What is Digg?191 Comments
- ChromaVita, on 04/09/2008, -7/+75On the front page with no comments? QUICK! Think of something witty to say!
- AlienX3.5, on 04/09/2008, -14/+58But can it run Crysis?
- mattwalton56, on 04/09/2008, -3/+44...no....nothing can.
- harrypl0tter, on 04/09/2008, -3/+22LOL have you looked at your username recently?
- UncertainIntent, on 04/09/2008, -6/+25...and improving ten-fold upon them.
- cvrefugee, on 04/09/2008, -13/+31I didn't understand any of that *****.
- Dylson, on 04/09/2008, -7/+24Well smack my ass and call me sally!
- inactive, on 04/09/2008, -5/+22Not even...God.
- dallashigh, on 04/09/2008, -1/+17Essentially, there will not even BE a front-side bus anymore. Like AMD did with their HyperTransport technology, memory will be accessed directly by the processor. There will no longer be an external memory controller.
- bwdd, on 04/09/2008, -8/+22Oh god, just when I get a new PC.
- Hegemony, on 04/09/2008, -1/+14If AMD can't come up with a competitive architecture to go up against Nehalem they may never come back.
- dannyboy3020, on 04/09/2008, -1/+13Less than a minute too late :)
- BigglesPiP, on 04/09/2008, -4/+15Intel's copying of K8 architecture (but better) is complete.
- craftyguy, on 04/09/2008, -1/+12Satanael,
FSB is dead with the onset of QPI. QPI is essentially a direct connection from the CPU to the memory. The memory controller is built on the CPU. With FSB-oriented platforms, the memory controller was built on a separate chip (northbridge), so you essentially had one connection from the CPU to the memory controller, and then another connection from the memory controller to the memory.
QPI basically cuts out the 'middleman', and in doing so, negates the use of a FSB. - Spuy767, on 04/09/2008, -0/+11Basically, you should start getting an erection now, because with the Core 2's already taking everything else to the woodshed, and this looking to extend, rather generously, on those microarchitecture advances, this thing is looking to be a ***** barn burner.
- MrPig, on 04/09/2008, -0/+10Thanks! That helped so much!
- craftyguy, on 04/09/2008, -1/+11I'd say you would reach a limit when you no longer have anything to dish out in parallel. Some threads depend on others, and therefore cannot be executed in parallel to the thread they depend on. For example some character on screen doesn't start walking before it stands up just because the 'start walking' function/thread completed before the 'stand up' function/thread if they executed in parallel.
I've seen a few enterprise-level applications utilizing 16+ cores at 100%. Most modern games just do not have enough 'stuff' for that many cores to do at once, since they tend to be more GPU intensive as it is.
Also, if you are a game developer, wouldn't you rather develop for 2 cores/threads instead of 8, given that there's a greater majority of your customers running 2 cores/threads. - spinaltap87, on 04/09/2008, -1/+11insert comment about digg's front page algorithm
- inactive, on 04/09/2008, -0/+9A question for programmers: When you code in multi-threads, is there a limit to how many core's a program can actually utilize? For instance would Crysis be programmed with an ideal number of cores in mind such that after a certain number you don't gain marginal clockspeed? Or does one thread correspond with one core?
- mattwalton56, on 04/09/2008, -3/+12Well, maybe God.
- madbadger, on 04/09/2008, -1/+10cool, it comes with Dance Dance Revolution 3!
- dracostimpy, on 04/09/2008, -2/+11Donny, you're out of your element.
- Lane, on 04/09/2008, -0/+9Here comes the new mobo line ups.
- qwertycopter, on 04/10/2008, -0/+9Atleast he didn't ask if it will blend. *****.
- Bkaufman, on 04/09/2008, -1/+10I'm no expert (hardware engineer, not software), but I believe it would depend on the architecture of the software. The simplest way is to just split the main components of your program into a standard number, and then run those on separate cores (or run some on one core if there aren't enough available). You can also write code that will dynamically distribute work to all of the available cores, which is significantly more scalable, but also much more complicated.
- mcberg182, on 04/09/2008, -4/+12*slap on the ass* Sally!
- Owned1Up, on 04/09/2008, -3/+11Am I the only one having trouble pronouncing this?
- Satanael, on 04/09/2008, -2/+10I said what this means FOR the FSB, not what FSB means. Ass.
- matt646, on 04/09/2008, -0/+8perhaps
- inactive, on 04/09/2008, -8/+16I don't know about you guys but I'm getting tired of this...
- apzdsx, on 04/09/2008, -1/+9*smack* Sally!
You like that? - Bkaufman, on 04/09/2008, -2/+9Won't happen until they integrate the Floating point capability of a video card into the current chips that we have today. Just increasing the power/number of cores won't do it.
- inactive, on 04/09/2008, -0/+7Very informative indeed.
- Ksg89, on 04/09/2008, -4/+11Can it run doom then?
- Jumba990, on 04/10/2008, -0/+7So it's not worth getting a new computer now? =/|
*empties Newegg shoppingcart* - kaplanfx, on 04/09/2008, -0/+7Intel is never gonna let us down!
- inactive, on 04/09/2008, -0/+6Obsessing over the codename, aren't we?
- michaelschmitt, on 04/09/2008, -0/+6Nehalem 2. When it's time to replace your computer, you'll be able to buy something even better than Nehalem. Enjoy your new PC.
- hybridcreation, on 04/10/2008, -0/+6Someone obviously never learned simple economics.
- renegadeafk, on 04/10/2008, -0/+6That is a stupid statement. Without AMD there would be no competition and you would pay 500 bucks for a Pentium dual-core
- JayD16, on 04/10/2008, -0/+6You can't just "split the code into sections." You have to specifically write code that can be executed at the same time without causing errors. Putting function calls in another core is useless because the first core is waiting idle.
- noseeme, on 04/09/2008, -1/+7***** that stupid digg meme.
- inactive, on 04/09/2008, -1/+7I don't do all that much with threading but you can achieve higher efficiency by splitting things up (in a games case, Physics, Rendering, Input and Graphics, and Audio could all be split off)
One thread generally corresponds to one core, again, AFAIK. An easy way to test this is to run a rapidly CPU depleting application that does not use threading on a multicore system and see that the percentage of CPU usage is always 100/number of cores
Games don't usually do dynamic distribution, if ever. - hybridcreation, on 04/10/2008, -0/+5True.
Phenom is a bust and Nehalem might just be the most powerful CPU we have seen yet. The divide is massive now...this is only going to more it bigger. - Spuy767, on 04/09/2008, -1/+6The biggest thing is going to be the integrated memory controller.
- eqtitan, on 04/09/2008, -0/+5{QPI} I'm wondering what the bandwidth limit will be and if it can really be overclocked well beyond spec.
Octacore here I come.......
The word Nehalem is of Indian origin. One attributed meaning is "place where people live." - inactive, on 04/09/2008, -0/+5nuh-halem
- protogenxl, on 04/10/2008, -0/+5"The most amazing achievement of the computer software industry is its continuing cancellation of the steady and staggering gains made by the computer hardware industry."
-Henry Petroski - inactive, on 04/09/2008, -0/+5I sure as hell hope so.
- sarixe, on 04/10/2008, -0/+5he said think it, not say it
-
Show 51 - 100 of 191 discussions
The Digg Toolbar for Firefox lets you Digg, submit content, and keep track of Digg even when you're not on the Digg site. Download the official 
© Digg Inc. 2009
— Content created and posted by Digg users is