Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
A while ago, I hoped to upgrade to a flashed 1GB Radeon but the bandwidth cost was noticeable while working on the 2.5 GT/s speed.
So I am puzzled by this. I thought (mistakenly?) that the lower link speed was associated with injected cards but not flashed ones. Or is there a limited aspect to the Ati flash routines you have that only gets up to 2.5 (or did you deliberately turn it down somehow from 5 to see the effect)?
 
I didn't deliberately do anything - I think it's just a failing of flashed/injected cards to be properly recognized as native but they still load the driver.
 
I didn't deliberately do anything - I think it's just a failing of flashed/injected cards to be properly recognized as native but they still load the driver.

My guess is that there is an itty, bitty resistor that is in a different place on OEM Apple cards compared to all the rest. Apple kept many cards unflashable in the past this way. (Ti4600, GF3 Ti200 & 500, 3870, etc)

Since there are likely to be several resistors like this, finding the right one would be tricky. As Beige is the only person so far to quantify the difference, it is doubtful we will ever figure it out.
 
10THzMac- Thanks for the info! Sadly the 5.0 vs 2.5 speed is part of my issue being on a Mac Pro 2,1 I'm limited to the 2.5, with PCIe 1.0 slots. Being that I will have to use the the 2 video cards in 8x slots I'm curious what the Cuda processing on the injected card would be? If you could use the Expansion Utility located at /System/Library/CoreServices/ to switch the slots your video cards are in to 8X and then run those same benchmarks that would be great help!? Thanks.
 
missed this post.

point me to whichever CUDA benchmark you are familiar with and I will give it a shot

If you could run your cards in 8X and run both the XBench graphics benchmarks and the CUDA operations that 10thMhz did above that would be awesome help! Maybe even just the nbody simulation.
 
If you could use the Expansion Utility located at /System/Library/CoreServices/ to switch the slots your video cards are in to 8X and then run those same benchmarks that would be great help!? Thanks.
Will do - but I am away from that rig now till the weekend so give me a few days. Thanks to Rominator for the explanation that there is yet another complication here. The GTX 480 has now showed up so I will also try that at the weekend, though I am just aiming for bootcamp for now. It needs both a 6 and 8 pin connector (470s seems to be 2x6 like the 285) so I am going to have to fiddle with the power a bit as well.
 
Will do - but I am away from that rig now till the weekend so give me a few days. Thanks to Rominator for the explanation that there is yet another complication here. The GTX 480 has now showed up so I will also try that at the weekend, though I am just aiming for bootcamp for now. It needs both a 6 and 8 pin connector (470s seems to be 2x6 like the 285) so I am going to have to fiddle with the power a bit as well.

Thanks 10THzMac. The info will be super helpful. I hope Rominator can run some of the same tests as well...he has the same system as I. Thanks again.
 
Cinebench OpenGl Mystery

OK, I downloaded latest Cinebench.

Talk about confusing results !!!

With the GTX285 as primary card, I got 22.2 FPS. When I moved DVI connnector to 8800GT in a 4x slot, I got 22.2 FPS.(And Cinebench still reported GTX285 as the graphics card)

When I rebooted with JUST the 8800GT in, Cinebench reported 8800GT as the card, and STILL got a 22.2 FPS for the OpenGl test. Something isn't right.

Also, I have no idea where this "MonteCarlo" test is or how to run it.

Hit me with a link and I will do it.

Meanwhile, anybody got an idea why the 8800GT in a 4x lane scores IDENTICALLY to a GTX285 in a 16x lane in CInebench?
 
OK, I downloaded latest Cinebench.

Talk about confusing results !!!

With the GTX285 as primary card, I got 22.2 FPS. When I moved DVI connnector to 8800GT in a 4x slot, I got 22.2 FPS.(And Cinebench still reported GTX285 as the graphics card)

When I rebooted with JUST the 8800GT in, Cinebench reported 8800GT as the card, and STILL got a 22.2 FPS for the OpenGl test. Something isn't right.

Also, I have no idea where this "MonteCarlo" test is or how to run it.

Hit me with a link and I will do it.

Meanwhile, anybody got an idea why the 8800GT in a 4x lane scores IDENTICALLY to a GTX285 in a 16x lane in CInebench?

That is some interesting results. Yeah I'm not sure about the MonteCarlo stuff either...just wanted some CUDA numbers for comparison. Could you also run the XBench benchmark with just the video card based tests. Same thing both cards normally and both cards in an 8X slot. Thanks.

http://www.xbench.com/
 
OK, I downloaded latest Cinebench.

Talk about confusing results !!!

With the GTX285 as primary card, I got 22.2 FPS. When I moved DVI connnector to 8800GT in a 4x slot, I got 22.2 FPS.(And Cinebench still reported GTX285 as the graphics card)

When I rebooted with JUST the 8800GT in, Cinebench reported 8800GT as the card, and STILL got a 22.2 FPS for the OpenGl test. Something isn't right.

Also, I have no idea where this "MonteCarlo" test is or how to run it.

Hit me with a link and I will do it.

Meanwhile, anybody got an idea why the 8800GT in a 4x lane scores IDENTICALLY to a GTX285 in a 16x lane in CInebench?

24.46 here with a native Mac GTX285 - Cinebench favours ATI cards and Nvidia's Mac drivers aren't very good.
 
Sorry if the code I was citing is a bit obscure. It's one of the examples in the CUDA 3.0 toolkit that you can download at

http://developer.nvidia.com/object/cuda_3_0_downloads.html

You need a driver, toolkit and SDK.

Ran my 480 in a regular PC yesterday and it was only turning out 118000 on the MonteCarloMultiGPU on a first run alone. With 480 cores I would have expected more like 180000+. But it might be the old story of CUDA not necessarily firing the card up to top speed, which is what we had with the Mac 285 at first. On the nbody sim it was same speed as the 285. Hmmm. Will do 8x PCI as soon as I can on my home rig.
 
Sorry if the code I was citing is a bit obscure. It's one of the examples in the CUDA 3.0 toolkit that you can download at

http://developer.nvidia.com/object/cuda_3_0_downloads.html

You need a driver, toolkit and SDK.

Ran my 480 in a regular PC yesterday and it was only turning out 118000 on the MonteCarloMultiGPU on a first run alone. With 480 cores I would have expected more like 180000+. But it might be the old story of CUDA not necessarily firing the card up to top speed, which is what we had with the Mac 285 at first. On the nbody sim it was same speed as the 285. Hmmm. Will do 8x PCI as soon as I can on my home rig.

Thx 10ThzMac some benchmarks of your GTX285's in an 8x mode would be really helpful. Especially if you could get the bandwidth at 2.5...that way it would be a closer scenario to how I would be running. Thanks.
 
hi, Rominator

Which rom you have Flashed with your GTX 285? The EVGA's 156k rom ?
MAC can recognize the card? Need install driver or not?:confused:
 
need fixed driver....

Dear all.
I am trying to get my GTX 285 to run CUDA on my hackintosh.
The fixed driver is no longer available from the rapidshare link.
Could someone please post or pm the fixed driver, can't wait to try CUDA!

NVIDIA_Retail_Mac_Driver_Installer_18.5.2f16.mpkg.zip.html
 
Dear all.
I am trying to get my GTX 285 to run CUDA on my hackintosh.
The fixed driver is no longer available from the rapidshare link.
Could someone please post or pm the fixed driver, can't wait to try CUDA!

NVIDIA_Retail_Mac_Driver_Installer_18.5.2f16.mpkg.zip.html

driver is for 10.5.7/8 only

driver exists in OS since then

you need to go to a HACKINTOSH forum

i feel so tired
 
Thanks 10THzMac. The info will be super helpful. I hope Rominator can run some of the same tests as well...he has the same system as I. Thanks again.

I could not get the utility you mentioned in Core services up on my 08 machine - it came up with an annoying dialog saying it was not designed to work on this set up and I had to shut it down.

I have posted my 480 results to the 480 thread - bootcamp is fine with it.
 
OK, I downloaded latest Cinebench.

Talk about confusing results !!!

With the GTX285 as primary card, I got 22.2 FPS. When I moved DVI connnector to 8800GT in a 4x slot, I got 22.2 FPS.(And Cinebench still reported GTX285 as the graphics card)

When I rebooted with JUST the 8800GT in, Cinebench reported 8800GT as the card, and STILL got a 22.2 FPS for the OpenGl test. Something isn't right.

Also, I have no idea where this "MonteCarlo" test is or how to run it.

Hit me with a link and I will do it.

Meanwhile, anybody got an idea why the 8800GT in a 4x lane scores IDENTICALLY to a GTX285 in a 16x lane in CInebench?

Hey Rominator,
So my ATI Radeon X1900 XT is on it's way out and I have to boot multiple times before I can get any of my monitors working. This is obviously accelerating my need to replace it and figure out the future DaVinci Resolve issue. I like the idea of spending $600 for a couple video cards instead of $6500 for a new machine. So I am going to try the injected approach very soon. I wanted to ask specifically what GTX285 you were running? I'm assuming the 1GB version from EVGA? If the 2GB can work in the Mac Pro 2,1 I would love to get that one instead?
I had a crazy idea this weekend that could solve my concerns. If this setup with both cards running in 8x mode for working in Resolve works...then it would be sweet to setup a switcher that is fed by both video cards. That way when running DaVinci Resolve I could switch over to using the 8800GT to run the monitors and then in the Expansion Utility switch the cards into the 8x, 8x mode (to accomodate the Raid and Kona card that need to run in the other slots at 4x). Then when I'm done working in DaVinci flip the switcher back to running the monitors off the GTX card and use the Expansion Utility to switch the GTX into 16X mode and the 8800GT into 1X mode (that way the other 2 slots stay in 4x mode) and I have a power house for After Effects and 3D work. Any thoughts on this idea? Can the whole setup work with the 8800GT running in 1X and the GTX in 16X with the monitors hooked up to the GTX? Thanks again for all the great info.

J
 
you need to read through this thread again

1x ain't gonna do it

and for now, at least, 2 Gig GTX won't either

search for "2 gig bug"

research rules

Yeah I already did the research and saw the 2GB bug but was curious if anything had updated about that yet. And the 1x thing was just a random thought...the 1x slot being the 8800GT that I wouldn't be using when utilizing the GTX285. I know it needs a minimum of 4x but I was curious if it would indeed boot anyhow with the GTX at 16X and the 8800GT at 1X? Has anyone actually tried it?
 
Hi,

I've come across an XFX and/or BFG GTX 275 for sale quite cheap, would this work or would i need an EVGA version?

Thanks!
 
The 275 seems to have some problems - find the posts by peskaa earlier - even when properly injected following a boot with an Nvidia card like the 120. A good read of this thread would help.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.