Tuesday, November 6th 2018

AMD Unveils World's First 7 nm GPUs - Radeon Instinct MI60, Instinct MI50

AMD today announced the AMD Radeon Instinct MI60 and MI50 accelerators, the world's first 7nm datacenter GPUs, designed to deliver the compute performance required for next-generation deep learning, HPC, cloud computing and rendering applications. Researchers, scientists and developers will use AMD Radeon Instinct accelerators to solve tough and interesting challenges, including large-scale simulations, climate change, computational biology, disease prevention and more.

"Legacy GPU architectures limit IT managers from effectively addressing the constantly evolving demands of processing and analyzing huge datasets for modern cloud datacenter workloads," said David Wang, senior vice president of engineering, Radeon Technologies Group at AMD. "Combining world-class performance and a flexible architecture with a robust software platform and the industry's leading-edge ROCm open software ecosystem, the new AMD Radeon Instinct accelerators provide the critical components needed to solve the most difficult cloud computing challenges today and into the future."
The AMD Radeon Instinct MI60 and MI50 accelerators feature flexible mixed-precision capabilities, powered by high-performance compute units that expand the types of workloads these accelerators can address, including a range of HPC and deep learning applications. The new AMD Radeon Instinct MI60 and MI50 accelerators were designed to efficiently process workloads such as rapidly training complex neural networks, delivering higher levels of floating-point performance, greater efficiencies and new features for datacenter and departmental deployments.

The AMD Radeon Instinct MI60 and MI50 accelerators provide ultra-fast floating-point performance and hyper-fast HBM2 (second-generation High-Bandwidth Memory) with up to 1 TB/s memory bandwidth speeds. They are also the first GPUs capable of supporting next-generation PCIe 4.02 interconnect, which is up to 2X faster than other x86 CPU-to-GPU interconnect technologies, and feature AMD Infinity Fabric Link GPU interconnect technology that enables GPU-to-GPU communications that are up to 6X faster than PCIe Gen 3 interconnect speeds.

AMD also announced a new version of the ROCm open software platform for accelerated computing that supports the architectural features of the new accelerators, including optimized deep learning operations (DLOPS) and the AMD Infinity Fabric Link GPU interconnect technology. Designed for scale, ROCm allows customers to deploy high-performance, energy-efficient heterogeneous computing systems in an open environment.

"Google believes that open source is good for everyone," said Rajat Monga, engineering director, TensorFlow, Google. "We've seen how helpful it can be to open source machine learning technology, and we're glad to see AMD embracing it. With the ROCm open software platform, TensorFlow users will benefit from GPU acceleration and a more robust open source machine learning ecosystem."

Key features of the AMD Radeon Instinct MI60 and MI50 accelerators include:
  • Optimized Deep Learning Operations: Provides flexible mixed-precision FP16, FP32 and INT4/INT8 capabilities to meet growing demand for dynamic and ever-changing workloads, from training complex neural networks to running inference against those trained networks.
  • World's Fastest Double Precision PCIe 2 Accelerator5: The AMD Radeon Instinct MI60 is the world's fastest double precision PCIe 4.0 capable accelerator, delivering up to 7.4 TFLOPS peak FP64 performance5 allowing scientists and researchers to more efficiently process HPC applications across a range of industries including life sciences, energy, finance, automotive, aerospace, academics, government, defense and more. The AMD Radeon Instinct MI50 delivers up to 6.7 TFLOPS FP64 peak performance1, while providing an efficient, cost-effective solution for a variety of deep learning workloads, as well as enabling high reuse in Virtual Desktop Infrastructure (VDI), Desktop-as-a-Service (DaaS) and cloud environments.
  • Up to 6X Faster Data Transfer: Two Infinity Fabric Links per GPU deliver up to 200 GB/s of peer-to-peer bandwidth - up to 6X faster than PCIe 3.0 alone4 - and enable the connection of up to 4 GPUs in a hive ring configuration (2 hives in 8 GPU servers).
  • Ultra-Fast HBM2 Memory: The AMD Radeon Instinct MI60 provides 32GB of HBM2 Error-correcting code (ECC) memory6, and the Radeon Instinct MI50 provides 16GB of HBM2 ECC memory. Both GPUs provide full-chip ECC and Reliability, Accessibility and Serviceability (RAS)7 technologies, which are critical to deliver more accurate compute results for large-scale HPC deployments.
  • Secure Virtualized Workload Support: AMD MxGPU Technology, the industry's only hardware-based GPU virtualization solution, which is based on the industry-standard SR-IOV (Single Root I/O Virtualization) technology, makes it difficult for hackers to attack at the hardware level, helping provide security for virtualized cloud deployments.
Updated ROCm Open Software Platform
AMD today also announced a new version of its ROCm open software platform designed to speed development of high-performance, energy-efficient heterogeneous computing systems. In addition to support for the new Radeon Instinct accelerators, ROCm software version 2.0 provides updated math libraries for the new DLOPS; support for 64-bit Linux operating systems including CentOS, RHEL and Ubuntu; optimizations of existing components; and support for the latest versions of the most popular deep learning frameworks, including TensorFlow 1.11, PyTorch (Caffe) and others. Learn more about ROCm 2.0 software here.

Availability
The AMD Radeon Instinct MI60 accelerator is expected to ship to datacenter customers by the end of 2018. The AMD Radeon Instinct MI50 accelerator is expected to begin shipping to data center customers by the end of Q1 2019. The ROCm 2.0 open software platform is expected to be available by the end of 2018.
Sources: Radeon Instinct MI60, Radeon Instinct MI50
Add your own comment

45 Comments on AMD Unveils World's First 7 nm GPUs - Radeon Instinct MI60, Instinct MI50

#26
jabbadap
medi01team overpriced green fan got bh over your comment.

So saddening AMD needs 331mm² 7nm 300W chip to take on300W 815mm2 12nm chip.
Needing 300W for 331 mm² card does not sound optimal clocks for the arch. Without competition this would be some ~200W card at ~6.66TFlops of fp64 compute power. Now they are running it at too high clock that arch can give them just to be "top" on fp64 power on pcie. Vega itself can be very high efficient if not pushed too far. Granted this will give them some design wins on HPC market(All ready got few). So in that pov it's not the bad product.
SteevoReally I hate the "First PCIe 4.0 7Tflop bla bla bla" if it weren't for the PCIE 4.0 they couldn't say that, and PCIe 4.0 is currently unsupported in reality and isn't going to be supported before this card hits the market, so the spin on this proves they very carefully worded it and I would rather just see the numbers, is it going to take less than the 300W the Nvidia uses to get the same 7Tflop performance? Is it going to do some other fancy faster math? Is it going to do something more or the same at a lower cost.

AMD is trying really hard in the server market, but I don't think 2018 will be their year to take any crown, and neither will 2019. Maybe 2020 if they keep up with Zen2 and Navi is impressive. But that will also require thousands of hours to write the tools to make their supposed cards faster, or to make the same speed cards as fast and easy to use, which if Lisa is in the know she will already have people working on, but if not we will know the reason why they fail. Given AMD's vaporware issues, where they build hardware for software that isn't ready, or software that has great implementation of either ease of use, speed, or functionality, and you can only choose one......

I think AMD is playing their cards right for the midsize guys where a few IT guys run the show and want to save thousands to put into software development for long life peak performance, they will survive and their prosumer, gaming and server business will work out in the end, they will never be as big as Nvidia or Intel though. The same reason the Ford F-150 sells so many shitty trucks, its the king, its the classic standard of equal to the neighbors. AMD is the ful featured but still slightly odd Holden, the loud and hot Corvette versus the supercars, its second dog to the CPU and GPU business mostly due to mismanagement, Im just glad they are here to keep us from paying thousands more that Intel and Nvidia would charge if they could.
Well epyc2 will have pcie 4.0. Nvidia Xavier(Just for jokes) has one and IBM Power9(i.e. Raptor Talos II) though I don't know if amd have any drivers for arm or power machines.
Posted on Reply
#27
HTC
Correct me if i'm wrong but, like other PCIe versions, PCIe 4.0 is backwards compatible, right?

If so, then it can be used with any current PCIe 3.0 solutions but the full benefit will only come with boards that are PCIe 4.0 capable, much like current 2700X CPU runs on a B350 / X370 board but runs better on a B450 / X470 board (PBO and stuff like that).
Posted on Reply
#28
R0H1T
HTCCorrect me if i'm wrong but, like other PCIe versions, PCIe 4.0 is backwards compatible, right?

If so, then it can be used with any current PCIe 3.0 solutions but the full benefit will only come with boards that are PCIe 4.0 capable, much like current 2700X CPU runs on a B350 / X370 board but runs better on a B450 / X470 board (PBO and stuff like that).
In case of AMD the motherboard is mainly just for show, all the major functions are within the chip since it's a full SoC. The PCIe 4.0 Zen2 will work just fine with current boards, like Zen 1 & x370 but the additional lanes from the motherboard will be limited by the chipset. I can't say how the (CPU) lanes will be routed though.
Posted on Reply
#29
jabbadap
HTCCorrect me if i'm wrong but, like other PCIe versions, PCIe 4.0 is backwards compatible, right?

If so, then it can be used with any current PCIe 3.0 solutions but the full benefit will only come with boards that are PCIe 4.0 capable, much like current 2700X CPU runs on a B350 / X370 board but runs better on a B450 / X470 board (PBO and stuff like that).
Yup, that's true. I'm not sure if ZEN2 Desktop Ryzens and i.e. Navi cards would run at pcie 4.0 on current am4 board. They will probably just work as pcie 3.0 on those and for pcie 4.0 new chipset and motherboards are needed.
Posted on Reply
#30
londiste
Current boards are unlikely to support PCI-e 4.0. CPU and cards will just fall back to PCI-e 3.0
Posted on Reply
#31
R0H1T
londisteCurrent boards are unlikely to support PCI-e 4.0. CPU and cards will just fall back to PCI-e 3.0
Current boards just route the PCIe lanes, at least for Zen don't they? The GPU or storage will need to be PCIe 4.0 compliant though.
On November 29, 2011, PCI-SIG preliminarily announced PCI Express 4.0, providing a 16 GT/s bit rate that doubles the bandwidth provided by PCI Express 3.0, while maintaining backward and forward compatibility in both software support and used mechanical interface.
en.wikipedia.org/wiki/PCI_Express#PCI_Express_4.0
Posted on Reply
#32
londiste
That wiki quote sounds optimistic. The PCI-SIG FAQ it links to as source does say cards are compatible both ways.
There have been comments from manufacturers that boards for PCI-e 4.0 (and 5.0) will need to be built with tighter tolerances and possibly additional components.
Posted on Reply
#33
HTC
R0H1TCurrent boards just route the PCIe lanes, at least for Zen don't they? The GPU or storage will need to be PCIe 4.0 compliant though.

en.wikipedia.org/wiki/PCI_Express#PCI_Express_4.0
But they should be able to run @ PCIe 3.0 capabilities when placed in a PCIe 3.0 capable board, right?

Pretty much like you can use a RTX 2080Ti card and run it @ PCIe 2.0, as described here, even though it's a PCIe 3.0 card.
Posted on Reply
#34
jabbadap
londisteThat wiki quote sounds optimistic. The PCI-SIG FAQ it links to as source does say cards are compatible both ways.
There have been comments from manufacturers that boards for PCI-e 4.0 (and 5.0) will need to be built with tighter tolerances and possibly additional components.
That mechanical interface probably just means the port sizes will be the same(x16, x8, x4, x1). There's nothing about electrical compatibility on wikipedia.
Posted on Reply
#35
R0H1T
londisteThat wiki quote sounds optimistic. The PCI-SIG FAQ it links to as source does say cards are compatible both ways.
There have been comments from manufacturers that boards for PCI-e 4.0 (and 5.0) will need to be built with tighter tolerances and possibly additional components.
Those are electrical characteristics which need to be adhered to, so for instance just because every Z370 or Z390 board support 9900k, it doesn't imply that every board will support say 125W (or more) power consumption with all cores loaded. Some boards will, while others will not.
Posted on Reply
#36
XXL_AI
my experience with amd can be summarized like this,
-they don't have a proper software,
-they consume a lot of power,
-they heat too much that can melt the fan's cables,
-they draw too much power from motherboard,
-they lack several features that has been used in video/photo editing for years,
-they never deliver the performance they claim,
and they always fail within a year.
Posted on Reply
#37
Readlight
how about fixsing broken economy and trade?
i hope its not Made to spy everyone.
Posted on Reply
#38
Tomorrow
What's your sample size on this?
XXL_AI-they don't have a proper software,
Actually this has improved significantly over the years. That might have been true back when CCC was the control panel but not now.
XXL_AI-they consume a lot of power,
Yes they do consume more power than their equivalent performing Nvidia counterparts.
XXL_AI-they heat too much that can melt the fan's cables,
I've yet to see that happen on any card. It's pretty impossible actually. It would mean heatsink itself has to be over 100c on the card.
XXL_AI-they draw too much power from motherboard,
This was the case with RX480 when it came out. It was fixed.
XXL_AI-they lack several features that has been used in video/photo editing for years,
Such as?
XXL_AI-they never deliver the performance they claim,
Example?
XXL_AIand they always fail within a year.
Again, sample size?
Posted on Reply
#40
jabbadap
Steevoen.m.wikipedia.org/wiki/Volta_(microarchitecture)

Shows Dec 2017 release date, there were some benchmarks I found that used engineering samples in late 2017, and the white paper for V100 from Nvidia was released June 2018.
Well the specs were announced in summer, like these cards are announced now availability is market later on. December was a release date for Titan V. First Tesla V100 were sold September 2017 in DGX systems.
Posted on Reply
#42
jabbadap
ToxicTaZNvidia Tesla T4 (Tu100) Turning architecture on 12nm now

www.nvidia.com/en-us/data-center/tesla-t4/

Can't emphasize enough with Nvidia future 7nm replacement.

Why are people trying to compare with old Volta (V100) architecture?
Tesla T4 is 70W tdp version of cut down tu104. There's no full fp64 capability version of Turing architecture.

So full fp64 card for full fp64 card. Turing might get full fp64 compute capability due to die shrink, but current Nvidia's HPC arch is Volta. Sure in Datacenter/ML/DL/AI these card has to compete with current Turings too. And Nvidia will probably release T40 from tu102 chip too, which is more line of datacenter use with Mi60/50 than T4.
Posted on Reply
#44
Vya Domus
XXL_AImy experience with amd can be summarized like this,
-they don't have a proper software,
-they consume a lot of power,
-they heat too much that can melt the fan's cables,
-they draw too much power from motherboard,
-they lack several features that has been used in video/photo editing for years,
-they never deliver the performance they claim,
and they always fail within a year.
Troll much ?
Posted on Reply
#45
John Naylor
Who cares ? Don't care about cores, don't care about nm .... might as well say my piza will now get delivered in a different size box
Posted on Reply
Add your own comment
Mar 15th, 2025 22:20 EDT change timezone

New Forum Posts

Popular Reviews

Controversial News Posts