4 ms·
I have a retina 15 inch MacBook Pro with 16gb of RAM and also an iMac that is dedicated for work. I was hoping to get more into AI and OpenCV with c++.
by jason_slack 10y ago
I have a retina 15 inch MacBook Pro with 16gb of RAM and also an iMac that is dedicated for work.
I was hoping to get more into AI and OpenCV with c++.
- ralmeida 10y agoI've done computer vision/machine learning with a much less powerful MBPRetina, you'll be fine even if you start now.
- jason_slack 10y agofine on the hardware I have or fine if I buy a Mac Pro even thought they might upgrade it soon.
- yompers888 10y agoIn case parent doesn't return: fine with current hardware, based on >...done computer vision/machine learning with a much less powerful MBPRetina...
- ralmeida 10y agoYep, that's what I meant, fine with the hardware you already have.
- jason_slack 10y agoCan you clarify one thing for me. My rMBP has a AMD Radeon R9 M370X 2048 MB this means no CUDA, but OpenCL should be just fine?
- brudgers 10y agoIf those application domains are driving a hardware purchase, a workstation that can be stuffed with GPU's might make more sense. With SSH all the interaction could be through just about any device.
- jason_slack 10y agoI thought about this too (an Ubuntu workstation), but there are so many options as to what to buy. Seems overwhelming. Any suggestions?
- benmcnelly 10y agoI am afraid I am net an expert in AI and OpenCV, but I do know how to build a custom desktop. I would research what hardware other people are finding effective, and then just double check Ubuntu/Debian will have the driver support (its pretty good nowadays) for what you want. Building a rig is a great experience and quite fun too!
- jason_slack 10y agoa recommendation on a motherboard that supports a number of GPU's and a decent amount of RAM? When I built a Bitcoin mining rig a long while back I had to use riser boards and this took a log of custom work on my end :-) Looking to avoid that now.
- brudgers 10y agoThis describes the machine I am in the midst of building for my boy: https://news.ycombinator.com/item?id=11884519 https://news.ycombinator.com/item?id=11884519 I'm in the US and it the used Dell Precision T7500 [1] was about $300 including shipping. There's another ~$40 apiece in a new 120gb ssd and a new Nvidia GeForce 710 GPU from NewEgg. The 710 does ~350 gigaflops which is about $0.11 per, and better than anything else at a price that's rounding error on a high end GPU. So why am I a fan of the Dell's? They're seriously robust machines. I've got a dual E5405 T7400 that's 8 years old. Last month on a whim I looked at the SMART stats for the original disk drives: more than five years powered on. Still whisper quiet. Never a hardware problem. These are machines Dell ships with a three years Next Business Day warranty. They're designed and built accordingly. [1]: http://www.dell.com/downloads/global/products/precn/Dell_Precision_T7500_Brochure_1_en.pdf http://www.dell.com/downloads/global/products/precn/Dell_Pre...
- lhl 10y agoIf you're looking to do primarily CUDA/OpenCV work you'd be crazy to not build your own PC/drop in a new Pascal card. $450 will get you a GTX 1070 that should easily do 7 TFLOPs of SP (w a slight overclock), and for $600-700 a GTX 1080 that could push 9+ TFLOPs. As far as price/performance, a high end Mac is only worthwhile if you have very specific software requirements (mostly, if you use Mac-only media production software).
- jason_slack 10y agoAny help on a board that supports multiple cards and lots of RAM?
- inlineint 10y agoHow much RAM do you need? I can recommend SuperMicro boards but it's just a guess because it's better to know what exactly do you want.