3DMarks & GFXBenchmark

We don't use 3DMark to draw GPU performance conclusions but it does make for a great optimization target. Given what we've seen thus far, and Intel's relative inexperience in the performance GPU space, I wondered if Iris Pro might perform any differently here than in the games we tested.

It turns out, Iris Pro does incredibly well in all of the 3DMarks. Ranging from tying the GT 650M to outperforming it. Obviously none of this has any real world impact, but it is very interesting. Is Intel's performance here the result of all of these benchmarks being lighter on Intel's weaknesses, or is this an indication of what's possible with more driver optimization?

3DMark: Ice Storm

3DMark: Ice Storm Extreme

3DMark: Cloud Gate

3DMark: Fire Strike

3DMark: Fire Strike Extreme

3DMark 11 - Performance Defaults

3DMark 06

I also included GFXBenchmark 2.7 (formerly GL/DXBenchmark) as another datapoint for measuring the impact of MSAA on Iris Pro:

GFXBenchmark 2.7 T-Rex HD

GFXBenchmark 2.7 T-Rex HD - 4X MSAA

Iris Pro goes from performance competitive with the GT 650M to nearly half its speed once you enable 4X MSAA. Given the level of performance Iris Pro offers, I don't see many situations where AA will be enabled, but it's clear that this is a weak point of the microarchitecture.

Synthetics Compute Performance
Comments Locked

177 Comments

View All Comments

  • whyso - Saturday, June 1, 2013 - link

    They are completely different systems making power consumption values irrelevant.
  • codedivine - Saturday, June 1, 2013 - link

    Hi folks. Can you post the OpenCL extensions supported? You can use something like "GPU Caps viewer" from Geeks3d.
  • tipoo - Saturday, June 1, 2013 - link

    Interesting that the compute is punches above it's game performance weight. I wonder if they could put more EUs in a chip, maybe a larger eDRAM, and put it on a board as a compute card.
  • lmcd - Saturday, June 1, 2013 - link

    They already have a compute card called Xeon Phi if I remember correctly.
  • Klimax - Sunday, June 2, 2013 - link

    Different Arch (X86 in Phi)
  • tipoo - Sunday, June 2, 2013 - link

    I'm aware, but the Xeon Phi requires completely different programming than a GPU like this which can just use OpenCL.
  • Soul_Master - Saturday, June 1, 2013 - link

    What's your point for comparing desktop GPU with middle-range mobile GPU? CPU on both devices are not equal.
  • Soul_Master - Saturday, June 1, 2013 - link

    Sorry. I misunderstood about i7 4950HQ process, a high-end quad-core processor for laptops.
  • Ryan Smith - Sunday, June 2, 2013 - link

    It's what we had available. We wanted to test a DDR3 version of GK107, and that's what was on-hand.
  • tipoo - Saturday, June 1, 2013 - link

    Hmm, so it's heavily hinted at that the next rMBP will ditch discreet graphics. The 5200 is good, but that would still be a regression in performance. Not the first time Apple would have done that, there was the Radeon cut out of the Mini, the 320M to the 3000, even the bottom rung of the newest iMac with the 640m. I wonder if it would at least be cheaper to make up for it.

Log in

Don't have an account? Sign up now