mattgadient.com

iMac 21.5″ i3-3.2Ghz with the Radeon 5670 (Mid-2010) in gaming

If you hate 27″ screens and wanted to get the strongest video card Apple offers in the 21.5″ model, this is probably the machine you went for (or… the machine you’re thinking of getting).

In any case, the Radeon HD 5670 offered in the iMac is actually an ATI Mobility 5730m. Higher model number, but mobile chips are clocked a little lower, so you’ve got roughly the power of a desktop 5670. If you’re wondering why Apple goes with mobility GPU’s, they’ve got pretty good reasons:

  • Mobility chips use a lot less power. It’s tough to find a desktop machine that competes with the iMac’s power consumption. This makes the tree-huggers happy, as well as the people with high electric bills. Incidentally, it also means Apple can go a little less crazy on the power supply in the iMac. Everybody wins.
  • Mobility chips generate less heat. Related to the power thing above. Less heat means the fans don’t need to sound like jet engines. Apple doesn’t have to worry about people burning their hands on the aluminum. Your air-conditioning doesn’t have to work as hard in the summer. Good plusses here.
  • Mobility chips tend to go on small circuit boards. Since these are the things you find in laptops, they’re smaller. There’s a lot packed into the guts of an iMac so this helps immensely.

Moving on, because this is a mid-range card and I’m used to high-end cards, I was curious to see how this would fare in games. I ran it through World of Warcraft (on the Mac side), and through Dragon Age: Origins and Mass Effect 2 in Windows through BootCamp.

Here are the results:

You’ll notice 2 resolutions were used, adjusting the Quality Slider in WoW for different test points. This was done in Dalaran in the wee hours of the morning (few people on) in a location where the frame rates were typically low. Framerates differ massively throughout Dalaran, and I wanted consistent numbers that weren’t very optimistic.

To give a playability perspective, I hopped into a battleground (Alterac Valley), so that I could see if things felt smooth with 40 players.

The “Good” quality setting in AV was absolutely perfect. “High” was very strong too, although there were a couple moments where it wasn’t 100% smooth (but only for a moment). I’d recommend starting off at “Good” quality, and going to High/Ultra if you really feel you’re missing out. You can always back down again if you find things get spiky at inopportunte times.

In terms of Windowed mode, the game isn’t handled quite as well (even though the FPS dictates it’s about the same). Spinning in circles, it’s not 100% smooth all the time. It’s still perfectly playable, and many people probably won’t even notice the minor spikes (which look like a single dropped frame).

Regardless, in the case of WoW, I’d recommend playing at full screen, full resolution, and medium to high settings. It looks great. Use VSync though, or you will get tearing during the times/locations when the framerate skyrockets above 100fps.

Next up is Dragon Age: Origins (bootcamp/windows 7). You’ll notice a couple things:

  1. It scales pretty evenly as you lower the resolution.
  2. The runs where Frame Buffer Effects was enabled (the top section) are a fair bit lower in framerate.

I’ll start by saying that all these resolutions felt playable. However, I have very strong recommendations (and I’ll explain why for each):

  • Set Graphics Detail to Low. The reason is that low is bright, and Medium/High/VeryHigh are dark. The higher settings make everything look dark, and in particular, shadowed areas look way too dark. The higher settings also takes away from the crisp-ness. It makes things look smooth, but not in a good way. Use low. The pros in terms of looks outweigh the cons. As a side benefit, you get slightly better FPS.
  • Set Frame Buffer Effects to Off. There’s a scene in the Post Coronation where sunlight comes through the window. Frame Buffer Effects actually make the sunlight look really good. So why turn it off? Because everything else looks so white it’s washed out. Faces look white and bland. Bright items are just too bright. The effects seem to be applied to *everything* which more often than not is a bad thing. These effects also suck up a lot of video-card juice. Turn it off.
  • Set Texture Detail to High. It costs you about 2fps at most (closer to 1 and a half), and you get improved quality out of it.
  • Anti-Aliasing doesn’t seem to make much of a difference visually. It might at a lower resolution, but at the default, the couple things that were slightly jaggy before AA was turned on were still jaggy after. It’s possible that it has more of an effect during conversations (where you have a close-up view), but I’d be inclined to leave it off unless you’re YouTubing one of the chats or something and want to give it a try.
  • Use VSync. Less tearing is good (and it doesn’t seem to hamper performance just by being on)

A couple critical notes if you decide not to listen to me, and are determined to use Frame Buffer Effects anyway:

  • Frame Buffer Effects + Full Screen Mode + VSync = 30FPS max. Try that combination, and you’ll be limited to 30FPS. You either have to turn off VSync or use Windowed mode. Don’t ask me how or why I managed to find that deadly combination.
  • Frame Buffer Effects + Low Graphics Detail = blindingly white. If you use Frame Buffer Effects, you’ll want to use Medium/High/VeryHigh.

And finally, if you choose to go with Medium/High/VeryHigh for Graphics Detail (which makes everything quite dark), you can try enabling Frame Buffer Effects so you can see a little better. Again, you’re better off choosing Low and keeping Frame Buffer Effects off.

TLDR version: Use Low/0xAA/High/Off, turn on VSync, and the game will both look and feel great at max resolution (and play just fine in a lower resolution mode if you’re looking to play in windowed).

Finally, we have Mass Effect 2 (bootcamp/Windows 7). During the opening scene (when you walk through the firey ship to save your friend), I actually experienced some really odd behavior. Every resolution from 1920×1080 down to 1280×1024 seemed to hug the 30FPS mark. Dropping 1 more spot to 1280×768 however instantly started hugging the 60FPS mark. It’s not a big deal (it’s a minor scene), and I’m not sure why it was like this, but it’s an oddity I thought I’d mention.

In any case, once you’ve started playing the game (right after you wake up), you’ll see fairly typical frame rates, with steady FPS improvements as you lower the resolution. ME2 actually looks good in full screen regardless of the resolution you pick, so if you find the framerates aren’t quite what you like, lower the resolution some.

During my testing, the mouse seemed very quick/inaccurate at the higher resolutions (which made the gameplay feel choppy), but this can probably be remedied by changing the mouse settings. Chances are that I’ll play at a lower resolution though for better framerates, since the game doesn’t seem to suffer visually from non-native resolutions.

One thing to note is that the video options (High Quality Bloom, Dynamic Shadows, Light Environment Shadows, Film Grain) don’t seem to have any discernable impact on the frame rates. Like WoW, the frame rates seem to flutter (although not as drastically), so it’s hard to tell for certain, but I suspect that you’re looking at a difference of perhaps 1-2 FPS.

Conclusion

The 21.5″ iMac with the i3-550 and ATI Radeon HD 5670 (5730m) seems to perform pretty well overall in games. You’re not going to be cranking everything to maximum, but you certainly won’t be stuck playing games at the minimum settings. Games will generally look good, and play well.

Certainly, you can put together a custom PC  that outperforms the iMac for less, but then you lose out on OSX (when you’re not gaming), and a few of the minor niceties like the low power consumption and simple all-in-one design.

One flaw that does present itself however is the Bluetooth “Apple Magic Mouse”. It’s magic in that it magically stops working at every opportunity in Windows. It continually isn’t recognized when booting Windows, and when waking from Windows sleep. If you plan to grab this iMac an install Windows through BootCamp, invest in a mouse too (and make sure you have another USB keyboard around that you can use for the actual Windows BootCamp install). Maybe in the future, Apple will fix the BootCamp software. Then again, it’s possible they’re trying to punish Microsoft for making terrible Mac Apps by making a faulty Windows one.

  • Kevin

    Thanks for the post – I have the same machine being delivered this morning and I can’t wait to jump on. I have a 360 for any ‘major’ gaming that I do, but I wanted this to be able to run Starcraft 2 and (hopefully one day…) Diablo 3.

  • Steve

    I was looking to replace my macbook 13 for this iMac and was wondering how it would fare with the upcoming FFXIV, could you possibly run a benchmark on bootcamp/windows 7?

    benchmark can be found here http://www.finalfantasyxiv.com/media/benchmark/na/download.html

    thanks

    • Steve,

      Ran the Final Fantasy XIV benchmark. Here’s what I got. Note that it used windowed mode (no visible way to change that), and I used the default character (Hyur). I also had to manually install d3dx9_41.dll (I’d assume they’ll include it in the actually directx installer for the game):

      HIGH – 1920×1080
      Score: 982 Load Time: 23677ms
      Scene 1: about 18-20fps on average
      Scene 2: same, with some early sections hitting mid-20’s, and low teens during the rain effects
      Scene 3: low-mid teens, hit 20fps a couple times
      Scene 4: low-mid 20’s, high teens
      Overall at that large resolution it seemed a little choppy at those framerates. Not slide-show by any means, but not smooth.

      LOW – 1280×720
      Score: 1909 Load Time: 19504ms
      Scene 1: 35-40fps
      Scene 2: 35-50fps, mainly high 20’s during rain to low 30’s
      Scene 3: mid 20’s-mid 30’s
      Scene 4: 30-40fps
      Seemed pretty smooth throughout. Again, looked good, but again this was windowed mode.

      I’ll note that the framerates I listed were the average rates that it hung around. Often you’d get large spikes into the 50’s for a brief few seconds during some easy scenes. I’ll also note that these really looked like cutscenes, not actual gameplay. I have no idea if this’ll be reflective of what will be seen in game, but hopefully it’s enough to help you compare against others who might be posting their FF results. It would be nice if there were a switch to see how full-screen mode went (I’d assume it’d pick up a few fps over windowed), but alas there wasn’t one.

  • Steve

    Thanks Matt, greatly appreciated.

    I posted your scores here http://www.psu.com/forums/showthread.php?t=212184&p=5130829&viewfull=1#post5130829 in quoted text with link to your blog, hope you don’t mind.

    Considering I had crummy results in the past with FFXI benchmark but my PC could still run it at sustained 30fps in crowded area, this iMac setup should do just fine while waiting on a PS3 version. Square-Enix does not seem to know how to code for PCs.

    Thanks again for taking the time to run these tests.

    Steve

  • Matt

    Do you have numbers for this machine runnign Counter-Strike:Source in both Bootcamp and OS X? Thanks!

    • Sorry, I don’t own CS:Source, so can’t give any numbers offhand. If Valve’s got a benchmarking demo for it, send the link. If it’s reasonably small I can probably give it a run (if it’s large I may not be able to though – our ISP have severe data transfer limits and we’ve already used up most of what we get this month).

  • Colt Fireball

    Thanks for those benchmarks. i wasnt sure about the 21.5“ iMac but this looks awesome. Have you played Starcraft 2? Would love to see those numbers.

  • Joe

    Starcraft 2 runs just fine on this Mac in OSX. I run medium settings without any performance or frame rate issues.

  • Colt Fireball

    Only medium? sad but thank you

  • Joe

    I know – not as good as my PC running a Wolfdale and 8800GT, but still not bad. I haven’t tried turning up the settings because, strangely, medium on the iMac screen still looks great. Might be the higher res and glossy screen playing tricks on me. Who knows.

  • Jelmer

    Thanks Matt for this great review, I bought one of these after reading what you wrote and I’m very happy with yet another Apple product.

    PS: SC2 runs perfect on ‘high’ settings but you’ll notice some frame rate issues on ‘ultra high’.

  • martyn

    Great review, I’m now running this myself. If you buy the parallels software and run windows as a virtual machine the magic mouse issue goes away! It runs the best of both worlds on one screen… Happy days.

  • Steven

    I really like this article but one thing is wrong:

    You say that the mobility 5730 is roughly the power of a 5670 desktop but that is not true.
    On http://www.videocardbenchmark.net they give the 5730m a rating of 700 and the 5670 desktop a rating of 1250 ! A big difference!

    Of course I don’t give benchmarks too much weight, so I cheeked out some FPS comparisons, and I would give the 5730m a rating of 700 then I would give the 5670 desktop a rating of 1000.

    Most desktop 5670 have GDDR5 memory, where as the 5730m only has GDDR3 memory.

    The desktop 5670 has a memory speed of 1000Mhz, where the 5730m has a 800Mhz speed.

    The desktop 5670 has a GPU core clock of 775mhz, where the 5730m has a 650mhz core clock.

    They both have the same GPU, so the same around of stream processors (400), same texture units (20), and same ROPs (8).

    5730m : having the slower GDDR3 memory, and a slower core clock, it is a much cooler and energy efficient video card, but runs about 30% slower on average, compared to the desktop version (5670 GDDR5 version)