I guess using parallel cores could predict the linier portions of an image as precharged/preloaded? This would make it look better, but not work faster? I guess that means The preload is not faster, the out come is not faster, the response is faster. The more that can be predicted, the faster the response could be? I'm glad I got the 64bit then.
Wouldn't we want to see obsolete core processors then? Making core processors, consumer specific products? Like gaming hand held what they use to be. Only now with low power processing with high user input and high resolution display. But what will replace computers then? Quantum Computers?
I'm sorry, this doesn't prove to me that Amdahl's Law is universal, as it makes too many assumptions. What it is true to say is that adding parallel processors does not speed up the serial portion of your program. A truer statement is that: if task 'delegation + processing' for all parallel tasks completes faster than the serial portion, you won't notice any speed increase. As for the task delegation speed penalty, take a look at the Transputer concept, there's more than one way to organise.
youldn't it be a good idea to make the CPU asymetrical so you have like 1 or 2 large cores that have a high clock and high IPC and a lot of small ones that arn't clocked that high but that are meant to do the parallel work or is this a like an apu
That was fun. Is there a law that relates to the amount of work to process vs. number of processors? If the lawn was much bigger, the overhead would become much less significant.
That wasn't just a simplified analogy, from the assumptions to the analysis and all the way through to the conclusions you drew... You replicated the marginal productivity theory (of labor) verbatim!
Not bad, you got it across. I would have pointed out that somewhere along the curve is the optimal # of parallel processes to shoot for, a max benefit point.
I'd also give a real world example of a process that maximizes say at 3.14 cores, I wouldn't do the math or explain the process itself, just mention that there is process XYZ that you're better off running on a dual core rather than quad core system. (all other parameters being equal)
lol keep...doing stuff what your doing
Mrgreatestfreakout 1 week ago
I guess using parallel cores could predict the linier portions of an image as precharged/preloaded? This would make it look better, but not work faster? I guess that means The preload is not faster, the out come is not faster, the response is faster. The more that can be predicted, the faster the response could be? I'm glad I got the 64bit then.
Samurninsai 1 month ago
Wouldn't we want to see obsolete core processors then? Making core processors, consumer specific products? Like gaming hand held what they use to be. Only now with low power processing with high user input and high resolution display. But what will replace computers then? Quantum Computers?
Samurninsai 1 month ago
Comment removed
AvenueRiseMusic 3 months ago
I wonder if John Deere approves of this analogy ? " Buy less mowers from us, and get more done !!! " (-;
mikecarter79 4 months ago 4
I'm sorry, this doesn't prove to me that Amdahl's Law is universal, as it makes too many assumptions. What it is true to say is that adding parallel processors does not speed up the serial portion of your program. A truer statement is that: if task 'delegation + processing' for all parallel tasks completes faster than the serial portion, you won't notice any speed increase. As for the task delegation speed penalty, take a look at the Transputer concept, there's more than one way to organise.
blahdelablah 6 months ago
"Look... Process Boy." This guy is hilarious. Kinda reminds me of Larry David.
mahcuz 8 months ago 3
lmao .....very funny and informative...... what do people do when they're not mowing lawns ?????
charlestwoo 9 months ago
youldn't it be a good idea to make the CPU asymetrical so you have like 1 or 2 large cores that have a high clock and high IPC and a lot of small ones that arn't clocked that high but that are meant to do the parallel work or is this a like an apu
themassau 9 months ago
great video, thanks for this
SirCrest 10 months ago
i thinks it practically not good by applying this law.
kashifkhi 10 months ago
Woo Hoo! now I can go do stuff :D
gldouglass 10 months ago
That was fun. Is there a law that relates to the amount of work to process vs. number of processors? If the lawn was much bigger, the overhead would become much less significant.
fullerrobin 11 months ago
That wasn't just a simplified analogy, from the assumptions to the analysis and all the way through to the conclusions you drew... You replicated the marginal productivity theory (of labor) verbatim!
You're a labor economist!
DOHC2L 11 months ago 2
John Deere sucks.
treat666 11 months ago 2
Not bad, you got it across. I would have pointed out that somewhere along the curve is the optimal # of parallel processes to shoot for, a max benefit point.
I'd also give a real world example of a process that maximizes say at 3.14 cores, I wouldn't do the math or explain the process itself, just mention that there is process XYZ that you're better off running on a dual core rather than quad core system. (all other parameters being equal)
BTW the lawn example is useful in many areas.
t1nym1ght 11 months ago 2
For set-up time, shouldn't it be 9 mins and 39 mins since you aren't including yourself? It's just how you've labelled the columns
canadafreakazoid 11 months ago