Flashed GTX 780 - only 40-60fps????

[Xylez]

Hi All,

I recently purchased a GTX 780 with a flashed EFI from ebay, Exact one here.

The only reason I did this one, is because their is a $200+ price difference between MacVidCards and this one, why is the difference so huge? With that being said, I think maybe I may now know why.

I received a Zotac GTX 780, installed the cards, installed the drivers everything is working and detected just fine. Now, when playing dota2, wow, insurgency and many other games I was expecting FPS over 100 with settings on "Fastest" or "lowest", not the case.

My fps is anywhere from 40-80 fps depending on whats going on in the game, on the lowest settings possible.

My Radeon 5770 seemed to almost be running better, is this something I should return, bite the bullet, purchase the GTX 970 from Mac Vid Cards and be on my way? (Would not repurchase a 780 for $200 more, would just buy the GTX 970)

Just not sure what the issue is. This gaming is done in Mac OSX, haven't tested windows yet.

 

[dpny]

You're likely CPU-limited.

 

[Xylez]

You're likely CPU-limited.

I was thinking the same thing, I have a 2010 Quad-core Nehalem processor right now, If thats what you mean by CPU-Limited.

There are some nice single six core processor that I am eye-balling.

 

[dpny]

Any cMP is going to be CPU-limited when gaming. It's what happens when you have a five or six year old CPU.

 

[Xylez]

Any cMP is going to be CPU-limited when gaming. It's what happens when you have a five or six year old CPU.

Thanks, Will look to upgrade that now

 

[owbp]

Totally agree on cMP and CPU bottleneck, but @Xylez is talking about Core2Duo games that run great on much worst cards that 780 is. Im hitting similar FPS with cMP1,1 and HD6850 in maxed out Skyrim (without any mods though).

You could do some GPU only benchmarks and compare results done with the same card if you still have doubts about card you bought?

 

[Xylez]

Totally agree on cMP and CPU bottleneck, but @Xylez is talking about Core2Duo games that run great on much worst cards that 780 is. Im hitting similar FPS with cMP1,1 and HD6850 in maxed out Skyrim (without any mods though).

You could do some GPU only benchmarks and compare results done with the same card if you still have doubts about card you bought?

Thank you for the response, yes the FPS rate seems to have stayed the same, maybe only a 10-20 bump if that. I understand the CPU-Limiter, which I am going to take care of shortly, but such minimal increase? I mean we are talking about Dota2 here which doesn't take many resources. I guess I can load up fallout 4 and see what my FPS looks like there

I have been searching around online, but there are so many different programs, do you have one you recommend? I can start doing that now.

Appreciate the help

 

[owbp]

No prob, I'm not expert on this, so hope that others will correct me if I'm wrong, but you shouldn't see any bottleneck on games you mentioned. Only on modern, CPU intensive games, when GPU starts working very hard, you could see it stuck at some point (not being used in full) and thats the sign that bottleneck is happening.

Fallout 4 isn't gonna be a good test because it still has huge frame drops even on maxed out PCs but it's fun to play
Mine, absolutely outdated, config is going from 60 (indoors) to 1 (stuttering in cities during the day) FPS (1920x1080, everything on low).

I know for sure that Lexmark (but its OpenCL only) can do only GPU bench and it is widely used, so you could compare results. v3 is the newest, but i think that majority of benchmarks are done on v2 "Sala scene" so see what you can find.

 

[Machines]

I was thinking the same thing, I have a 2010 Quad-core Nehalem processor right now, If thats what you mean by CPU-Limited.

There are some nice single six core processor that I am eye-balling.

Your 2010 cMP has a single Quad 2.8 . Try upgrading to a X5677 Quad 3.46 GHz and install 3 x 4GB 1333 MHz ECC memory modules in slots 1 ,2 and 3 . The chip costs around 75 bucks and if you see no improvement then getting its 6 core bigger brother won't likely help . A key advantage is to install memory in sets of three .

 

[Xylez]

Your 2010 cMP has a single Quad 2.8 . Try upgrading to a X5677 Quad 3.46 GHz and install 3 x 4GB 1333 MHz ECC memory modules in slots 1 ,2 and 3 . The chip costs around 75 bucks and if you see no improvement then getting its 6 core bigger brother won't likely help . A key advantage is to install memory in sets of three .

Theoretically speaking it should be a huge improvement with 1333 MHZ ecc memory and a CPU upgrade, no? With the new GPU i should be around 60fps on Ultra in some games.

Right now I'm 40-80 fps instead of 30-50.

 

[Fl0r!an]

I'd first check with some common benchmarks (e.g. Unigine Heaven, Luxmark, Octane) that your GPU is working as expected and you didn't receive a badly flashed one.

Btw, a good GPU usually won't speed up your games when running in low graphics settings. In this case, you're usually CPU limited. A powerful GPU comes into play at high resolutions.

 

[h9826790]

Theoretically speaking it should be a huge improvement with 1333 MHZ ecc memory and a CPU upgrade, no? With the new GPU i should be around 60fps on Ultra in some games.

Right now I'm 40-80 fps instead of 30-50.

1333 memory, around 2% improvement, so not significant at all.

2.8 upgrade to 3.46 CPU, it's about 23%.

So, by numbers, overall will be about 25% improvement, if you are CPU limiting both before and after the CPU / memory upgrade. Otherwise, the improvement should be less than that.

 

[thornslack]

I game all the time with my cMP and CPU limiting is something I almost never encounter, although certain titles are reliant on single thread performance. Most modern games are rarely capable of saturating these older processors with only a single GPU present. Faster memory will provide almost no tangible difference in performance so long as you have enough memory to run things. I would strongly recommend installing windows via bootcamp and gaming there. Performance is wildly better than under OSX. If you're still disappointed you can always go the hardware upgrade route.

 

[flowrider]

To the OP:

Run some GPU Tests so we can really see what's going on. Something like Valley, Heaven or LuxMark.

Lou

 

[Machines]

1333 memory, around 2% improvement, so not significant at all.

2.8 upgrade to 3.46 CPU, it's about 23%.

So, by numbers, overall will be about 25% improvement, if you are CPU limiting both before and after the CPU / memory upgrade. Otherwise, the improvement should be less than that.

Do you know there is an upgrade in the Processor Interconnect Speed of the factory Quad Nehalem Mac Pro (4.8 GT/s), when the processor is upgraded to a X5677 (6.4 GT/s) ? The performance gain at the System level is not just arithmetic between the two processor frequencies . It's greater . We don't exactly use buses with the Nehalem Mac Pros - we use interconnects .

 

[Xylez]

To the OP:

Run some GPU Tests so we can really see what's going on. Something like Valley, Heaven or LuxMark.

Lou

I am trying to use Luxmark, but as I am new with this, I am unsure what mode/scene to run. Recommendations so I can get this posted here?

Thanks everyone for responses.
[doublepost=1457894527][/doublepost]

I game all the time with my cMP and CPU limiting is something I almost never encounter, although certain titles are reliant on single thread performance. Most modern games are rarely capable of saturating these older processors with only a single GPU present. Faster memory will provide almost no tangible difference in performance so long as you have enough memory to run things. I would strongly recommend installing windows via bootcamp and gaming there. Performance is wildly better than under OSX. If you're still disappointed you can always go the hardware upgrade route.

I have bootcamp for many games, I am going to boot there now, and see what kind of difference this is.
[doublepost=1457894623][/doublepost]

Your 2010 cMP has a single Quad 2.8 . Try upgrading to a X5677 Quad 3.46 GHz and install 3 x 4GB 1333 MHz ECC memory modules in slots 1 ,2 and 3 . The chip costs around 75 bucks and if you see no improvement then getting its 6 core bigger brother won't likely help . A key advantage is to install memory in sets of three .

Is it difficult to Change the chipset? I have been upgrading PC's/repairing for a very long time, just stepped in the Mac Relm about a year ago. I mean I will obviously watch some video's before I think about preforming an upgrade but, just curious
[doublepost=1457894996][/doublepost]

Do you know there is an upgrade in the Processor Interconnect Speed of the factory Quad Nehalem Mac Pro (4.8 GT/s), when the processor is upgraded to a X5677 (6.4 GT/s) ? The performance gain at the System level is not just arithmetic between the two processor frequencies . It's greater . We don't exactly use buses with the Nehalem Mac Pros - we use interconnects .

View attachment 620950

Did you upgrade your 2.8 to the 3.46? Was it a significant, noticible difference?

I have an option here, $375 for the Six-core if i do a tray-exchange or I install the x5677 myself for $75. And save myself $300, I am really only gaming here, nothing else on this machine.

 

[h9826790]

I am trying to use Luxmark, but as I am new with this, I am unsure what mode/scene to run. Recommendations so I can get this posted here?

Thanks everyone for responses.
[doublepost=1457894527][/doublepost]

I have bootcamp for many games, I am going to boot there now, and see what kind of difference this is.
[doublepost=1457894623][/doublepost]

Is it difficult to Change the chipset? I have been upgrading PC's/repairing for a very long time, just stepped in the Mac Relm about a year ago. I mean I will obviously watch some video's before I think about preforming an upgrade but, just curious

Use Luxmark 3, no need to choose anything, just open it and it will automatically run a 2min GPU benchmark with the default setting. Then you will see something like this.

You can simply capture the screen and post the result at here.

 

[Xylez]

Use Luxmark 3, no need to choose anything, just open it and it will automatically run a 2min GPU benchmark with the default setting. Then you will see something like this.
View attachment 620962
You can simply capture the screen and post the result at here.

Thank you! Here is my benchmark;

 

[Machines]

Is it difficult to Change the chipset? I have been upgrading PC's/repairing for a very long time, just stepped in the Mac Relm about a year ago. I mean I will obviously watch some video's before I think about preforming an upgrade but, just curious .

Not the chipset . The Xeon processor . Basically , it is easy to upgrade a single processor CPU Tray . But you also should re-thermal paste the Northbridge controller chip heatsink . That would be considered very difficult and tricky . I'm going to write it up some day when I have some spare time . But, it really needs to be done with these older Macs or they'll die eventually from lack of maintenance .

Did you upgrade your 2.8 to the 3.46? Was it a significant, noticible difference?

I have an option here, $375 for the Six-core if i do a tray-exchange or I install the x5677 myself for $75. And save myself $300, I am really only gaming here, nothing else on this machine.

Read my job description in my profile . And , I think games rarely use more than 4 processor cores .

 

[h9826790]

Thank you! Here is my benchmark;

So now the others can help.

For OpenGL (more related to gaming in OSX). Unigine Valley Extreme Preset is a good choice. once you choose the Extreme preset, all other options will be locked. And this test run in fullscreen, so should be more accurate for others to compare.

 

[Demigod Mac]

Run Valley, Extreme HD preset.
http://unigine.com/en/products/benchmarks/valley

 

[Xylez]

So now the others can help.

For OpenGL (more related to gaming in OSX). Unigine Valley Extreme Preset is a good choice. once you choose the Extreme preset, all other options will be locked. And this test run in fullscreen, so should be more accurate for others to compare.

I am downloading that now, thanks, here is a new benchmark ran without anything in the background. Didn't realize I had some games and such open, looks like it did better.

 

[h9826790]

Oh yes, I want to say Extreme HD preset, not extreme preset.

 

[Xylez]

So now the others can help.

For OpenGL (more related to gaming in OSX). Unigine Valley Extreme Preset is a good choice. once you choose the Extreme preset, all other options will be locked. And this test run in fullscreen, so should be more accurate for others to compare.

Run Valley, Extreme HD preset.
http://unigine.com/en/products/benchmarks/valley

Here are 2 benchmarks, one on medium, one on ultra
[doublepost=1457900069][/doublepost]

Oh yes, I want to say Extreme HD preset, not extreme preset.

Here is the Extreme HD Benchmark!

 

[thornslack]

That might be about right. Here are results from my system using a flashed GTX 780 classified and unflashed GTX 980 and Titan X. It's a dual processor 4,1 upgraded with x5677s, 24 GB 1333. In OS X the NVIDIA drivers haven't matured since Kepler, which is why a 980 scores worse than my upclocked 780. I included a generic 780 score for reference as well.