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 ... 69
GeeXLab - english forum / GeeXLab 0.20.x.x released
« on: January 09, 2018, 08:35:31 PM »
GeeXLab has been released today. Main new features:
- PhysX 3.4.1 + GRB (Windows + Linux)
- ASUS Tinker Board support


Release notes:

GeeXLab - english forum / Re: GeeXLab 0.19.x.x released
« on: January 09, 2018, 05:35:44 PM »
Sorry, there is no offline doc available.

But you can read these short articles to get an idea of how GeeXLab works:


You can also test and modify the learn code samples available in the full code sample pack (in the learn/ folder).

And the reference quide for all functions:

3D-Tech News Around The Web / AMD CPU/GPU Roadmap 2018
« on: January 09, 2018, 04:13:09 PM »
Roadmap for 2018:
- CPU: Ryzen desktop APU (CPU + GPU), Ryzen 2nd generation CPU
- GPU: Vega 64/56 for desktop and Radeon Vega for notebook

Roadmap for 2020:


Intel unveiled its new processors that pack a Radeon RX Vega GPU (the pGPU), an Intel HD 630 GPU (the iGPU) and a 4C/8T CPU on the same chip.

Intel Core i7-8809G
- CPU: Kaby Lake, 4C/8T, 3.1GHz (base) / 4.2GHz (boost)
- pGPU: Radeon RX Vega M GH, 24 CUs (1536 shader cores)
- pGPU clock speed: 1063MHz (base) / 1190MHz (boost)
- pGPU memory: 4GB HBM2, 1024-bit
- iGPU: Intel HD 630
- Package TDP: 100W

Intel Core i7-8709G
- CPU: Kaby Lake, 4C/8T, 3.1GHz (base) / 4.1GHz (boost)
- pGPU: Radeon RX Vega M GH, 24 CUs (1536 shader cores)
- pGPU clock speed: 1063MHz (base) / 1190MHz (boost)
- pGPU memory: 4GB HBM2, 1024-bit
- iGPU: Intel HD 630
- Package TDP: 100W

Intel Core i7-8706G
- CPU: Kaby Lake, 4C/8T, 3.1GHz (base) / 4.1GHz (boost)
- pGPU: Radeon RX Vega M GL, 20 CUs (1280 shader cores)
- pGPU clock speed: 931MHz (base) / 1101MHz (boost)
- pGPU memory: 4GB HBM2, 1024-bit
- iGPU: Intel HD 630
- Package TDP: 65W

Intel Core i7-8706G
- CPU: Kaby Lake, 4C/8T, 3.1GHz (base) / 4.1GHz (boost)
- pGPU: Radeon RX Vega M GL, 20 CUs (1280 shader cores)
- pGPU clock speed: 931MHz (base) / 1101MHz (boost)
- pGPU memory: 4GB HBM2, 1024-bit
- iGPU: Intel HD 630
- Package TDP: 65W

Intel Core i5-8305G
- CPU: Kaby Lake, 4C/8T, 2.8GHz (base) / 3.8GHz (boost)
- pGPU: Radeon RX Vega M GL, 20 CUs (1280 shader cores)
- pGPU clock speed: 931MHz (base) / 1101MHz (boost)
- pGPU memory: 4GB HBM2, 1024-bit
- iGPU: Intel HD 630
- Package TDP: 65W


GeeXLab - english forum / Re: simple Lua framework over GeeXLab Lua API
« on: January 06, 2018, 07:33:26 PM »
Sorry for the delay, I've worked on the new version of GeeXLab and as usual it took lot of time. I will add the small framework in the code sample...
And happy new year too!

3D-Tech News Around The Web / Vulkan API specifications 1.0.67 released
« on: January 06, 2018, 07:30:18 PM »
Change log for January 5, 2018 Vulkan 1.0.67 spec update:

  * Bump API patch number and header version number to 67 for this update.
  * Update copyright dates to 2018

Github Issues:

  * Fix texture lookup functions in `GL_KHR_vulkan_glsl` specification
    (public pull request 363).
  * Clarify the state waited semaphores are left in when a call to
    flink:vkQueuePresentKHR fails (public issue 572).
  * Cleanup descriptions of slink:VkObjectTablePushConstantEntryNVX and
    slink:VkObjectTableDescriptorSetEntryNVX (public issue 583)
  * Remove redundant flink:vkCmdSetDiscardRectangleEXT valid usage
    statements (public pull 586).
  * Make dynamic state array length valid usage statements implicit for
    flink:vkCmdSetViewportWScalingNV, flink:vkCmdSetDiscardRectangleEXT, and
    flink:vkCmdSetViewport (public pull 589).
  * Clarify meaning of window extent (0,0) in slink:VkSwapchainKHR for the
    Windows and X11 platforms, in their respective extensions (public issue
  * Allow flink:vkGetPastPresentationTimingGOOGLE to return
    ename:VK_INCOMPLETE (public issue 604).
  * Add synchronization valid usage statements to flink:vkAcquireNextImage
    (public pull 611).
  * Fix some broken external links and internal xrefs (public pull 613).
  * Clean up slink:VkViewport valid usage statements in the presence or
    absence of relevant extensions (public pull 623).
  * Remove
    token from VK_KHR_maintenance2 from the non-extension VU path for
    slink:VkGraphicsPipelineCreateInfo (public issue 628).
  * Miscellaneous minor markup fixes - extension name strings (public pull
    631), Notes (pull 633), queue names emitted by generator scripts (pull
    634), block formatting in slink:VkDescriptorUpdateTemplateEntryKHR (pull
    641), quotes and apostrophes (pull 643),
  * Miscellaneous minor grammar fixes (public pull 644).
  * Fix markup macros so usage like ptext:*Src* works (public pull 647).

Internal Issues:

  * Clarify in the `VK_KHR_surface` and `VK_KHR_swapchain` extensions that
    parameter combinations which aren't supported for normal images are also
    unsupported for presentable images, even if the parameter values are
    individually supported as reported by the surface capability queries
    (internal issue 1029).
  * Fixed XML typo in the valid value field of the pname:sType member of
    slink:VkPhysicalDeviceExternalMemoryHostPropertiesEXT (internal issue

Other Issues:

  * Add memory semantics validity rules to the <<spirvenv-module-validation,
    Validation Rules within a Module>> section of the SPIR-V environment
    appendix, and specify that sequentiality consistency is not supported.
    This forbids certain cases like "`Load+Release`" that we don't expect to
    ever be meaningful.
  * Document mapping of OpenGL Shading Language barriers to SPIR-V scope and
    semantics in the `GL_KHR_vulkan_glsl` specification.

New Extensions:

  * `VK_EXT_conservative_rasterization`


Geeks3D's GPU Tools / Re: GPU Caps Viewer 1.37.0 released
« on: January 03, 2018, 08:00:38 PM »
CPU OpenCL demos are supported by AMD and Intel only. NVIDIA does not provide CPU OpenCL and supports only GPU OpenCL.  You can try to install the AMD OpenCL SDK on you system to get software OpenCL.

Snelly is a system for physically-based SDF (signed distance field) pathtracing in a web browser.

A Snelly scene consists of 3d objects defined by a mathematical signed distance function (SDF) written in GLSL code. The SDF gives the distance to the surface from any given point in space, where the distance is positive in the exterior of the shape and negative in the interior (and of course zero on the surface). In each scene there can (currently) only exist three such specified objects, with different rendered material properties: a Metal, a Dielectric, and a general purpose plastic-like Surface ("uber" material). These three materials can freely intersect and embed one another.

It is generally quite challenging to find SDF functions which correspond to interesting shapes. We provide some example scenes (and this library of sample scenes will be added to over time). A lot of interesting examples and resources can be found on the web, at for example shadertoy. Fractal surfaces in particular are quite easy to define as SDFs, as described for example here.


3D-Tech News Around The Web / Reverse Z Cheat Sheet
« on: January 03, 2018, 07:53:08 PM »
Just recently I’ve started looking into ways to optimize the depth buffer precision for large draw distances and one specific approach caught my eye again. A technique which is now commonly referred to as Reverse Z. While it comes with just a few minor changes, the results can be quite considerable.

That considerable everyone should just go ahead and use it.

The general idea itself is actually very simple: instead of mapping the interval between the near and far plane [zn,zf] to [0,1], a special projection matrix is constructed in a way that it is being mapped to [1,0] instead.

Why this actually increases the depth buffer precision is not directly obvious, but I will also not go into detail here. I’ve added some references to articles on this topic at the end of this post.


Here are some details posted on Intel India website (now removed) about the Intel Core i7 8809G, a quad-core CPU (4C/8T) that comes with two GPUs: an Intel HD 630 (iGPU) and a Radeon Vega (pGPU):


3D-Tech News Around The Web / Re: (Shadertoy) Turn n' Burn
« on: December 12, 2017, 08:14:01 PM »
I ported this demo to GeeXLab few days ago. It's available in the full code sample pack here:


The demo needs several seconds (around 10 sec) before displaying the first frame...

3D-Tech News Around The Web / Re: HWiNFO32 + HWiNFO64 v5.70
« on: December 12, 2017, 08:00:17 PM »

Geeks3D's GPU Tools / Re: FurMark 1.19.1 released
« on: December 11, 2017, 07:33:20 PM »
Here's the problem in detail.

I have a laptop with i7-8700K CPU, Nvidia GTX 1070 graphics card and windows 10.
The Furmark v1.19.1.0 will show me a 140 FPS and a 99% of GPU usage after running stress test.
After I ran a CPU stress test together as the GPU burner bulid-in Furmark app.(or Prime95)
The FPS and GPU usage will immediately drop down to 108 FPS and 65% of GPU usage.
In the meantime, No any limit reasons were found in Furmak.
But it will come back to normal as running Furmark alone if I reset the affinity(CPU) in the device manager.

Yes there a minor thread affinity issue in the current FurMark but it will be fixed in the next release.

Geeks3D's GPU Tools / Re: FurMark 1.19.0 released
« on: December 11, 2017, 07:24:18 PM »
Any plans in including artifact scanner into the new releases like ROG FurMark has? I would love this.

Currently, the artifact scanner feature is only for graphics cards makers (ASUS, EVGA, MSI, etc.). Maybe in the future...

Geeks3D's GPU Tools / Re: FurMark 1.19.1 released
« on: December 11, 2017, 07:22:12 PM »
Why is there no database on Furmark 1.19.1 for 4k resolution. I have a very decent score on my RX Vega 56 of 7555 and would like to know how that stands comparitively to other 4k resolution scores.

There is a link at the bottom of the FurMark scores page. But most of scores seem fake. I just re-benchmarked my GeForce GTX 1080 and I get a score of 2790 points (47 FPS) for preset P2160 (3840x2160). I haven't tested the RX Vega 56 but 7555 seems a very high score for a single Radeon. Maybe it's a crossfire score?  For the preset P1080, I get 7295 points (122 FPS) with the GTX 1080.

I will update asap this page with 4k scores.

The 14TB models use an innovative 9-disk, helium-sealed design to deliver massive capacity that fits into standard 3.5 inch SATA drive bays.

The MG07ACA Series utilizes Toshiba precision laser welding technology to seal helium inside the 9-disk mechanics for the life of the drive.  The helium-sealed design reduces aerodynamic drag to deliver a lower operational power profile, helping to achieve TCO objectives for cloud-scale and software-defined data center infrastructure.
The massive 14TB and 12TB model capacity is delivered using proven CMR/PRM recording technologies providing optimum compatibility and data reliability.  Equipped with SATA 6.0 Gbit/s interface, the MG07ACA models integrate easily into standard 3.5-inch SATA drive bays to help reduce the footprint and operational burden of cloud-scale storage infrastructure, and business critical servers and storage systems.


Irvine, CA – December 7, 2017 – Toshiba Electronic Devices & Storage Corporation today announced the launch of the MG07ACA Series, the world’s first1 enterprise 14TB2 Conventional Magnetic Recording (CMR) HDD. Using a 9-disk, helium-sealed design, the new MG07ACA Series provides the power-efficient capacity and storage density needed by cloud-scale and enterprise storage solution providers to achieve their TCO objectives.

“We have raised the bar with the new MG07ACA Series 9-disk helium-sealed design,” said Akitoshi Iwata, Vice President of Storage Products Division, Toshiba Electronic Devices and Storage Corporation. “By utilizing an innovative design, we continue to improve the benefits that high-capacity disk storage can deliver to our broad global customer base.”

The MG07ACA Series features both 14TB 9-disk and 12TB 8-disk models. The helium-sealed 3.5-inch3 mechanical design realizes better storage density and a lower HDD operating power profile than the previous MG06ACA Series for optimal TCO in cloud-scale infrastructures. The series also utilizes Toshiba Group’s laser welding technology to ensure the helium remains securely sealed inside the drive enclosure. The drives support a SATA 6Gbit/s interface and 7,200rpm access performance. The 9-disk 14TB models achieve a 40% increase in maximum capacity over previous MG06ACA 10TB models. Additionally, the 14TB models improve power efficiency by over 50% (W/GB)4.

“Toshiba’s first helium-sealed nearline drive intercepts the market at a class-leading 14 TB capacity with CMR,” said John Chen, industry analyst at Trend Focus. “Its early time-to-market for this capacity positions the company well to meet the storage needs of large hyperscale and cloud companies. Additionally, the company’s choice of a 9-disk platform paves the way to achieving higher capacities in future product generations.”

“While enterprise server and storage customers realize that shingled magnetic recording (SMR) technology can improve HDD capacity, the adoption of SMR HDD products into server and storage systems is a transition that will take several years,” said John Rydning, Research Vice President for hard disk drives at IDC. “Toshiba’s new helium-sealed enterprise HDD is the world’s first 14TB of storage capacity using conventional rather than shingled magnetic recording technology, giving enterprise customers the highest capacity HDD available in the market today for existing server and storage system architectures.”


Are you curious about how Oats Studios created such high-fidelity sets, characters and costumes in real time for ADAM: The Mirror and ADAM: Episode 3? Read our in-depth, behind-the-scenes blog posts on lighting, Alembic support, clothing simulation, Timeline, shaders, real-time rendering and more.

- Lighting tips & tricks in the ADAM films
- ADAM: Episode 2 @ youtube
- ADAM: Episode 3 @ youtube

Sad news, everyone.

After months of deliberation, I have decided to stop the development of Classic Shell.

It was a great adventure for me that lasted 8 fun and exciting years. It started back in 2009 as a simple weekend project and over time grew to be a hugely popular software brand with many millions of downloads. It was in large part due to all of you, the active Classic Shell community, that reported issues, suggested features, provided translations, developed new skins and participated in forum discussions. My thanks go to you for your continued support and encouragement.

And of course, special thanks to Gaurav Kale, who stuck with me from the early days to the end. His passion for all things Windows was instrumental to the Classic Shell success.

There were few factors that led to my decision:
1) Lack of free time. I have other hobbies that demand my time, some related to software and some not. It requires a lot of effort to add new major features to Classic Shell and keep it relevant. Even keeping it running on newer version of Windows is a lot of work. That leads me to point #2

2) Windows 10 is being updated way too frequently (twice a year) and each new version changes something that breaks Classic Shell. And

3) Each new version of Windows moves further away from the classic Win32 programming model, which allowed room for a lot of tinkering. The new ways things are done make it very difficult to achieve the same customizations

- Download latest Classic Shell 4.3.1:

3D-Tech News Around The Web / C++ 17 new features overview
« on: December 08, 2017, 09:03:56 AM »
C++ language is constantly evolving, and for us, as for developers of a static analyzer, it is important to track all its changes, in order to support all new features of the language. In this review article, I would like to share with the reader the most interesting innovations introduced in C++17, and demonstrate them with examples.


Pages: 1 2 [3] 4 5 ... 69