Perhaps this applies to only 95%, with 5% holding out. That's a small job, and computer scientists prefer to think in terms of rates of growth. Make the job larger, and that 95% becomes 99.9%, and we're craving more than 20 cores.
Could you go into these a bit more please? I don't really understand.The virtual cores of Core i7 are a second order benefit, a scheduling convenience. It used to be that using all cores would slow down a parallel computation, as the odd other running jobs held back one core. Now, using 5, 6, or 7 of the 8 virtual cores of a Core-i7 2600K is pretty much the same throughput, with the effect of cleanly using all 4 physical cores with no scheduling slowdown. In other words, where I used to use 3 cores of a Q6600, I now effectively use all 4 cores of a 2600K.
In February 2010, researchers at IBM reported that they have been able to create graphene transistors with an on and off rate of 100 gigahertz, far exceeding the rates of previous attempts, and exceeding the speed of silicon transistors with an equal gate length. The 240 nm graphene transistors made at IBM were made using extant silicon-manufacturing equipment, meaning that for the first time graphene transistors are a conceivablethough still fancifulreplacement for silicon.
Huh if you really think that Haskell with IO monads (well functional languages and IO, a never ending story really) and whatnot is becoming the norm, you'll be extremely disappointed by the harsh reality, because if you look around 99% of all people using functional programming languages are academics and that has been true for the last 40years and I wager won't change in the next decadesI
Functional programming is destined to become the norm, as we cope with multiple cores. It will take a long time, with lots of kicking and screaming, as programming languages are like religions. These aren't reborn Lispers preaching what they'd like to believe. This comes down to the need for read-only memory to avoid cache thrashing, and is begrudgingly acknowledged by people who aren't pleased by this news.
No, that just assumes a completely wrong understanding of Amdahl's law. The maximum speedup is 20x (obviously you can never get rid of the 5% time that have to be executed by exactly one CPU), NOT that only 20 cores will give a speedup. Those two are completely different statements. Although there are many algorithms that don't scale at all or to more than a few dozen cores, Amdahl's law has nothing to do with that.
Other than that: If Intel/Amd would know what else to do with their transistors they surely would love to do that, but there's hardly an alternative for more cores. At least for clients who don't run dozens different programs at the same time, more cores won't do good for long (that is, if you don't spend all your time encoding videos..)
Ah sorry, yep there we totally agree - though really, I didn't doubt that one second, because everyone getting monads right, surely won't have any problem with that kind of theory *personally a big fan of functional programming himself*You misunderstood my post. (Or perhaps I could have wrote it better.) What I meant is what you said, the maximum speedup is 20x.
the x86 core wars could end by 2013. What war is going to start and if not already started the integration of the I/0 and such as Ethernet etc etc
When CPUs move off of silicon onto something new (like graphene), then clocks will increase rapidly. Until then, we are only going to see very moderate increases in clock speeds with die shrinks.
Well alone the fact that nobody so far was even able to prove that the brain works non-deterministic (and many experts wouldn't agree with that assumption) shows that you're making some pretty large assumptions hereWriting software to work like the brain does is a very complex subject, but the main thing is that computers would no longer be predictably deterministic. The exact same input would only give approximately the same output, not exactly the same, depending on small differences in timing. Catastrophic failures where the network handles an input improperly can't freeze the whole system, but they can lead to various other nasty conditions.
but I imagine it is at least 5-10 years away from seeing any consumer processors that are on something other than silicon.
Maybe. If we take the mods number of 64 cores, let's see how long that'll take.
2 cores: Intel Pentium D 820 EE 5/5/2005, AMD Athlon X2 5/31/2005
4 cores: Intel Core 2 Quad QX6700 11/2/2006,
Phenom II "Barcelona" 11/9/2007
6 cores: Intel Core i7 980X 3/11/2010, Phenom II X6 4/27/2010
8 cores: AMD code-name Bulldozer Q2/Q3 2011, Intel Ivy Bridge-E Q2/Q3 2012?
10 cores: Q2/Q3 2012?
Intel
2 to 8 cores: 7 years
2 to 6 cores: 5 years(As of post)
2 to 4 cores: 1.5 years
AMD
2 to 10 cores: 7 years
2 to 8 cores: 6 years(Later this year)
2 to 6 cores: 5 years(As of post)
2 to 4 cores: 2.5 years
Aside from the superfast transition from 2 to 4 cores, the transition is much slower.
Assuming 29% increase(midpoint of 25% and 33%) from now, and say the average it takes to increase by that much takes 1 year. 64 cores is say 7x the amount(8x Intel, and 6.4x AMD).
It looks like at the current trend we might see 64 cores from AMD in 2020.
2012 10
2013 12
2014 16
2015 20
2016 25
2017 30
2018 40
2019 50
2020 64
However, I don't think we'll see that, rather we'll see heterogenous cores around 2013 timeframe, so we'll have lots of small cores + few big ones.
You could say Llano has reached 4 big cores + 80 small cores(400 "SPs" are really 80 5-width, albeit simple cores) already. Intel with that metric is at 4 big cores + 12 small cores.
Intel and AMD both already have 8-core processors and they have for a while.
Also, Intel just released their 10-core Xeons last week.
Well if we are counting the server market, get ready for 16-core CPUs from AMD
Are there any estimates on minimum die size needed to achieve certain high core numbers such as 32, 64, 128, 256, etc?
Depends on the core complexity and the desired latency/throughput of the cores.
Today's GPUs have crazy numbers of "cores"...but they can't do much beyond a narrow range of compute functions.
Most people also have no more than 2, and won't have more than 2 for 1-2 more years. Atom, Brazos, Core i3, dually i5s...it's really only AMD that's going into trying to get them quads, and I'm pretty sure they'll still have plenty of 2- and 3-core cut-down Llano CPUs to sell.I still dont see the purpose of "core wars" for consumer desktops. Most people dont even need more than 2.....