said:
I remember FluidStack doing hourly @lentro but I think they're bare metal instead of VMs.

Iirc Scaleway also has GPU instances

Out of alternatives, you could always go with Beff Jezos.

Thanks! I actually left Fluidstack and am building my own cloud GPU service for ML. It’ll be called TensorDock.

I’ll launch it in the far future but here is pricing if you’re interested:
https://tensordock-website.pages.dev/#pricing

Customize per resource what you need. I’ve got a few racks in Boston, Chicago (SBA Edge), and Las Vegas (Switch).

It’ll be cheaper than anything else on the market.

If you’re interested, PM me and I can get you set up on the private beta platform.

Billing is done with hourly pricing and then pro-rated to a per minute basis.

Is serverless compute (ala Lambda) on the roadmap for release? or are they just VMs (for now)?

I have a (currently dead) TF workload that I may or may not want to migrate.

lentro

@jmgcaguicla said:

@lentro said:

@jmgcaguicla said:
I remember FluidStack doing hourly @lentro but I think they're bare metal instead of VMs.

Iirc Scaleway also has GPU instances

Out of alternatives, you could always go with Beff Jezos.

Thanks! I actually left Fluidstack and am building my own cloud GPU service for ML. It’ll be called TensorDock.

I’ll launch it in the far future but here is pricing if you’re interested:
https://tensordock-website.pages.dev/#pricing

Customize per resource what you need. I’ve got a few racks in Boston, Chicago (SBA Edge), and Las Vegas (Switch).

It’ll be cheaper than anything else on the market.

If you’re interested, PM me and I can get you set up on the private beta platform.

Billing is done with hourly pricing and then pro-rated to a per minute basis.

Is serverless compute (ala Lambda) on the roadmap for release? or are they just VMs (for now)?

I have a (currently dead) TF workload that I may or may not want to migrate.

Just VMs for now, as that’s my expertise

In a year or two might add serverless, but for now it’d be fully dedicated VMs with preinstalled ML software and stuff

k4zz

Did you check https://www.paperspace.com ?