Jump to content

Hopper (microarchitecture)

From Wikipedia, the free encyclopedia
(Redirected from H100 Tensor Core GPU)

Hopper
LaunchedSeptember 20, 2022; 2 years ago (2022-09-20)
Designed byNvidia
Manufactured by
Fabrication processTSMC N4
Product Series
Server/datacenter
Specifications
L1 cache256 KB (per SM)
L2 cache50 MB
Memory supportHBM3
PCIe supportPCI Express 5.0
Media Engine
Encoder(s) supportedNVENC
History
PredecessorAmpere
VariantAda Lovelace (consumer and professional)
SuccessorBlackwell
4 NVIDIA H100 GPUs

Hopper is a graphics processing unit (GPU) microarchitecture developed by Nvidia. It is designed for datacenters and is used alongside the Lovelace microarchitecture. It is the latest generation of the line of products formerly branded as Nvidia Tesla, now Nvidia Data Centre GPUs.

Named for computer scientist and United States Navy rear admiral Grace Hopper, the Hopper architecture was leaked in November 2019 and officially revealed in March 2022. It improves upon its predecessors, the Turing and Ampere microarchitectures, featuring a new streaming multiprocessor, a faster memory subsystem, and a transformer acceleration engine.

Architecture

[edit]

The Nvidia Hopper H100 GPU is implemented using the TSMC N4 process with 80 billion transistors. It consists of up to 144 streaming multiprocessors.[1] Due to the increased memory bandwidth provided by the SXM5 socket, the Nvidia Hopper H100 offers better performance when used in an SXM5 configuration than in the typical PCIe socket.[2]

Streaming multiprocessor

[edit]

The streaming multiprocessors for Hopper improve upon the Turing and Ampere microarchitectures, although the maximum number of concurrent warps per streaming multiprocessor (SM) remains the same between the Ampere and Hopper architectures, 64.[3] The Hopper architecture provides a Tensor Memory Accelerator (TMA), which supports bidirectional asynchronous memory transfer between shared memory and global memory.[4] Under TMA, applications may transfer up to 5D tensors. When writing from shared memory to global memory, elementwise reduction and bitwise operators may be used, avoiding registers and SM instructions while enabling users to write warp specialized codes. TMA is exposed through cuda::memcpy_async[5]

When parallelizing applications, developers can use thread block clusters. Thread blocks may perform atomics in the shared memory of other thread blocks within its cluster, otherwise known as distributed shared memory. Distributed shared memory may be used by an SM simultaneously with L2 cache; when used to communicate data between SMs, this can utilize the combined bandwidth of distributed shared memory and L2. The maximum portable cluster size is 8, although the Nvidia Hopper H100 can support a cluster size of 16 by using the cudaFuncAttributeNonPortableClusterSizeAllowed function, potentially at the cost of reduced number of active blocks.[6] With L2 multicasting and distributed shared memory, the required bandwidth for dynamic random-access memory read and writes is reduced.[7]

Hopper features improved single-precision floating-point format (FP32) throughput with twice as many FP32 operations per cycle per SM than its predecessor. Additionally, the Hopper architecture adds support for new instructions, including the Smith–Waterman algorithm.[6] Like Ampere, TensorFloat-32 (TF-32) arithmetic is supported. The mapping pattern for both architectures is identical.[8]

Memory

[edit]

The Nvidia Hopper H100 supports HBM3 and HBM2e memory up to 80 GB; the HBM3 memory system supports 3 TB/s, an increase of 50% over the Nvidia Ampere A100's 2 TB/s. Across the architecture, the L2 cache capacity and bandwidth were increased.[9]

Hopper allows CUDA compute kernels to utilize automatic inline compression, including in individual memory allocation, which allows accessing memory at higher bandwidth. This feature does not increase the amount of memory available to the application, because the data (and thus its compressibility) may be changed at any time. The compressor will automatically choose between several compression algorithms.[9]

The Nvidia Hopper H100 increases the capacity of the combined L1 cache, texture cache, and shared memory to 256 KB. Like its predecessors, it combines L1 and texture caches into a unified cache designed to be a coalescing buffer. The attribute cudaFuncAttributePreferredSharedMemoryCarveout may be used to define the carveout of the L1 cache. Hopper introduces enhancements to NVLink through a new generation with faster overall communication bandwidth.[10]

Memory synchronization domains

[edit]

Some CUDA applications may experience interference when performing fence or flush operations due to memory ordering. Because the GPU cannot know which writes are guaranteed and which are visible by chance timing, it may wait on unnecessary memory operations, thus slowing down fence or flush operations. For example, when a kernel performs computations in GPU memory and a parallel kernel performs communications with a peer, the local kernel will flush its writes, resulting in slower NVLink or PCIe writes. In the Hopper architecture, the GPU can reduce the net cast through a fence operation.[11]

DPX instructions

[edit]

The Hopper architecture math application programming interface (API) exposes functions in the SM such as __viaddmin_s16x2_relu, which performs the per-halfword . In the Smith–Waterman algorithm, __vimax3_s16x2_relu can be used, a three-way min or max followed by a clamp to zero.[12] Similarly, Hopper speeds up implementations of the Needleman–Wunsch algorithm.[13]

Transformer engine

[edit]

The Hopper architecture was the first Nvidia architecture to implement the transformer engine.[14] The transformer engine accelerates computations by dynamically reducing them from higher numerical precisions (i.e., FP16) to lower precisions that are faster to perform (i.e., FP8) when the loss in precision is deemed acceptable.[14] The transformer engine is also capable of dynamically allocating bits in the chosen precision to either the mantissa or exponent at runtime to maximize precision.[15]

Power efficiency

[edit]

The SXM5 form factor H100 has a thermal design power (TDP) of 700 watts. With regards to its asynchrony, the Hopper architecture may attain high degrees of utilization and thus may have a better performance-per-watt.[16]

Grace Hopper

[edit]
Grace Hopper GH200
Designed byNvidia
Manufactured by
Fabrication processTSMC 4N
Codename(s)Grace Hopper
Specifications
ComputeGPU: 132 Hopper SMs
CPU: 72 Neoverse V2 cores
Shader clock rate1980 MHz
Memory supportGPU: 96 GB HBM3 or 144 GB HBM3e
CPU: 480 GB LPDDR5X

The GH200 combines a Hopper-based H100 GPU with a Grace-based 72-core CPU on a single module. The total power draw of the module is up to 1000 W. CPU and GPU are connected via NVLink, which provides memory coherence between CPU and GPU memory.[17]

History

[edit]

In November 2019, a well-known Twitter account posted a tweet revealing that the next architecture after Ampere would be called Hopper, named after computer scientist and United States Navy rear admiral Grace Hopper, one of the first programmers of the Harvard Mark I. The account stated that Hopper would be based on a multi-chip module design, which would result in a yield gain with lower wastage.[18]

During the 2022 Nvidia GTC, Nvidia officially announced Hopper.[19] By 2023, during the AI boom, H100s were in great demand. Larry Ellison of Oracle Corporation said that year that at a dinner with Nvidia CEO Jensen Huang, he and Elon Musk of Tesla, Inc. and xAI "were begging" for H100s, "I guess is the best way to describe it. An hour of sushi and begging".[20]

In January 2024, Raymond James Financial analysts estimated that Nvidia was selling the H100 GPU in the price range of $25,000 to $30,000 each, while on eBay, individual H100s cost over $40,000.[21] As of February 2024, Nvidia was reportedly shipping H100 GPUs to data centers in armored cars.[22]

H100 accelerator and DGX H100

[edit]

Comparison of accelerators used in DGX:[23][24][25]

Model Architecture Socket FP32
CUDA
cores
FP64 cores
(excl. tensor)
Mixed
INT32/FP32
cores
INT32
cores
Boost
clock
Memory
clock
Memory
bus width
Memory
bandwidth
VRAM Single
precision
(FP32)
Double
precision
(FP64)
INT8
(non-tensor)
INT8
dense tensor
INT32 FP4
dense tensor
FP16 FP16
dense tensor
bfloat16
dense tensor
TensorFloat-32
(TF32)
dense tensor
FP64
dense tensor
Interconnect
(NVLink)
GPU L1 Cache L2 Cache TDP Die size Transistor
count
Process Launched
B200 Blackwell SXM6 N/A N/A N/A N/A N/A 8 Gbit/s HBM3e 8192-bit 8 TB/sec 192 GB HBM3e N/A N/A N/A 4.5 POPS N/A 9 PFLOPS N/A 2.25 PFLOPS 2.25 PFLOPS 1.2 PFLOPS 40 TFLOPS 1.8 TB/sec GB100 N/A N/A 1000 W N/A 208 B TSMC 4NP Q4 2024 (expected)
B100 Blackwell SXM6 N/A N/A N/A N/A N/A 8 Gbit/s HBM3e 8192-bit 8 TB/sec 192 GB HBM3e N/A N/A N/A 3.5 POPS N/A 7 PFLOPS N/A 1.98 PFLOPS 1.98 PFLOPS 989 TFLOPS 30 TFLOPS 1.8 TB/sec GB100 N/A N/A 700 W N/A 208 B TSMC 4NP
H200 Hopper SXM5 16896 4608 16896 N/A 1980 MHz 6.3 Gbit/s HBM3e 6144-bit 4.8 TB/sec 141 GB HBM3e 67 TFLOPS 34 TFLOPS N/A 1.98 POPS N/A N/A N/A 990 TFLOPS 990 TFLOPS 495 TFLOPS 67 TFLOPS 900 GB/sec GH100 25344 KB (192 KB × 132) 51200 KB 1000 W 814 mm2 80 B TSMC 4N Q3 2023
H100 Hopper SXM5 16896 4608 16896 N/A 1980 MHz 5.2 Gbit/s HBM3 5120-bit 3.35 TB/sec 80 GB HBM3 67 TFLOPS 34 TFLOPS N/A 1.98 POPS N/A N/A N/A 990 TFLOPS 990 TFLOPS 495 TFLOPS 67 TFLOPS 900 GB/sec GH100 25344 KB (192 KB × 132) 51200 KB 700 W 814 mm2 80 B TSMC 4N Q3 2022
A100 80GB Ampere SXM4 6912 3456 6912 N/A 1410 MHz 3.2 Gbit/s HBM2e 5120-bit 1.52 TB/sec 80 GB HBM2e 19.5 TFLOPS 9.7 TFLOPS N/A 624 TOPS 19.5 TOPS N/A 78 TFLOPS 312 TFLOPS 312 TFLOPS 156 TFLOPS 19.5 TFLOPS 600 GB/sec GA100 20736 KB (192 KB × 108) 40960 KB 400 W 826 mm2 54.2 B TSMC N7 Q1 2020
A100 40GB Ampere SXM4 6912 3456 6912 N/A 1410 MHz 2.4 Gbit/s HBM2 5120-bit 1.52 TB/sec 40 GB HBM2 19.5 TFLOPS 9.7 TFLOPS N/A 624 TOPS 19.5 TOPS N/A 78 TFLOPS 312 TFLOPS 312 TFLOPS 156 TFLOPS 19.5 TFLOPS 600 GB/sec GA100 20736 KB (192 KB × 108) 40960 KB 400 W 826 mm2 54.2 B TSMC N7
V100 32GB Volta SXM3 5120 2560 N/A 5120 1530 MHz 1.75 Gbit/s HBM2 4096-bit 900 GB/sec 32 GB HBM2 15.7 TFLOPS 7.8 TFLOPS 62 TOPS N/A 15.7 TOPS N/A 31.4 TFLOPS 125 TFLOPS N/A N/A N/A 300 GB/sec GV100 10240 KB (128 KB × 80) 6144 KB 350 W 815 mm2 21.1 B TSMC 12FFN Q3 2017
V100 16GB Volta SXM2 5120 2560 N/A 5120 1530 MHz 1.75 Gbit/s HBM2 4096-bit 900 GB/sec 16 GB HBM2 15.7 TFLOPS 7.8 TFLOPS 62 TOPS N/A 15.7 TOPS N/A 31.4 TFLOPS 125 TFLOPS N/A N/A N/A 300 GB/sec GV100 10240 KB (128 KB × 80) 6144 KB 300 W 815 mm2 21.1 B TSMC 12FFN
P100 Pascal SXM/SXM2 N/A 1792 3584 N/A 1480 MHz 1.4 Gbit/s HBM2 4096-bit 720 GB/sec 16 GB HBM2 10.6 TFLOPS 5.3 TFLOPS N/A N/A N/A N/A 21.2 TFLOPS N/A N/A N/A N/A 160 GB/sec GP100 1344 KB (24 KB × 56) 4096 KB 300 W 610 mm2 15.3 B TSMC 16FF+ Q2 2016

References

[edit]

Citations

[edit]
  1. ^ Elster & Haugdahl 2022, p. 4.
  2. ^ Nvidia 2023c, p. 20.
  3. ^ Nvidia 2023b, p. 9.
  4. ^ Fujita et al. 2023, p. 6.
  5. ^ Nvidia 2023b, p. 9-10.
  6. ^ a b Nvidia 2023b, p. 10.
  7. ^ Vishal Mehta (September 2022). CUDA Programming Model for Hopper Architecture. Santa Clara: Nvidia. Retrieved May 29, 2023.
  8. ^ Fujita et al. 2023, p. 4.
  9. ^ a b Nvidia 2023b, p. 11.
  10. ^ Nvidia 2023b, p. 12.
  11. ^ Nvidia 2023a, p. 44.
  12. ^ Tirumala, Ajay; Eaton, Joe; Tyrlik, Matt (December 8, 2022). "Boosting Dynamic Programming Performance Using NVIDIA Hopper GPU DPX Instructions". Nvidia. Retrieved May 29, 2023.
  13. ^ Harris, Dion (March 22, 2022). "NVIDIA Hopper GPU Architecture Accelerates Dynamic Programming Up to 40x Using New DPX Instructions". Nvidia. Retrieved May 29, 2023.
  14. ^ a b Salvator, Dave (March 22, 2022). "H100 Transformer Engine Supercharges AI Training, Delivering Up to 6x Higher Performance Without Losing Accuracy". Nvidia. Retrieved May 29, 2023.
  15. ^ "Nvidia's Next GPU Shows That Transformers Are Transforming AI - IEEE Spectrum". spectrum.ieee.org. Retrieved October 23, 2024.
  16. ^ Elster & Haugdahl 2022, p. 8.
  17. ^ "NVIDIA: Grace Hopper Has Entered Full Production & Announcing DGX GH200 AI Supercomputer". Anandtech. May 29, 2023.
  18. ^ Pirzada, Usman (November 16, 2019). "NVIDIA Next Generation Hopper GPU Leaked – Based On MCM Design, Launching After Ampere". Wccftech. Retrieved May 29, 2023.
  19. ^ Vincent, James (March 22, 2022). "Nvidia reveals H100 GPU for AI and teases 'world's fastest AI supercomputer'". The Verge. Retrieved May 29, 2023.
  20. ^ Fitch, Asa (February 26, 2024). "Nvidia's Stunning Ascent Has Also Made It a Giant Target". The Wall Street Journal. Retrieved February 27, 2024.
  21. ^ Vanian, Jonathan (January 18, 2024). "Mark Zuckerberg indicates Meta is spending billions of dollars on Nvidia AI chips". CNBC. Retrieved June 6, 2024.
  22. ^ Bousquette, Isabelle; Lin, Belle (February 14, 2024). "Armored Cars and Trillion Dollar Price Tags: How Some Tech Leaders Want to Solve the Chip Shortage". The Wall Street Journal. Retrieved May 30, 2024.
  23. ^ Smith, Ryan (March 22, 2022). "NVIDIA Hopper GPU Architecture and H100 Accelerator Announced: Working Smarter and Harder". AnandTech.
  24. ^ Smith, Ryan (May 14, 2020). "NVIDIA Ampere Unleashed: NVIDIA Announces New GPU Architecture, A100 GPU, and Accelerator". AnandTech.
  25. ^ "NVIDIA Tesla V100 tested: near unbelievable GPU power". TweakTown. September 17, 2017.

Works cited

[edit]

Further reading

[edit]