Nvidia: Moore's Law is Dead, Multi-core Not Future
Nvidia chief scientist says that everyone needs to rethink processors in order for Moore's Law to continue.
Bill Dally, the chief scientist and senior vice president of research at Nvidia, wrote an article for Forbes purporting that Moore's Law, the theory that transistor count and performance would double every 18 months, is dead.
The problem, according to Dally's paper on Forbes, is that current CPU architectures are still serial processors, while he believes that the future is in parallel processing. He gives the example of reading an essay, where a single reader can only read one word at a time – but having a group of readers assigned to a paragraph each would greatly accelerate the process.
"To continue scaling computer performance, it is essential that we build parallel machines using cores optimized for energy efficiency, not serial performance. Building a parallel computer by connecting two to 12 conventional CPUs optimized for serial performance, an approach often called multi-core, will not work," he wrote. "This approach is analogous to trying to build an airplane by putting wings on a train. Conventional serial CPUs are simply too heavy (consume too much energy per instruction) to fly on parallel programs and to continue historic scaling of performance."
"Going forward, the critical need is to build energy-efficient parallel computers, sometimes called throughput computers, in which many processing cores, each optimized for efficiency, not serial speed, work together on the solution of a problem. A fundamental advantage of parallel computers is that they efficiently turn more transistors into more performance," Dally added.
Dally also posed that focusing on parallel computing architectures will help resurrect Moore's law, "Doubling the number of processors causes many programs to go twice as fast. In contrast, doubling the number of transistors in a serial CPU results in a very modest increase in performance--at a tremendous expense in energy."
One big driver of the current processor design are the programs written to run on current chips. Dally said that the long-standing, 40-year-old serial programming practices are ones that will be hard to change, and that programmers trained in parallel programming are scarce.
"The computing industry must seize this opportunity and avoid stagnation, by focusing software development and training on throughput computers - not on multi-core CPUs," Dally concluded. "Let's enable the future of computing to fly--not rumble along on trains with wings."
- Graphics,
- Hardware,
- Nvidia,
- Intel,
- moores-law ,
- multicore ,
- serial ,
- parallel ,
- programming
- Internet Explorer Market Share at 10-Yr Low Point
- Fallout New Vegas Screens, Details Released
- External HDD Combines Storage, Speaker, USB
- Apple Could Face Antitrust Probe from DOJ, FTC
- ATI FirePro RG220 Shares GPU on Networks
- 15 Years of Intel CPUs in One Picture
- QNAP Launches 8-Drive Rack-Mounted NAS
- Fermi Heading to Laptops With GTX 480M?
- Rage Screens Released; PS3 Better Than X360
- Deals for May 5: HP dv8t, Free .com, AC PSP-3000
- Medal of Honor Reboot Info, New Screenshots
- Intel Launches Moorestown, Supports Android
- HP Envy 14 and 17 Official, Starting at $999
- VIDEO: Flash Running on Tegra 2 Android Tablet
- Ellen DeGeneres Made Apple Angry With Fake Ad
- Speedy Chrome 5 Tested Against Potato Cannon
- Intel Shows Light Peak Laptop Pushing 2 HD Videos
- A Dating Site Made Just for Apple Fans (iDate?)
A good analogy, why then is his company pushing Fermi, a hark back to the era of the Valve, hot, power hungry and inefficient?
As good as his ideas might sound, does he take into account how much effort it would take to turn them into reality? How many programs written today can run on a system he proposes? The gains made by more efficient computing won't outweigh the effort needed for a transition, at least not for home use. Servers and super computers are a different matter.
And not to mention the many years in which both serial and parallel systems will run alongside each other, meaning every app would have to be coded twice...
And a company that took years to design the piece of junk Fermi really is, isn't exactly in a position to tell what computing should evolve into. If there is a single company not in a position to speak about energy efficiency, it's Nvidia. 480's using nearly as much power as 5970's but at significantly lower performance aren't good examples of efficient computing.
A good analogy, why then is his company pushing Fermi, a hark back to the era of the Valve, hot, power hungry and inefficient?
He said it himself; Not many are proficient in parallel programming. I expect the support doesn't exist. Nvidia sending out a product that doesn't work with anything would be... dumb?
" In contrast, doubling the number of transistors in a serial CPU results in a very modest increase in performance--at a tremendous expense in energy."
Aaaawwwww, look! How cute! The guy is worried about our electricity bill!
Psssssttt, don´t look now, but your new card guzzles down Watts like it´s made of candy.
Would not be surprised if the 480 had something to spark Bill Dally's idea. Essentially the 480 was crammed with a couple billion more transistors than its predecessor to gain only a "modest increase in performance--at a tremendous expense in energy."
Maybe he doesn't know about Fermi yet ...
we thats exactly what i wanted the cpus have stopped improving since pentium 4 , so i guess its time for a change
"Doubling the number of processors causes many programs to go twice as fast."
I assume this guy has some prior software engineering experience? If so then'll he'll realise what an absurd statement that is. Performance will only 'double' if the software is written specifically to take advantage of those extra cores. And like SLI or Crossfire scaling that is an idealised figure at best- all multicore software will fall short of that figure (save some minor anomalies) because of the extra overhead (synchronisation locks etc.) required to make the software parallel in the first place.
Furthermore not every problem is embarrassingly parallel like graphics. In some cases having 1000's of cores is no good because the problem simply cannot be split across that many nodes at once. The needs of general purpose computing are not the same as the needs of graphics; Intel, AMD, etc. are adopting the right approach when it comes to their multi-core CPUS. If the game changes and we need some new architecture then I'm sure they'll be the among first to go down that route.
Can't help but think this sound-bite is yet another sneaky dig at Intel, especially at Larrabee. Maybe NVIDIA would be better focused on getting it's product line in order and restoring it's competitive edge in the graphics market, rather than making childish jibes at the competition.
The empty can rattles the most...
Good point darraghcoy
you'd need a new framework before parallel CPUs are of any use. But I don't think that this is a jibe at Intel.