Nvidia Pascal GP100 GPU Specs - Massive 4096bit Bus And 3X More Bandwidth Than GTX Titan X

Nvidia's big Pascal GPU code named GP100 will feature a massive 4096bit bus and four HBM2 stacks each up to 8-Hello. The upcoming Nvidia flagship Pascal bit set to debut on TSMC'due south 16nm FinFET process later next year. Nosotros take confirmed with our sources that the GPU will be made with two unlike variations of stacked HBM2 solutions, notwithstanding both will characteristic a massive 4096bit retentivity interface just like AMD'due south flagship Republic of the fiji islands GPU launched final month.

The first variation will pack four HBM2 stacks, each will exist 4-How-do-you-do and will be clocked at 1Ghz.  This volition go into the traditional consumer GeForce line of GP100 based products. The second variation is too equipped with four HBM2 stacks clocked at 1Ghz, even so each will be 8-Howdy.

In HBM stacking #-Howdy denotes to the number of stacked DRAM dies, however this system does not take into account the additional base dice which incorporates logic and retentivity PHY.  And so information technology'south only meant to specify how many DRAM dies are in the stack rather than the total number of chips in the stack. GP100 packages with 8-Hi HBM stacks will exist limited to professional products, including Quadro and TESLA GPUs, where huge memory capacities are essential.

Nvidia's Big Pascal Slated For Release Side by side Twelvemonth

We've heard just recently that Nvidia's flagship Pascal code named GP100 has really taped out on TSMC's 16nm FinFET manufacturing process last month. Interestingly nosotros've too learned only three days ago that AMD has also taped out two FinFET chips last quarter. It's admittedly not a coincidence that both companies completed their designs at exactly the same time. Both are aggressively pushing for a debut of their next generation FinFET based GPUs next yr.

TSMC's new 16nm FinFET process promises to be significantly more ability efficient than planar 28nm. It also promises to be a bring a considerable improvement to transistor density. Which would enable Nvidia to build faster, significantly more than complex and more power efficient GPUs.

TSMC'south 16FF+ (FinFET Plus) technology can provide above 65 percent higher speed, effectually 2 times the density, or 70 per centum less power than its 28HPM technology. Comparing with 20SoC technology, 16FF+ provides extra 40% higher speed and threescore% ability saving. By leveraging the feel of 20SoC technology, TSMC 16FF+ shares the same metallic backend process in order to speedily improve yield and demonstrate process maturity for time-to-market value.

Although notably, unlike Nvidia which has confirmed that its Pascal GPUs volition exist manufactured using TSMC's 16nm FinFET process, AMD has withal to announced whether the Arctic Islands family of GPUs volition be made on TSMC's 16nm or Samsung's 14nm process. Both nodes are very similar, so which process AMD ends up using will be primarily dictated by yields and time-to-market place.

Also different Nvidia, AMD has a much more powerful incentive to launch its next generation of FinFET GPUs start. This is because the company has priority to HBM2 capacity - which is going to be limited initially - equally a result of co-inventing the technology with Hynix. Past pushing its graphics products to launch first AMD can establish two competitive advantages over its rival. The start obvious reward is beingness first to market place past launching its products earlier than its rival. Only most importantly this enables AMD to capture much of that initial HBM2 capacity away from Nvidia and extend its time-to-market lead essentially. This could create an interesting market dynamic only whether it can succeed remains to be seen.

Obviously Nvidia realizes that this play is in the cards, no pun intended, and will undoubtedly bide its time wisely honing its chips. GP100 has already been taped out so there's not much that can exist done to the chip's floorplan, even so Nvidia tin still use that extra fourth dimension on post-silicon work. Nvidia tin also spend more than time working on its smaller Pascal chips which oasis't been taped out as of yet.

Apart from HBM2 and 16nm there is i big compute-centric characteristic that Nvidia will debut with Pascal. And information technology's NVLink. Pascal will be the first GPU from the company to support this new proprietary server interconnect.

The technology is aimed at GPU accelerated servers where the cantankerous-chip communication is extremely bandwidth express and a major system bottleneck. Nvidia states that NV-Link will exist up to 5 to 12 times faster than traditional PCIE iii.0 making it a major step forward in platform atomics. Earlier this year Nvidia announced that IBM will be integrating this new interconnect into its upcoming PowerPC server CPUs.

NVIDIA® NVLink™ is a high-bandwidth, energy-efficient interconnect that enables ultra-fast communication between the CPU and GPU, and between GPUs. The engineering allows data sharing at rates five to 12 times faster than the traditional PCIe Gen3 interconnect, resulting in dramatic speed-ups in awarding performance and creating a new breed of high-density, flexible servers for accelerated computing.

There's clearly a distinct deviation between the goal that Nvidia is trying to reach with Pascal every bit opposed to Maxwell. With Maxwell it was all almost power efficiency no matter the cost and this meant huge sacrifices in compute performance. With Pascal however, Nvidia is shifting its focus back on compute and this is done through a myriad of ways. Loftier Bandwidth Retentivity is going to play a huge function in providing the necessary bandwidth and memory capacity needed for the HPC and enterprise markets. NV-Link is also going to play a very crucial part in allowing clients to realize their performance targets with Pascal accelerator clusters. Finally mixed precision ways even more power efficiency for the mobile workload. So it'due south clear Pascal is much more than Maxwell and Nvidia is pursuing a more comprehensive multi-faceted approach this time around.

Source: https://wccftech.com/nvidias-gp100-pascal-flagship-pack-4096-bit-memory-bus-8hi-hbm-stacks/

Posted by: trevinohirly1986.blogspot.com

Share this post