Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Messages - JeGX

Pages: 1 2 [3] 4 5 ... 32
41
English forum / GPU PhysX on Linux
« on: October 18, 2014, 04:07:32 PM »
The new PhysX SDK version 3.3.2 adds the GPU PhysX acceleration on Linux (see THIS NEWS).  I updated GLSL Hacker with this new SDK (GLSL Hacker v0.7.2.0) and I added a new particle demo in the code sample pack:

host_api/PhysX3/Pool/demo_gl2_v1.xml

This is a simple particle/fluids demo that fills a pool with particles:



On Linux, you can start the demo with the command line:
Code: [Select]
$ ./GLSLHacker /demofile=\"path_to_code_sample_pack/host_api/PhysX/Pool/demo_gl2_v1.xml\"


Currently, I didn't manage to get the GPU acceleration on Linux (Mint 17 64-bit). I tested with latest R331.104 and with R340.xx. There is a cuInit failure (I installed the latest CUDA toolkit v6.5.14). The cuInit failure is also present with PhysX SDK samples.

Here are some benchmark numbers with GPU PhysX (currently only under Windows) and CPU PhysX (Windows and Linux).

To force CPU PhysX, just edit the demo file (demo_gl2_v1.xml) and update the line 182:
Code: [Select]
gpu_physx = 0


Benchmark settings: 6000 particles, 1280x720 windowed.

On Windows with a GeForce GTX 660 (R337.50) + Intel Core i5 2320 @ 3GHz:
- GPU PhysX: around 420 FPS
- CPU PhysX: around 150 FPS


On Linux Mint 17 64-bit, with a GeForce GTX 680 (R331.104) + AMD FX 6100 @ 3.3GHz:
- GPU PhysX: not available (cuInit failed)
- CPU PhysX: around 60 FPS (this CPU sucks!)

As soon as the GPU PhysX will be enabled, we should see a jump in the FPS on Linux (> 100 FPS on my Linux box).


42
English forum / Re: OpenGL 4 Shader Subroutines - NVIDIA/OSX bug?
« on: October 18, 2014, 03:40:51 PM »
I will try to update the NV driver, but I don't know if the Quadro driver will be ok for the GT650M.

Will try also OSX 10.10 asap...

43
Quote
qu3e is a compact, light-weight and fast 3D physics engine in C++. It is has been specifically created to be used in games. It is portable with no external dependencies other than various standard c header files (such as cassert and cmath). qu3e is designed to have an extremely simple interface for creating and manipulating rigid bodies.

qu3e is of particular interest to those in need of a fast and simple 3D physics engine, without spending too much time learning about how the whole engine works. In order to keep things very simple and friendly for new users, only box collision is supported. No other shapes are supported (capsules and spheres may be added in the future if requested).

Since qu3e is written in C++ is intended for users familiar with C++. The inner-code of qu3e has quite a few comments and is a great place for users to learn the workings of a 3D physics engine.

qu3e stands for "cube", since the 3 looks slightly like the letter b and boxes (or cubes!) are the primary type of collision object.

link: https://github.com/RandyGaul/qu3e

44
English forum / OpenGL 4 Shader Subroutines - NVIDIA/OSX bug?
« on: October 17, 2014, 06:32:29 PM »
I struggled many hours recently with subroutines and I didn't manage to make them ok under OS X with a GeForce GPU.
I added a new simple demo that shows a pqtorus rendered three times, each time with a different subroutine (texture, phong, phong+texture).

The demo is available in the code sample pack:
host_api/gl-400-arb-shader-subroutine/demo_gl4_v3.xml

The latest GLSL Hacker 0.7.2.0 is recommended.

The demo works fine on Windows. On OS X, the demo works fine for Intel and AMD GPUs but not for NVIDIA GPUs. With the GT650M / OSX, the same subroutine is used to shade all meshes and that's the bug. I suspect a bug on OSX with GeForce GPUs because the demo works fine everywhere else. But since it's a NVIDIA related issue, my code can be also guilty...


OK - NVIDIA GeForce GTX 660 - R337.50 - Win7 64-bit


OK - Intel HD 4000 - OS X 10.9


OK - AMD Radeon HD 6870 - OS X 10.9 (Hackintosh)


ERROR - NVIDIA GeForce GT 650M (Macbook Retina Mid-2012) - OS X 10.9


45
English forum / GLSL Hacker 0.7.2.0 released
« on: October 17, 2014, 03:16:11 PM »
A new version of GLSL Hacker is ready, this time for all OSes (Windows, OSX and Linux) at the same time!

The PhysX plugin has been updated with the latest PhysX SDK v3.3.2

Downloads: http://www.geeks3d.com/glslhacker/download.php

The code sample pack has been updated too: http://www.geeks3d.com/glslhacker/cs/

Quote
Version 0.7.2.0 - 2014.10.17
----------------------------
! improved the FBX plugin (Windows and OS X) for loading Autodesk FBX files.
! updated the PhysX 3 plugin (Windows, Linux and OS X) with
  latest PhysX SDK v3.3.2.
+ added a new OBJ loader (for testing purposes) and a way to change
  current OBJ loader with gh_model.set_current_3d_obj_loader(name):
  "ObjLoaderV1" or "ObjLoaderV2". Default is "ObjLoaderV1".

Full changelog is available here: http://www.geeks3d.com/glslhacker/changelog.php

46
3D-Tech News Around The Web / Qt 5.4 Beta Available
« on: October 17, 2014, 02:52:46 PM »
Quote
I am extremely happy to announce that Qt 5.4 Beta is now available for download. There are a lot of new and interesting things in Qt 5.4 and I will try to summarize the most important highlights in this blog post.

Qt 5.4 brings capability to dynamically select during the application startup whether to use ANGLE or OpenGL on Windows. It is possible to use either opengl32.dll or ANGLE’s OpenGL ES 2.0 implementation in Qt applications without the need for two separate builds of the binaries. This significantly simplifies the task of creating Qt Quick applications for Windows PCs. Dynamic GL switching is not yet enabled in the prebuilt Qt 5.4 Beta binaries. In addition to these, there is a large number of smaller improvements and bug fixes for the Windows port in Qt 5.4.

...

QOpenGLContext is now able to adopt existing native contexts (EGL, GLX, …). This allows interoperability between Qt and other frameworks, such as game engines.

Link: http://blog.qt.digia.com/blog/2014/10/17/qt-5-4-beta-available/

47
English forum / GLSL Hacker 0.7.1.5 released
« on: October 10, 2014, 05:43:15 PM »
A new dev version of GLSL Hacker is ready for Windows 64-bit only

You can download it from this page:
http://www.geeks3d.com/glslhacker/download.php

Full changelog:
http://www.geeks3d.com/glslhacker/changelog.php

Changelog:
Quote
Version 0.7.1.5 - 2014.10.10
----------------------------
! updated pixel format support for image2D writing (GL_ARB_shader_image_load_store)
  PF_U8_RGBA (rgba8 in GLSL) and PF_U8_R (r8 in GLSL) added (gh_texture.image_bind()).

48
English forum / GLSL Hacker 0.7.1.4 released
« on: October 10, 2014, 05:38:03 PM »
A new dev version of GLSL Hacker is ready for Windows 64-bit, and OS X.

You can download it from this page:
http://www.geeks3d.com/glslhacker/download.php

Full changelog:
http://www.geeks3d.com/glslhacker/changelog.php

Changelog:
Quote
Version 0.7.1.4 - 2014.10.10
----------------------------
! improved shader subroutines support with several uniform subroutines per GPU program.
! improved GPU programs speed (better management of uniform variables).


Version 0.7.1.3 - 2014.10.05
----------------------------
+ added webcam_get_num() to the gh_utils lib (Lua/Python).
! improved the management of multiple webcams in the Windows version.


Version 0.7.1.2 - 2014.10.04
----------------------------
* bugfix: crash when ZOMBIE scripts are stored in separate file
  instead of in the XML.
+ added network_message_pop() to gh_utils.


49
English forum / Re: RGBA8 texture is unsuported in Compute shaders
« on: October 10, 2014, 05:29:16 PM »
It's not a bug and what you are doing is ok. It's just a pixel format for image2D that was not handled by GLSL Hacker. I added it in GLSL Hacker 0.7.1.5 (which should be downloadable in few minutes).

Now you can use PF_U8_RGBA in Lua and layout(rgba8) in GLSL. Same thing with PF_U8_R and layout(r8). I added a new demo (demo_v02_rgba8.xml).

I will add 16-bit channel texture support in one of the next versions.

Update: GLSL Hacker 0.7.1.5
http://www.geeks3d.com/forums/index.php/topic,3723.msg4470.html#msg4470


50
Quote
When it comes to AMD FirePro™ S-Series GPUs, there is never “too much of a good thing.” During my conversations at trade shows and conventions, customers have been asking for a server GPU solution that not only provides the speed to accomplish computational tasks, but they also want accuracy for precise results. Earlier this year, we announced the top-of-line AMD FirePro™ S9150 – the most powerful server GPU ever built for High Performance Computing1 measured by performance/watt.  But customers also told us they want a product for mainstream, budget conscious projects that still delivers equally exceptional performance for HPC.
 
To meet that demand, we’ve expanded our server GPU portfolio. By infusing our most current iteration of the AMD GCN architecture2 into AMD FirePro™ server GPUs, we have been able to offer two products that deliver leading edge double-precision floating-point rates, and operate with tremendous efficiency.
 
The new AMD FirePro™ S9100 server GPU is supreme in its class, with high peak single- and double-precision floating point performance, including  2.11 TFLOPS peak double-precision and up to 9.4 GFLOPS per watt TDP double-precision performance3. Compute-intensive workloads are where this card excels. With 2,560 stream processors and 12GB of GDDR5 memory, the AMD FirePro S9100 is equipped to handle supercomputing tasks. For those working in scientific compute, research, or structural analysis, it would be fair to say this GPU is a compute powerhouse. On top of the performance, the card’s maximum TDP comes in at 225W, and its passive cooling solution makes it ideal for deployment in server environments.

Complete story:
http://community.amd.com/community/amd-blogs/amd-business/blog/2014/10/06/why-amd-gpus-are-right-for-the-data-center-a-closer-look-at-amd-firepro-s9100

51
Supports the new GeForce GTX 980M and GTX 970M GPUs, based upon the second-generation Maxwell architecture:

- R344.24 Win7/Win8 64-bit
- R344.24 Win7/Win8 32-bit

53
English forum / GLSL Hacker 0.7.1.1 released
« on: September 29, 2014, 03:42:03 PM »
A new dev version of GLSL Hacker is ready for Windows 64-bit, and OS X.

You can download it from this page:
http://www.geeks3d.com/glslhacker/download.php

Full changelog:
http://www.geeks3d.com/glslhacker/changelog.php

v 0.7.1.1 changelog:
Quote
Version 0.7.1.1 - 2014.09.29
----------------------------
! updated network script updater (v0.2.0, Windows and OSX) with
  live coding checkbox and execute button.
! fixed some bugs and improved the live coding in the Windows version.
+ added auto-save of scripts and shaders if necessary (read: they have been
  modified).
+ added a button in script and shader editors to save the code source in
  the scripts/ or shaders/ folder of GLSL Hacker (Windows version only).
* script and shaders errors are now correctly redirected to the output window.
+ added an option to disable live coding in script and shader editing windows
  (Windows version only).
* fixed some potential crashes in gh_utils functions that take strings as
  parameters (like font_render() or font_render_3d().
! updated the dialog box / output window for the traces (Windows version only).
+ added circle_create(), circle_update_radius(), line_create(),
  line_set_start_position(), line_set_end_position(),
  line_set_start_color() and line_set_end_color() to gh_utils lib.
! updated the Lua immediate mode: now the Lua state of the first script
  is used (not the case so far!).
* improved the GPU monitoring plugin when running on NVIDIA Optimus systems.
+ added gh_leap (Leap Motion) library in the Python plugin.
+ updated Python plugin: gh_renderer.get_capability_4i(),
  gh_renderer.query_xxxx() (GL_ARB_pipeline_statistics_query GL4.5), gh_utils...

55
Quote
  • GL_NV_framebuffer_mixed_samples and GL_EXT_raster_multisample - target-independent multisampling control and mixed samples: this feature adds a lot of flexibility to the multi-sampled rasterization, and decouples the rasterization sampling frequency (which can be set explicitly) from the actual framebuffer storage. It enables rasterization to operate at higher sampling frequency than the number of samples in the target color render buffers, and it supports both depth and stencil testing at this frequency, if the corresponding depth and stencil buffers are sampled accordingly (it must be a multiple of the number of samples in the color buffers).
  • GL_NV_fragment_coverage_to_color - coverage to color conversion: this allows using ROP to automatically convert the post depth-/stencil-/alpha- test coverage mask into a color and write it into a color render target.
  • GL_EXT_post_depth_coverage - post-depth coverage: this extension allows the fragment shader to get the post depth-test coverage mask of the current fragment as input (gl_SampleMaskIn[]).
  • GL_NV_sample_mask_override_coverage - multisample coverage override: the standard GL behavior for FS output coverage mask (gl_SampleMask[]) is to AND it with the actual primitive input coverage mask. This extension disables this operation.
  • GL_NV_sample_locations - programmable sample locations: Allows applications to explicitly set the location of sub-pixel samples for multisample rasterization, providing fully programmable sampling patterns.
  • GL_NV_conservative_raster - conservative rasterization:  It allows rasterization to generate fragments for any pixel touched by a triangle, even if no sample location is covered on the pixel.
  • GL_NV_fragment_shader_interlock - fragment shader interlock: This extension exposes an hardware-accelerate critical section for the fragment shader, allowing hazard-free read-modify-write operations on a per-pixel basis.
  • GL_NV_fill_rectangle: screen-space bounding-box rasterization: this allows rasterizing the axis-aligned screen-space bounding box of submitted triangles, disregarding the actual triangle edges.
  • GL_NV_geometry_shader_passthrough - geometry shader passthrough: this extension allows to write more efficient geometry shaders in the case there is a one-to-one mapping between input and output primitives.
  • GL_EXT_sparse_texture2: Enhanced sparse textures: this new extension adds the ability to retrieve texture access residency information from GLSL, to specify minimum allocated LOD to texture fetches and to return a constant zero value for lookups into unallocated pages.
  • GL_EXT_texture_filter_minmax - texture Filter min/max: exposes a new sampler parameter allowing to perform a min or max reduction operation on the values sampled inside a texture filtering footprint.
  • GL_NV_shader_atomic_fp16_vector - FP16 global atomics: This extension provides a set of new atomic operations operating on 2 and 4 components vectors of 16b floating point values for images, bindless pointers to global memory and storage buffers.

Complete story: http://blog.icare3d.org/2014/09/maxwell-gm204-opengl-extensions.html

New extensions available with R344.11

56
3D-Tech News Around The Web / GALAX = GALAXY + KFA2
« on: September 19, 2014, 11:22:05 AM »
Quote
BORN FROM NEARLY TWO DECADES OF DEDICATED DESIGN AND MANUFACTURING, NOW UNIFIED AS A ONE-WORLD BRAND.

Born from nearly two decades of dedication to the design and manufacturing of the world’s best performing gaming products, GALAXY and its European brand KFA2 today announced their global unification, officially merging to form GALAX.

Originally founded in 1994, GALAXY built its reputation as the behind-the-scenes designer and manufacturer of many of the most popular OEM-branded graphic card products on the market. Since 2000, GALAXY introduced numerous critically-acclaimed products including the world renowned "Hall of Fame" series with custom white PCBs and record breaking performance.

One of the most exciting benefits of the merger for GALAX customers will be an all new level of global brand consistency and product availability. With a single unified brand, GALAX is now able to offer every customer in every market the same world class service and support along with access to GALAX’s entire product portfolio. Our customers guide everything we do at GALAX, and it’s this commitment that defines us as a company.

GALAX will continue providing all current levels of service and support to all Galaxy/KFA2 customers, and will of course honour all Galaxy/KFA2 warranties to their full extent.

Link: http://www.galax.net/europe/news_press_launch.html

58
Quote
Intel® Processor Graphics refers to the technology that provides graphics, compute, media,
and display capabilities for many of Intel’s SoC products. Within Intel, architects colloquially
refer to Intel® Processor Graphics architecture as simply “Gen”, short for Generation. A specific
generation of the Intel® Processor Graphics architecture may be referred to as “Gen6” for
generation 6, or “Gen7” for generation 7, etc. The branded products Intel® HD Graphics 4600,
Intel® Iris™ Graphics 5100, and Intel® Iris™ Pro Graphics 5200 are all derived from instances
of Intel® Processor Graphics Gen7.5 architecture. Intel® HD Graphics 5300 is the first released
product derived from an instance of Intel® Processor Graphics Gen8 architecture. This
whitepaper focuses on just the compute architecture within Intel® Processor Graphics
Gen8. For shorthand, in this paper we may use the term Gen8 compute architecture to refer
to just those compute components. The whitepaper also briefly discusses the instantiation of
Intel® Processor Graphics Gen8 within the recently released Intel® Core™ M Processor for low
power form factors. Additional processor products that include Intel Processor Graphics Gen8
may be released in the near future.


Link to the PDF: https://software.intel.com/en-us/file/compute-architecture-of-intel-processor-graphics-gen8pdf

via: http://www.geeks3d.com/20140630/intel-hd-graphics-drivers-v10-18-10-3652-with-opengl-4-3-support/comment-page-1/#comment-58676

59
3D-Tech News Around The Web / Zotac GeForce GTX 970
« on: September 11, 2014, 04:28:10 PM »

60
English forum / Re: Cant start GLSL Hacker 0.7.0.3
« on: September 08, 2014, 05:18:55 PM »
Maybe there is another plugin that perturbs GLSL Hacker startup. Have you tried to rename the plugins folder (that allows to disable all plugins)?

Pages: 1 2 [3] 4 5 ... 32