3DMark’s new benchmark: Steel Nomad and Steel Nomad Light.

I’m seeing some people (not here) being surprised by the delta between the M4 Max and the 5090. I think it’s to be expected given the delta in compute and power between the two.
 
casperes1996 said:
Being big on a platform like YouTube is less about your technical knowledge and accuracy and more about your presentation style and camera-personality. The very biggest channels are often less knowledgable and accurate than the ones a few ticks smaller
Click to expand...
True but even knowing that I wasn’t really impressed. 🙃

Edit: Okay to flesh that out beyond me just being jealous and petty, he’s basically a standard TechTube clone in that regard. As far as I can tell, he doesn’t have anything that really differentiates him.

I mean for a mistake like the one in the video, in fairness, not the worst mistake in the world though usually you’d get a visual edit over it (we’ve all seen that where the YouTuber puts the right number with an asterisk in big bold font every time they said the wrong thing or put up a chart with it). And no doubt having to churn out video after after video such a mistake would eventually happen (edit of an edit: holy shit he’s posted 6 videos in the last 2 days alone). Truthfully the grind of maintaining such a channel must be pretty intense, especially since it’s not like YouTube pays a ton unless you’re huge not just big.

I can’t remember what about the previous video I’d seen that annoyed me, but yeah it was probably technical too. I think it was a video posted on the other site.
 
Last edited:
Jimmyjames said:
It appears that the full version of Steel Nomad will be released for macOS soon. Here is a pre-release score from this video. Around RTX 3070 level.

View attachment 34406
Click to expand...

Steel Nomad and Steel Nomad light is the same test at different resolutions.

It looks like 3dmark is making a macOS launcher for the existing tests. It doesn't seem like they are going to bring over the existing feature tests. Someone said they were working on a new mobile RT test, though they would probably be better served just bringing Port Royal or Speedway over.
 
diamond.g said:
Steel Nomad and Steel Nomad light is the same test at different resolutions.

It looks like 3dmark is making a macOS launcher for the existing tests. It doesn't seem like they are going to bring over the existing feature tests. Someone said they were working on a new mobile RT test, though they would probably be better served just bringing Port Royal or Speedway over.
Click to expand...
It’s just steel nomad. The other tests are considered old/out of date I believe.
 
Jimmyjames said:
It’s just steel nomad. The other tests are considered old/out of date I believe.
Click to expand...
Speedway "just" came out as the replacement for Port Royal. The only thing it is missing is a path traced feature test. With DXR 1.2 being announced/released I suppose they are going to make a new RT test to take advantage of it. I presume Metal already has the same feature set?
 
diamond.g said:
Speedway "just" came out as the replacement for Port Royal. The only thing it is missing is a path traced feature test. With DXR 1.2 being announced/released I suppose they are going to make a new RT test to take advantage of it. I presume Metal already has the same feature set?
Click to expand...
By other tests, I meant Time Spy etc. Speedway is a DirectX specific benchmark afaik.
 
Jimmyjames said:
I’m seeing some people (not here) being surprised by the delta between the M4 Max and the 5090. I think it’s to be expected given the delta in compute and power between the two.
Click to expand...
I see what you mean:

forums.macrumors.com

Apple is falling behind, M4 Max nowhere close to RTX 5090.

When the M1 Max launched, Apple said that it rivaled the flagship NVIDIA GPU at the time, the RTX 3080. However, in 2025, when comparing the M4 Max MacBook Pro to PC laptops equipped with the flagship NVIDIA GPU, the RTX 5090, which costs the same as a M4 Max MacBook Pro, the MacBook Pro gets...
forums.macrumors.com forums.macrumors.com

:)
 
edit: Wondering if somebody with an iPhone 16 Pro Max would mind running the benchmark. I'm getting about 12-13fps on my iphone 16 pro after a clean boot, whereas I'm getting average of about 17 fps on a S25 ultra phone. Wondering if the smaller size of my Pro is making the a18 throttle somewhat in the bench. Wondering if the Max variant would put in a better score close to the S25.
 
Last edited:
Jimmyjames said:
It appears that the full version of Steel Nomad will be released for macOS soon. Here is a pre-release score from this video. Around RTX 3070 level.

View attachment 34406
Click to expand...

Interestingly Notebookcheck has the 5060 Ti scoring 15%-20% better than the M4 Max's Steel Nomad score, but the reverse is true in Blender:


opendata.blender.org

Blender - Open Data

Blender Open Data is a platform to collect, display and query the results of hardware and software performance tests - provided by the public.
opendata.blender.org opendata.blender.org

The size of the 5060 Ti GPU should be similar, in fact smaller than a M4 Max:

www.techpowerup.com

NVIDIA GeForce RTX 5060 Ti Specs

NVIDIA GB206, 2572 MHz, 4608 Cores, 144 TMUs, 48 ROPs, 8192 MB GDDR7, 1750 MHz, 128 bit
www.techpowerup.com www.techpowerup.com

(technically the TFLOPs of the 5060 Ti are higher, but as discussed previously actual performance is closer to TFLOPs/2*1.3)

This discrepancy between the results could be a function of multiple (non-mutually exclusive) things: 1) the Steel Nomad Mac benchmark is in beta, it's possibly still be worked on. 2) The 5060 Ti's drivers aren't set and for whatever reason Blender suffers worse. 3) Blender has ray tracing while SN is pure raster. 4) The SN engine is just better suited to Nvidia hardware for some reason. Given the size of the two GPUs the Blender results make slightly more sense.
 
Last edited:
dada_dave said:
Interestingly Notebookcheck has the 5060 Ti scoring 15%-20% better than the M4 Max's Steel Nomad score, but the reverse is true in Blender:


opendata.blender.org

Blender - Open Data

Blender Open Data is a platform to collect, display and query the results of hardware and software performance tests - provided by the public.
opendata.blender.org opendata.blender.org

The size of the 5060 Ti GPU should be similar, in fact smaller than a M4 Max:

www.techpowerup.com

NVIDIA GeForce RTX 5060 Ti Specs

NVIDIA GB206, 2572 MHz, 4608 Cores, 144 TMUs, 48 ROPs, 8192 MB GDDR7, 1750 MHz, 128 bit
www.techpowerup.com www.techpowerup.com

(technically the TFLOPs of the 5060 Ti are higher, but as discussed previously actual performance is closer to TFLOPs/2*1.3)

This discrepancy between the results could be a function of multiple (non-mutually exclusive) things: 1) the Steel Nomad Mac benchmark is in beta, it's possibly still be worked on. 2) The 5060 Ti's drivers aren't set and for whatever reason Blender suffers worse. 3) Blender has ray tracing while SN is pure raster. 4) The SN engine is just better suited to Nvidia hardware for some reason. Given the size of the two GPUs the Blender results make slightly more sense.
Click to expand...
I honestly have no idea how well optimized 3DMark’s benchmarks are for macOS. I suspect not very well optimized. but I have nothing to back that up.
 
Jimmyjames said:
I honestly have no idea how well optimized 3DMark’s benchmarks are for macOS. I suspect not very well optimized. but I have nothing to back that up.
Click to expand...
Macs and iPhones generally perform really well in the older Wild Life Extreme and okay in SNL if I remember right. But this result is definitely odd.
 
Jimmyjames said:
I honestly have no idea how well optimized 3DMark’s benchmarks are for macOS. I suspect not very well optimized. but I have nothing to back that up.
Click to expand...
I was under the impression UL writes the bench to an API, and don't include vendor specific paths.


I wonder why I cannot find any vulkan scores for the 5060ti in Steel Nomad (Light as well).
 
diamond.g said:
I was under the impression UL writes the bench to an API, and don't include vendor specific paths.


I wonder why I cannot find any vulkan scores for the 5060ti in Steel Nomad (Light as well).
Click to expand...
I am sure that’s true. But even writing general DirectX or Metal or whatever code there’s still usage patterns that are more or less friendly to a given micro architecture
 
Back
Top