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

[Jimmyjames]

3DMark have just released their new benchmark designed as a replacement for Time Spy. There are two variants: Steel Nomad, designed for desktops and laptops; and Steel Nomad Light aimed primarily at mobile devices, although you can run it on deskops/laptops as well.

The full version is currently only out for Windows, with macOS and Linux arriving later. The Light version is out now for Windows, iOS, Android. You can run the light version on macOS it seems.

I tried it out on my 15 Pro Max and it is definitely much more demanding than Wildlife Extreme, averaging 10fps! I’d be interested to see other device scores.
Here is the link for the iOS version:

‎3DMark

‎Test and compare the performance of the latest iPhones and iPads with this free and easy-to-use benchmarking app. 3DMark Steel Nomad Light is our heaviest non-raytracing benchmark for iOS. Steel Nomad Light lets you compare the performance of the latest iPhones and iPads in heavy gaming...

apps.apple.com

 

[dada_dave]

I tried it out on my 15 Pro Max and it is definitely much more demanding than Wildlife Extreme, averaging 10fps! I’d be interested to see other device scores.

I have an iPhone 12 … mini. If you’re getting 10fps I think I’ll just be sitting this one out.

 

[Jimmyjames]

I have an iPhone 12 … mini. If you’re getting 10fps I think I’ll just be sitting this one out.

I should have mentioned, it seems like it only runs on phones with 8GB of ram or more!! So nothing before the 15 Pro.

 

[Jimmyjames]

iPad Air (2022] M1 = 14.2 fps.

 

[Jimmyjames]

Interesting to see a Samsung S24+ with an Exynos 2400 has a score of 13.52 fps.

x.com

x.com

I would have thought my iPhone would have scored higher than this, but clearly not. Difficult to say why. Are there bugs? This is a more sophisticated test than Wildlife Extreme, so should more accurately reflect the power of the devices but maybe the Exynos is better in important ways that this test measures?

 

[dada_dave]

Interesting to see a Samsung S24+ with an Exynos 2400 has a score of 13.52 fps.

x.com

x.com

I would have thought my iPhone would have scored higher than this, but clearly not. Difficult to say why. Are there bugs? This is a more sophisticated test than Wildlife Extreme, so should more accurately reflect the power of the devices but maybe the Exynos is better in important ways that this test measures?

So if my info is correct the Exynos GPU is 3.4 TFLOPs compared to your iPhone’s 2.1. Now TFLOPs aren’t everything but that’s a big GPU - AMD based. If anything the iPhone is over performing compared to its TFLOPs either because of its architecture or thermal capacity of the Samsung phone. How long a test is it?

Samsung Exynos 2400: specs and benchmarks

Samsung Exynos 2400: performance tests in benchmarks (AnTuTu 10, GeekBench 6). Battery life and full specifications.

nanoreview.net

 

[Jimmyjames]

So if my info is correct the Exynos GPU is 3.4 TFLOPs compared to your iPhone’s 2.1. Now TFLOPs aren’t everything but that’s a big GPU - AMD based. If anything the iPhone is over performing compared to its TFLOPs either because of its architecture or thermal capacity of the Samsung phone. How long a test is it?

Yeah it does seem quite variable in terms of scores. It’s about a minute long.

I’ve done a few runs now and the iPhone has gotten as high as 13fps. Just need to find some liquid nitrogen now!

 

[Jordanmarkstone]

Yeah it does seem quite variable in terms of scores. It’s about a minute long.

I’ve done a few runs now and the iPhone has gotten as high as 13fps. Just need to find some liquid nitrogen now!

My run had highs of 17fps, but lows of 7. So thermals definitely play a big part.

 

[Jimmyjames]

My run had highs of 17fps, but lows of 7. So thermals definitely play a big part.

Sounds similar to mine.

 

[casperes1996]

M1 Pro work laptop with 16GB of RAM (though I had Mail and iTerm2 open alongside Wireguard and Stats that I never close - Can also try on M1 Max personal laptop if people are interested. Don't think it can run on Intel Macs since it is just the iOS app I ran there. - It also has a ray traced test it seems. Interested to see how that one performs M3+/A17 vs. prior. That ray traced test seems to be older and less taxing than Steel Nomad even though it's ray traced, I just never heard of it before. Here's the work laptop results for it (with Teams and Safari opened back up so not necessarily that good results, but I doubt it matters all too much)

My 16"MBP work laptop had its fans off entirely for the duration of both tests but it also never really heated up that much. - The RAM usage never seemed to go above 3.x GB where I'm unsure exactly where x peaked, but something low like 3.1-3.2 ish GB. I see no reason it requires 8+GB of RAM honestly.

 

[Jimmyjames]

View attachment 29567

M1 Pro work laptop with 16GB of RAM (though I had Mail and iTerm2 open alongside Wireguard and Stats that I never close - Can also try on M1 Max personal laptop if people are interested. Don't think it can run on Intel Macs since it is just the iOS app I ran there. - It also has a ray traced test it seems. Interested to see how that one performs M3+/A17 vs. prior. That ray traced test seems to be older and less taxing than Steel Nomad even though it's ray traced, I just never heard of it before. Here's the work laptop results for it (with Teams and Safari opened back up so not necessarily that good results, but I doubt it matters all too much)
View attachment 29568

My 16"MBP work laptop had its fans off entirely for the duration of both tests but it also never really heated up that much. - The RAM usage never seemed to go above 3.x GB where I'm unsure exactly where x peaked, but something low like 3.1-3.2 ish GB. I see no reason it requires 8+GB of RAM honestly.

Great stuff, thanks so much!

I’d definitely be interested in the results for the M1 Max.

 

[casperes1996]

Great stuff, thanks so much!

I’d definitely be interested in the results for the M1 Max.

The Stats view is near the end of running Steel Nomad.

16" MacBook Pro with M1 Max, 32GB of RAM, 24 Core GPU model, so it's the cut-down M1 Max. Maximum frame rate of Solar Bay was probably limited by the ProMotion Display as it stopped at 120. Could've run again with the Unlimited mode but that test not priority, just a side note, and most of the samples were below 120 by a fair bit anyway so didn't bother. - Basically just a sample or two at the beginning at 120FPS.

In my Stats screenshot, the two 0s are the fan speeds. The number above system wattage is GPU temp and the other is CPU temp (hottest measurements)
I expanded the RAM view as it otherwise doesn't mean much. As you can see, I didn't close everything else but didn't have anything "active" running. Nothing that really eats CPU or GPU.

 

[Jimmyjames]

View attachment 29573

View attachment 29574View attachment 29575

View attachment 29576

The Stats view is near the end of running Steel Nomad.

16" MacBook Pro with M1 Max, 32GB of RAM, 24 Core GPU model, so it's the cut-down M1 Max. Maximum frame rate of Solar Bay was probably limited by the ProMotion Display as it stopped at 120. Could've run again with the Unlimited mode but that test not priority, just a side note, and most of the samples were below 120 by a fair bit anyway so didn't bother. - Basically just a sample or two at the beginning at 120FPS.

In my Stats screenshot, the two 0s are the fan speeds. The number above system wattage is GPU temp and the other is CPU temp (hottest measurements)
I expanded the RAM view as it otherwise doesn't mean much. As you can see, I didn't close everything else but didn't have anything "active" running. Nothing that really eats CPU or GPU.

Awesome. Many thanks for that.

 

[Jimmyjames]

From Twitter

x.com

x.com

x.com

x.com

 

[diamond.g]

I get 5 fps on my Pixel 7 and 44 on my 14" M1 Max 24c GPU. I guess it won't run on my iPhone 14 Pro...

I get 40fps on my 6900xt on the full test. Haven't tried the light one yet

 

[Yoused]

iPad mini: 2883 – 17.3fps

I mean, it mostly looked quite good. I am not a gamer, I do not immerse myself in a 4K 5.1 (or whatever sound is hip today) experience. I might possibly see that higher than 60fps could matter, but when they start talking in the 100+ range, no, you really do not need to go higher and higher.

 

[mr_roboto]

M1 Max 32c GPU results:

 

[Jimmyjames]

Here is a list of results from many other systems.

3DMark Steel Nomad: Ergebnisse im Überblick

Der neue Test Steel Nomad ist ab sofort kostenlos für 3DMark verfügbar. Jetzt seid ihr gefordert – mit PC, Notebook, Tablet oder Smartphone.

www.computerbase.de