Congrats on the launch! My co-founder and I always joke that there are only two hair-on-fire problems in 2023 and they can be summarised in 6 letters: GPU & PMF. Really love what you're building.
What's PMF?
Product Market Fit
Show HN: Shadeform – Single Platform and API for Provisioning GPUs
21–26 of 26 posts
Re: Show HN: Shadeform – Single Platform and API for Provisioning GPUs
#22Congrats on the launch! My co-founder and I always joke that there are only two hair-on-fire problems in 2023 and they can be summarised in 6 letters: GPU & PMF. Really love what you're building.
Appreciate it! We're hoping we can help solve one while achieving the other.
Re: Show HN: Shadeform – Single Platform and API for Provisioning GPUs
#23Surprisingly little engagement with this post. I'm not in the market but can people who use gpus but didn't find their offering attractive explain why?
The pricing or lack there of is a huge turn off. We use Replicate and Runpod. On Replicate everyone just shares the GPU cloud. On Runpod I had to bundle 8 together into my own cloud but I can't see that this service solves my problems (queueing image generation and restarting jacked stable diffusion nodes).
Re: Show HN: Shadeform – Single Platform and API for Provisioning GPUs
#24Surprisingly little engagement with this post. I'm not in the market but can people who use gpus but didn't find their offering attractive explain why?
The pricing or lack there of is a huge turn off. We use Replicate and Runpod. On Replicate everyone just shares the GPU cloud. On Runpod I had to bundle 8 together into my own cloud but I can't see that this service solves my problems (queueing image generation and restarting jacked stable diffusion nodes).
This is great feedback for us. We'll make it more transparent. We charge whatever the underlying provider charges + 5% when you deploy into one of our accounts. All provider prices are displayed on our main platform page.
Re: Show HN: Shadeform – Single Platform and API for Provisioning GPUs
#25Problem for our use case is saving on gpus is pointless if we have to keep paying egress fees for our 250 TB training dataset.
The single interface for any cloud GPU is cool, but hard to imagine it taking off without some additional features.
I think for lots of shops the hardest part isn't the compute but moving the data around. Ie for us, we use s3, some lustre caching and spot instance nodegroups. We are a deep learning research team that spends roughly 40-50k/month on aws compute for training jobs. I imagine this is somewhat mid tier, maybe more than some but certainly far less than others.
For inference, data egress costs could be less of an issue, but your service would really need to be almost invisible. It probably would be pretty complicated for a number of reasons, but if you could design a "virtual on-demand nodegroup"™ that I could add to my existing clusters and then map to whatever k8s stuff I want, that would probably be useful. I would need to be able to auto deploy a base image to the machine and then provision the node and register with my cluster.
Just some unorganized thoughts. Good luck and have fun.
Re: Show HN: Shadeform – Single Platform and API for Provisioning GPUs
#26Problem for our use case is saving on gpus is pointless if we have to keep paying egress fees for our 250 TB training dataset. The single interface for any cloud GPU is cool, but hard to imagine it taking off without some additional features. I think for lots of shops the hardest part isn't the compute but moving the data around. Ie for us, we use s3, some lustre caching and spot instance nodegroups. We are a deep le…
What are your use case that justify a cost 40-50k/month just for training?