The 660 Ti Thread: Questions and Answers Here!

[zeosrule]

Today I installed 10.8.3 and now when I wake from sleep I get no video.

EDIT: Reverted to 10.8.2. I am now fine again.

 

[idaanx]

I'm getting artefacts/screen freeze on Flash videos with VDADecoder enabled by setting sysdef to macmini 5,1, also tried a few other options. In 10.8.2 I fixed this by enabling HD4000 in the bios. This doesn't work anymore in 10.8.3 since I'm getting a kernal panic at boot with both GPUs enabled.

Is there a way to get this working without going back to macpro 3,1 and disable the hardware acceleration of videos? Running VDADeconderCheck reports its fully supported.

 

[morg]

A little follow up to this...

Updated the BIOS on my Sabertooth x79 mobo to the most current and it seems many problems were fixed.

However, discovered that the display port connection on my 660ti only connected at 1080p as a max resolution under windows, and under MacOS I would get an elaborate color show of artifacts. Now have an RMA approved so guessing it was a down to a faulty card in the end...

 

[mundmc]

Greetings all! I am in need of some instructions on how you all have installed your 660 Ti, EVGA GTX 660 Ti SC in specific. I am no dummy when it comes to installation and troubleshooting, yet can not get my card working stable. I am seriously thinking of doing an RMA on the card. I don't want to, so before I do I will give it one last try.

What I am looking for is:
1) Which AGPM GPU device plist do I use, and what edits (if any, besides adding my device ID) are needed?
2) Does Finder QuickLook work with your mp4 videos, and do HD videos in YouTube work alight?
3) I would also like to know if your games are working ok and not crashing/freezing; I am not looking for fps, just stability. I play only Diablo III, StarCraft 2, and WoW in OS X so your experience with any or all of them is appreciated.

So far Finder and Safari freezes are rare; they usually happen when I have a GPU crash aka

NVDA(OpenGL): Channel exception! exception type = 0xd = GR: SW Notify Error IOVendorGLContext::ReportGPURestart.​

Last year at this exact time, shortly after I upgraded to a 6870HD, I had, if I recall correctly, almost the same exact problems. I RMAd the card and all was working perfect after the replacement. This is why I don't want to RMA yet another card, and desperately need help.

I'm having a similar problem; while everything initially appeared fine, playing Starcraft II led to random seeming white-screens/ black-screens/ gray screens requiring a turn off from my power supply. I suspected the 660 TI was overheating, as the fan doesn't throttle, and the vent gets REALLY hot. I have to reboot to see a temp, and i'm not sure how much it has cooled down by then, but it's usually high 60's. Running FurMark, the card hit's 74 on stress test with no increase in fan speed and plateau's there.

I did a clean install with Multibeast 5.2 using OS 10.2 updated to 10.3 (combo update, re-installed TRIM, graphics=no, symbios 10,3).

I don't know if it's covered under warranty in a hackintosh, as i have started the RMA procedure. Help very appreciated.

 

[caraguia]

I have installed a fresh mountain Lion 10.8.3. No install fakeFMC. CUDA DRIVERS installed. Don't installed any multibeast opengl graphics option.I don't use the integrated videocard ( one monitor). In the bios I have selected the 660ti and IGPU in auto.

I have edited AGPM for my device 660 ti.

In Macpro 5,1 youtube work very well but only can you see 16x and 39x states of CPU
.
Mac mini 6,2 definition give me 16 22 25 29 35 39. It's very good. But youtube in safari and firefox, don't work and
looks pink and green and after the system frozen. in chrome youtube woks perfectly. Why?

in the console it can see when I'm watching a youtube video:

02/04/13 01:37:52,000 kernel[0]: NVDA(OpenGL): Channel exception! exception type = 0x20 = PBDMA Error
02/04/13 01:37:52,000 kernel[0]: 0000006e
02/04/13 01:37:52,000 kernel[0]: 00000001 0000902d 00000000 00000000
02/04/13 01:37:52,000 kernel[0]: 00000000 00000100 00000000 0000000d
02/04/13 01:37:52,000 kernel[0]: 00000000 00000000 00000000
02/04/13 01:37:52,000 kernel[0]: 00000000 00000000
02/04/13 01:37:52,000 kernel[0]: 0000006e
02/04/13 01:37:52,000 kernel[0]: 00000001 0000902d 00000000 00000000
02/04/13 01:37:52,000 kernel[0]: 00000000 00000100 00000000 0000000d
02/04/13 01:37:52,000 kernel[0]: 00000000 00000000 00000000
02/04/13 01:37:52,000 kernel[0]: 00000000 00000000



Help me please.

iMac13,2 works as well. In fact, I see smoother performance (CPU and GPU) using iMac13,2 over MacPro5,1. As for my previous issues, it seems one pair of RAM is wonky, and XMP was setting the timings wrong. Manually setting the sticks corrected things.

Since updating to 10.8.3 I too have seen LuxMark scores almost cut in half. 650 vs 1015....:

Have you in mac pro 5,1 definition many CPU states?

 

[dakotaht]

FCPX 660TI and 10.8.3 Rock Solid! Finally . . . Let the party begin.

Since the start of my Hack project I've always had issues with video cards. OSX
would simply lockup and I'd cold start: over and over.

10.8.2 installed in the way you expect it, then I did an update to 10.8.3 and a scratch install.

First I created a new Unibeast Flash Drive using my MacBook Pro, downloaded a new ML install 10.8.2 and the 10.8.3 combo update.
Next I formatted the USB flash drive and created 2 partitions, the second partition to hold Multibeast and the combo update.

My hack has a SSD 2.5 inch external dock in one of the bays, specifically the Antec EasySwap USM, originally designed to swap a Seagate GoFlex but the dock also handles bare 2.5 drives and SSDs. The system board Gigabyte GA-Z77X-UP5 TH has esata in the back, pretty handy for an external esata dock.
Set the bios to internal graphics and plugged my apple Cinema 30 into the DVI connection.

Started up with the Unibeast install, and created a start disk. The video is a EVGA Nvidia 660TI which I had practically given up on.
I have a number of small laptop hardisks I have collected over the years so I installed to one of them.

Installed the 10.8.3 combo update then immediately ran Multibeast, edited the Chameleon plist (in extra folder) so GraphicsEnabler=no.
Restarted and everything worked perfectly. At first the infamous blue rectangles appeared then the rectangles cleared and the computer locked. A cold start got it running again, and it hasn't locked since!

Let me say that again...it has not locked. . . period.
FCPX was running for 3 hours, there was an occasional flicker but it never locked.

The trick is to install Multibeast immediately after the update and BEFORE
restarting. The prior update from 10.8.1 to 10.8.2 worked the same way.

Don't install Nvidia drivers or CUDA.

Cloned the drive to an SSD, reran Multibeast and fixed permissions (also edited the Chameleon plist file) . . . perfect.

Apparently Apple included fixes for Nvidia cards in 10.8.3.
ATI was a bitter experience for me: the company sells Radeon video cards. AMD bought ATI but the
company has never had any skills in writing drivers which I believe is the most
difficult part of selling hardware.

Lately Nvidia has become, in my experience, a suspect also. Every half hour or so the original Hack would
lockup, changed cards (also Nvidia because Adobe is using Nvidia cards for
acceleration) and that would lockup. I noticed that the display showed white specks and I realized that video memory was corrupt; and of course blue rectangles!

Absolutely nothing worked. I jerked the card and went back to the 460. At least I could get some work done before it locked up, this time after 2-3 hours.

I digress:

This install I did NOT install Nvidia drivers (bundled in the update). So far, no lockups. Apple always
does a great job with drivers, so much so that OSX is unkillable (is there such
a word?). Conversely Microsoft's biggest problem is lousy drivers from third
party vendors.

Updating FCPX to 10.8 requires 10.8.3 update.

=============
Update: Another SSD updated but that SSD had CUDA Installed. Usual issue: blue rectangles but this time the machine locked. Cold started then deleted the preference pan with app delete, problem disappeared. Could there be an issue with CUDA and FCPX?

=============
Just when I thought 10.8.3 was stable with the 660TI I decided to install Diablo III (Heck of a way to waste a hlaf an hour or so). Diablo video froze and the sound continued, then the machine locked up.
Sigh . . . will this ever end?
I have a Radeon I pulled from my Mac Pro (2008) I'm thinking it might work on this machine.

=============
Another update - RMA'ed the card back to EVGA. 3 months of hassling has convinced me the card is defective. The final clue was a very small horizontal line on the Cinema 30 which indicated to me a memory problem with the card itself.

More to come, no doubt.

 

[Bimmer325Ci]

Here is what I've found with my system regarding "NVDA(OpenGL): Channel exception!". Whenever I choose an Ivy Bridge SMBIOS, I get the "NVDA(OpenGL): Channel exception!" issues. I'd like to know why, if anyone has an answer??? The MacPro SMBIOS's work great for me, expect for power management (i.e...only 2 p states). What SMBIOS are you getting the fault with???

 

[caraguia]

I had the same problems. I selected the imac 13.2 profile with my graphics injection in AGPM and so far so it's working.

Thanks to BANSAKU. He told me.


You can also choose mac pro 5.1 with graphic inyection in the AGPM kext.

 

[Bansaku]

I'm having a similar problem; while everything initially appeared fine, playing Starcraft II led to random seeming white-screens/ black-screens/ gray screens requiring a turn off from my power supply. I suspected the 660 TI was overheating, as the fan doesn't throttle, and the vent gets REALLY hot. I have to reboot to see a temp, and i'm not sure how much it has cooled down by then, but it's usually high 60's. Running FurMark, the card hit's 74 on stress test with no increase in fan speed and plateau's there. Help very appreciated.

Your temps are fine. 74'c is normal, especially during and stress tests. What you might want to check out is your RAM. All of my issues went away after I took out one of my pairs (2x4GB). I had initially ran MemTest (via USB boot) one each of my 4 sticks individually, and paired. All passed. However, I did notice that one pair were posting at 9-11-10, as opposed to stick's rating of 9-11-9. Manually setting the timings did not help. Removing the bad pair fixed everything.

I am not suggesting your timing is off, but I would consider investigating your RAM. Many system problems, from hard drives acting strange, apps quitting, to what seems like GPU issues, can all be traced back to potentially bad RAM. If it wasn't for Windows, by pure chance, giving me a visible BSOD with a memory warning (happened so fast I am amazed I caught it before reboot) I would have never thought to take another look at my RAM.

 

[Midphase]

For the most part my system is running pretty well...however I occasionally get strange freezes where my mouse pointer and the entire screen are stuck and the only way out of it is to press reset and restart. At first I thought it might be due to my overclocking, or the CPU getting too hot, but the temperatures seem to be pretty cool (45C-ish) and I'm starting to suspect that this is a GPU issue.

I'm running an MSI 660ti on a Sabertooth X79 with an 3930k with 10.8.3

These freezes seem to happen when I'm running apps which are leveraging the GPU, like today for instance when I was rendering out of Premiere Pro, or when I am doing some graphically intensive work in After Effects.

As I said, the mouse pointer and the screen freezes on me, but otherwise the computer "sounds" like it's still running and not overheating.

Any advice would be much appreciated!