31–40 of 114 posts
I'm holding out for a better price on GPU instances (for parallel processing). My 'back of the napkin' calculations indicate it would take very little time for it to pay for itself by purchasing your own NVIDIA GPU(s) and running them from a colo. Especially if your requirements are for the instance(s) to run 24x7.
Will Google Cloud follow up?
https://cloud.google.com/blog/products/compute/google-comput...
I'm holding out for a better price on GPU instances (for parallel processing). My 'back of the napkin' calculations indicate it would take very little time for it to pay for itself by purchasing your own NVIDIA GPU(s) and running them from a colo. Especially if your requirements are for the instance(s) to run 24x7.
Why would they ever be not the case ? If you ignore sysadmin costs AWS is always more expensive (even when reserved for three years) than any machine you maintain yourself
Are they still extorting customers by price gouging bandwidth? Maybe one day I’ll wake up to news they’ve decided to charge for capacity instead of transfer.
Actually, I do genuinely wonder if some day this will be seen as anticompetitive behavior (a form of lock-in via higher than reasonable egress fees).
If you have a massive amount of data, you can always get AWS to transfer it to you via a Snowball for $30/TB + a $300 fee.
I'm holding out for a better price on GPU instances (for parallel processing). My 'back of the napkin' calculations indicate it would take very little time for it to pay for itself by purchasing your own NVIDIA GPU(s) and running them from a colo. Especially if your requirements are for the instance(s) to run 24x7.
Isn't the whole point of cloud that you don't own the machines? The calculus probably isn't as good when you factor in support, maintenance, and power costs right?
In the many years I've used cloud I've never requested support, so in my experience I don't need to calculate that in.
And even when you consider machine costs, as long as you can get at least 1 yr (not an unreasonable assumption) out of your equipment you're more than making up for what it would've cost in the cloud.
I'm holding out for a better price on GPU instances (for parallel processing). My 'back of the napkin' calculations indicate it would take very little time for it to pay for itself by purchasing your own NVIDIA GPU(s) and running them from a colo. Especially if your requirements are for the instance(s) to run 24x7.
Earlier quoted context omitted.
It's nice to see that the increasing amount of hyperbole here isn't impacting everyone.
It's not just hyperbole, it's grossly unjust. Amazon quite rightly charge a higher price for some portions of AWS--transparently, up front, with tools to assess usage and estimate fees beforehand--because they created a product that many organizations believe provides extraordinary value. And for the sin of earning a profit through voluntary trade, a vocal minority of HN damns them every single time AWS is mentioned.…
Just to be clear: Amazon is not paying a variable cost for how many bits are transferred over their wires. And the standard, prior to cloud, and still in colocation, was to charge for capacity (i.e. $/gbps) rather than transfer (i.e $/gb). They are making massive, massive profits by this pricing arrangement. The cost of sustaining 100 gbps transfer for 30 days on Amazon is orders of magnitude higher than paying for one month of 100 gbps of IP transit.
They’ve normalized a pricing structure that is disconnected from any actual cost basis. You’re free to pay for it, but it’s quite amusing to say criticizing that is unjust.
I'm holding out for a better price on GPU instances (for parallel processing). My 'back of the napkin' calculations indicate it would take very little time for it to pay for itself by purchasing your own NVIDIA GPU(s) and running them from a colo. Especially if your requirements are for the instance(s) to run 24x7.
Can you actually buy GPUs now? An advantage of Amazon in the past was that they were actually available there, where they weren’t available to buy in practice.
The demand right now has also fallen dramatically even from regular customers.
https://www.tomshardware.com/news/coronavirus-motherboard-gr...
I'm holding out for a better price on GPU instances (for parallel processing). My 'back of the napkin' calculations indicate it would take very little time for it to pay for itself by purchasing your own NVIDIA GPU(s) and running them from a colo. Especially if your requirements are for the instance(s) to run 24x7.
Can you actually buy GPUs now? An advantage of Amazon in the past was that they were actually available there, where they weren’t available to buy in practice.
You can even buy Tesla V100s that power AWS accelerated instances.