TweakTown NewsRefine News by Category:
The leaks of AMD's upcoming GPUs continues, with a full line up of SAPPHIRE products leaked. VideoCardz.com has picked it up, where SAPPHIRE is expected to have many cards to offer consumers.
Starting with the Hawaii XT-based Radeon R9 390X Tri-X OC 8GB, followed by the Hawaii PRO-based Radeon R9 390 Nitro 8GB. The R9 390X is expected to feature a new Tri-X cooler, with a factory overclock that is sure to impress. The R9 390 Nitro is a new brand for SAPPHIRE, which will also feature 8GB of VRAM and the Tri-X cooler, but the Hawaii PRO core.
Under these two flagship cards we have the Radeon R9 380 Nitro, Radeon R9 380 ITX, Radeon R7 Nitro, Radeon R7 370 and Radeon R7 360. SAPPHIRE will use a mix of its Dual-X and stock cooling on these cards. We should expect to see SAPPHIRE unveil these cards in the coming weeks.
We are getting closer and closer to the launch of the next generation video cards from AMD with each passing day, where today some leaked specifications on the Radeon Fury X, of which we had a world exclusive on last week. It was only hours ago that we posted some OpenCL benchmarks of the Fury X, where it managed to keep up with and beat the Titan X.
The Radeon Fury X will reportedly feature 4096 stream processors, 64 GCN Compute Units, 128 ROPs, 256 TMUs, 4GB of HBM on a 4096-bit memory bus, a GPU clock speed of 1050MHz or more, and an effective memory bus of 1GHz, providing around 512GB/sec of memory bandwidth. All of these specs of the purported Fury X have it being around 54% faster than the Hawaii XT-based Radeon R9 290X, and around 48% more power efficient.
Speaking of power efficiency, the Fury X has a TDP of 300W with around 28.7 GFLOPS/watt of performance versus the 19.4 GFLOPS/watt that the Radeon R9 290X has on its 290W TDP. Overall performance has the Fury X capable of over 8.6 TFLOPS of performance, blasting out the 5.6 TFLOPS that the Radeon R9 290X managed.
With Star Wars: Battlefront shaping up to be one of the biggest releases of the year, it shouldn't come as too much of a surprise to hear the rumor that AMD will be bundling the game with its Fury branded video cards.
AMD will reportedly be releasing the HBM-powered Radeon Fury X and Radeon Fury cards at E3 2015 next week, but what better way of selling their flagship cards than by bundling one of the biggest games coming out this year. As for the Radeon 300 series, which are rebrands of the Radeon 200 series, they will reportedly include a copy of the new Dirt game coming out later this year.
With the leaked benchmarks from quite a while ago most likely out of date now, and with our world exclusive reveal on the Fury X branding from AMD, we're seeing some leaked OpenGL performance numbers from CompuBench.
The leak shows that the Fiji XT-based, HBM-powered Radeon Fury X beats NVIDIA's GM200-based, Maxwell-powered GeForce GTX Titan X from NVIDIA by a decent amount. But then, the R9 280X seems to not be that far behind the Titan X.
The Fury X loses quite horribly to the Titan X in the 'face detection' test on CompuBench, while it beats the Titan X once again in the 'TV-L1 Optical Flow' test.
When it comes to potential gaming performance, the Fury X scores 102FPS on the Manhattan benchmark found in GFXBench, with the Titan X beating it easily with 137FPS. We're sure drivers will play a key part, as well as other benchmarks like 3DMark, Heaven and real-world gaming performance. We should see the Fury X and Radeon R9 390X unveiled next week at E3 2015.
ASUS has just announced Auto-Extreme, its new production process for making video cards. Auto-Extreme is "the industry's first fully-automated production process" reports Tom's Hardware.
Normally, video cards are constructed by hand with the help of machines, but there is always the chance of human error. The new process from ASUS looks to eliminate human error, with full automation from PCB manufacturing, all the way through to manufacturing the MOSFETs. Some of the PCB components used to be made by hand, but now ASUS has fully automated the process, ushering in a new wave of video card production.
The automation process also allows for a higher precision, made possible because humans aren't touching the process. Installation of the components can also be completed without oxidation, and in environments that aren't healthy for humans. ASUS should be able to offer consumers a lower failure rate, bolstered by reduced production costs which should hopefully spin down to consumers, and longer lifetime on cards.
With the new Radeon video cards launching in the next couple of weeks, most have been wondering what pricing AMD would stick to. Well, it looks like the MSRP on the Radeon R9 390X could be just $389.
There are two enthusiast cards in the Radeon R9 390X and Radeon R9 390, with the 'enhanced Hawaii XT' in the R9 390X with an MSRP of $389 and the 'enhanced Hawaii PRO' under the Radeon R9 390 with a price of $329. This should compete directly with the GeForce GTX 980 from NVIDIA which dropped to $499 thanks to the release of the GTX 980 Ti at $649. These two cards will better compete with the GTX 900 series, while the Fury range of cards will compete with the GTX 980 Ti and Titan X.
Will you be buying the Radeon R9 390X, or will you be waiting for the higher-end, HBM-based Fury cards?
We've had to pull our ASUS Radeon R9 390X post, as well as our PowerColor Radeon R9 390X post, so here's hoping that news and images on the XFX Radeon R9 390X Double Dissipation 8GB doesn't need to be pulled.
The XFX branded card is the latest in a slew of Radeon R9 390X video cards that has popped up, a rebranded R9 290X with 8GB of VRAM. It sports a slick looking cooler with unlocked voltage, and a diamond-shaped texture which looks great. There's 7 extended heat pipes which should keep the XFX Radeon R9 390X Double Dissipation 8GB nice and cool.
The XFX Radeon R9 390X Double Dissipation 8GB output configuration is as follows: DVI-I, DVI-D, HDMI and DisplayPort.
After our world exclusive news that AMD would be releasing its Fury X as the watercooled, HBM-based flagship GPU, we began hearing about Fury Nano. We didn't want to post the news and possibly have it backfire on our source, so we waited on someone else to hear about it - and here we are.
We can now reveal that there will be three different Fury cards released, the Fury X as the flagship, Fury, which will be joined by Fury Nano. Fury Nano will be a tiny video card, which could arrive as the flagship. Right now it's all up in the air, but we do know there will be three different Fury cards, followed by a rebrand that will arrive as the Radeon R9 300 series.
We had a world exclusive with the first Radeon R9 390X spotted thanks to PowerColor, but now details have leaked on the ASUS Radeon R9 390X DirectCU II OC. The ASUS variant will feature 8GB of GDDR5 RAM, which is in line with our exclusive story on the Fiji XT-based Fury X and Fury featuring HBM, while the 300 series cards will be powered by GDDR5.
The details leaked tease that the ASUS Radeon R9 390X DirectCU II OC will feature 8GB of GDDR5, 1070MHz Core, 6GHz on the 8GB of VRAM spread on a 512-bit memory bus. Connectivity wise, we'll have one DisplayPort, one HDMI, and two DVI-D. The ASUS Radeon R9 390X DirectCU II OC will sport 2816 stream processors, 176 texture mapping units (TMU) and 64 ROPs.
The 8GB of RAM being clocked at 6GHz is a change from the 5GHz on the R9 290X, so that's most likely what we can expect from the other Radeon R9 390X cards when they launch. ASUS has two 8-pin PCIe connectors on the card, and an estimated price of around $449 when it launches later this year.
While AMD is about to launch its Fiji XT-based Radeon R9 Fury X and the respin that will arrive as the Radeon R9 390X, it looks like NVIDIA is already playing around with its next-gen GPU: GP100. GP100 will reportedly rock between 4500 and 6000 CUDA cores, making it NVIDIA's biggest GPU yet.
Right now we have GM200, with the 'M' standing for Maxwell, so the GP100 and its 'P' standing for Pascal. Pascal is NVIDIA's next generation architecture, with the GPU being built on the 16nm process. Not only will Pascal be baked onto 16nm, but it will arrive with support for HBM2 memory, which should see memory bandwidth scaling up to an insane 1.2TB/sec or 1200GB/sec. Considering the GeForce GTX 980 Ti has 336GB/sec, the GTX 1080 Ti (or whatever NVIDIA calls it) could have up to 1.2TB/sec bandwidth, a near 400% increase in memory bandwidth alone.
The news is coming from a source on the Beyond3D forums who says that the 'big Pascal' chip (GP100) has been taped out on TSMC's 16nm process, with a 'target release' window of Q1 2016. We don't know if this is true or not, but I would be pretty sure that NVIDIA is playing around with Pascal right now. I've asked many of my NVIDIA sources about Pascal, 16nm and HBM2 and all I get back are smiles... we should be more excited about the next-gen GPU from NVIDIA than any other release from the company, ever.