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

jmho

macrumors 6502a
Jun 11, 2021
502
996
Not to sound too difficult, but why would game developers waste resources (and money) with high quality textures and shadows and things like ray tracing if "none" of their userbase can utilize it?
I worked as an engine / graphics programmer in the games industry for 8 years and the main reason for it is that 99% of devs are in it for passion not money.

I was earning half what I could make outside of games, and the main reason I put up with it so long was that I could work on cutting edge, interesting stuff and get paid to do it. Lots of people in the industry work on their own projects for free too just because they love doing it. I guarantee that almost every game that currently supports ray tracing started with the programmers begging the business people to let them do it and not the other way around.

It's the same with the artists. They want to make mind-blowing high quality art because they love making mind-blowing high quality art. If you asked them to spend all day quickly knocking up ugly low resolution art most of them would quit.
 

Lihp8270

macrumors 65816
Dec 31, 2016
1,143
1,608
Not to sound too difficult, but why would game developers waste resources (and money) with high quality textures and shadows and things like ray tracing if "none" of their userbase can utilize it?
Games are often released that require several years to pass before computers can finally run them reasonably at “max” with high resolutions.
 

mi7chy

macrumors G4
Oct 24, 2014
10,625
11,296
If you want performance per watt, you get the Mac.

Not true for GPU. 3060 70W scores 86390 on Geekbench 5 OpenCL so 86390/70W = 1234 OpenCL/W.

https://browser.geekbench.com/v5/compute/3555295

While M1 Max 60W scores 60167/60W = 1003 OpenCL/W.

So, 3060 70W is 17% more watt for 43% more performance.

Too bad there isn't a 60W GPU power limit on Legion Slim 7 to confirm 3060 is still faster.
 
Last edited:

leman

macrumors Core
Oct 14, 2008
19,521
19,679
Not true for GPU. 3060 70W scores 86390 on Geekbench 5 OpenCL so 86390/70W = 1234 OpenCL/W.

https://browser.geekbench.com/v5/compute/3555295

While M1 Max 60W scores 60167/60W = 1003 OpenCL/W.

So, 3060 70W is 17% more watt for 43% more performance.

Too bad there isn't a 60W GPU power limit on Legion Slim 7 to confirm 3060 is still faster.

Apple GPUs don't perform very well in Geekbench for some reason. For example, when you look at the (base) M1 it only scores 20k — significantly less than a mobile 1650 that has the same theoretical peak flops. It would be interesting to study why this is so. It's similar for AMD GPUs by the way, e.g. an RtX 3060 mobile consistently scores ~20% higher than an RX 6700M despite the GPUs are closely matched for compute specs and the 6700M actually being faster in games. I don't know whether current Nvidia GPUs are just so much better for compute or whether there is some sort of issue with Geekbench itself...
 
  • Like
Reactions: Jorbanead and Homy

l0stl0rd

macrumors 6502
Jul 25, 2009
483
420
Not true for GPU. 3060 70W scores 86390 on Geekbench 5 OpenCL so 86390/70W = 1234 OpenCL/W.

https://browser.geekbench.com/v5/compute/3555295

While M1 Max 60W scores 60167/60W = 1003 OpenCL/W.

So, 3060 70W is 17% more watt for 43% more performance.

Too bad there isn't a 60W GPU power limit on Legion Slim 7 to confirm 3060 is still faster.
OpenCL does not matter in a way and there are no Metal Nvida benchmarks?

Gfx Bench 5 it seen to be as fast as an 3070 / 3080

Need better stuff to compare, I wish Blender with Metal would be out.

Also would like to se 3dmark wild life extreme scores…
 

iBug2

macrumors 601
Original poster
Jun 12, 2005
4,540
863
Not true for GPU. 3060 70W scores 86390 on Geekbench 5 OpenCL so 86390/70W = 1234 OpenCL/W.

https://browser.geekbench.com/v5/compute/3555295

While M1 Max 60W scores 60167/60W = 1003 OpenCL/W.

So, 3060 70W is 17% more watt for 43% more performance.

Too bad there isn't a 60W GPU power limit on Legion Slim 7 to confirm 3060 is still faster.
Sorry but isn't the 60W the total power of the M1 Max SoC while 70W is only for the 3060 GPU?
 

mi7chy

macrumors G4
Oct 24, 2014
10,625
11,296
Sorry but isn't the 60W the total power of the M1 Max SoC while 70W is only for the 3060 GPU?

30W for CPU + ~60W for iGPU.

https://www.apple.com/newsroom/2021...the-most-powerful-chips-apple-has-ever-built/

Apple_M1-Pro-M1-Max_CPU-Performance_10182021_big.jpg.large.jpg


Apple_M1-Pro-M1-Max_M1-Max-GPU-Performance-vs-PC_10182021_big_carousel.jpg.large.jpg
 

dmr727

macrumors G4
Dec 29, 2007
10,674
5,835
NYC
Need better stuff to compare, I wish Blender with Metal would be out.

This is what I've been waiting for as well. Benchmarks and marketing BS are all fine and good for arguing on message boards, but I'm excited to see the results in real use scenarios. I'm not convinced the Max will be able to genuinely slug it out with an RTX 3070/3080 mobile, but if it can get within the ballpark while running cool and mostly silent, I'd be pretty happy.
 

Irishman

macrumors 68040
Nov 2, 2006
3,449
859
Good luck with that unified memory of yours when real GPU process kick in like rendering in Fusion which eats up RAM as much as it can and at the same time uses GPU for most of it's tools.

That sounds like a Fusion problem that can be fixed.
 

Irishman

macrumors 68040
Nov 2, 2006
3,449
859
What they won't tell you is you have to get 64GB for multi-channel memory bandwidth otherwise it probably diminishes with every step down.

They literally just told us that in the video.

How exactly is this a bad thing?
 

ElfinHilon

macrumors regular
May 18, 2012
142
48
Not true for GPU. 3060 70W scores 86390 on Geekbench 5 OpenCL so 86390/70W = 1234 OpenCL/W.

https://browser.geekbench.com/v5/compute/3555295

While M1 Max 60W scores 60167/60W = 1003 OpenCL/W.

So, 3060 70W is 17% more watt for 43% more performance.

Too bad there isn't a 60W GPU power limit on Legion Slim 7 to confirm 3060 is still faster.
You know, I'm starting to think more and more that these leaked benchmarks are actually the 24 core GPU. The numbers line up just far too well for it not to be.
 
  • Like
Reactions: l0stl0rd

ElfinHilon

macrumors regular
May 18, 2012
142
48
And you can add +15% to the openCL score to estimate the Metal score. openCL run on emulation on the M1.
This is only a compute task, I'm more interested in rendering tasks.

Anyway, that openCL score seems too low. There is no reason why it shouldn't be 4x the M1 score, so >70000.
There are sub-test where the M1 Max is about 4x faster, and some others where it's just a bit more than 2x faster. Not sure why that is, but I'rather see Metal scores. Who uses openCL on the Mac these days?
I'm only seeing 2-3 scores where it's actually 4x the score. The majority of them seem to be around 3x scaling. I genuinely am starting to believe that this is actually the 24 core GPU and NOT the full 32 core GPU. If our theories are correct, this would easily explain the "poorer" scaling we are seeing where nearly all the numbers correspond with 3x the score of the base M1, vs the much closer to 4x we are expecting with the 32 core GPU. I guess we will need to wait and see what actually happens.
 

EntropyQ3

macrumors 6502a
Mar 20, 2009
718
824
Not to sound too difficult, but why would game developers waste resources (and money) with high quality textures and shadows and things like ray tracing if "none" of their userbase can utilize it?
Because Nvidia pays them to.
And even does some of the work.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.