Mac pro is insanely overpowered!!!

[FluJunkie]

Yours looks impressive!

Mine on the other hand, looks very unimpressive, but KILLS machines!

two-dimensional LaGrangian particle tracking of particles dispersed in a carrier phase... see attached
(3-d simulation HAS to happen on a cluster with 40+ cores, or it's just not feasible)

100% across 8 cores and 100% memory utilization... took 3 days to run

time is money, and when 100mhz means hours... you get the point

Mine is even less impressive looking. The end result, with week long cluster runtimes?

A small CSV file, that will be expressed with a histogram.

 

[itsmrjon]

Mine is even less impressive looking. The end result, with week long cluster runtimes?

A small CSV file, that will be expressed with a histogram.

A week long for a histogram? What kind of simulations are you running? And how parallelized is your code?

 

[vladi]

Hilarious. Attached is a vector illustration I did for a client that took nearly 30 hours. Everything in the image was produced in Illustrator, down to the depth of field effect, so it could be scaled and utilized in any medium. They wanted a free-floating flip clock that looked to be counting down. It was going to be part of their logo -- and the referrer that sent them my way sent them to me because of the complicated vector illustration I provided for them (they wanted something that looked like it was designed by Sofa).

All that work and nearly four days later, and they ended up using a logo that took me about five minutes to throw together.

You are an overachiever and don't you ever change. You strive for the mastery and people like you are hard to find.

I understand your frustrations. Just how many hours and even days I wasted for designs and composits that client just couldn't understand and in the end just wanted something that I could have done in matter of minutes.
90% of the clients don't have clue about technology or process and most of the time they have no clue what they want but they have a vision! Their vision usually sucks big time but you do it for the money anyway and don't go showing it off to everybody.

If you can tell with what kind of a person you are dealing with you could save a lot of time. If they are not familiar with the matter they probably want simple stuff and you give it to them because they will most likely never ever appreciate complex designs simply because they dont understand.

But someday you will stumble upon one of those clients that belong to other 10% who will appreciate what you do and they will never ever let you go. You could be making a living of one client only (which is less important) and have granted autonomy to do what ever you think is right and become a master in whatever you do (more important).

 

[Commonmind]

You are an overachiever and don't you ever change. You strive for the mastery and people like you are hard to find.

I understand your frustrations. Just how many hours and even days I wasted for designs and composits that client just couldn't understand and in the end just wanted something that I could have done in matter of minutes.
90% of the clients don't have clue about technology or process and most of the time they have no clue what they want but they have a vision! Their vision usually sucks big time but you do it for the money anyway and don't go showing it off to everybody.

If you can tell with what kind of a person you are dealing with you could save a lot of time. If they are not familiar with the matter they probably want simple stuff and you give it to them because they will most likely never ever appreciate complex designs simply because they dont understand.

But someday you will stumble upon one of those clients that belong to other 10% who will appreciate what you do and they will never ever let you go. You could be making a living of one client only (which is less important) and have granted autonomy to do what ever you think is right and become a master in whatever you do (more important).

Amen. I've already garnered several very loyal, generous and appreciative clients. The irony is that, for every one of those clients, there are a dozen more that require hefty management of expectations (scope creep is my monster in the closet) and an education in what makes a good design, a good design. I often have to remind them why they hired me to begin with, as they start asking for certain aspects of the image to be changed, and then don't understand when they receive revisions which are less aesthetically pleasing than the files they requested to be altered. "Make this bigger," "Can you do this in bright red?" "I don't like how the logo is simply and iconic; can we add some gradients to it, throw on some wings -- like those Affliction fellas -- everyone loves their sequined muscle shirts; oh, and would it be possible to add one of those glossy reflections underneath our font, like Apple did five years ago -- that seems to be popular still!"

Thankfully I've gotten to the point where I've needed to take on a partner and hire other help, and I can now rely on others to manage those expectations for me, as I have no taste for it myself. I find it more difficult putting a client in their place and telling them why properly kerning the font in their logo is a good idea -- not because I'm a nice guy, but because I have little patience for such things (i.e., I don't have the time nor the energy for it). I'd rather be crunching code or spending time with my Winsor Series 7's and a nice piece of 11x17 bristol.

I think I've derailed the conversation quite a bit already, so I'll stop there. Needless to say, I think we've gotten the point across to any pot stirrers. Here's to hoping some new Mac Pros are announced this or next month, so I have a good reason to upgrade.

 

[FluJunkie]

A week long for a histogram? What kind of simulations are you running? And how parallelized is your code?

Sluggish computation of a network centrality measure, plus many, many simulations of disease spread on that network, along with some survival analysis of the simulations.

What you get out at the end is a single regression estimate from a Cox proportional hazard model for each iteration. 10,000 or so iterations...CSV file full of estimates.

Which gives you a nice mean and simulated confidence interval, best displayed as...a histogram