aiqbench analysis
This paper provides a rigorous, empirical analysis of bypassing NVIDIA driver-level throttling on RTX 2080 Ti GPUs to unlock full FP16 Tensor Core throughput for LLM pre-training. The study quantifies the resulting speedups across different model sizes while confirming that training dynamics and output quality remain bit-identical, offering practical guidance for researchers constrained by hardware budgets.
Unlocking Turing Tensor Cores: A Measured Throughput Account of From-Scratch Pre-training on a Single 22G RTX 2080 Ti
arXiv:2611.05419v1 Announce Type: cross Abstract: The RTX 2080 Ti (TU102) ships full second-generation Tensor Cores in hardware, yet the vendor throttles FP16 matrix multiplication at the driver level, leaving usable tensor throughput well below the hardware peak. turing-unlock lifts this restriction with a kernel module that, before the nvidia driver binds the device, writes the FP16 speed selectors (SS0=0x88888888, SS1=0x8) and releases the ACR mutex; published measurements report cuBLAS FP16 matmul throughput rising from 55.7 to 91.4 TFLOPS (about 1.6x). On the same 22G 2080 Ti and under the same from-scratch pre-training protocol as the companion memory ledger (five tiers of Chinese encoder-decoder models, 2,451 steps per tier, 16,384 tokens per step, seed 42), we obtain a before/after throughput account: the small tier (75.7M parameters, 512-wide, 8+8 layers, activation checkpointing on) improves in steady-state throughput from 6,480 to 7,246 tok/s (+11.8%) with wall time cut from 1.77h to 1.61h (9.0% saved); the tiny tier goes from 15,150 to 15,866 tok/s (+4.7%) with wall time savings of 3.8%; the three smaller tiers (pico-micro) gain at most +0.3%; per-tier peak memory stays bit-identical, and final quality metrics are bit-identical or consistent within 10^-3. We conclude that the unlock gain is a function of matrix size, rising monotonically with tier -- imperceptible for pico-micro, visible from tiny upward, and cashed out as a double-digit percentage at small -- and that it is a pure throughput gain that leaves training dynamics untouched.
Abstract:The RTX 2080 Ti (TU102) ships full second-generation Tensor Cores in hardware, yet the vendor throttles FP16 matrix multiplication at the driver level, leaving usable tensor throughput well below the hardware peak. turing-unlock lifts this restriction with a kernel module that, before the nvidia driver binds the device, writes the FP16 speed selectors (SS0=0x88888888, SS1=0x8) and releases the ACR mutex; published measurements report cuBLAS FP16 matmul throughput rising from 55.7 to 91.4 TFLOPS (about 1.6x). On the same 22G 2080 Ti and under the same from-scratch pre-training protocol as the companion memory ledger (five tiers of Chinese encoder-decoder models, 2,451 steps per tier, 16,384 tokens per step, seed 42), we obtain a before/after throughput account: the small tier (75.7M parameters, 512-wide, 8+8 layers, activation checkpointing on) improves in steady-state throughput from 6,480 to 7,246 tok/s (+11.8%) with wall time cut from 1.77h to 1.61h (9.0% saved); the tiny tier goes from 15,150 to 15,866 tok/s (+4.7%) with wall time savings of 3.8%; the three smaller tiers (pico-micro) gain at most +0.3%; per-tier peak memory stays bit-identical, and final quality metrics are bit-identical or consistent within 10^-3. We conclude that the unlock gain is a function of matrix size, rising monotonically with tier -- imperceptible for pico-micro, visible from tiny upward, and cashed out as a double-digit percentage at small -- and that it is a pure throughput gain that leaves training dynamics untouched.
Hardware Architecture (cs.AR); Computation and Language (cs.CL); Distributed, Parallel, and Cluster Computing (cs.DC); Performance (cs.PF)
C.4; I.2.1; I.2.6
arXiv:2611.05419 [cs.AR]
(or arXiv:2611.05419v1 [cs.AR] for this version)
https://doi.org/10.48550/arXiv.2611.05419 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 340 words