I got the email notification this morning. The Mac Pro can be ordered now and ships in 1-2 weeks.
So who is blowing a cool $20k on a base model? Or an eye watering $100k on the full fruit option?
1 Like
If you wanted one — mainly for the bragging rights, not because you’re going to do any of what Apple thinks people are going to use them for — $10,000 over 10 years isn’t such a crazy pill to swallow.
But $160 per wheel? That’s crazy town.
2 Likes
…pats the MP1.1 and MP3.1 purring under the desk. “Love you guys. Looks like you will have to keep working for another decade or so”.
There is definitely a few options in there that leave you wondering how they came up with pricing. Did they just ask a group of toddlers?
faint … dont wake me until Mac prices become affordable again.
Well, the unit is a price of a good car, so that’s about the right price for good wheels/ tyres. 
2 Likes
You can max spec it to nearly $90,000…
The 1.5TB of ram costs $40,000
Nice looking machine, but lack of Nvidia GPU support makes it a non-starter for me.
I’m a medical professional and perform a lot of whole genome sequencing on my Mac Pro’s. I would classify myself as someone who is in the professional space, requiring a Pro machine; but as I do not use FCPX I’m not considered Pro anymore.
I’m currently in front of a 2010 5,1 Mac Pro with 12 x 3.33GHz running an RX580 running 10.13.6 which is my data gathering machine, whilst to my right is a 2009 4,1 Mac Pro running Ubuntu 16.0.4 with an Nvidia 1080GTX crunching my data using CUDA.
Both machines I feel are nearing their EOL in terms of reliability and support.
Early next year I will be replacing them both with a 64 core threadrippper and dual 2080RTX Ti cards running Ubuntu as my main machine.
It will be a sad day when I retire my 4,1 and 5,1 Mac Pro’s but I’ll just use them at home for photography and music storage.
Why is the Nvidia GPU specifically required over others out of interest sake? Obviously ignoring the stupidly high price tag of the machine
CUDA compute and machine learning in the medical and scientific space is still huge.
In my case I am sequencing entire genomes of viruses.
In my workflow, it takes 3 hours to process 40GB of raw genome data (data that is just electrical signals) on a $400 NVIDIA 1080GTX into nucleic acid sequence files.
Same data on 12 core Mac Pro using CPU only takes 2 months using all available threads.
No support for AMD in this space, and all tools are CUDA only.
For those interested the technology I’m using is called an Oxford Nanopore MinION
2 Likes
That sounds like pretty amazing work!
Would the $100k 1.5Tb RAM Mac Pro make what you do much faster? Able to give an example of how long an analysis takes now, and how much quicker it might be? Just curious as to a real world application effect by using the top spec version.
It would be quicker than my current Mac Pro but I would have to rely on CPU only.
Something that may take my current Mac Pro 12 weeks to process using only CPU may take 9-10 on the new Mac Pro.
Ethics and governance overlays covering the work I do prevent this type of analysis being uploaded to cloud compute services, so I am stuck with having to do everything on in-house machines.
CUDA processing on Nvidia cards results on processing speed ups of 2-3 orders of magnitude.
So that same dataset would take from 2 to maybe 5 hours depending on which graphics card you use.
The big problem is Apple only using AMD cards.
My current 10 year old Mac Pro with a 3 years old graphics card will absolutely slaughter a maxed out Mac Pro due to my ability to use CUDA.
This isn’t a full dataset but it gives an idea of what CUDA can do.
https://esr-nz.github.io/gpu_basecalling_testing/gpu_benchmarking.html
Someone else’s opinion on how CUDA speeds things up.
That sounds impressive considering the first thing I’d install on the new MacPro would be Boxer with my Win311/Civ2 setup.
Romani rectores orbis terrarum!!!