Tuesday, May 22nd 2007

IBM Unleashes The World's Fastest Processor

IBM today simultaneously launched what it claims is the fastest processor in the world and an ultra-powerful new computer server that leverages the chip's many breakthroughs in energy conservation and virtualization technology. IBM's new POWER6 chip is a 64 bit, dual-core processor with 790 million transistors running at up to 4.7GHz and 8MB of on chip L2 cache. At 4.7GHz, the dual-core POWER6 processor doubles the speed of the previous generation POWER5 while using nearly the same amount of electricity to run and cool it. This means customers can use the new processor to either increase their performance by 100 percent or cut their power consumption virtually in half. Also announced today is the IBM's new 2- to 16-core server which offers three times the performance per core of the HP Superdome machine. The new server is also the first ever to hold all four major benchmark speed records for business and technical performance.
Source: IBM
Add your own comment

30 Comments on IBM Unleashes The World's Fastest Processor

#26
kwchang007
mdm-adphYou know, I've always found it rather odd that the three processors in the three top gaming systems (Wii, XBox 360, PS3) are PowerPC based, while everything else seems to be going to Intel.

Does Intel not like the gaming market or something? Or are Power processors simply better suited for some reason?
the three gaming systems cpu's are pretty much meant to be powerful but cheap. If you could make a cell processor run on a microsoft based system, you would see that core 2 quad will destroy the cell. why? because the cell is kind of like a celeron. they load it up with very little cache (i think it's like 64 kb per core, compared to intel's 2mb per core), not only that they also do not have a sophisticated branch predictor. Between these too problems, plus the problem of not being able to run on windows, it just kills powerpc based cpu's.

These chips must have alot of power saving techniques, doubling clock speed and still using that same amount of energy is not an easy task. just look at p4 compared to p3. while the clock speed is nice and all, i would like to see benchmarks, perhaps on a linux based work station between opteron, xeon, and power pc, all using the same amount of cores and the highest spec'd chips (no overclocking) then a test with all of them overclocked. My guess would be xeon would die in something memory intensive with power pc in second, and amd in first. In all eles..i would think xeon would win or be close. sorry for the long post guys.
Posted on Reply
#27
Steevo
^ Pretty good post.



They are made to run looping instructions or instructions that are pipe fed. Not graphics or sound or any other high variable process. And when they are in high variable process (think huge database) they use alot of system RAM and fast SAS disks to make up for the CPU's latency.



Yes they are super fast when fed a long stream of instructions that don't branch, or call for a whole active set to be flushed then reloaded a few cycles later. We on the other hand use our CPU's for graphics setup, sound, and many other out of order processes.
Posted on Reply
#28
kwchang007
Steevo^ Pretty good post.



They are made to run looping instructions or instructions that are pipe fed. Not graphics or sound or any other high variable process. And when they are in high variable process (think huge database) they use alot of system RAM and fast SAS disks to make up for the CPU's latency.



Yes they are super fast when fed a long stream of instructions that don't branch, or call for a whole active set to be flushed then reloaded a few cycles later. We on the other hand use our CPU's for graphics setup, sound, and many other out of order processes.
thanks. it was from an article somewhere, it was explaining why a pentium d :eek: or athlon 64 would be better than what the 360 has or the cell. i just condensed it a bit. they also explained how this makes game produces jobs hard as anything. because the branch predictors are so primative (compared to computer cpu's) the coders have to code big hints to the processor to keep the pipe full. i think that console makers focus to much on specs of what the system can run. :wtf: maybe with folding on the ps3 you can push it to the max....but with a game, no way, not to mention the thing has 8 (not including the power pc core) cores (some high number i think it's 8) and it's hard to keep them fed with info. look at pc's..how hard is it to keep 2 cores loaded with info. you have to run...two prime 95's, maybe two movies encoding with a virus scan running in the back ground. not many games can take that advantage, and even though you have the cpu in the console doing the load for sound, that takes one core at the max. so if you really put alot of manhours into a game, you get one thread for the sound, maybe two for cpu players, two for physics, and two for graphics. OMFG you get 7 threads, amazing what you could do with that, but now put that into how many manhours of work you need....that's a crap load of time. that's why most games on consoles do not take extensive use of multi core, and on the computer, multi threaded games are hard to come across, right now at least ;). sorry for the long post again, lol
Posted on Reply
#30
FatForester
malwareIBM today simultaneously launched what it claims is the fastest processor in the world and an ultra-powerful new computer server that leverages the chip’s many breakthroughs in energy conservation and virtualization technology. IBM's new POWER6 chip is a 64 bit, dual-core processor with 790 million transistors running at up to 4.7GHz and 8MB of on chip L2 cache. At 4.7GHz, the dual-core POWER6 processor doubles the speed of the previous generation POWER5 while using nearly the same amount of electricity to run and cool it. This means customers can use the new processor to either increase their performance by 100 percent or cut their power consumption virtually in half. Also announced today is the IBM’s new 2- to 16-core server which offers three times the performance per core of the HP Superdome machine. The new server is also the first ever to hold all four major benchmark speed records for business and technical performance.



Source: IBM
That is NOTHING like netburst! Everyone knows P4's were never this efficient with power/heat. This is like saying a 4.7ghz P4 has the same power/heat efficiency as a 2.2 ghz P4! YEA RIGHT. These Power6's aren't as efficient clock per clock, but they ARE faster, and use about the same amount of power of the past generation POWER5's. They may be running into a brick wall with their clock speeds, but what they did with power and heat is remarkable. And yea, its more expensive, but what new technology isn't?!

.. Sorry about the needless rant. I read some of the posts and felt like saying something. :laugh:
Posted on Reply
Add your own comment
Aug 25th, 2024 08:16 EDT change timezone

New Forum Posts

Popular Reviews

Controversial News Posts