3D Mark announces Steel Nomad benchmark, replaces Timespy

[nlel1975]

not bad

Anything above 6000 / 60FPS is really good at the moment. Come back in 7-8 years and it will be a different story.

 

[humbug]

With tight memory timings, i've had to bring the memory frequency down a bit to 2551 but its still posted a slightly higher score, it says 2574, ignore that. Reasons below.

It did not load hardware monitoring for whatever reason, it was hanging on loading the benchmark run, it says 2124 for the clock, no it wasn't, i set it the same as the last run just stick with 2999.

Last time i do this with this GPU, i think...

Humbug, 7800 XT, core 2999, Mem 2551, Score 4654

I scored 4 654 in Steel Nomad

AMD Ryzen 7 5800X, AMD Radeon RX 7800 XT x 1, 32768 MB, 64-bit Windows 11}

www.3dmark.com

 

[MadMatty]

The Systeminfo module hanging is a real annoyance. If it takes more than 10-15 seconds its likely not going work and give an error at the end whilst giving best ever score! It would be better if it determined that sysinfo failed at the start and abort the benchmark. I think there is something weird going on as it battles the amd drivers polling the cpu/gpu for the same info. I never had such an issue before with previous R9 290 with its older drivers and no metrics..

 

[DK83]

cant validate for some reason, 5075 on dx12, 5694 on vulkan on my 5800x and undervolted 4070ti

 

[Cob]

RX 6700XT @ 2750/2150 & 5700X3D

DX12 -

I scored 2 610 in Steel Nomad

AMD Ryzen 7 5700X3D, AMD Radeon RX 6700 XT x 1, 32768 MB, 64-bit Windows 11}

www.3dmark.com

Vulcan -

I scored 2 671 in Steel Nomad

AMD Ryzen 7 5700X3D, AMD Radeon RX 6700 XT x 1, 32768 MB, 64-bit Windows 11}

www.3dmark.com

 

[mrk]

I feel as though the benchmark is a bit too short given all the hype for it! Ok you can do the exploration bit and do stuff but I think I was expecting something much longer Timespy for example has a lot going on.

I've done my run and am bored so uninstalled, probably wasted a fiver on it haha but I'll reinstall when they release another update I guess.

 

[JediFragger]

I feel as though the benchmark is a bit too short given all the hype for it! Ok you can do the exploration bit and do stuff but I think I was expecting something much longer Timespy for example has a lot going on.

Looks a bit crap too imo, like something from 2010...

 

[L0rdMike]

7900XT
DX12

I scored 6 100 in Steel Nomad

AMD Ryzen 7 7800X3D, AMD Radeon RX 7900 XT x 1, 32768 MB, 64-bit Windows 11}

www.3dmark.com

Could prob get a bit more out of the card but cba benchmark is rubbish

Vulkan

I scored 6 105 in Steel Nomad

AMD Ryzen 7 7800X3D, AMD Radeon RX 7900 XT x 1, 32768 MB, 64-bit Windows 11}

www.3dmark.com

 

[mrk]

Looks a bit crap too imo, like something from 2010...

I wouldn't go that far But yes it's a bit limited!

 

[Grim5]

I wouldn't go that far But yes it's a bit limited!

Most GPU benchmarks don't look particularly great; I suppose its because your benchmark doesn't need to look good to be demanding and being demanding for modern hardware is the goal of the benchmark.

The last time I was impressed by the visuals of a benchmark was when Heaven was released in 2009

 

[jp7152]

A whopping 2388 points here with an overclocked 4060

I scored 2 388 in Steel Nomad

AMD Ryzen 7 5800X3D, NVIDIA GeForce RTX 4060 x 1, 32768 MB, 64-bit Windows 11}

www.3dmark.com

 

[nlel1975]

Heavily overclocked 4090 or GTFO!

Here's a video of it running on a 4090. The FPS are significantly higher than on my 7900 XTX but I wouldn't upgrade to one as I would lose too much on my 7900 XTX. I'll wait a few years.

âï¸ 3DMark Steel Nomad on a 4090

60 seconds · Clipped by Nick Le Lievre · Original video "3DMark Steel Nomad Benchmark | RTX 4090 | Ryzen 9 7950X3D" by Xenthorx

www.youtube.com

Also does anyone know how to enable sound in the benchmark, mines completely silent and I can't see any options to enable sound.

 

[JediFragger]

and I can't see any options to enable sound.

This may be an actual clue....

 

[nlel1975]

This may be an actual clue....

How did the guy in the above video have sound on his?

 

[JediFragger]

How did the guy in the above video have sound on his?

 

[nlel1975]

Here it is...

Never noticed it before for some reason lol

 

[JonRGV250]

I can see how you missed it.
Its tiny and not at all slap bang in the middle of the screen.

 

[Jay-G25]

Hall of Fame is up but is full of bugged scores so pretty much useless for DX12.... people posting 11 to 12K scores with silly low core clocks and high temps

3DMark Steel Nomad DX12 Hall of Fame

The 3DMark.com Overclocking Hall of Fame is the official home of 3DMark world record scores.

www.3dmark.com

 

[nlel1975]

Hall of Fame is up but is full of bugged scores so pretty much useless for DX12.... people posting 11 to 12K scores with silly low core clocks and high temps

I think a lot of those won’t be stable, i.e. have artifacts etc and other visual discrepancies but still give a score which can be uploaded.

 

[Audioboxer]

I got 7375 in that new 3DMark test with a 4080

I scored 7 375 in Steel Nomad

AMD Ryzen 9 7950X3Dâ, NVIDIA GeForce RTX 4080 x 1, 32768 MB, 64-bit Windows 11}

www.3dmark.com

Unfortunately, it was explained to me the HOF has quite a few glitched runs in it just now, especially with 4090s. Some way to cheat or something with VRAM glitching? Think it might involve maxing out the memory slider or something.

Anyway, the above required like 1.185v at 3.255ghz and 630w lol.

My daily profile isn't far off it and is "only" at 1.125v

I scored 7 298 in Steel Nomad

AMD Ryzen 9 7950X3Dâ, NVIDIA GeForce RTX 4080 x 1, 32768 MB, 64-bit Windows 11}

www.3dmark.com

This score for example, is probably "faked"

I scored 8 145 in Steel Nomad

Intel Core i7 processor 14700KF, NVIDIA GeForce RTX 4080 x 1, 32768 MB, 64-bit Windows 11}

www.3dmark.com

There's a way to prove that using the versus function, apparently, but also, just look at my core clocks above and compare it to that run.

I'm not really clock stretching either as you can see from this earlier run. So, people with higher scores and lower frequency could be legit, I mean, I even got to 7375 with roughly the same settings, all I increased a bit more was the power draw to nearer 630w.

But going to 8xxx with 3015mhz on the core of a 4080

edit - Yup, just confirmed, all it takes to cheat the benchmark is pushing your memory clock till it causes some graphics artefacts and then your FPS will go quite a bit higher

So yeah, watch out for fake runs where people push their memory slider till they get artefacts during the test.

That 8151 is with my daily profile, all I did was increase memory from a known stable for me, +1300, to +1500.