Our cards work exactly like an "official" card except for a single cosmetic issue. Every post "official" card has come from us.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Frequently Asked Questions About NVIDIA PC (non-EFI) Graphics Cards
- Thread starter Asgorath
- WikiPost WikiPost
- Start date
- Sort by reaction score
- Status
- Status
Our cards work exactly like an "official" card except for a single cosmetic issue. Every post "official" card has come from us.
I note that the new uber performance monster the GTX 1080 consumes max 180W power
This is great news!!!
I'm doing some saving up for the GTX-1080 MVC edition from their UK reseller!
I've waited very long before taking the plunge by exchanging my ATI 2600 HD & GTX-670 for a single EFI card, but now that this new model is almost around the corner: "Elk nadeel heb zn voordeel!" / "Each dis-advantage has his advantage!" ~ JC
Hope it will arrive soon!
Cheers
These are heavenly cards for us cMP users!!! Let's hope the webdriver will be as good as it can get too!
Cheers
Exciting times for cMP, going to leave the 6,1 even further in the dust.
Just upgraded from a EVGA GTX 760 2GB SC to an ASUS Strix GTX 970 4GB OC on a Mac Pro 3,1. The card has 1x 8-pin connector so the connection was pretty easy (2x6-pin to 1x8-pin adaptor). Interesting thing that the System recognises the card as x16 lane, the GTX 760 was only 8x lane (did anyone ever get a 16x on a non-flashed card?). Gaming in Windows is much smoother now!
Attachments
Last edited:
From the trend, those card most likely will be supported, but the web driver won't be optimised for the new card.
I don't know about driver, but for Nvidia, little modification on the current driver to give the new card some basic support should not be too hard.
No, just my personal guess from the trend. Web driver support 6xx, 7xx, 9xx..., even OSX has zero support to the Maxwell card.
And from other post on this forum, it seems the driver works but nothing really optimised for the Maxwell card (doesn't like on the Windows side, has new function supported). The performance gain from 6xx to 9xx in OSX also not that significant (I mean compare to Windows driver).
In the following screenshot, I see massive improvements from one version of the the Nvidia web drivers over the previous version for the Maxwell based cards. I would call that optimization.
Source: http://barefeats.com/gtx980ti.html
I would not expect built-in support for Maxwell based cards in OS X since Apple never shipped any computers with them. Sometimes lack of improvements from the 600-series cards to 900-series cards can be attributed to weakness of OpenGL which Nvidia doesn't have any control over. However, based on the above screenshot, I believe Nvidia does what they can on their drivers.
Source: http://barefeats.com/gtx980ti.html
I would not expect built-in support for Maxwell based cards in OS X since Apple never shipped any computers with them. Sometimes lack of improvements from the 600-series cards to 900-series cards can be attributed to weakness of OpenGL which Nvidia doesn't have any control over. However, based on the above screenshot, I believe Nvidia does what they can on their drivers.
I think i'd said "compare to windows". Anyway, I am not convinced. To me, that review is a confirmation that no optimisation for Maxwell card, isn't it? The improvement is significant between 346.02 and 346.01, but not between 980 and 680 (except Ocean OpenCL). The 680 even perform better then 980 Ti in Dirt 3. IMO, obviously no optimisation in OpenGL for Maxwell card. I am not even a beginner in API, but I think OpenGL is not any low level stuff, driver should play an important role. If not driver's fault about low performance in the same API (but much better hardware), then it's the OS's fault?
However, I agree that I should correct myself. Not "nothing optimised for Maxwell card", the Ocean OpenCL test is a prove that Nvidia improving the OpenCL support. However, on the Windows side, 980Ti can perform 3x better than 680 in compute task. In OSX, that screenshots shows that the improvement is not even reach 2x. So, again, IMO, it's "not good enough" or "not well optimised" (but they do "some" optimisation).
You can disagree my personal comment, but I think we can't disagree the facts that Maxwell card perform much better in Windows, in both OpenGL and computation. If OSX really limit the GPU performance to only 60-70% of the same card can do in Windows, then I am wrong, Nvidia do their best already. However, I really doubt if the OS itself make such a big difference.
Do you understand what it means for a game to be CPU limited? Do you realize that those benchmarks are using a 6-year old computer (MacPro5,1 from 2010) with PCIe Gen2? The main reason the Maxwell GPUs don't shine more in these games is the simple fact that they're all completely CPU limited and the GPU is no longer a bottleneck. This is easily verified via tools like the OpenGL Driver Monitor, which reports GPU utilization percentages.
The OS (or really the Apple driver model) really does make that big of a difference. There's a large software component controlled by Apple that the NVIDIA (and AMD/Intel) driver sits beneath. This does not exist on Windows, where NVIDIA (and AMD/Intel) get to implement the API directly. Sooner or later, this large software component starts getting in the way and affecting performance.
Hopefully all of this will improve with Metal, which is a much lighter-weight API with less of a CPU overhead problem.
There's a reason why gaming on Windows is preferable over OS X. It's because DirectX is far, far better than OpenGL. It doesn't help that OS X is still on OpenGL version 4.1 which was released back in 2010. The most current version of OpenGL is 4.5. Things will probably get worse with the adoption of Vulkan.
I don't know how much better a GTX 980 Ti is over a GTX 680 in compute. I'm too lazy to dig up my old GTX 680 and swap video cards, but here are my LuxMark results in Windows and OS X with my GTX 980:
They look pretty close to me... If this is without any optimization for Maxwell cards, I'm pretty impressed.
Further evidence of optimizations through versions can be seen in the following post:
https://forums.macrumors.com/threads/fcpx-amd-vs-nvidia.1956128/page-2#post-22657893
12884 vs 12036 hardly seems like 60-70%...
I'm not debating whether it's better to run these video cards in Windows or OS X. I just disagree with your assertion that Nvidia doesn't optimize their drivers for new cards.
Last edited:
I understand what is CPU limiting. However, not applicable in the case.
Please read again post 3096. If those games benchmark are CPU limiting. No matter how Nvidia work on the driver, the gaming performance CANNOT be improved. HOWEVER, they do. 356.02 obviously works much better then 346.01. That means the games were not CPU limiting from the very beginning.
If Dirt 3 is CPU limiting, 980 should perform same as the 680, but not 10% slower.
We can always blame the old CPU in the cMP, however, does not fit in this case.
On the other hand, after 356.02 released, I cannot deny that the gaming performance may hit the point that's now CPU limiting. But, 980 perform ~10% worse than 680 still the fact, which should not happen if the poor performance purely due to CPU limiting.
[doublepost=1462916589][/doublepost]
Why suddenly talk about DirectX? I NEVER compare OpenGL to DirectX. I was talking about OpenGL (Windows) vs OpenGL (OSX). However, I made a mistake that I have absolute no idea about the version of OpenGL in Windows. If Windows' OpenGL version is 4.5, is if possible to test the 4.1 performance and compare it to the OSX OpenGL 4.1 result? Or when it go to 4.5, the 4.1 part will also be improved?
Besides, I must admit that the number in my mind is just from the user benchmark.com, not from my personal test. So, the data may not accurate at all. What I see is that 980 Ti is about 2x better than 680 in the Nbody calculation test. That makes me believe that the 980Ti can do much better than 680 in compute. And I can't see that difference in OSX yet.
And one more issue that catch my attention is the "validation failed" message in OSX Luxmark. Which does not occur on the Windows side. Will that make the result not accurate? Anyone know what is it actually mean? Is that driver related?
Anyway, I am totally OK that you disagree my opinion (the web driver is not optimised). I am OK to accept you opinion. That's just an opinion. For me, 80% optimise may means bad, and 80% for you may means good. So, let's put that down and move on. I have no intention to persuade you that the driver is not optimised. And I don't want to see that the driver is not optimised. Let's find out more facts. I am more than happy to see that we can eventually prove that Maxwell card do as good as (or even better) in OSX then Windows for exactly the same job. Of course, I don't want to spend all your time to prove something to me. So, may be we can slowly gathering more info. Or do more test when you have time.
Last but not least, thank you for correcting my mistake. I jump to conclusion too early, and miss quite a few important factors. However, I am really not that stupid to compare DirectX to OpenGL, you get me wrong on this
Last edited:
- Status