I'm sorry if this sounds a little nooby, but why the hell would you need all that processing power? what could you actually do with it that would help you in "everyday" life?
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
New 324-Node Mac Pro Cluster In The Works At Va Tech
- Thread starter MacRumors
- Start date
- Sort by reaction score
find better formulas or actually discover formulas to better make 'stuff' hahahaha
Maybe Frank Beamer can use some of that power to help out the football team this year 
By the way, when they dismantled the G5 system, they sold off each computer. My neighbor has one that still runs perfectly to this day. He got a certificate of authenticity and actually framed it.
Leo Laporte picked one up too and as far as I know still uses his to this day as well.
Alumni here as well. Go Hokies!
By the way, when they dismantled the G5 system, they sold off each computer. My neighbor has one that still runs perfectly to this day. He got a certificate of authenticity and actually framed it.
Leo Laporte picked one up too and as far as I know still uses his to this day as well.
Alumni here as well. Go Hokies!
Let's go Hokies! (Alumni)
Good to see the project get an upgrade, but to "only" be in the top 10 whereas before they were third is a little disappointing. If I recall, the bigger news about this was the cost/benefit ratio since, at the time, the number 1 and 2 spots cost considerably more.
Good to see the project get an upgrade, but to "only" be in the top 10 whereas before they were third is a little disappointing. If I recall, the bigger news about this was the cost/benefit ratio since, at the time, the number 1 and 2 spots cost considerably more.
When you start need GigaFLOPS of power, then GPUs suddenly get very relevant if you can program them (CTM, CUDA, OpenCL).
That is aewsome!!! i just do not see a use for such a complex computer
"I think there is a world market for maybe five computers..."
Good point.
I wonder how many years in the future a computer with this much capability will fit in your hand? Will it be 10, 20, 50 or 100 years?
because the idea of a supercomputer is that it isn't for everyday life, it's for advanced scientific research and calculations. for example, an imac couldn't even begin to process all the data involved in this:
http://public.web.cern.ch/Public/en/LHC/LHC-en.html
Exactly.
Still it is fun to fantasize having that power on your desktop.
Computing power is evolving faster and faster. Fifty years ago, the desktop computer did not exist. Back then it was Big Iron (mainframes) for the most part.
Storage was via large magnetic tapes and huge disk drives.
Imagine what we will have 50 years from now. Almost incomprehensible.
Sex on Mercury would be HOT at least in the daytime as it reaches 427 °C.
Nightime would be a different matter, -180 °C.
You takes your pick...
sounds like my set up at home
you need to get your central heating / Air Con fixed then
Oh, you meant the Mac Pro setup.
Nehalem is a significant improvement, but it won't increase theoretical FLOPS at all.
That being said, I agree with waiting for Nehalem.
LOL. Good joke.
The task would be to:
- Read 5 to 8 GB from the DVD
- Cut into 2592 chunks (about 3 seconds of video per core)
- Distribute the 2592 "jobs" and data to the system
- Wait for all of the pieces to come back
- Assemble the 2592 pieces in order into the output video stream and write it to disk
In other words, a good illustration of the difficulties of parallel processing - note that only step "4" runs in parallel.
Parallel processing frameworks (like the future "Grand Central") make it easier to do this, but cannot solve the inherent problems of sections of algorithms which must be serialized, and the overhead of cutting jobs into threads.
See http://en.wikipedia.org/wiki/Amdahl's_law
Intel's Tera-scale project aims for > 1 teraFLOPS sometime in the 2010s (I think 2015) for regular computers. It will most likely take another decade or so to get 29 teraFLOPS, and maybe another decade to get a handheld device to 29 teraFLOPS. So I'm thinking about 30 years.