Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

iMac 5K GPU Performance

robertojorge

robertojorge

macrumors regular
Original poster
May 6, 2014
129
25
Portugal
Hey guys, so i know its still very early to be able to make real comparisons on the GPUs r9 m290x and r9 m295x but the 1st unboxing video shows a benchmark NOVABENCH witch im not sure how accurate it is as i dont use it.

Usually i use geekbench, cinebench and passmark to make my calculations.

Anyhow if we take a look at the GPU score of the new Imac r9 m290x
ngvt79.jpg


And compare it to the one i just did on my system
2czobdg.jpg


We will be able to see quite a diference there on the GPU, just disregard everything else.

Now according to passmark:

GTX 760 - 4984
GTX 880m - 4694
GTX 780m - 4340

So if the m290x its close to the 780m shouldn't the score be a lot higher?

I wish im wrong or that this benchmark is not very accurate, but it seems that the m290x has a values way too low.

Could any of you guys that have the Late 2013 with 780m do this benchmark so we could compare it to the m290x ?

Or can anyone tell if this bechmark is any good at all comparing performance of systems?

In my case im really on the defensive just because of the GPU and probably others too so i think that these comparisons may help at least to have an idea of what performance the gpu has.
 
Its interesting. I just ordered the one with m295x, I hope it will perform well.
Im not sure if those tests were done in 5k resolution using a 3d benchmark test. In that case Im not suprised at all. Nobody can exepct to run games in this resolution. But if they were benchmarking with the same res as you did, then its worrysome.
 
iMcLovin said:
Its interesting. I just ordered the one with m295x, I hope it will perform well.
Im not sure if those tests were done in 5k resolution using a 3d benchmark test. In that case Im not suprised at all. Nobody can exepct to run games in this resolution. But if they were benchmarking with the same res as you did, then its worrysome.
Click to expand...

That a good though, perhaps the 5k res is pushing the scores down.

I dont want games either at any res as i dont play at all :) All i do is AE and M5 so rendering times in M5 are the reason i want to upgrade.

You said you ordered yours. Once you get it are you willing to render 2 of my templates so that i could be sure of its performance?
 
At 5k no card will compete, even the twin d700's in the Mac Pro when dell start selling their 5k monitor or Apple finally make a tb2 for all the Pro users who are rightfully peeved they are left with an expensive black can and no 5k, crazy.

Halving the resolution and benching will be a far more accurate guide
 
robertojorge said:
That a good though, perhaps the 5k res is pushing the scores down.

I dont want games either at any res as i dont play at all :) All i do is AE and M5 so rendering times in M5 are the reason i want to upgrade.

You said you ordered yours. Once you get it are you willing to render 2 of my templates so that i could be sure of its performance?
Click to expand...

sure, ...templates in m5 or AE... what is M5 ? I can render in AE if you like. though rendering speed will have no negative effect from the high res of the screen, its only graphic intensive programs and games using the gpu a lot. And AE is hardly using the gpu except for programs such as Element 3D, so it will be still be the fastest iMac to work on in AE, Photoshop and such.
 
iMcLovin said:
sure, ...templates in m5 or AE... what is M5 ? I can render in AE if you like. though rendering speed will have no negative effect from the high res of the screen, its only graphic intensive programs and games using the gpu a lot. And AE is hardly using the gpu except for programs such as Element 3D, so it will be still be the fastest iMac to work on in AE, Photoshop and such.
Click to expand...

Im not really worried about AE because if it was only that i wouldnt have tried the "Mac Ecosystem" 2 years ago. Although as you correctly said E3D and Raytracing really improves with cuda, the rest is all CPU + Ram

Now Apple Motion 5 is a diferent story. I sold my mac mini just because the more complex the projects are becoming the mini just cant handle rendering in a effect time window. Thats my only reason to upgrade to a better GPU.

The Mac Pro being an overkill i had decided going for the Imac full spec expecting the release of the 980m witch we know more or less the performance. Now the r9 295x i cant find anything that really helps to decide.

Heres a thread that i started in june and could gather some results and might help understand my concern in the GPU.

EDIT heres the thread https://forums.macrumors.com/threads/1743631/
 
robertojorge said:
I wish im wrong or that this benchmark is not very accurate, but it seems that the m290x has a values way too low.

Could any of you guys that have the Late 2013 with 780m do this benchmark so we could compare it to the m290x ?
Click to expand...

Interesting. I have a mid-2012 rMBP with 650M GPU. Using gfxCardStatus I forced it into using the discrete GPU and changed resolution using QuickRes.

My results using NovaBench:-

Resolution / Score / 3d frames per second

2880*1800 / 147 / 441
1440 * 900 / 148 / 443

So it appears that resolution does not impact on this benchmark (unless the rMBP always renders at high resolution and then scales), or I'm bottlenecked somewhere. That implies that the M290X is indeed crap, only slight better than my 650M. Or the resolution is indeed a factor and and the M290X is just having a hard time pushing all those pixels, because comparing the raw stats the M290X should be kicking the 650M's backside.

http://gpuboss.com/gpus/Radeon-R9-M290X-vs-GeForce-GT-650M-Mac
 
Last edited:
steve23094 said:
Interesting. I have a mid-2012 rMBP with 650M GPU. Using gfxCardStatus I forced it into using the discrete GPU and changed resolution using QuickRes.

My results using NovaBench:-

Resolution / Score / 3d frames per second

2880*1800 / 147 / 441
1440 * 900 / 148 / 443

So it appears that resolution does not impact on this benchmark (unless the rMBP always renders at high resolution and then scales), or I'm bottlenecked somewhere. That implies that the M290X is indeed crap, only slight better than my 650M. Or the resolution is indeed a factor and and the M290X is just having a hard time pushing all those pixels, because comparing the raw stats the M290X should be kicking the 650M's backside.

http://gpuboss.com/gpus/Radeon-R9-M290X-vs-GeForce-GT-650M-Mac

You don't say what system you benchmarked on.
Click to expand...

Really scratching my head how this is possible. Yep it should be way superior and lets just hope that it is but somehow the benchmark is not giving accurate results because of resolution.

My system you can check in the 2nd picture uploaded. I7 4770K, Gtx 760, 32Gb Ram Win 8. I run triple monitors so my GTX is pushing 5760x1080
 
robertojorge said:
Really scratching my head how this is possible. Yep it should be way superior and lets just hope that it is but somehow the benchmark is not giving accurate results because of resolution.

My system you can check in the 2nd picture uploaded. I7 4770K, Gtx 760, 32Gb Ram Win 8. I run triple monitors so my GTX is pushing 5760x1080
Click to expand...

I can't wait to see some proper benchmarking done on this thing. It's so frustrating that all these unboxings on Youtube have a unit in hand but don't do any proper analysis.

I also want to know all about the M295X, but that will take a bit longer because it's a BTO.
 
So it seems the graphic inside retina iMac is bad for pushing 5K pixels. The fact that it's not much different from GTX 775M/780M from last year's high end is putting me off.

Might have to wait 1 or 2 years to see what Apple could offer to conquer 5K.
 
I would be VERY careful about reading into any single benchmark at this point... I'm sure the good people at Anandtech are putting the machine through it's paces right now.

According to what's already known, the M290X should be just below the 780m in speed, but above the 775m from the last bunch of iMacs.

http://www.notebookcheck.net/AMD-Radeon-R9-M290X.108643.0.html

The M295X should easily beat the 780m.

Also remember the AMD chips were probably chosen for compute performance, not for the best frame rate in games.
 
Last edited:
JordanNZ said:
I would be VERY careful about reading into any single benchmark at this point... I'm sure the good people at Anandtech are putting the machine through it's paces right now.

According to what's already known, the M290X should be just below the 780m in speed, but above the 775m from the last bunch of iMacs.

http://www.notebookcheck.net/AMD-Radeon-R9-M290X.108643.0.html

The M295X should easily beat the 780m.

Also remember the AMD chips were probably chosen for compute performance, not for the best frame rate in games.
Click to expand...

Maybe M290x and M295x will make a good upgrade over current 775M/780M if only it's just a usual, annual updates. But you know it's not just the usual. It also comes with 5K display and that changes everything.

It was like iPad 3 all over again, a slightly overclocked A5 rebranded as A5x chip to feed the new retina display with 4 times more pixels. It will struggle to do that and it will be a beta testing product before the real deal comes.

So again I would skip this retina iMac and wait for 2 or 3 years to see if Apple could actually use 5K to do intensive tasks. Not just to look pretty.
 
iSayuSay said:
Maybe M290x and M295x will make a good upgrade over current 775M/780M if only it's just a usual, annual updates. But you know it's not just the usual. It also comes with 5K display and that changes everything.

It was like iPad 3 all over again, a slightly overclocked A5 rebranded as A5x chip to feed the new retina display with 4 times more pixels. It will struggle to do that and it will be a beta testing product before the real deal comes.

So again I would skip this retina iMac and wait for 2 or 3 years to see if Apple could actually use 5K to do intensive tasks. Not just to look pretty.
Click to expand...

For UI stuff, and web browsing it'll be fine, as well as video/photo editing. No one is going to be playing games at 5k anytime soon.. Not for years.
 
Is that an expected CPU score? Seems the overall NovaBench score is really low - my 2009 Mac Pro does 1159 (816 on the CPU tests and 223 on RAM). Makes me wonder if upgrading is really worth it.
 
It could be an issue on the app's side. It was not updated in a while so it doesn't natively support Yosemite.
 
You also can't compare OS X graphics performance to Windows graphics performance.

And honestly, results at 5K resolution are utterly pointless. Nobody will be gaming at that res on the iMac.
 
I really dont rely much on Novabench, but its the only thing that i could find to try and compare it as i could not find any other bechmarks.

Regarding gaming im not worried at all i simply dont play :)

What i would like to see while we dont have any other info is if any of the actual non retina Imac late 2013 780m could do the Novabench test so we could get some comparisons :)

Regarding the diferent OS im aware that iw works diferently but i was trying to get an early read by comparing the results and then using them on the Passmark charts to see where it would fit so we could get a closer idea of where it might stand.

My wallet was ready and still is but before i drop 3.2k€ i need to make sure im getting GPU rendering improvements :)

PS: If any of you find other info online post it here so we can have more info.
 
http://www.3dmark.com/3dm11/8832926

This is Alienware laptop with Radeon R9 M295X GPU(everything so far points to it...).

Check the Graphics score. Its 20% better than Nvidia GTX980M GPU and slightly above average scores for desktop GTX970.

If AMD and Apple pulled that amount of power from 100W GPU...
http://forum.notebookreview.com/gaming-software-graphics-cards/747646-radeon-r9-m295x-104.html
Source for all of this.

P.S. Notice how CPU bottlenecks the GPU.

Proceso de fabricación de TSMC a 28 nanómetros.
5000 millones de transistores.
2048 Shader Processors que funcionan a 750 Mhz (800 Mhz Boost).
128 TMUs (Unidades de Textura).
32 ROPs (Unidades de Renderizado).
4 GB de memoria GDDR5 con bus de 256 bits, que funcionan a 5,5 GHz efectivos.
TDP de 100W.
Click to expand...
I think the clock might be a little to low for 3.5 Tflops, its more 850 MHz on Core and 5500 Mhz on GPU.
And its exactly same performance as D700 from Mac Pro.
 
Look at Graphics score.

R9 M295X is 20% faster than GTX980M and is on par with desktop GTX970.
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back