Years ago I wanted to run a specific analysis for a client but the home office for the brokerage I worked for said they don't do that and any analysis they are offering takes a few weeks.
Fast forward a few years and I've taught myself a little python, a deeper understanding of Unix, git and likely burned through the CPU on too many MacbookPro's and one iMac.
So I've purchased a MacPro 5,1 as a new home system to handle my ever-growing iTunes collection of movies, music and the first Geforce 1060 which does support 1280 Cuda cores. Eventually, I'll run two of these.
I've done some reading on StackExchange with this post.
[https://ai.stackexchange.com/questions/6855/thoughts-on-apple-mac-pro-vs-gcp-aws-for-deep-learning][1]
and this one
[enter link description here][2]
These two builds ( I believe ) are going in the right direction using the wrong hardware. Since I can't afford to pay Uncle Bezos crazy cash for AWS (nor do I have time to learn it) as I am an undergrad... plus I have some experience in building computers.
Yes, I understand that Google now offers GPU's and I plan to use them. There is also a need to have a local Postgresql server running for my own needs and like I said the iMac has been cooked while I learn how to train hyperparameters of Neural nets. There have been many mistakes. =(
So with the 5,1 that I've purchased, it came with Mojave but I need to wind it back to high sierra which is where the Nvidia drivers do work (yet deprecated in further versions of MacOS)
I have created one problem, now the machine gets to a boot screen and it won't restart, even though the screen says click here to shutdown. Nothing happens.
I figure I would post the screenshot and ask for help.
cMP (classic MacPro) Compatability List
cMP upgrade Sticky, now a Wiki
I'm liking the idea of having my build notes in one location, is it ok if I paste my todo list here and work through it, once complete I will edit the font to be something like strikethrough?
[1]: https://ai.stackexchange.com/questions/6855/thoughts-on-apple-mac-pro-vs-gcp-aws-for-deep-learning
[2]: https://datascience.stackexchange.c...card-to-make-deep-learning-with-a-macbook-pro
[3]:
Fast forward a few years and I've taught myself a little python, a deeper understanding of Unix, git and likely burned through the CPU on too many MacbookPro's and one iMac.
So I've purchased a MacPro 5,1 as a new home system to handle my ever-growing iTunes collection of movies, music and the first Geforce 1060 which does support 1280 Cuda cores. Eventually, I'll run two of these.
I've done some reading on StackExchange with this post.
[https://ai.stackexchange.com/questions/6855/thoughts-on-apple-mac-pro-vs-gcp-aws-for-deep-learning][1]
and this one
[enter link description here][2]
These two builds ( I believe ) are going in the right direction using the wrong hardware. Since I can't afford to pay Uncle Bezos crazy cash for AWS (nor do I have time to learn it) as I am an undergrad... plus I have some experience in building computers.
Yes, I understand that Google now offers GPU's and I plan to use them. There is also a need to have a local Postgresql server running for my own needs and like I said the iMac has been cooked while I learn how to train hyperparameters of Neural nets. There have been many mistakes. =(
So with the 5,1 that I've purchased, it came with Mojave but I need to wind it back to high sierra which is where the Nvidia drivers do work (yet deprecated in further versions of MacOS)
I have created one problem, now the machine gets to a boot screen and it won't restart, even though the screen says click here to shutdown. Nothing happens.
I figure I would post the screenshot and ask for help.
- 1. How can I move forward to install high sierra? From having a Mojave install when I keep getting this screen and it just loops back around to it. To clarify, when I click the triangle in the circle it does nothing, my wifi is connected but I can click the restart or shutdown under the apple icon in the top right of the screen. [![enter image description here][3]][3]
- Eventually, I will expand this machine to the 12 core 24 thread 3.42Ghz processor system with 96 gigs of ram, but what is a checklist of things to set up a new Mac? For reference: My shell will likely be bash, Anaconda, using GitHub as version control, Postgresql as backend server and python3 for a scripting language which I am still learning. This might get broken up into other posts but I figure post here as I can't be the only person wanting to get into a medium-sized home system for GPU deep learning. So I figure open-source my concept and share it with the SE community. Edit: There is a boot USB drive that is used to get the machine to this point, many of the how-to MacPro tutorials talk about using keystrokes on the Mac keyboard which I don't have, just an old but reliable Microsoft keyboard. So if you know how to do a firmware update on 2010 MacPro I would appreciate your advice.
cMP (classic MacPro) Compatability List
cMP upgrade Sticky, now a Wiki
I'm liking the idea of having my build notes in one location, is it ok if I paste my todo list here and work through it, once complete I will edit the font to be something like strikethrough?
First Challenge - Uninstall Mojave and move back to 10.13.6. COMPLETED- Second Things - As this build evolves I'll likely edit this and welcome recommendations to expand this for others like the two links that this forum has been building. It will be easier to post everything to one place and I guess I could create sub links to others like the compatibility list and the upgrade wiki which are awesome and thanks to each author and contributors for making those.
- Third Challenge - I've been able to get the high sierra running after using a USB SATA cable to create a bootable drive in an old 2.5 500 gig drive. The challenge is now my Boot drive is externally USB plugged and I need to move it inside.
- Fourth Challenge - The new(ish) mac GPU arrives next week but before then I need to figure out how to share (centralize) data among my 2tb MBP 13", my 1tb iMac 27" and my 2tb cMP 5,1. I have an a1254 airport extreme that I can use where I could use the time machine to back up the other computers but there is a ton of it I want to be migrated to the cMP. Would the migration assistant be a good candidate for that? Or what is a way to have maximum access to the data I have on each machine, in a central location
- Airport Challenge - Have any great tutorials or progressive steps about these to see if I can create an ssh connection for data access? Is that even a good idea? Something like this is what I've found but not sure if it best practice, course it all depends right? NAS vs Airport
- I have access to the internet currently via wifi and my cMP (classic MacPro) is connecting via wifi board on the machine but should I connect the airport extreme to the wall socket for ethernet about the same speed as wifi. I'm wanting to get firewalls and secure access to the airport but enable access to it for me only via these three machines the MPB, iMac, and cMP respectively.
- Each of the three machines is connected to the airport extreme via ethernet and this coming black Friday I'll get a few 10Tb drives or maybe some 4tbsshd that WD has been talking about to expand the capacity but at what point does the Airport extreme need to be migrated or linked to a NAS in terms of size or functionality. This is part of a large question so I am looking for links and advice.
- Fifth Challenge - CUDA GPU is my (first of many GPUs) Quadro 4000 arrives from eBay next week so that is Cuda capable, but not as fast as the GTX 1060 but small steps first. I need to get Pycharm setup with anaconda and git version control, then once we have a few tests setup I need to get Concourse to help with integrating the stack and docker will be added into that.
- Six Challenge - Are there any links someone can recommend for a checklist of things to set up on a new mac or ubuntu machine for data science work? There is the list of 12 things that Joel recommends for level 10 software developers but I am talking strictly "Setting up a new mac" Things like
- Create an admin account and subsequent user accounts.
- Create a log of your PATH and use virtual machines with requirements.txt files to set up the machine. I know this is moving into software but I'm hoping I can keep this focused on the build I am doing, is it also ok to link between here and the StackOverflow account as they are more purely focused on the code vs hardware like you guys/girls?
- The long term goal of this is to create a somewhat automated list of things that get set up on any subsequent machines so this doesn't have to get recreated ever new machine or ubuntu server we add for the project.
- Seventh Challenge - security. How to secure the ports of my computer with some sort of monitoring software like Datadog, Wireshark and or a python based system. This next chunk was sourced at fullstackpython.com which is a great source for best practices in scaling efficient python projects.
- Monitoring Layers - There are several important resources to monitor the operating system and network level of a web stack.
- CPU utilization
- Memory utilization
- Persistence storage consumed versus free
- Network bandwidth and latency
- Application-level monitoring - encompasses several aspects; server time, server timecost, resources dedicated to each aspect will vary based on whether an application is read-heavy, write-heavy, or subject to rapid swings in traffic.
- Application warnings and errors (500-level HTTP errors)
- Application code performance
- Template rendering time
- Browser rendering time for the application
- Database querying performance
- Monitoring Layers - There are several important resources to monitor the operating system and network level of a web stack.
[1]: https://ai.stackexchange.com/questions/6855/thoughts-on-apple-mac-pro-vs-gcp-aws-for-deep-learning
[2]: https://datascience.stackexchange.c...card-to-make-deep-learning-with-a-macbook-pro
[3]:
Attachments
Last edited:
