Webdriver for GTX 1080&1070

[prodigyman]

Just got my 1080 Ti, not seeing any improvements at all even @4k. Also playing games like WoW, FPS did not improve using Metal under MacOS

LuxMark did go up from 17k 980ti to 22k with 1080ti

 

[SoyCapitanSoyCapitan]

Just got my 1080 Ti, not seeing any improvements at all even @4k. Also playing games like WoW, FPS did not improve at all using Metal under MacOS

LuxMark did go up from 17k 980ti to 22k with 1080ti

Pretty bad considering the 980Ti is now cheap and has a much lower clock speed.

 

[Asgorath]

Just got my 1080 Ti, not seeing any improvements at all even @4k. Also playing games like WoW, FPS did not improve using Metal under MacOS

LuxMark did go up from 17k 980ti to 22k with 1080ti

You only posted one set of 4K results, and as discussed at great length, the default 1600x900 resolution is CPU limited in a cMP. What resolution/settings are you using with WoW? It is also a very CPU-limited game, especially in a cMP. How did you measure FPS with WoW exactly?

Edit: Here's a really simple test, grab iStat Menus and check your GPU utilization when running those tests, especially WoW. It should be pegged at 100% for the entire time, or else your CPU can't keep up with the GPU (i.e. CPU limited).
[doublepost=1492105476][/doublepost]

Pretty bad considering the 980Ti is now cheap and has a much lower clock speed.

LuxMark is known to favour the AMD architecture, and if all you care about is LuxMark performance, then go and buy a 980 Ti (or an AMD card) then. We get it, you hate NVIDIA and will talk crap about their drivers at any available opportunity. Maybe you should go and use an AMD Vega card instead. Oh wait, they still haven't released them yet.

 

[prodigyman]

You only posted one set of 4K results, and as discussed at great length, the default 1600x900 resolution is CPU limited in a cMP. What resolution/settings are you using with WoW? It is also a very CPU-limited game, especially in a cMP. How did you measure FPS with WoW exactly?

Edit: Here's a really simple test, grab iStat Menus and check your GPU utilization when running those tests, especially WoW. It should be pegged at 100% for the entire time, or else your CPU can't keep up with the GPU (i.e. CPU limited).

Yeah I use iStat, @4k heaven GPU was @ 100% and CPU broken down to each core + HT were spread out across all 12 cores, primary being @ 30-40%

WoW 4K windowed or full screen GPU was only @ 35-40%.

WoW 980Ti testing I did was to log in while above a city, shutdown install new 1080 Ti and login @ same spot and time very similar FPS, @ 4k esp in new areas its well below 30 FPS.

I only did scaled res I use on my 4k Screen on 980 ti but doing same test on 1080ti shows about 64% score increase. Once I put in 980ti back in I'll run some true 4k tests.

 

[Asgorath]

Yeah I use iStat, @4k heaven GPU was @ 100% and CPU broken down to each core + HT were spread out across all 12 cores, primary being @ 30-40%

WoW 4K windowed or full screen GPU was only @ 35-40%.

WoW 980Ti testing I did was to log in while above a city, shutdown install new 1080 Ti and login @ same spot and time very similar FPS, @ 4k esp in new areas its well below 30 FPS.

Okay, so you answered your own question -- if the GPU is only running at 40%, even at 4K resolution, then you are totally limited by the slow CPUs in your cMP for that game. As such, it's totally expected that you would get the same FPS with a 980 Ti and 1080 Ti. If you care about WoW performance, the best thing you can do is switch to a Hackintosh with a fast CPU, as that might actually be able to keep up with your 1080 Ti card.

 

[prodigyman]

Okay, so you answered your own question -- if the GPU is only running at 40%, even at 4K resolution, then you are totally limited by the slow CPUs in your cMP for that game. As such, it's totally expected that you would get the same FPS with a 980 Ti and 1080 Ti. If you care about WoW performance, the best thing you can do is switch to a Hackintosh with a fast CPU, as that might actually be able to keep up with your 1080 Ti card.

I've edited the post above with more benchmarks. I knew the new card wouldn't increase gaming preformance much, just posting benchmarks for others. I had almost double 4k FPS on cMP in bootcamp.

 

[Asgorath]

I've edited the post above with more benchmarks. I knew the new card wouldn't increase gaming preformance much, just posting benchmarks for others. I had almost double 4k FPS on cMP in bootcamp.

Sure, I was mostly responding to SCSC's usual "zomg the NVIDIA drivers are terrible" posts. It's expected that macOS OpenGL will be significantly slower than Windows (both OpenGL and DirectX), it's dead as far as Apple is concerned and they haven't updated it in years (and their design has always had way more overhead than Windows). My main point here is that there's always context behind benchmark scores, so when people post that they aren't seeing an improvement going from a 980 Ti to a 1080 Ti, the simplest answer is that you are CPU limited in the cMP (not that the GPU is slow or that the NVIDIA drivers are terrible, which is what most people seem to conclude).

 

[SoyCapitanSoyCapitan]

The guy who wants to be Nvidia guru always says I hate Nvidia and almost every GPU I had for the last two decades has been Nvidia - from 1999 to 2017. How many Radeon cards have I ever bought...just 1. I posted this retort 100 times and even have created a thread where we all listed the GPUs we have ever used. He doesn't listen.

It's a bad choice on the Mac. We don't get to see the proper generation on generation improvements. The APIs and drivers don't let it sing properly and the only time it's worth buying is when prices drop 50%.

And for balance, AMD cards also can't sing properly on macOS. Apply the same buying decision: Always go for previous gen or discounted cards if you use a Mac platform.

 

[Asgorath]

The guy who wants to be Nvidia guru always says I hate Nvidia and almost every GPU I had for the last two decades has been Nvidia - from 1999 to 2017. How many Radeon cards have I ever bought...just 1. I posted this retort 100 times and even have created a thread where we all listed the GPUs we have ever used. He doesn't listen.

It's a bad choice on the Mac. We don't get to see the proper generation on generation improvements. The APIs and drivers don't let it sing properly and the only time it's worth buying is when prices drop 50%.

And for balance, AMD cards also can't sing properly on macOS. Apply the same buying decision: Always go for previous gen or discounted cards if you use a Mac platform.

What are the alternatives? A low-end GPU from AMD? Using GPUs from 5 years ago? If that works for you, great. I'm very happy with my Hackintosh and now a 1080 card, upgrading from a Maxwell card before that. I've had zero issues in the applications/games I use. Are the drivers perfect? Of course not, but neither are the drivers supplied by Apple for the Intel and AMD GPUs either. At least NVIDIA is attempting to enable their users to use macOS by providing their own drivers. There are plenty of usage cases where a Pascal card is a clear upgrade over everything else, but your posts make it sound like it's always the wrong choice.

You may claim it's a bad choice, but my point is that it's the least bad choice that we have, as all the other alternatives are worse (IMO). So, you can suck it up and accept that things will never be perfect, or just keep living with your official products from 5+ years ago (or super-low-end GPUs officially provided by Apple as part of their current lineup).

 

[ancelrick]

Then, after a new MacOS release, perform your install via Terminal without restart. Then install Nvidia WebDriver.

Just turn off autoupdates , and let others experiment for you. After the experimentation has happened , you'll know if it will work or not.

Only update the OS AFTER the web drivers have caught up.

 

[Norbert Mikołajczyk]

I've edited the post above with more benchmarks. I knew the new card wouldn't increase gaming preformance much, just posting benchmarks for others. I had almost double 4k FPS on cMP in bootcamp.

Can you please post more comparison benchmarks from games in bootcamp?
I'm wondering if there will be any fps growth under Windows in games.

 

[Prince134]

Tested LuxMark with two Titan X (pascal alone and together with Maxwell). Obviously Titan X Pascal performs better than 1080ti in the territory of compute capability. Be noted this is not Titan Xp. Gosh, so confusing!

Also, APPLE just released its FCP update to 10.3.3 tonight. Both Titan X Pascal as well as Maxwell got 30 seconds in the BruceX benchmark in Sierra 10.12.4 (Before tonight's update, with FCP 10.3.2, the result in Sierra was 47 seconds). This is Apple's quick response to the new Pascal web driver, which is a good sign between Nvidia and Apple, IMO, although the update is for a bunch of other features. But unlike AMD 7970 (aka D700) Nvidia cards are not scalable in FCP, ie two Titan cards are getting exactly the same result as one. Odd that LuxMark and Octane are scaled almost linear.

On another note between El Capitan and Sierra comparison, on my other test with Titan X Maxwell, in El Capitan BruceX was 20 seconds. This has been the case since Sierra OS was introduced. It's a puzzle to me why new Sierra OS get worse result in FCP.

 

[devynbrew]

Tested LuxMark with two Titan X (pascal alone and together with Maxwell). Obviously Titan X Pascal performs better than 1080ti in the territory of compute capability. Be noted this is not Titan Xp. Gosh, so confusing!

Also, APPLE just released its FCP update to 10.3.3 tonight. Both Titan X Pascal as well as Maxwell got 30 seconds in the BruceX benchmark in Sierra 10.12.4 (Before tonight's update, with FCP 10.3.2, the result in Sierra was 47 seconds). This is Apple's quick response to the new Pascal web driver, which is a good sign between Nvidia and Apple, IMO, although the update is for a bunch of other features. But unlike AMD 7970 (aka D700) Nvidia cards are not scalable in FCP, ie two Titan cards are getting exactly the same result as one. Odd that LuxMark and Octane are scaled almost linear.

On another note between El Capitan and Sierra comparison, on my other test with Titan X Maxwell, in El Capitan it was 20 seconds. This has been the case since Sierra was introduced. It's a puzzle to me why new OS get worse result in FCP.

I talked with a guy who's somewhat apart of the NVidia community and he believes because of the recent 2018 Pro announcement that Apple will be working with NVidia a lot more to make their cards widely available on Mac OS into the future!

 

[flowrider]

For all you folks considering GTX 1080 cards of any ilk, I'd advise you to stay current here:

http://www.macvidcards.com/blog

and here:

http://forum.netkas.org/index.php/board,6.0.html

MVC is spending much time developing the EFI drivers and testing out various Pascal cards, as well as posting his findings. Not all are working well (Titan-X Xp). MVC appears to be the only Mac enthusiast with the chops to work through the issues. And he, IMHO, is the only guy that has kept the 5,1 cMP relevant.

Lou

 

[Prince134]

For all you folks considering GTX 1080 cards of any ilk, I'd advise you to stay current here:

http://www.macvidcards.com/blog

and here:

http://forum.netkas.org/index.php/board,6.0.html

MVC is spending much time developing the EFI drivers and testing out various Pascal cards, as well as posting his findings. Not all are working well (Titan-X Xp). MVC appears to be the only Mac enthusiast with the chops to work through the issues. And he, IMHO, is the only guy that has kept the 5,1 cMP relevant.

Lou

I am not sure what he suspected that the PSU could be under too much load for Titan Xp. But I've got two Titan X (pascal + Maxwell) running Octane and LuxMark fine without external power. But the 2 6pin powers for each card from SATA bays though.

 

[xsmi123]

 

[omegafilm]

On another note between El Capitan and Sierra comparison, on my other test with Titan X Maxwell, in El Capitan BruceX was 20 seconds. This has been the case since Sierra OS was introduced. It's a puzzle to me why new Sierra OS get worse result in FCP.

[doublepost=1492346548][/doublepost]Sierra stopped supporting Metal in MacPro5.1
That's why we need Hackin MacPro6.1 or go back to El Capitan as for FCPX user.

List of Mac computers that support Metal
https://support.apple.com/en-us/HT205073

 

[tomvos]

Sierra stopped supporting Metal in MacPro5.1
That's why we need Hackin MacPro6.1 or go back to El Capitan as for FCPX user.

List of Mac computers that support Metal
https://support.apple.com/en-us/HT205073

Funny — on my Mac Pro 5,1 with a 680 GTX and the default Apple drivers for 10.12.4 Metal is supported.

 

[StuAff]

'Apple supported' has long been a rather shorter list than 'works on your Mac'.

 

[Squuiid]

Single slot 1070...
http://www.techspot.com/news/68947-galax-announces-world-first-single-slot-geforce-gtx.html

 

[flowrider]

Sierra stopped supporting Metal in MacPro5.1
That's why we need Hackin MacPro6.1 or go back to El Capitan as for FCPX user.

List of Mac computers that support Metal
https://support.apple.com/en-us/HT205073

Not true at all. It depends on the GPU installed in the cMP 5,1. The stock Radeon 5770, does not support Metal. I don't know if the optional Radeon 5870 does or not. My GTX 780 certainly does as will my upcoming GTX 1080.

Lou

 

[Filin]

Single slot 1070...
http://www.techspot.com/news/68947-galax-announces-world-first-single-slot-geforce-gtx.html

Seems heatsink blowing heat inside case, not outside

 

[owbp]

Seems heatsink blowing heat inside case, not outside

Yep, very similar to 4,1's GT120.

 

[Squuiid]

Seems heatsink blowing heat inside case, not outside

Yes, but the cMP case can easily handle that heat without breaking a sweat! It's a very well designed case as far as cooling is concerned.

 

[namethisfile]

Not true at all. It depends on the GPU installed in the cMP 5,1. The stock Radeon 5770, does not support Metal. I don't know if the optional Radeon 5870 does or not. My GTX 780 certainly does as will my upcoming GTX 1080.

Lou

I have the HD 5770 and it does not support Metal. Since the HD5870 is in the same family as the HD5770, then, I am 99.9% sure it, also, does not support Metal.

I think the cutoff point as Apple states is year 2012. It seems like iMacs and GPU's made that year and afterwards will support Metal. HD7950 and GTX 680 family of GPU's withstanding, which support Metal. So, basically any GPU's from HD7xxx series and upward from AMD and GTX6xx series and upward from Nvidia will support Metal.

My MBP, which has Nvidia GT330m dGPU does not support Metal. But, my Mac Pro with RX 460 does....