Early Benchmark Result Shows NVIDIA Titan X Pascal is 51% Faster Than GTX 1080 On Triple 4K Setup

The earlier leaked benchmark from videocardz.com showing that the NVIDIA Titan X Pascal is 29% better in performance over the GTX 1080 appears to be untrue as the latest Unigine Heaven benchmark result from TweakTown is showing a massive 51.4% better in performance. 

TweakTown's attempt to push the Titan X Pascal to its limit with a triple 4K monitor setup with a total resolution of 11520 x 2160. With the Titan X Pascal already taking the lead at such mammoth resolution, we can now set our expectation high on the irresponsible performance which NVIDIA has promised to deliver.

(Source: TweakTown)

$1200 NVIDIA Titan X Pascal Sold Out Within Hours After Launch

The new NVIDIA TITAN X Pascal has a whopping 11 TFLOPS of power, 3584 CUDA Cores, 12GB of 10Gbps GDDR5X VRAM, a 384-bit Memory Interface, 480 GB/sec of Memory Bandwidth, boosts to 1.5GHz out of the box, utilises superior heat dissipation vapour chamber cooling technology, and features a stunning design. 

Although it comes with a cut-throat price tag at $1200, the card was sold out within hours just after its launch. It is unknown when the Titan X Pascal will be available again but here's the worst part - the Titan X Pascal is currently available only at NVIDIA webstore.

(Source: GeForce.com)

Leaked 3DMark Performance Shows NVIDIA Titan X Pascal Is 29% Faster Than The GTX 1080

Earlier leaked synthetic benchmark of the new NVIDIA Titan X Pascal suggests that reviewers around the world have already started receiving the Titan X Pascal samples from NVIDIA, but the most recent leak suggests that NVIDIA has decided to not doing that on a large scale. So, don't be surprised if you can't find many reviews for the Titan X Pascal on its launch date.

For those of you who has been preparing your cash for the Titan X Pascal, here's a leaked 3DMark Performance chart from videocardz.com that shows the performance difference between the Titan X Pascal with the currently available GTX 1060, GTX 1070 and GTX 1080.

According to the chart, the NVIDIA TITAN X Pascal yields around 1.3x performance of stock GTX 1080 and 1.13x of overclocked GTX 1080. It’s almost 1.5 times faster than GTX 1070 and 1.87x faster than overclocked GTX 1060. 

(Source: videocardz)

Alleged Synthetic Benchmark of NVIDIA Pascal Titan X Surfaces

NVIDIA's Pascal Titan X launch is just around the corner and it seems that NVIDIA has started sending out the Pascal Titan X samples and some have already arrived in the hands of reviewers and relevant organization or individual. 

The alleged synthetic benchmark on Chiphell shows that the Pascal Titan X performs much better than the Maxwell Titan X. But, the difference in numbers isn't reflecting any of the gaming performance difference between the Maxwell Titan X and Pascal Titan X, but the synthetic CUDA-based software that uses different and more architecture optimized cuDNN4 and cuDNN5 libraries, which can be used in deep learning neural networks.  

(Source: Chiphell)

NVIDIA Titan X With Pascal GPU Unleashed - 60% Faster Than Previous Titan X, Available This August 2nd For $1200

If the GTX 1080 is the new king, this new Titan X with GP102 Pascal GPU will be the new emperor for consumer graphics. That's right folks, NVIDIA has officially announced their upcoming Pascal Titan X today and it will be available on August 2nd with the specifications that aren't exactly the same as what we've expected from the rumor. 

The Titan X GP102 GPU features 12 billion transistors that are almost the double of the transistors count of 7.2 billion transistors on the GTX 1080 GP104, higher performance with 11 TFLOPs compared to 9 TFLOPs on the GP104, more 3584 CUDA Cores clocked at 1417MHz and boost clock of 1531MHz that can easily overwhelm the GP104 with 2560 CUDA cores even if it comes with a higher clock speed of 1607MHz and boost clock of 1733MHz.

HBM2 isn't something that we can expect from the Pascal Titan X at the moment but the 12GB GDDR5X with 10 GB/s Memory Speed on a 384-bit bus and a total memory bandwidth of 480 GB/s that comes near to HBM1's memory bandwidth of 512 GB/s will still be able to deliver quite a monstrous performance.

With all that amount of performance, the GP102 Titan X requires a 6+8 pin power to run and features a total TDP of 250W.

Appearance wise, the Pascal Titan X will be having the same founders edition blower style cooler as both GTX 1070 and GTX 1080, except that it's in black. One of our reliable sources suggests that NVIDIA doesn't plan to bring the Pascal Titan X to its AiB partners, which means that it will only be available as a Founders Edition card that can only be purchased directly from NVIDIA's website at the price of $1200. 

NVIDIA Doubles Performance for Deep Learning Training

New Releases of DIGITS, cuDNN to Deliver 2x Faster Neural Network Training; cuDNN to Enable More Sophisticated Models

SINGAPORE — July 8, 2015 — NVIDIA today announced updates to its GPU-accelerated deep learning software that will double deep learning training performance. 

The new software will empower data scientists and researchers to supercharge their deep learning projects and product development work by creating more accurate neural networks through faster model training and more sophisticated model design.

The NVIDIA® DIGITS™ Deep Learning GPU Training System version 2 (DIGITS 2) and NVIDIA CUDA® Deep Neural Network library version 3 (cuDNN 3) provide significant performance enhancements and new capabilities.

For data scientists, DIGITS 2 now delivers automatic scaling of neural network training across multiple high-performance GPUs. This can double the speed of deep neural network training for image classification compared to a single GPU.

For deep learning researchers, cuDNN 3 features optimized data storage in GPU memory for the training of larger, more sophisticated neural networks. cuDNN 3 also provides higher performance than cuDNN 2, enabling researchers to train neural networks up to two times faster on a single GPU.

The new cuDNN 3 library is expected to be integrated into forthcoming versions of the deep learning frameworks Caffe, Minerva, Theano and Torch, which are widely used to train deep neural networks.

“High-performance GPUs are the foundational technology powering deep learning research and product development at universities and major web-service companies,” said Ian Buck, vice president of Accelerated Computing at NVIDIA. “We’re working closely with data scientists, framework developers and the deep learning community to apply the most powerful GPU technologies and push the bounds of what is possible.”

DIGITS 2 – Up to 2x Faster Training with Automatic Multi-GPU Scaling

DIGITS 2 is the first all-in-one graphical system that guides users through the process of designing, training and validating deep neural networks for image classification.

The new automatic multi-GPU scaling capability in DIGITS 2 maximises the available GPU resources by automatically distributing the deep learning training workload across all of the GPUs in the system. Using DIGITS 2, NVIDIA engineers trained the well-known AlexNet neural network model more than two times faster on four NVIDIA Maxwell™ architecture-based GPUs, compared to a single GPU1. Initial results from early customers are demonstrating better results.

“Training one of our deep nets for auto-tagging on a single NVIDIA GeForce GTX TITAN X takes about sixteen days, but using the new automatic multi-GPU scaling on four TITAN X GPUs the training completes in just five days,” said Simon Osindero, A.I. architect at Yahoo's Flickr. “This is a major advantage and allows us to see results faster, as well letting us more extensively explore the space of models to achieve higher accuracy.”

cuDNN 3 – Train Larger, More Sophisticated Models Faster

cuDNN is a GPU-accelerated library of mathematical routines for deep neural networks that developers integrate into higher-level machine learning frameworks.

cuDNN 3 adds support for 16-bit floating point data storage in GPU memory, doubling the amount of data that can be stored and optimising memory bandwidth. With this capability, cuDNN 3 enables researchers to train larger and more sophisticated neural networks.

“We believe FP16 GPU storage support in NVIDIA’s libraries will enable us to scale our models even further, since it will increase effective memory capacity of our hardware and improve efficiency as we scale training of a single model to many GPUs,” said Bryan Catanzaro, senior researcher at Baidu Research. “This will lead to further improvements in the accuracy of our models.”

cuDNN 3 also delivers significant performance speedups compared to cuDNN 2 for training neural networks on a single GPU. It enabled NVIDIA engineers to train the AlexNet model two times faster on a single NVIDIA GeForce® GTX™ TITAN X GPU.2

Availability

The DIGITS 2 Preview release is available today as a free download for NVIDIA registered developers. To learn more or download, visit the DIGITS website. 

The cuDNN 3 library is expected to be available in major deep learning frameworks in the coming months. To learn more visit the cuDNN website.