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 ... 88 89 [90] 91 92 ... 95
take a GTS 250, it's not expensive and can run latest CUDA demos  ;D


Ubuntu 9.10 had its beta release last week and the final release is coming in just three weeks, but this late in the release cycle, it has been decided to pull in the final Mesa 7.6 version. Pulling in this newer code that provides OpenGL acceleration on Linux provides numerous bug-fixes along with support for a greater number of OpenGL extensions on different hardware and other improvements.

3D-Tech News Around The Web / Exploring the Mandelbrot Set with your GPU
« on: October 08, 2009, 11:02:07 AM »

The Mandelbrot set is a good candidate for GPU implementation: it's simple, uses vector math, and is embarrassingly parallel.  Doing so, however, generally adds a significant amount of complexity to a program.  The programmer is forced to twist all computations to fit within the graphics pipeline, and a significant amount of ceremony surrounds even the simplest operation. 

The GPGPU framework in Penumbra is designed to get hide away all these rough edges, with the goal of making it easier to use the GPU for these sorts of tasks than it would be to use plain Clojure.  It's not quite there yet, but in this particular example it proves to be extremely useful.

3D-Tech News Around The Web / The Black Art of Optimising
« on: October 08, 2009, 10:56:52 AM »

Optimising program code for maximum performance is often considered to be some sort of black art. Also, quite a number of programmers believe in micro-optimisations that rarely ever improve program execution time, but nearly always introduce subtle bugs or obfuscate the code, or both.

In the following we will have a look at optimisation possibilities that help cut down execution time considerably while still keeping the code clear and easily maintainable.

3D-Tech News Around The Web / ATI Cypress GPU and Architecture analysis
« on: October 08, 2009, 09:18:25 AM »

The guys at Beyond3D offer us a deep investigation into what makes Cypress a powerfull GPU.

Ah, the winds of change are once again upon us. A new DirectX iteration is about to enter the fray with Windows 7, and new, fully compliant hardware is there to bring its glory to screens everywhere. One such piece of hardware is ATI's new Cypress GPU.

We'll get into the details shortly, but it's a fully DX11-compliant beast that is the result of countless man hours and has made more than one engineer gain a few extra grey hairs.

3D-Tech News Around The Web / Python is Unix
« on: October 08, 2009, 08:34:11 AM »

Ryan Tomayko’s [I like Unicorn because it’s Unix] should be required reading for anyone doing anything involving networks or unixes these days. Like Ryan, I share a deep appreciation for the dark art of Unix system calls, and like Ryan I’m a bit dismayed to see them relegated to the dusty corners of our shiny dynamic languages.
Since I’m a Python hacker, though, and since I had a couple minutes, I thought it’d be an interesting exercise to port Ryan’s code to Python. So, with no further ado, here we go:

General Discussion / GeeXLab 0.1.10 available
« on: October 08, 2009, 08:27:22 AM »

General Discussion / Re: porting from demoniak3d
« on: October 07, 2009, 08:14:34 PM »
The new update of GeeXLab should solve some of your problems:

I added 3 new examples:

1 - sound phaser: for your sound effect "zaap". Just load once all audio files you need, depending on the type of sound, set the play and loop params. As soon you need to produce a sound, just call HYP_Sound.Play(). Look at the demo. If you press the SPACE key, you'll heard the phaser for around 1 second.

2 - Draw line: there are two demos that draw a line. The first demo draws a line using a primitive line as you did. That works and it's simple. But I recommend you to use the second technique: immediate mode. In this mode you talk directly with the renderer and then you have a full control over what's going on.

3 - Cloning model: there was a bug in mesh cloning code so cloning a model was impossible. Now it's ok. But if your models are simple (one sub mesh only), I recommend you to use mesh instancing to manage huge amount of instances. Just remove the sub mesh from the model hierarchy (see this code sample) and use mesh instancing techniques like in this code sample.
I added in GeeXLab 0.1.10 one useful function for instances management:
Code: [Select]
HYP_Mesh.SetInstanceRenderState(instanceIndex, state)


General Discussion / Re: porting from demoniak3d
« on: October 06, 2009, 04:41:29 PM »
Nice news Alex!
Don't hesitate to post some screenshots (or send me them).

For what do you use HYP_Object.Clone()? for Model? 

In GeeXLab, transparent surfaces have a material with an opacity < 1.0. That was not necessary in Demoniak3D.
So check this detail and let me know.

I will release this evening an update that includes a bugfix for mode bounding volumes. Maybe that will solve your problem.

General Discussion / Re: introducing crew
« on: October 05, 2009, 11:43:31 AM »
nice intro post SPRINGER  ;D


The demo that we were shown behind closed doors was of course an MSI Big Bang motherboard that was running Windows 7 and ATI Radeon HD 4890 and GeForce GTX 260 graphics cards at the same time with both cards being load balanced. Notice that the Lucid HYDRA 200 load balancing solution has no interconnects, bridges, external dongles or adapters that are needed. Both cards were running fine and we were able to change which card had the monitor output on it from the Windows 7 desktop, which was pretty slick. Lucid gave us a chance to try out PC Game titles F.E.A.R. 2: Project Origin and Bioshock to see what the future of scalable graphics looks like with the Hydra 200. We must admit that we were fairly impressed with what we saw as the technology appeared to be working without any issues on these two game titles...

3D-Tech News Around The Web / ATI RV740 GPU and Architecture Analysis
« on: September 28, 2009, 11:39:45 AM »


The HD 5870 already measures in at an unheard of 11″ long, bigger than what a lot of case manufacturer’s have budgeted for (standard spacing is 10.5″).  Images obtained this morning seem to depict that the x2 version could be even bigger!

While it is difficult to get an accurate length just from pictures, with a second die , it would be nearly impossible to squeeze it into the same sized PCB as the 5870. From our basic visual calculations, the card appears to be extended by at least 1″ – 1.5″, giving it a total footprint of an estimated 12″-12.5 inches long. Best of luck fitting this behemoth into any mid sized tower, or even a smaller full tower.


This multi-GPU beast more appropriately known as Radeon HD 5870 X2 was shown in a meeting room running CryTek’s latest CryEngine 3 development build at very smooth framerates. More specifically, the engine was running at 1920x1200 with AA turned off (as CryEngine has never needed manual AA) and the Screen Space Ambient Occlusion (SSAO) rendering technique was significantly more pronounced than in Crysis and Crysis Warhead.

3D-Tech News Around The Web / AMD Demos Dual Cypress Graphics Card - Hemlock
« on: September 28, 2009, 08:06:25 AM »

While the enthusiasts are getting crazy over the Radeon HD 5870 and Radeon HD 5850 cards, AMD keeps up the hype by showcasing us a machine fitted with a Hemlock graphics card. Due to launch later this year in Nov-Dec, a Hemlock card has two "Cypress" ASICs on it and it is priced below US$500.

3D-Tech News Around The Web / Sapphire Radeon HD 5870 1GB in CrossFire
« on: September 24, 2009, 09:00:53 PM »

We're not going to blabber on too much here, because we're sure you don't want to read about us having two HD 5870s. You just want to know what happens when we put those two HD 5870s into the one system and run our flurry of benchmarks.


We loved the HD 5870 in single form and we love it even more now. CrossFire manages to scale extremely well and if this is an idea of what we can be expecting from the HD 5870 X2, then NVIDIA had better be worried! Once companies like Sapphire start attacking the model with new cooling and overclocks, it's only going to get better for the model.

3D-Tech News Around The Web / Larrabee: first public demonstration
« on: September 23, 2009, 11:11:14 AM »

Intel has given the first ever public demonstration of its long-awaited Larrabee graphics system.

"Larrabee gives you a fully programmable rendering pipeline. So you can use DirectX, or OpenGL - or your own pipeline."

"These reflections here are just ten lines of code," he stated. "To do something like this on a conventional GPU would be really quite painful."


Developed by LucidLogix and dubbed HYDRA, the technology was set to revolutionise multi-GPU graphics by eliminating the need for specific, limited driver support from either ATI or NVIDIA in the form of CrossFireX or SLI, respectively. Rather, sitting in-between the DirectX API and graphics cards, Lucid employed a manufacturer-agnostic chip and associated software that latched on to the PCIe bus to mete out the necessary data to two-, three-, and four-card combinations.

So, a year or so later, LucidLogix is back with a newer HYDRA 200 chip. This time around, however, the company has leveraged the support of a tier-one motherboard vendor, and we'll be seeing the HYDRA-powered board in a few weeks' time.

See also:


There are several key features in DirectX 11 that will make graphics on the screen look closer to reality. Tessellation is used to increase the polygon count in an image. The more polygons, the more realistic the image will look. Gone will be the days of blocky looking characters as the polygon count will increase significantly with the next generation of DirectX 11 hardware.

Multithreaded rendering will allow Direct3D processes to run across multiple CPU cores. Most games are using dual core CPUs, but multithreaded rendering will make gaming on triple and quad cores finally worth the cost. A faster processing pipeline and increased scaling are only some of the benefits.

DirectCompute allows access to the shader cores and pipeline. It allows for non-proprietary physics implementations, which some open-source physics projects are looking to take advantage of. Video transcoding will also take a significant leap due to access to the many processors on a modern GPU.

Pages: 1 ... 88 89 [90] 91 92 ... 95