Quote from: GPUHoarder on May 12, 2018, 08:19:32 PM

Quote from: senseless on May 12, 2018, 08:13:43 PM

Quote from: GPUHoarder on May 12, 2018, 07:38:08 PM

@senseless - Yes I’ve got my own RTL and bitstreams - as well as some of my own hardware. I’m pondering your shell/publishing concept, but I haven’t decided if it makes sense yet. I can see the VCU1525 and derivatives being produced in much greater quantity once there is interest though, and that leads to price reductions and makes it difficult for other options to be competitive.

I am in talks with someone who's in the process of developing a 250-300A vccint version of the VCU1525. It's quite possible that may be the version that ends up getting sold. I would like it to be, but again, it just depends on how things go. I should know more in the next couple of weeks. The price should remain within the same ranges depending on QTY. The 9P is the most mass produced chip, there's no getting around the fact it will be the lowest cost / logic ratio in large quantity.

If you have any questions RE the shell, send me a ping. Would love to hear your feedback and any problems you might have with it.

I need to fire up an F1 instance and refamiliarize myself with how AWS does it. It has been sometime since I initially touched those and to be honest I never got deep into it.

You can run the HDK on a much cheaper instance than the F1, do the dev work there and package the bitstream into an AFI (Amazon FPGA Image) that you load onto the F1 instance. This link is useful . https://aws.amazon.com/ec2/instance-types/f1/