OpenGL Performance & Performance-Per-Watt For NVIDIA GPUs From The Past 10 Years

[Michael]

For those that have been wanting 1080p results, will run a smaller comparison for an article in the next few days.

[efikkan]

Thank you Michael for another great benchmark!

I am still amazed at how much the good 'ol GTX 680 is still a relative beast.

I can see that my dear GTX 680 was a good investment
In terms of GPU architecture, Kepler was a major change from Fermi so it appears this helped Nvidia's OpenGL implementation more than Direct3D. I actually upgraded from a GTX 580 to a GTX 680, which sounds silly since it's only about ~7% faster in Direct3D, but retrospectively I can clearly see that it's way more powerful in OpenGL

[Jumbotron]

Yeah.....it seems that Kepler was a real paradigm shift and that Maxwell is like the Intel "Tick Tock" development cycle where Kepler was the real architectural change or "Tick" and Maxwell is the refinement stage or the "Tock" in the upgrade cycle. I mean....sure....efficiency increases and additional hardware video decode and encode upgrades are nice and legitimate innovations. But it's not a paradigm shift. Hopefully Pascal is another paradigm shift.

[Ardje]

Wow... I'm amazed by the lack of progress in heat dissipation. Almost all cards suck air from inside the machine, and blow it back into the machine, probably almost directly back to the fan of the card, or to the motherboard. Some even do a little (but not much) hot air to the outside.
Why has there been no progress (except in servers) in airflow and heat dissipation desing in pc's?
Doing SLI is as good as impossible, because all cards suck the hot air from the card next to them.
Even the silent-pipe-3 is bad in design, although it probable is the best among them.
I want a simple design: air is sucked in from the back of the card and it is exhausted at the outside of the card. This can either be passive (mandatory fan in the case needed) or active.
In case of a riser card with a 90 degrees angle flow from the bottom of the card to the top would als be acceptable.
But all these cards just suck in hot air from inside the machine and exhaust it at all possible points from the card back inside the machine.

So yes, I am very dissappointed with PC design these days. No progress for 20+ years.

[slacka]

I agree a more "classical" resolution like 1080 would let older cards feel better, but changing resolution for each group of cards... definitely no.

If Michael does what you propose this forum would be filled by people whining that comparisons make no sense at all because resolution / details are different from one bench to another...

No you misunderstood me. Of course he needs to keep the resolution the same. My point was he should have tested at BOTH high and low resolutions.

[drSeehas]

I used the cards I have.

You complain in every graphics card review you get no card from AMD unlike nVidia where you get every card from.
You didn't get a single GK208 card from nVidia?

2x G84
3x G92
2x GF119
2x GK107
2x GM107
3x GK104
2x GM206
2x GM204
2x GM200
and
0x GK208?
Are you serious?

[tigerroast]

I agree. I'm not really sure why companies like Intel are wasting so much time and money trying to do die shrinks when they should be looking into other elements or types of transistors.

You should check out what Stanford's doing with researching a new processor architecture using carbon nanotubes. This should pique your interest.

For reference, scientists have found that graphene (the material used to make carbon nanotubes) is a damn near perfect conductor of electricity. And heat, moreso than copper. There's two major problems that need to be worked out with carbon nanotubes/carbon nanotube transistors though:
1) mass production (that has yet to be figured out yet)
2) reducing its danger as a health hazard (they're small enough to accumulate in your body and cause necrosis, so mainly a hazard to workers unless someone wants to create asbestos 2.0)

[schmidtbag]

tigerroast Yup, I'm well aware of the wonders of carbon nanotubes and graphene. Both of which I personally deem as of the greatest materials ever conceived. Many people know stuff about graphene's conductivity or that it acts as a super-capacitor, but what people don't usually know are the slew of other amazing abilities it has such as converting thermal energy into electricity, or converting photons into electricity.

[tigerroast]

Yup, I'm well aware of the wonders of carbon nanotubes and graphene. Both of which I personally deem as of the greatest materials ever conceived. Many people know stuff about graphene's conductivity or that it acts as a super-capacitor, but what people don't usually know are the slew of other amazing abilities it has such as converting thermal energy into electricity, or converting photons into electricity.

Pardon my unnecessary crash course.

Mass-producing it would make it revolutionary, and refining it for various uses would make it game-changing. Sourcing carbon certainly isn't an issue, refining it even less so. Making a computer, hell even an FPGA, out of CNT transistors would be savage if the theoretical performance gains over silicon are really that promising.