Mobile IGP Comparison

I narrowed down the platforms for our mobile IGP comparison. I simulated a Core i3 350M by taking an i3-530, underclocking it (I couldn't do anything about the 4MB vs. 3MB L3 cache) and capping its GPU frequency at 667MHz. This is the best case scenario for the i3-350M, and as you'll see below, it doesn't really matter. I also paired a 2.2GHz Pentium Dual-Core with a G45 motherboard, agian simulating the cheaper mobile Pentium platform. Finally I installed Windows 7 on the 1.6GHz Core 2 Duo based 11-inch MacBook Air with its GeForce 320M to give you an idea of the upper bound for mobile performance with what might as well be a low end discrete GPU.

Updated: I've added performance results from a simulated Core i3-330UM, the E-350's competition in ultra portables.

We'll start with Modern Warfare 2:

Mobile IGP Comparison - Modern Warfare 2 - 1024 x 768 - Low Quality

The E-350 puts the i3-350M, i3-330UM and Pentium DC to shame, delivering 67% better performance. The frame rate is just shy of being totally smooth however. I found that in most modern games 1024 x 768 would result in frame rates just under 30 fps.

Mobile IGP Comparison - BioShock 2 - 1024 x 768 - Low Quality

BioShock 2 showed a similar performance advantage. Again we're not able to break 30 fps but the performance advantage is huge compared to the Intel platforms with integrated graphics.

Mobile IGP Comparison - Dragon Age: Origins - 1024 x 768 - Low Quality

Dragon Age: Origins is mostly CPU bound at low quality settings and thus there's no real advantage to the E-350's Radeon HD 6310 GPU. It's faster than the Pentium/G45 platform, but significantly slower than the i3-350M. I expect most games however to be GPU bound at these settings across the board.

Mobile IGP Comparison - World of Warcraft - 1366 x 768 - Fair Quality

World of Warcraft continued the trend. The E-350 ended up 57% faster than the i3-350M, although still fell short of a discrete GPU.

Of course I wondered how well Brazos would play Starcraft 2:

Mobile IGP Comparison - Starcraft 2 GPU Test - 1366 x 768 - Low Quality

The GPU handles SC2 just fine, however the game is very CPU dependent and thus you see a pretty big advantage from the mainstream i3 system. The comparison is a lot closer when we look at the i3-330UM. The E-350 won't be able to play SC2 as well as a $500 mainstream notebook, but it'll be comparable to an ultraportable running ULV Arrandale.

We don't have numbers for the G45 platform here because the system wouldn't run our benchmark (our tests use an older version of SC2 which apparently had issues with the G45 drivers).

Our SC2 CPU test gives you an idea of the lower end of performance in large multiplayer battles:

Mobile IGP Comparison - Starcraft 2 CPU Test - 1366 x 768 - Low Quality

The E-350 offers only 58% of the performance of the i3-350M system. The Bobcat cores do hold the platform back from time to time. Again, compared to the i3-330UM there's no performance difference at all.

Just for kicks I also ran the Civilization V benchmark, which gave us two datapoints: GPU performance and no-render/CPU performance.

Mobile IGP Comparison - Civilization V - DX10/DX11 - 1366 x 768 - Low Quality

The benchmark doesn't score well on either platform, although AMD does hold a 72% performance advantage over the i3 and G45 platforms. The CPU test puts the E-350 at about 55% of the speed of the Pentium dual core platform.

Mobile IGP Comparison - Civilization V - DX10/DX11 - 1366 x 768 - Low Quality

Civ V is one area where the Arrandale CPU advantage wins out over GPU performance.

Overall, the E-350 has no problems outperforming any of the current Intel integrated graphics offerings in 3D games. In CPU bound titles the E-350 loses out to the mainstream i3, but is competitive with ultra low voltage i3s. Just as with Atom, you'll have to sacrifice performance vs. a mainstream notebook, but compared to low voltage Arrandale the E-350 can hold its own.

Desktop IGP Comparison: Faster than Clarkdale Final Words
Comments Locked

207 Comments

View All Comments

  • legoman666 - Tuesday, November 16, 2010 - link

    are missing.
  • Anand Lal Shimpi - Tuesday, November 16, 2010 - link

    Power consumption was listed at the end of the first page. The systems used throughout the rest of the article were based on desktop hardware, so it didn't make sense to compare to those numbers (the Brazos numbers include the test platform's LCD display). We'll have to wait for final notebook hardware to do real comparisons to competing notebook platforms. But at 25 - 30W for the entire system (including display) under load, Brazos is definitely in the right league.

    Take care,
    Anand
  • Sc4freak - Tuesday, November 16, 2010 - link

    Still seems to be missing something from page 3:

    "When it comes to power consumption however, the E-350 can't be touched. I measured max system power consumption at "

    And then it just cuts off.
  • Anand Lal Shimpi - Tuesday, November 16, 2010 - link

    er, oops...fixed :)

    Take care,
    Anand
  • jensend - Tuesday, November 16, 2010 - link

    Hm. From the third page:
    "the rest of the desktop platforms here consume much more than that at idle (much less under load)."

    Unless the desktop platforms really suck much more than 25W idling and much less than 25 under load, this isn't exactly fixed.

    I find it odd that you're even troubling to compare this to systems which suck tons more juice and not even throwing up a power consumption chart. I mean, we could certainly find that Brazos performance is rather limited compared to a Gulftown with 3xSLI GF 580s, but that's not going to tell us a lot about what makes a better laptop. If you did a performance vs. power draw scatter chart or something that'd be a lot more interesting.

    Sure, it's a little bit apples-to-oranges since you've chosen to compare it to desktop systems. But maybe that should make you reconsider what you're comparing it to (don't you have a good number of laptops around?) rather than leaving power consumption data out of the picture.
  • quiksilvr - Tuesday, November 16, 2010 - link

    Soprano'd
  • sinigami - Wednesday, November 17, 2010 - link

    uh, since there were so many missing parts to the article, maybe you are missing some charts on page 4 or something? because i only see three, and against the Clarkdale, the zacate wins only the first one, ties on the second one, and loses on the third. With only one win, that page's headline should NOT say it's faster than Clarkdale!

    you could say that the Clarkdale IGP wins or ties the zacate on two out of three gaming benchmarks, and declare it faster.

    sheesh, is that some biasedness, or just wishful thinking?
  • wiak - Tuesday, November 16, 2010 - link

    is it even suposed to be competive with E4400 anyway?, isnt the whole point of it to be power efficient, faster, use less power and totally own atom in graphics/video department?

    Danube is amd's mainstream competition to E4400, Core 2, Core i3, i5 etc
    Brazos is amd's low power competition to atom
  • wiak - Tuesday, November 16, 2010 - link

    correction: E2200 but E4400 aint that far away :P
  • Dark_Archonis - Tuesday, November 16, 2010 - link

    ... Except it does not use less power than an Atom platform. It uses more power than an Atom platform. Also as the benchmarks show, performance-wise it's barely faster than an Atom D510 in some situations.

Log in

Don't have an account? Sign up now