NVIDIA's upcoming GeForce RTX 5060 Ti performance segment graphics card appeared in alleged Furmark online database entries. These reference the card by its device ID, 10DE-2D04. There are three instances of the card in the database, one of which sees the benchmark being run in a remote desktop session, where the card scores 5634 points or 93 FPS on average, but with its core boosting only up to 1238 MHz. The second, more plausible result sees the card score 10242 points or an average of 170 FPS, with its core going up to 2656 MHz. Both these tests are run at 1080p. The third result sees the GPU score 4411 points or 73 FPS, but at 4K Ultra HD resolution—again, plausible given its score with 2656 MHz maximum boost. The second result has Furmark read the card's maximum TDP to be 180 W.

The GeForce RTX 5060 Ti debuts the new "GB206" GPU, which probably has 36 or 40 SM. The RTX 5060 Ti is configured with 36 SM for 4,608 CUDA cores, 144 Tensor cores, 36 RT cores, 144 TMUs, and an unknown number of ROPs (probably 64). The GPU has a 128-bit wide GDDR7 memory interface, and comes with 16 GB and 8 GB memory variants. We gather from these Furmark screenshots that the GPU boosts over 2600 MHz, and it's possible that the memory ticks at 28 Gbps, yielding 448 GB/s of memory bandwidth. The 180 W TDP shown in one of these screenshots means that some if not all custom-design RTX 5060 Ti cards could feature 8-pin PCIe power connectors (150 W from the connector, plus 75 W from the slot). All rumors point to a mid-April launch of the RTX 5060 Ti.

Although it has only been a few days since the RDNA 4-based GPUs from Team Red hit the scene, it appears that we have already been granted a first look at the 3D Mark performance of the highest-end Radeon RX 9070 XT GPU, and to be perfectly honest, the scores seemingly live up to our expectations - although with disappointing ray tracing performance. Unsurprisingly, the thread has been erased over at Chiphell, but folks have managed to take screenshots in the nick of time.

The specifics reveal that the Radeon RX 9070 XT will arrive with a massive TBP in the range of 330 watts, as revealed by a FurMark snap, which is substantially higher than the previous estimated numbers. With 16 GB of GDDR6 memory, along with base and boost clocks of 2520 and 3060 MHz, the Radeon RX 9070 XT managed to rake in an impressive 14,591 points in Time Spy Extreme, an around 6,345 points in Speed Way. Needless to say, the drivers are likely far from mature, so it is not outlandish to expect a few more points to get squeezed out of the RDNA 4 GPU.

FurMark's development team, Geeks3D, seems to be relieved after work was completed on version 2.1.0's public release—according to release notes: "It took me more time than expected but it's there!" The Beta version was made available back in December 2022 (through Geeks3D's Discord)—a milestone achievement for the Furmark dev team, since no major updates had been implemented since 2007. The GPU stress test and benchmarking tool was improved once again—last August, when the Beta was upgraded to v2.0.10.

Main author, JEGX, provided a little bit of background information: "FurMark 2 is built with GeeXLab. The GUI is a pure GeeXLab application while the furmark command line tool is built with the GeeXLab SDK. GeeXLab being cross-platform, this first version of FurMark 2 is available for Windows and Linux (the Linux 32-bit version is also available, I will re-compile it for the next update). I plan to release FurMark 2 for Raspberry Pi (I just received my Raspberry Pi 5 board!) and maybe for macOS too." He states that feedback is welcome, and requests for OpenGL 2.1 and 3.0/3.1 support will be considered. The full timeline of changelog updates can be found here.

The Geeks3D team has been working quietly on a new version of FurMark, a highly popular benchmarking application as used by many in the enthusiast overclocking community. The GPU stress test and benchmarking tool is set to receive its first major upgrade (public release) since 2007, although a beta version has been made available to download since December 2022—accessible via the Geeks3D Discord group. A late August update had the beta reaching a new milestone in version 2.0.10—fixing various bugs, expanding GPU support and adding more resolution options.

The new version of FurMark supports OpenGL or Vulkan graphics APIs, and its three preset benchmark options include 1080p, 1440p, and 2160p resolutions. Users can also customize parameters for different graphical and resolution settings. A results database is already active, but Tom's Hardware notes that it consists of data submitted by beta users, thus it does not log: "results from the previous version of the app." The development team has teased a loose later this month launch window for FurMark 2.0 in public form.

PC enthusiast and overclocker VIK-on pulled off a daring memory chip mod on his Palit GeForce RTX 3070 GamingPro OC graphics card, swapping its 8 GB of 14 Gbps GDDR6 memory with 16 GB of it, using eight replacement 16 Gbit chips. The modded card is able to recognize the 16 GB of memory, is able to utilize it like other 16 GB graphics cards (such as the Radeon RX 6800), and is fairly stable with benchmarks and stress tests, although not initially stable. It did spring up some black-screens. VIK-on later discovered that locking the clock-speeds using EVGA Precision-X stabilizes the card, so it performs as expected.

The mod involves a physical replacement of the card's stock 8 Gbit memory chips with 16 Gbit ones; and shorting certain straps on the PCB that let it recognize the desired memory chip brand and density. After the mod, the GeForce driver and GPU-Z are able to read 16 GB of video memory, and the card is able to handle stress tests such as FurMark. The card was initially underperforming in 3DMark, putting out a TimeSpy score of just 8356 points; but following the clock-speed lock fix, is able to score around 13000 points. The video presentation can be watched from the source link below. Kudos to VIK-on!

PowerColor today released an optional "Unlocked BIOS" for its Radeon RX 480 RED Devil graphics card. Available through PowerColor DevilClub (and TechPowerUp), the BIOS is optional because it is intended only for power-users. The BIOS increases power-limits on the card, facilitating higher overclocking headroom and clock speed sustainability.

The obvious trade-offs here are higher power-consumption, temperatures, and noise. In fact, PowerColor strongly recommends against using the modded card with FurMark as the company found that it will damage even non-overclocked cards with the new BIOS. PowerColor, however, stated that flashing your card with the optional BIOS will not void your warranty.

DOWNLOAD: PowerColor RX 480 Unlocked BIOS

Geeks3D released the latest version of every GPU's worst nightmare, FurMark. Version 1.10 adds formal support for real-time monitoring of frequencies and power figures for NVIDIA's GeForce GTX 680. It also includes an improved interface, relocation of some key settings, support for new resolutions, and support for GPU-Z 0.6.0 API.DOWNLOAD: Geeks3D FurMark 1.10

The complete change-log from the developer follows.

The enthusiast community's favourite trial-by-fire for graphics cards, FurMark, has been updated to versions 1.8.5 and 1.9.1. Its developers maintain two branch versions of the program, since version 1.8 is popular, and 1.9 is the latest, bleeding-edge in the suite of stress test routines it has. FurMark is an OpenGL-based 3D graphics hardware stress-testing and benchmarking program. Its stress-testing functions are widely used to test stability of overclocked graphics hardware, and is dreaded by GPU manufacturers.

Version 1.8.5 is a maintenance release, which packs improved score submission, an updated number of window resolutions available, and improved graphics hardware detection using an updated ZoomGPU code. The FurMark 1.7 mode is removed. A bugfix filters the spikes in temperature graphs. Minimizing the window now can't be done in full-screen mode. Version 1.9.1, on the other hand, adds support for new AMD and NVIDIA GPUs with updated detection code, supports GPU-Z 0.5.1 and GPU Shark 0.5.1; added a workaround for a bug in AMD Catalyst GLSL compiler, which led to wrong lighting of the furry donut.DOWNLOAD: FurMark 1.8.5 and 1.9.1

Internationally renowned graphics card and mainboard manufacturer MSI officially launches the N480GTX Twin Frozr II graphics card. Built on the latest 40nm manufacturing process, the N480GTX Twin Frozr II is equipped with 1536MB of high-speed GDDR5 memory. The exclusive, award-winning Twin Frozr II thermal design meets the performance and thermal demands of graphics enthusiasts.

Along with various exclusive technologies from NVIDIA such as 3D Vision Surround, PhysX, and CUDA, the N480GTX Twin Frozr II allows games to be played at the highest resolutions. The Twin Frozr II dual-fan thermal design from MSI reduces temperature by 14°C and noise by 4dB when compared to the the reference design. Without a doubt, the N480GTX Twin Frozr II is the only and best solution for superb performance, low operating temperature, and quiet performance.

A Chinese tech community MyMyPC.com member pictured, and tested a Radeon HD 5750 graphics accelerator, which is intended to be AMD's flagship product in the sub $150 market segment. Unlike the Radeon HD 5770, which is pictured with a different cooler that resembles that of the Radeon HD 4770 (AMD reference), and Radeon HD 3870, this card makes do with a GPU cooler essentially similar to that of the more common Radeon HD 4770 cooler design, except for a few changes with the fan and the cooler shroud. Under the black, egg-shaped shroud is a heatsink with radially-projecting fins, in which is nested a red 80 mm fan. Everything else on the card, relies on its air-flow.

Nearly a year since its first release, the creators of FurMark have released a new version of their Physx based FluidMark, which like Furmark can be used as either a stress test or benchmarking tool. FluidMark is designed to run in hardware mode for all CUDA ready NVIDIA cards, which includes the 8, 9 and 200 series of graphics cards. For all others the program runs in software mode, which can be forced even if your system has one of the above mentioned cards. The creators say in software mode, the tool "is an excellent CPU benchmark." The latest graphics drivers and Physx version 9.09.0428 is recommended to be installed prior to running this. For more information and to download go here

FluidMark 1.1.0 changelog:

• New: simple color point rendering mode (press the R key in stability test).
• New: use the code base than FurMark.
• New: command line parameter (/no_sph) to disable SPH fluid simulation method.
• Change: compiled with NVIDIA PhysX SDK 2.8.1.
• Bugfix: minors bugs fixed

Nearly four months after its previous version, the chaps at oZone3D have released Furmark 1.7.0. This release packs a host of nifty new features, and a number of bug fixes. For starters, FurMark is able to work along with GPU-Z to provide real-time readings on the graphics card's temperatures, voltages and VDDC current (for cards that support it). An experimental feature allows you to Twit your score onto your Twitter account. While the stability test or benchmark is running, the main GUI stays minimized, so you needn't have to start another instance to run several tests.

With multiple GPUs doing the rendering, each GPU is given its own temperature graph. You can start or stop the rendering by hitting the space key without having to close the window. A number of new resolutions have been added, and the application is now also available in Castilian, Bulgarian, Polish, Slovak, and Spanish, thanks to translations. Issues relating to temperature updates in the graph, and the application's multithreading management are resolved. Give your graphics cards a sunbath.DOWNLOAD: FurMark 1.7.0