Just a thought I had just now. They aren't great for anything but flat images. So when do you think that sort of thing will happen for intel chips?
I'll guess 25 years.
Well, considering that development cycles can take up to a decade- they may already exist, or something close. I don't imagine Intel can justify the cost vs. demand for something like that.
I think they go with demand, which for the average user, means playing 1080p video, maybe some League of Legends. That doesn't take much.
Plus you know, controlling the market, making/taking as much $$$$ as they can from us, lol.
1 Like
Until Zen APUs with HBM2 on die probably
4 Likes
25 years is too long, my Intel HD 4400 godstomps my parent's 10 year old ATI X1300 30x over. At least I can play most games at HD (except for the most graphically demanding, which outright won't be playable, makes me sad). I remember having to play games like Spore at 480p Minimum settings and still got 20 FPS on that old computer when I can play it at 1080p with most settings at High (Spore is a bad example, but it's the only reference I can make).
Although this also depends on simply the limitations of our current technology as well as far as how far we can go. Right now, the best iGPU is around an R7 250(X?) equivelent on AMD's side and Intel Iris Pro 580 is comparable to a GT 940M. We will probably see GTX 980 levels in 4-5 years on integrated graphics.
AMD could pull that off within a few years. They got GPUs, they got CPUs and consoles are getting stronger every year.
Expensive solution:
R9 Fury with HBM with two Zen 4 core packs.
Less expensive solution:
Wait for console peasents to demand 1440p 60FPS. Then add two years.
4 Likes
Man, it would be a wet dream for me to have a Surface Pro-like tablet but with the Zen/HBM APU solution. Think of the performance that I can get in such a small but versatile package. Would be a suitable upgrade form a Surface Pro 2, I don't think the Vaio Z Canvas is substantial enough.
1 Like
Well using my SUPER NON SCIENTIFIC method of looking at their last 3 years of iGPU flops and then finding the average % gains year to year and continuing that trend. Intel wont hit the 4612 GFLOPS until roughly 2028.
Obviously this is not a good way of doing this but here is my math.
From 2013 to 2015 the average jump in GFLOPS on the top Consumer iGPU was 11.15%. Assuming this average continues and nothing else changes the iGPU Intel will reach the 4612 GFLOPS needed around late 2028.
Mostly Real data:
2013: Iris Pro Graphics 6200: GFLOPS 883.2
2014: no data, I averaged between 2013 & 15
2015: Iris Pro Graphics 580: GFLOPS 1152
GUESS WORK:
2016 GFLOPS: 1278.72
2017 GFLOPS: 1419.38
2018 GFLOPS: 1575.51
2019 GFLOPS: 1748.82
2020 GFLOPS: 1941.19
2021 GFLOPS: 2154.72
2022 GFLOPS: 2391.74
2023 GFLOPS: 2654.83
2024 GFLOPS: 2946.86
2025 GFLOPS: 3271.01
2026 GFLOPS: 3630.82
2027 GFLOPS: 4030.22
2028 GFLOPS: 4473.54
2029 GFLOPS: 4965.62
2 Likes
Isn't the upcoming PS4 supposed to tote something akin to an RX 480?
So, if we include consoles, 980-grade iGPU is just around the corner.
... As for when these iGPU will be available for desktops, is another question.
Except the GTX 980 outclasses the RX 480 in most games, even on DX12, you need to gain a good overclock to put the RX 480 at GTX 980 levels and the PS4 is having it heavily underclocked to where it's just above a R9 380X.
It even outperforms the 1060. LOL!
I know, and yet people act like they are on the same level. Maybe it's cause lack of Async and stuff.
One of these:
Or two of these:
Screw you both get 6 GTX 9800's
3 Likes
RX 480 cause more Mega-Gigabytes... 4 GB ain't shit anymore..
You're right, as it stands, the GTX 980 performs measurably better than the RX 480 in current games, and, if we want, we can put it in a higher class. But I don't think that splitting that sorting that "step-up" as a separate class makes the difference in performance a world of difference - this is still performance levels approaching a 980 that will be coming out within the next year.
Like you said, it's probably only going to be a handful of years before we get there, or even surpass it.
There is no need for it for a long time yet until 4320p is a normal thing that comes as standard.
But if they wanted to I figure 3 generations more could have it easy enough.
I'll take 20 for Bit Coin Mining...jk, I'd take two over the GTX 980, but the GTX 1070 and 1080 are a thing and even 2 RX 480's have trouble beating them. But that was missing my point about the cards in performance.
I hope I got my point accross.
One costs double (or close to that), then you can compare them but the result is given by the question posed.
I would love to get a second Fury because they are cheaper then GTX 980s, but I got my FX-8320 in my path...
1 Like
I am really hoping for dual APU scalability. To go from 4 cores/1000 steam processors to 8 cores/2000 steam processors just plop in another APU:)
except for broadwell intel is more or less able to double gpu performance every generation. that said we only have 2 die shrinks left on silicone based chips. intel will likely need to create a new architecture to advance. much like the jump from core2duo to i-seres. if intel switches to graphene based chips and uses fiber-optic motherboard traces then odds are good intel will reach 980 level in 2nd gen of that architecture.which could be in the year 2022 given that in the intel roadmap they plan to reach 8nm by 2018.
1 Like