StriKeVillain
Banned
I just moved up from a 965 BE to a 4690k. I'm really liking the improvement.
-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
Intel: future CPUs to be slower but more efficient
- Thread starter Renekton
- Start date
I just moved up from a 965 BE to a 4690k. I'm really liking the improvement.
It's been suggested earlier in the thread that Intel spends $10 billion annually on research, and that the laws of physics are actually causing issues here.
I don't think it's fair to say they're "sitting on their asses" if those are true.
Freshmaker
I am Korean.
They should be making cube CPU's then. Their 2d thinking will hold gaming back forever.
...Not seeing the problem here. From a consumer perspective this is great in saving money in the long run. The only people that'd be disappointed are the ones that want to throw money out the window to render 16K Movies or some shit insanity like that.
BlackBuzzard
Banned
So i guess no need to replace my 2500K. I really just need to OC it. But im afraid to fuck something up and fry it.
OnionPowder
Member
Glad I just purchased my 5820K last year then.
Wondering if future advances in PC games will rely on GPGPU instead of the CPU.
Wondering if future advances in PC games will rely on GPGPU instead of the CPU.
Sectorseven
Member
Can't they just duct tape two Skylakes together?
Sure but even with AMD's promise of having at least 40% IPC improvement over their previous architecture they still won't reach parity up for another 2-2.5 years.
Roughly 15% IPC each year and then you have to factor in that reviews for low overhead APIs like Mantle and the new DirectX 12 show bigger performance gains for older Intel CPUs and current AMD CPUs than the newest INtel ones which means the overall gap is going to be closer again for the next year or 2. Software really hasn't done much to make Intel's newest chips a must buy.
Cute Dragon Here
Member
The minor CPU-side improvements do stack up somewhat, but most of the improvements went to the processor graphics (integrated graphics), platform power, and idle power usage. Most things that benefit laptop and tablet designs, yes, but today's desktop should also run very efficiently.
Though, the 2500K and 2600K are still more than fast enough for everything, so...
Saucycarpdog
Member
It'll be a while before they are fully competitive with ARM in terms of efficiency. The market wants compact, efficient processors for their mobile lives but even Intel's Atom series and "efficient" M Core series still are vastly bigger energy hogs compared to their ARM cousins.
MutonCommander
Banned
It makes sense, current CPUs are already good enough for pretty much everything the consumer wants to do.
Fredrik
Member
This is depressing if true. Can't they go for more cores or combining several CPUs on the same mother board or something like that? I never thought I'd see the day when computer tech stopped evolving toward better performance. How do we go forward now? Better programming? Faster memory?
MidgarBlowedUp
Member
I went from a 965BE to a W3690 oc'd to 4.0Ghz and it is just fine. I'm ready to upgrade my 4gb 760 though.
My other pc is still on a Q8200 & an old ass gtx260
TAJ
Darkness cannot drive out darkness; only light can do that. Hate cannot drive out hate; only love can do that.
I have one. It's not enough for some games that have terrible multi-threading performance unless you win the jackpot in the overclocking-potential lottery or don't care about 60fps.
Nexus Zero
Member
Nothing unexpected. This is what coming to the end of a computing paradigm is meant to look like. Let's see where we are in a decade though.
Some Nobody
Junior Member
Since games pretty much use consoles as a base for visuals, maybe by the time they've figured out a way to make faster CPUs we'll have learned how to pull the best visuals out of the PC CPUs we have now, since today's PC specs are tomorrow's console's.
But I'm firmly in favor of increasing efficiency, so let's get on that for a little while.
But I'm firmly in favor of increasing efficiency, so let's get on that for a little while.
RealityCheque
Banned
But they're fun and awesome. So the planet can just suck it up.
thekiddfran
Member
They are about to begin ramping 10nm technology and are testing 7nm technology. They are constantly innovating.
Plus their chips will become programmable in a few years thanks to the purchase of Altera
Plus their chips will become programmable in a few years thanks to the purchase of Altera
Beautiful Ninja
Member
They already do the more cores/cpu's thing in servers and supercomputers. Higher efficiency chips allow Intel to more cores in the same power envelope and size envelopes, which means higher performance for applications that can take advantage of the extra cores. There's still plenty of performance evolution going on, but it's not being aimed at single threaded IPC due to the laws of physics getting in the way.
The issue for performance arises for applications who do not take to paralleization well, which to my understanding includes gaming. While there is still a lot game engines can do to improve multithreading performance, DX12 helps here, it's still going to be a struggle to get huge leaps in CPU performance for gaming until the likes of Intel defy physics or move onto things like quantum computing.
SRTtoZ
Member
Trojita
Rapid Response Threadmaker
Yeah let's turn off all those server farms while we are at it /s
CrispyBoar
Banned
nvm
GillianSeed79
Member
Been wondering whether I should get the 5820K and invest in X99 or go Skylake and get an i7-6700K. Anyone have any suggestions?
Youre not thinking big enough. Make it so big, that we can comfortably fit inside. Then when you are living inside your processor thats also fully furnished, the access terminals will only require as much space as functionally desired.
Problem solved. House-puters.
Lan involves throwing a string with a can on either end into the neighbors window for voice chat.
I just upgraded from a 3570k to a 5820k and saw massive gains. I think the 2500k is adequate but outdated. My workflow has improved greatly with the new CPU and a lot of the slowdown I was getting in modern games has cleared up without changing my GPU.
Not necessarily but there are many other tasks that benefit greatly from increased CPU power. I spend a lot of time producing video content and any amount of CPU you can throw at it has a beneficial effect.
(This is going to sound kind of bitter but its just my 0.02$).
I get why you guys are happy that you will be able to keep your CPU's for a long time, but this has far reaching affects beyond gaming. Unless there is a material/approach switch, (be it the higher clocks of graphene, or the smaller density of a carbon nano-tube, or something even more exotic like optical/quantum computing), there may be a lot of stalls in the current progression of technology in many fields. Computing affects many other industries and particularly artificial intelligence (well the hunt for true artificial intelligence) will suffer (even if temporarily) from a delay. On the other hand there is a much needed switch off of silicon coming and the only real thing that may satiate the hunger for more powerful CPUs before a material/approach switch is the approach towards Amdahl's law (which even then we are already close to approaching the limit of). So in many ways this is bad, but it also has the potential for massive breakthroughs down the road.
However I do think Intel is calling it quits rather early as there have been successful 1nm transistors, although the yields on those will be beyond bad. So here's hoping for a materials/approach switch in the near future since at the moment this is rather bad for the progression of technology as a whole. Because in this Intel based future you could just power up more low power machines, but it does not replace out performing ten of those low powered machines for one of the power hungry ones (which you can then get ten of) by any margin, and the more computers you add to super computing clusters the more the network, coordination, messaging, and limits that apply from parallel scaling come into play.
I get why you guys are happy that you will be able to keep your CPU's for a long time, but this has far reaching affects beyond gaming. Unless there is a material/approach switch, (be it the higher clocks of graphene, or the smaller density of a carbon nano-tube, or something even more exotic like optical/quantum computing), there may be a lot of stalls in the current progression of technology in many fields. Computing affects many other industries and particularly artificial intelligence (well the hunt for true artificial intelligence) will suffer (even if temporarily) from a delay. On the other hand there is a much needed switch off of silicon coming and the only real thing that may satiate the hunger for more powerful CPUs before a material/approach switch is the approach towards Amdahl's law (which even then we are already close to approaching the limit of). So in many ways this is bad, but it also has the potential for massive breakthroughs down the road.
However I do think Intel is calling it quits rather early as there have been successful 1nm transistors, although the yields on those will be beyond bad. So here's hoping for a materials/approach switch in the near future since at the moment this is rather bad for the progression of technology as a whole. Because in this Intel based future you could just power up more low power machines, but it does not replace out performing ten of those low powered machines for one of the power hungry ones (which you can then get ten of) by any margin, and the more computers you add to super computing clusters the more the network, coordination, messaging, and limits that apply from parallel scaling come into play.
Mac@NeoGAF
Banned
Well anyone who's just played 'The Division' Beta knows the time when games are stretching your top end, overclocked CPU to the limit....has already arrived.
I don't know enough about Artificial Intelligence or the approaches for it (neural networks, deep learning) to say for sure, I just brushed up on the topic over time. However I think that on some level, some semblance of intelligence is possible. But my thinking basically boils down to this: Given enough bits, you can simulate a synapse, neuron, etc. Why would you not be able to simulate the brain given enough computing power? True intelligence or a computer that is able to reason beyond our capacity may be a stretch, (however at its core they would be inherently better even on a human's level due to the fact that they remember precisely, don't fatigue or lose focus). But I don't see how intelligence on some level is impossible for a computer. Given enough bits. (Which is a big given in and of itself).
do they look like real life yet?
BigTnaples
Todd Howard's Secret GAF Account
So is the future dead?
I mean. Where do we go next?
Surely this is not the pinnacle of computing.
I mean. Where do we go next?
Surely this is not the pinnacle of computing.