So what will happen to PC gaming after the year 2021?

>>58888338
Eh, then they'll switch transistor types, substrates, and get into die stacking.

That means my iMac will be relevant for years to come!

>>58888338
MoarCores.jpg?

PC gaming is already dead lad

the golden years were 2002-2008

pc gaming is officially a shit now

AAA games are more like interactive movies than they are games

>>58888338
>That's the year that most in the semiconductor industry agree will be when transistors will stop shrinking.
You're grossly misstating facts here.
About 2021 is when the ITRS sees an definitive end to log scaling of gate length for a silicon channel based device. That does not mean an end for the scaling of transistors themselves, it means new approaches will be taken. DNA based computing, self assembling structures, new materials, etc.

Area scaling will ultimately end, because matter has finite sizes, but logic density, and perf/watt will not stop improving.

>>58889319

Look at this newfag.
PC gaming golden age was from 1992 to 2004.

Better programmers, lots and lots, of programmers who know what they are doing and know how to optimize, gameplay to fixed hardware specifications, the pc may become like a console of it's own. But I have no idea.

>>58891542
twitch.tv? it's not dead it's changed.

>>58888338
>We are in the last years of rapid improvement in computer performance, that has enabled gigantic leaps in productivity and information creation for the last few decades. How does it feel /g/?

I haven't needed upgrade shit since fucking 2010 because vidja games today are pretty much the same shit as they were ten years ago, just in HD. Since a decent card back then could run the games of the day at full settings in 1080p 60fps, it can run today's games passably.

>>58888338
>>58888448
>>58888479
>>58889256
>>58889319
>>58889892
>>58891542
>>58891564
>>58891583
>>58891606
Moore's Law has been dead since the first dual core CPU. CPUs are not GPUs, you will not get double the speed just because adding cores unless the application is purely non-interractive (encoding for instance) and not interractive (e.g. gaming).
So, stop being an illiterate bastard and know it's all a fucking meme. Only GPUs can claim progress from shoving more transistors into them because by definition they work in multicore. Even then NVIDIA has to be overpriced with bigger dies if the transistor isn't easier to make.
Even when you claim a gayme "used all muh cores" you say nothing about how efficiently. Interactive applications have to use so many inefficient locks in their code to become thread-safe that there is nothing telling you how efficiently they did use them.

>>58891606
I played a handful of games and got good at them, they were around the era where an 8800 gt would be overkill. I experimented with GTA IV and V but they are a bit much for me. GTA IV is still fun on occasion though.

>>58889892
Nothing is going to beat silicon, it's amazing that it works as well as it does. Any other technology would require an enormous investment in R&D to even break even with Si.

>>58891542
-2003 you mean
'04 was only negatives, and some of the largest negatives the platform has ever faced

>>58891877
Silicon isn't going to flat out be replaced right away. Like right now we're starting to implement more SiGe in critical areas, IBM's new processes use it as a channel material. In a few years we will likely see silicon wafers with another 2D material grafted on top of them. Hybrid wafers.

>>58891921
>>58891542
>>58889319
You can still play ur old games because it's a PC and not a console :^)

>>58891877
>>58891957
What happened with the graphene meme? Is it dieded?

developers might need to actually work on optimization instead of relying on stupid amounts of power to run their shitty rehashes.

>>58892079
It's still in the baby research phase

>>58892091
It's all about algorithmic advances for a very long time. My friend who was first at his class of Computer Science and finished his course one year faster than normal called it to me in like 2007. "Look mike, it's dead, we can do algorithmic advances though".
He coded face recognition algos in MatLAB.
He is now a school teacher in a shitty town. Some harlot made him a baby, lol.

>>58888338

remember when 2010 was supposed to be the wall?

>>58891615
>Moore's Law has been dead since the first dual core CPU.

you don't know what moore's law actually says

>>58892130
I do you harlot. I say that "transistor count" is an autistic shitty metric. It's dead as anything meaningful for a CPU metric.
They are not GPUs.
That would assume everything you do on a dual core is like double the speed which utter bullshit for anything that isn't purely and nearly 100% non-interactive and therefore has no need for mutex locks.

>>58892079
Most of graphene's properties relevant to semiconductors are also found in other 2D materials which are easier to produce and work with. Samsung a couple years ago did manage to grow uniform graphene crystals, and did demonstrate a bit of research in graphene switching logic, but there hasn't been much invested in it by any large foundry. Its the type of thing far more popular in the popsci crowd.

Silicene, MoS2, HfS2, and other 2Ds and III-V materials will bear fruit in the semiconductor industry before graphene does. Though, we might actually see graphene in wide use, just not as a substrate, or part of any gate structure. Copper surrounded by graphene can sustain incredibly high signaling rates without blowing apart, and thats something that could be used in an interconnect fabric without a crazy long time to market.

>>58892190
I figured a good way to see what the secret plans of each are, to search google's Patents database. I've been finding some graphene-transistors patterns there from the big guys a couple of years ago.

>>58892212
Patents tend to be given out like candy. Companies file for thousands of patents a year just so they can keep them in a vault and claim ownership in the eventuality of someone else ever producing something even vaguely similar. Doesn't necessarily mean anything will materialize into a real product sometime soon.

Lowkey I think 450mm wafers were universally shelved by the industry years ago because they weren't going to be using bulk or even SOI at that point, and they wouldn't have the infrastructure to marry larger 450mm silicon wafers to another material. Hell, we could see deep channel trench FETs made from layers of HfS2 on top of silicon.

>>58892142
>I say that "transistor count" is an autistic shitty metric.

well thats what moores law is about, since he was concerned with fabrication costs when he wrote it.

you can claim its useless as a performance metric but it was never really meant to be that in the first place.

>>58892355
>its useless as a performance metric but it was never really meant to be that in the first place.
People took it as such, executives took it as such. It was actually as such for a time, but probably because it also came together with other transistor advances together with the count itself.
But when the dual core appeared that's when it all turned to shit.
After the dual core, it should never be mentioned again.
Only GPUs should care about a Law of their own of that kind because right now they do get almost double the speed with double the cores because that's how they work, but, if the transistor making process isn't advanced then you have phenomena like NVIDIA being extremely overpriced for the bigger chips and then kids saying shit like "AMD BTFO'ed!". Bullshit. AMD knows how to make exactly the same chips. They just choose to go on a cheaper segment of the market. Those two companies are practically identical. If a big maker like Intel or Samsung that have their own modern Foundries were into the game of GPUs that are discreet and fully featured, both NVIDIA and AMD would be dead 12 years ago. They hog the IP of the desktop GPU.

>>58891542
2003/2004 was the year steam was coming out. that is when PC gaming died.

>>58891615
Moore's law refers to transistor count, not speed you fucking mong

>>58888338
Feels good. Developers will begin to optimise their programs.

>>58891615
>Games don't use more than 1 core
>Games don't use more than 2 cores
>Games don't use more than 4 cores
>Games don't use hyperthreading
Every time the same singsong

>>58896992
>>58896992
Even when you claim a gayme "used all muh cores" you say nothing about how efficiently. Interactive applications have to use so many inefficient locks in their code to become thread-safe that there is nothing telling you how efficiently they did use them.

>>58888338
>prediction

>>58897035
a lot more efficiently than spending twice as many transistors to raise per core performance 20%

>>58897051
You have no fuck no fucking idea if that is true. Have you ever tried to code anything interactive on multiple threads? Every 10 fucking seconds you have a new variable that must be wrapped around slow as shit locks in order to become thread-safe.
Not everything is a X265 encoder.

>>58888338
It's a good thing, developpers will now have to optimize their code to make up for it