Amazing system, FFXIV taking FPS hit. Why?

[Archadius]

Well it sounds like that mobility card is just causing issues. Unfortunately MacPro is too expensive for me, it's mostly a whole lot of stuff I just don't need (to get the same processor, etc, it would cost me over $4000) and a lower graphics card. I think my best option would be to just deal until they release another version (next year or 2013) that doesn't use an M-grade card.

My old system didn't have problems, it just wasn't performing graphically as well as I wanted, so I saw the upgrade and thought "why the hell not". I never expected Apple to upgrade their processor etc but DOWNGRADE their GPU in the new iMac models. By what you're telling me, anything Mobility is really a downgrade. Unfortunately though, I don't really have another option right now since you can't select different GPUs when you customize the iMac. My graphics do look pretty good now, so I am ok waiting a year or two for an upgrade again, I had just hoped for better results from this card than I'm getting is all. Thought maybe it could be something simple like a driver fix, etc.

On a side note, I'm also fairly certain that Fraps is just plain broken. Based on the premise that windowed and full screen modes have a considerable FPS difference, I conducted some tests. Normally I run windowed, and I was using fair settings and seem to be running around 25 FPS average. So I tried full screen... Same FPS! From everything I've read in the forums and elsewhere, the difference between windowed and full screen should be a huge FPS hit... In conclusion I think I'm probably getting better FPS than I think I am, Fraps just sucks. lol

[Coldfire]

Actually the HD 6970M is as fast as a normal HD 6850 and is pretty highend. The problem lies elsewhere. You should be able to play the game in high settings with that card.

[Archadius]

Actually the HD 6970M is as fast as a normal HD 6850 and is pretty highend. The problem lies elsewhere. You should be able to play the game in high settings with that card.

Currently:

- 2560x1440 (this can't be changed or everything looks like ass, not sure why)
- Windowed
- General Drawing 8
- Background Drawing 3 or 4 (maybe my eyes suck but I notice no difference between 3 and 5)
- Shadow Detail standard or high (same as above, no difference between standard and highest)
- 8xQ MSAA
- Texture Detail high
- Texture Filtering highest
- Cutscene Effects on
- Ambient Occlusion off
- Hardware Mouse on

Additional Notes:
- All settings returned to defaults in ATI CCC
- Fraps results 20-30 FPS with occasional lag spikes

Does that seem reasonable? If not, then my quest continues... lol
There has got to be something I can do besides "buy a PC" or "upgrade your hardware again".

[origamikitsune]

I both agree and disagree with Coldfire. When Celeron chips first came out, they were as fast as the Pentium chips they were based on; however they performed poorly in comparison. For one, they lacked on-die cache. Point is, MOBILE chips of any kind are built around power saving, space saving, and heat saving; as such there are fundamental differences in HOW they operate. I'm not say it is a -bad- chip, I'm saying it is a -mobile- chip, and they are -not- made for gaming (usually they are marketed towards HD video and artists in my experience). You will notice most gaming laptops have full on GPUs, not mobile version of the GPUs.

As for FRAPS in windowed mode; I suffer about a 5FPS hit in cities if I switch from full screen to windowed, and a 0-2FPS hit outside of cities between full and windowed mode. How much FPS you loose varies very, very, widely between systems based on your card, driver version, number of monitors, etc., etc. In and of itself no change doesn't mean anything. It is suggest to switch to full screen because the majority of cards with the majority of drivers on the majority of machines it helps to a noticeable degree (though not necessarily to a significant one).

Also, at that resolution, it occurs to me that you aren't necessarily out of range of expected performance. For one, most people who are seeking optimal performance do run full screen, and they run at at 1080p (1920x1080) on General Draw quality 8. They do this because 1.) that's the native resolution and 2.) GDQ 8 sets the screen buffer equal to the screen resolution; that is, it draws the screen at the same size as the resolution and displays it.

By setting the draw quality to 9, the buffer is increased to 1.5x the resolution, and then shrunk to fit the screen (providing a smoother display through blending). This causes a noticeable hit to FPS.

By running at draw quality 10, the buffer is increased to 2x the solution, and the hit to FPS is even more extreme.

Now look at the numbers:
General Draw Quality 8 at 1080p = 1920x1080 = 2,073,600 (number of pixels drawn)
General Draw Quality 8 at your resolution = 2560x1440 = 3,686,400
General Draw Quality 9 at 1080p = 2880x1620 = 4,665,600

The thing to note here is that while your performance should not be expected to be as low as running a 1080p on GDQ9, there is a significant difference between your setup and what (most) people are running. And to top it off, it is done on a mobility. If you want proof that Mac hardware is decent, look no further than this. People have been overtaxing their GPUs and giving themselves heating and power issues running settings like that on PCs

Anyhow, I have a pet Mac expert that I talk to on the first Friday of every month (and the significance of that day for 1,000 geek points anyone...?); I'll poke him and see if he has any super secret mac-head techie tricks in his back pocket that might shed a little more light on the issue.

[Archadius]

As far as your data is concerned:

Looks like in theory I'd be running much better and the graphics would probably look about the same if I was able to do 1920x1080, but for whatever reason, every time I switch to that instead of my computer's native (2560x1440) the FFXIV display is crap, even at character selection screen (and believe me, I tried tons of different settings coupled with 1080p resolution to get a different result).

What you say about the Mobile chips makes a lot of sense to me, gaming is not the top priority on a Mac (they are working on improving it with each year though), HD video and graphic artistry (that's me!) are where it's at. This totally explains the fact that they chose a Mobility card and it's still considered an upgrade.

Maybe Fraps isn't lying to me then, based on your experiences. I doubt I would notice a 2-5 FPS hit difference from windowed and full screen, especially since I only did comparative testing for maybe 10 mins. I was expecting something far greater, like a 25-50% FPS drop from full screen to windowed.

Your post overall has given me a much needed wave of calm lol, and I greatly appreciate you asking your friend about tips/tricks in my situation. Please let me know what you discover!

Also, I don't get the geek reference of the first friday of each month, sorry lol

EDIT: Oh and if either you or your friend know of a good way to monitor/manipulate my GPU temps/fan speeds whilst running Windows 7, that would also be appreciated. I've tried HWMonitor and the ATI CCC, both of which are either inaccurate (HWM said my GPU was running at like 90 degrees after maybe 2hrs, no friggin way) or not functioning properly (CCC won't let me control my fan speed even with AMD Overdrive unlocked and it reads as always "0% fan speed").

[origamikitsune]

For Windows 7 I recommend HWMonitor by Cupid. It isn't the best display style, but it is small, low memory footprint, accurate data, and most importantly doesn't slow FF14 down.

If you were running 14 while using HWMonitor, yes, it is entirely possible things were running very hot. They usually do while running 14. It was that very issue that made us techies realize that there was an issue with running GDQ above 8. Also, is that C or F? Big difference there. I don't know if HWMonitor lets you control your fan speed though. Also, I don't know if there are differences in mac hardware that will throw off those readings or not. It isn't out of the range of possibility.

Just for reference, I run at about 69C on GDQ8, and run about 79C on 9; remember that you're at like, GDQ 8 and 1/2 relatively speaking because of the increased resolution. Also, with a mobility chip I wouldn't be surprised if some of the function of the GPU is shoved off onto the processor; I don't know enough about ATI Mobility chips to say for certain, but that could also explain some of the increased heat. Someone more experienced with ATI could probably answer that better than I.

[Coldfire]

Can you post a screenshot of "crap". Eh I mean of the game on a different resolution. And did you try 720p?

[Archadius]

I was running it while on FFXIV, it just seemed very hot to me in a very short amount of time. I believe the highest was 88C. I've heard that Mac stuff can run hotter than usual and still be fine, but that seemed high to me. I also need to give it a try using my new settings. For the record, HWMonitor doesn't let you control fan speed. I'd like to controlling this because I seriously believe my fan speed is not operating correctly whilst in Windows.

[origamikitsune]

88c is not too bad for an all-in-one system running 14 at that res. Its hot for sure, but not system breaking. Intel's website can tell you the recommended and/or maximum

[Archadius]

88c is not too bad for an all-in-one system running 14 at that res. Its hot for sure, but not system breaking. Intel's website can tell you the recommended and/or maximum

Yay! I spent some time looking for something that would give me an idea of recommended/max temps but didn't really get far... That should help, thank you!