GravityMark 1.70 Released With VK_EXT_mesh_shader, Vulkan ReBAR Support

[phoronix]

Phoronix: GravityMark 1.70 Released With VK_EXT_mesh_shader, Vulkan ReBAR Support

GravityMark 1.70 has been released as the multi-API graphics benchmark developed by Tellusim Technologies that was started by one of the former Unigine engine developers...

https://www.phoronix.com/news/GravityMark-1.70-Released

[Linuxxx]

Really good stuff!

Which makes me wonder:

What's up with Unigine these days?
Still no signs of Vulkan support, right?

Hopefully Tellusim finds more visible commercial success with their engine than Unigine ever did...

[smitty3268]

Vulkan Resizable Bar support

I'm curious what that means. Resizable Bar shouldn't be visible to the app, should it? Is it changing the memory usage in ways that require ReBar to be available in order to be optimized, or something?

[birdie]

One of the most useless 3D "benchmarks". Barely tests anything.

Just like the built-in CPU-Z "benchmark" - what it does and shows - no one knows.

Here, something more modern and which actually uses 3D capabilities of your GPU: https://store.steampowered.com/app/7...nch_Benchmark/

[frustum]

I'm curious what that means. Resizable Bar shouldn't be visible to the app, should it? Is it changing the memory usage in ways that require ReBar to be available in order to be optimized, or something?

Resizable BAR support requires a slightly different behavior from the dynamic buffer allocator. Otherwise, there is no difference.

[Linuxxx]

One of the most useless 3D "benchmarks". Barely tests anything.

Just like the built-in CPU-Z "benchmark" - what it does and shows - no one knows.

Here, something more modern and which actually uses 3D capabilities of your GPU: https://store.steampowered.com/app/7...nch_Benchmark/

To be fair, the reviews on Steam are not particularly enthusiastic about the benchmark you linked, either:

I really don't see the point of this software. OK, it's a benchmark/stress test that reduces the finest hardware to rubble. Maybe of interest to hardware geeks and overclockers. However, even in that role, it only gives a 'score' and an average FPS result. Not even a max/min figure, still less some detail to indicate what is the bottleneck on the specific system. Even the benchmark in Shadow of the Tomb Raider gives you CPU and GPU breakdowns. Although not an overclocker myself I imagine there are plenty of good stress tests that do a far better job

But, what is ti telling us about Unreal 5? Is it saying my shiny new (mid range) upgrade to 12th gen Intel and RTX will run an Unreal 5 game like a dead dog? It only managed 20 FPS average and dropped to around 12 at places. I thought Unreal 5 was supposed to provide high quality graphics 'even' on consoles and at high resolution. I also assumed Lumen was an alternative to ray-tracing but apparently this test does a lot of ray tracing.. So, what use my new 32", 1440p, 240 Hz monitor at only 20 FPS? OK, they say it is "not optimised". Will optimisation make a 10x difference? Will I be able to run Unreal 5 games acceptably, which is what I wanted to know?

For me, this software just muddies the waters.

[Linuxxx]

Resizable BAR support requires a slightly different behavior from the dynamic buffer allocator. Otherwise, there is no difference.

How are Your old colleagues at Unigine doing without You?

Are You still in contact with them?

And can You share with us Your impression of the state of Unigine?

Thanks for the impressive GravityMark, BTW!

Hopefully You'll find great success with Tellusium!

[frustum]

How are Your old colleagues at Unigine doing without You?
Are You still in contact with them?
And can You share with us Your impression of the state of Unigine?

Most of my old colleagues don't work in Unigine anymore.
And I don't know anything about the current state of the engine.

Thanks for the impressive GravityMark, BTW!
Hopefully You'll find great success with Tellusium!

Thank you!

[shmerl]

I'm curious what that means. Resizable Bar shouldn't be visible to the app, should it? Is it changing the memory usage in ways that require ReBar to be available in order to be optimized, or something?

Apparently application can make some decisions about memory allocation based on resizable BAR availability or not.

https://github.com/HansKristian-Work...md#performance
https://github.com/HansKristian-Work...source.c#L6500

[Eirikr1848]

Phoronix: GravityMark 1.70 Released With VK_EXT_mesh_shader, Vulkan ReBAR Support

I'll have up v1.70 benchmarks in my next round of Linux GPU benchmarks.

https://www.phoronix.com/news/GravityMark-1.70-Released

… well damn this tease!!! 😂 Why am I always so excited ‘bout benchmarks!?