CORSAIR iCUE LT100 Smart Lighting Towers Review
Gigabyte B550 Aorus Pro review
Asus ROG Strix XG27UQ review
be quiet! Pure Rock 2 review
Fractal Design Define 7 Compact review
Lian Li Galahad 360 liquid cooler review
Guru3D Rig of the Month - July 2020
Patriot Viper Steel DDR4 3600 MHz (64GB) review
Red Dead Redemption 2: PC graphics benchmark review (revisited)
Kioxia Exceria 960 GB SATA3 SSD review
Nvidia announces Turing architecture for gpu's Quadro RTX8000, 6000, 5000
Jensen Huang has let the cat out of the bag and has made a number of announcements over at SIGGRAPH, one of them. Turing is the new architecture for next-gen GPUs. In addition, the company has announced a number of graphics cards, not intended for the consumer market.
For the SIGGRAPH presentations, Nvidia focused mainly on ray tracing CAPABILITIES, and Turing gpu's will have specialized RT cores on board to enable ray tracing with minimal delay. Together with a new generation of Tensor cores a Turing gpu would offer performance increases, six time that of the previous Pascal generation. Nvidia plans to develop Turing with the goal of enabling real-time raytracing effects. This was expected after Microsoft added DXR into DirectX 12 and Nvidia's implementation of RTX at GDC 2018. Thus far, only Volta has been able to accelerate this in hardware. Turing can do that too - via dedicated circuits that you know as the Volta tensor cores for AI. For Turing NVIDIA adds RT cores, ray tracing cores - next to the classic CUDA cores, of course.
Quadro RTX 8000 Quadro RTX 6000 Quadro RTX 5000 Quadro GV100 Quadro P6000
Architecture
Turing
Turing
Turing
Volta
Pascal
FP32 ALUs
4608
4608
3072
5120
3840
Tensor cores
576
576
384
640
-
RT-cores
Yes
Yes
Yes
-
-
FP32 performance
16 TFLOPS
? TFLOPS
? TFLOPS
14.8 TFLOPS
12 TFLOPS
Memory
48 GB GDDR6
24 GB GDDR6
16 GB GDDR6
32 GB HBM2
24 GB GDDR5X
Turing GPUs make use of, or better yet, can make use of gddr6 memory at a maximum speed of of 14Gbit/s, and that is available from Samsung, SK Hynix and Micron. Nvidia announced four professional models, the Quadro RTX8000, 6000, 5000 and RTX Server. The flagship is the RTX8000 has 4608 shader-cores and 576 Tensor cores at a die size of 754 square millimeters. The new GPUs will be released this autumn.
“Turing is NVIDIA’s most important innovation in computer graphics in more than a decade,” said Jensen Huang, founder and CEO of NVIDIA, speaking at the start of the annual SIGGRAPH conference. “Hybrid rendering will change the industry, opening up amazing possibilities that enhance our lives with more beautiful designs, richer entertainment and more interactive experiences. The arrival of real-time ray tracing is the Holy Grail of our industry.”
Even though there is no information yet about the gaming graphics cards, it was already mentioned that the new GeForce models will also be based on Turing. Nvidia is expected to announce the consumer versions of the new GPUs next week at the GamesCom gaming fair in Cologne Germany.
NVIDIA AIB Manli registers GA104-400 - Ampere? And Lists GeForce GTX 2070 and 2080 - 08/02/2018 05:03 PM
Holy moley, we're in for some gossip and chatter. Manli Technology Group, yes from the graphics cards, has a EEC certification upcoming products. And if you read closely you'll notice the mention ...
NVIDIA adds far more visual option to not install GeForce Experience - 07/16/2018 07:13 PM
Do you remember how Geforce Experience first was a feature, then an optional install and then pretty much became a mandatory installation toed to cloud logins like Facebook etc? Well, there always has...
NVIDIA AI that can remove noise and text from photos - 07/11/2018 04:35 PM
Researchers from NVIDIA, Aalto University, and MIT developed a deep learning based method that can fix photos by simply looking at examples of corrupted photos only. ...
NVIDIA Announces Financial Results for First Quarter Fiscal 2019 - 05/11/2018 07:43 AM
NVIDIA reported record revenue for the first quarter ended April 29, 2018, of $3.21 billion, up 66 percent from $1.94 billion a year earlier, and up 10 percent from $2.91 billion in the previous quart...
NVIDIA Announces RTX technology a Raytracing API - 03/19/2018 07:20 PM
Much like anyone in the graphics arena, Nvidia has been working a long time on actual real-time raytracing, the primary rendering technology used in the film industry. Today things get a step closer....
Texter
Senior Member
Posts: 3143
Joined: 2008-11-03
Senior Member
Posts: 3143
Joined: 2008-11-03
#5574297 Posted on: 08/16/2018 10:49 AM
That's why ATAA is presented as next thing, because it is AA based on thing Turing improved and therefore something which may be used to knockout older HW.
Some Volta owners are probably under another impression, as the Quadro version isn't even six months old...some of them probably thought nVidia were selling them a top of the line GPU at $9k, not a skipped architecture.
Each architecture may have its own merits, dunno...
That's why ATAA is presented as next thing, because it is AA based on thing Turing improved and therefore something which may be used to knockout older HW.
Some Volta owners are probably under another impression, as the Quadro version isn't even six months old...some of them probably thought nVidia were selling them a top of the line GPU at $9k, not a skipped architecture.
Each architecture may have its own merits, dunno...
Denial
Senior Member
Posts: 12837
Joined: 2004-05-16
Senior Member
Posts: 12837
Joined: 2004-05-16
#5574342 Posted on: 08/16/2018 02:15 PM
But here, numbers are laughable. When basic investment is sub 3ms, you do not want to spend another 28~40ms on AA.
Secondly, improvement of regular SSAA seems to correlate exactly to base render, so no improvement there other than more massive GPU or higher clock.
That's why ATAA is presented as next thing, because it is AA based on thing Turing improved and therefore something which may be used to knockout older HW..
The Siggraph presentation was for pre-production film rendering and not gaming - that's why the AA quality is set so high. Also if you're going to ship a large chunk of a GPU with dedicated raytracing hardware you might as will build some value-add around it.
I think the bigger takeaway from that slide is that Turing is 750mm2 and is 50% faster than a Titan V (815mm2), Titan V is on average 25-30% faster than a 1080Ti.
Some Volta owners are probably under another impression, as the Quadro version isn't even six months old...some of them probably thought nVidia were selling them a top of the line GPU at $9k, not a skipped architecture. Each architecture may have its own merits, dunno...
We don't know the FP64 performance of the Quadro RTX series or if the GDDR6 has ECC support. Could be targeting one Quadro towards high precision workloads and the other towards compute, INT4 workloads.
But here, numbers are laughable. When basic investment is sub 3ms, you do not want to spend another 28~40ms on AA.
Secondly, improvement of regular SSAA seems to correlate exactly to base render, so no improvement there other than more massive GPU or higher clock.
That's why ATAA is presented as next thing, because it is AA based on thing Turing improved and therefore something which may be used to knockout older HW..
The Siggraph presentation was for pre-production film rendering and not gaming - that's why the AA quality is set so high. Also if you're going to ship a large chunk of a GPU with dedicated raytracing hardware you might as will build some value-add around it.
I think the bigger takeaway from that slide is that Turing is 750mm2 and is 50% faster than a Titan V (815mm2), Titan V is on average 25-30% faster than a 1080Ti.
Some Volta owners are probably under another impression, as the Quadro version isn't even six months old...some of them probably thought nVidia were selling them a top of the line GPU at $9k, not a skipped architecture.
Each architecture may have its own merits, dunno...We don't know the FP64 performance of the Quadro RTX series or if the GDDR6 has ECC support. Could be targeting one Quadro towards high precision workloads and the other towards compute, INT4 workloads.
pharma
Senior Member
Posts: 1308
Joined: 2003-09-10
Senior Member
Posts: 1308
Joined: 2003-09-10
#5574386 Posted on: 08/16/2018 04:16 PM
Usually when I question something I seek knowledge/advice from experts in the field.
I think if you merge the individual components that are required for this to work you can begin to realize it is a new architecture. The quote and link below pretty much describes the pieces that make up the architecture:
NVIDIA Breaks New Ground with Turing GPU Architecture
In a bid to reinvent computer graphics and visualization, NVIDIA has developed a new architecture that merges AI, ray tracing, rasterization, and computation. The new architecture, known as Turing, was unveiled this week by NVIDIA CEO Jensen Huang in his keynote address at SIGGRAPH 2018.
A key element in the Turing architecture is the RT Cores, a specialized bit of circuitry that enables real-time ray tracing for accurate shadowing, reflections, refractions, and global illumination. Ray tracing essentially simulates light, which sounds simple enough, but it turns out to be very computationally intense. As the above product chart shows, the new Quadros can simulate up to 10 billion rays per second, which would be impossible with a more generic GPU design.
The on-board memory is based on GDDR6, which is something of a departure from the Quadro GV100, which incorporated 32GB of HBM2 memory. Memory capacity on the new RTX processors can be effectively doubled by hooking two GPUs together via NVLink, making it possible to hold larger images in local memory.
As usual, the SM will supply compute and graphics rasterization, but with a few twists. With Turing, NVIDIA has separated the floating point and integer pipelines so that they can operate simultaneously, a feature that is also available in the Volta V100. This enables the GPU to do address calculations and numerical calculation at the same time, which can be big time saver. As a result, the new Quadro chips can deliver up to 16 teraflops and 16 teraops of floating point and integer operations, respectively, in parallel. The SM also comes with a unified cache with double the bandwidth of the previous generation architecture.
Perhaps the most interesting aspect to the new Quadro processors is the Turing Tensor Cores. For graphics and visualization work, the Tensor Cores can be used for things like AI-based denoising, deep learning anti-aliasing (DLAA), frame interpolation, and resolution scaling. These techniques can be used to reduce render time, increase image resolution, or create special effects
The Turing Tensor Cores are similar to those in the Volta-based V100 GPU, but in this updated version NVIDIA has significantly boosted tensor calculations for INT8 (8-bit integer), which are commonly used for inferencing neural networks. In the V100, INT8 performance topped out at 62.8 teraops, but in the Quadro RTX chips, this has been boosted to a whopping 250 teraops. The new Tensor Cores also provide an INT4 (4-bit integer) capability for certain types of inferencing work that can get by with even less precision. That doubles the tensor performance to 500 teraops – half a petaop. The new Tensor Cores also provide 125 teraflops for FP16 data – same as the V100 – if for some reason you decide to use the Quadro chips for neural net training.
https://www.top500.org/news/nvidia-breaks-new-ground-with-turing-gpu-architecture/
Idk the entire "revolutionary" bit is related to Raytracing - other than that it's just more of the same with some slight architecture tweaks. I don't really find the word revolutionary synonymous with affordable.
I think if you merge the individual components that are required for this to work you can begin to realize it is a new architecture. The quote and link below pretty much describes the pieces that make up the architecture:
NVIDIA Breaks New Ground with Turing GPU Architecture
In a bid to reinvent computer graphics and visualization, NVIDIA has developed a new architecture that merges AI, ray tracing, rasterization, and computation. The new architecture, known as Turing, was unveiled this week by NVIDIA CEO Jensen Huang in his keynote address at SIGGRAPH 2018.
A key element in the Turing architecture is the RT Cores, a specialized bit of circuitry that enables real-time ray tracing for accurate shadowing, reflections, refractions, and global illumination. Ray tracing essentially simulates light, which sounds simple enough, but it turns out to be very computationally intense. As the above product chart shows, the new Quadros can simulate up to 10 billion rays per second, which would be impossible with a more generic GPU design.
The on-board memory is based on GDDR6, which is something of a departure from the Quadro GV100, which incorporated 32GB of HBM2 memory. Memory capacity on the new RTX processors can be effectively doubled by hooking two GPUs together via NVLink, making it possible to hold larger images in local memory.
As usual, the SM will supply compute and graphics rasterization, but with a few twists. With Turing, NVIDIA has separated the floating point and integer pipelines so that they can operate simultaneously, a feature that is also available in the Volta V100. This enables the GPU to do address calculations and numerical calculation at the same time, which can be big time saver. As a result, the new Quadro chips can deliver up to 16 teraflops and 16 teraops of floating point and integer operations, respectively, in parallel. The SM also comes with a unified cache with double the bandwidth of the previous generation architecture.
Perhaps the most interesting aspect to the new Quadro processors is the Turing Tensor Cores. For graphics and visualization work, the Tensor Cores can be used for things like AI-based denoising, deep learning anti-aliasing (DLAA), frame interpolation, and resolution scaling. These techniques can be used to reduce render time, increase image resolution, or create special effects
The Turing Tensor Cores are similar to those in the Volta-based V100 GPU, but in this updated version NVIDIA has significantly boosted tensor calculations for INT8 (8-bit integer), which are commonly used for inferencing neural networks. In the V100, INT8 performance topped out at 62.8 teraops, but in the Quadro RTX chips, this has been boosted to a whopping 250 teraops. The new Tensor Cores also provide an INT4 (4-bit integer) capability for certain types of inferencing work that can get by with even less precision. That doubles the tensor performance to 500 teraops – half a petaop. The new Tensor Cores also provide 125 teraflops for FP16 data – same as the V100 – if for some reason you decide to use the Quadro chips for neural net training.
https://www.top500.org/news/nvidia-breaks-new-ground-with-turing-gpu-architecture/
-Tj-
Senior Member
Posts: 16882
Joined: 2012-05-18
Senior Member
Posts: 16882
Joined: 2012-05-18
#5574405 Posted on: 08/16/2018 05:20 PM
I like how they boosted tensor performance compared to 1st gen tensor cores.
So this is basically a 2080gtx tensor performance., I think those 4 and 8bit will really shine, no one would notice 100% accurate or 50%, 75% less while in motion.
If they can use it for shadows, AO, then maybe even for physx flex, smoke, particles, fluids
Usually when I question something I seek knowledge/advice from experts in the field.
I think if you merge the individual components that are required for this to work you can begin to realize it is a new architecture. The quote and link below pretty much describes the pieces that make up the architecture:
https://www.top500.org/news/nvidia-breaks-new-ground-with-turing-gpu-architecture/
I think if you merge the individual components that are required for this to work you can begin to realize it is a new architecture. The quote and link below pretty much describes the pieces that make up the architecture:
https://www.top500.org/news/nvidia-breaks-new-ground-with-turing-gpu-architecture/
I like how they boosted tensor performance compared to 1st gen tensor cores.
So this is basically a 2080gtx tensor performance., I think those 4 and 8bit will really shine, no one would notice 100% accurate or 50%, 75% less while in motion.
If they can use it for shadows, AO, then maybe even for physx flex, smoke, particles, fluids
Click here to post a comment for this news story on the message forum.
Senior Member
Posts: 10527
Joined: 2012-07-20
This means that for nV to cash on those improvements, games they sponsor will have them set to higher details.
(One can just hope that there will be option to set them to regular level at which previous generation does not lose 30% of fps just from having AO.)
But here, numbers are laughable. When basic investment is sub 3ms, you do not want to spend another 28~40ms on AA.
Secondly, improvement of regular SSAA seems to correlate exactly to base render, so no improvement there other than more massive GPU or higher clock.
That's why ATAA is presented as next thing, because it is AA based on thing Turing improved and therefore something which may be used to knockout older HW.
Funny thing is that this is also comparison of SSAA 16x vs ATAA 8x. It is like apples to oranges.