NVIDIA's Kepler SMX delivers three times the energy-efficient performance of the Fermi SM.
Mostly true True in substance, wrong in a detail.
Factual: Statements about the present or the past, checked against the record: did they hold up?
The second judge would not resolve this claim; the verdict shown is the first judge's.
Said for an organisation they lead or work for; it counts on both records.
What would count as true or false
Judged as of May 2012. True: NVIDIA's Kepler GK110/GK104 whitepaper or architecture documentation states that the Kepler SMX delivers about 3x the performance per watt of the Fermi SM, and independent measurements are broadly consistent with it. Mostly true: official figures support about 3x per watt for the SMX design (e.g., at peak FP32), while real-world gains were somewhat lower (around 2–3x). Mostly false: the 3x figure applies only to a narrow metric or workload and typical gains were substantially lower (under 2x). False: official or independent data show perf/watt gains far below the claim or no such comparison exists.
Set before the research began.
Evidence (4)
-
E1 NVIDIA Pioneers New Standard for High Performance Computing With Tesla GPUs Built on Kepler Architecture NVIDIA · 15 May 2012 Own record
“It delivers up to three times more performance per watt than the Fermi streaming multiprocessor, making it possible to build a supercomputer that delivers one petaflop of computing performance in just 10 server racks.”
-
E2 NVIDIA Launches Kepler Into HPC HPCwire · 15 May 2012
“In doing so, they were able to realize about three times the performance per watt of the Fermi GPU.”
-
E3 GTX 680 Kepler Whitepaper NVIDIA · 20 March 2012 Own record
“but with 1536 CUDA cores in GK104, the GeForce GTX 680 SMX provides 2x the performance per watt of Fermi’s SM (GF110).”
-
E4 NVIDIA Launches Kepler Into HPC HPCwire · 15 May 2012
“According to NVIDIA, initial testing with LAMMPS, NAMD, seismic processing, and certain integer-centric defense codes showed performance increases between 1.5X and 2.0X compared to the Fermi M2090.”
Reasoning
NVIDIA's May 2012 Kepler announcement says the SMX delivers "up to three times more performance per watt" than the Fermi SM (E1), and HPCwire repeats the figure of about 3x (E2). However, NVIDIA's own GK104 (GTX 680) whitepaper gives 2x the performance per watt of Fermi's SM (E3). According to NVIDIA, initial application tests showed 1.5x to 2.0x the performance of the Fermi M2090 (E4), but that is raw performance, not performance per watt. NVIDIA's official 3x figure for the Kepler SMX is stated as an upper bound ("up to") (E1), and NVIDIA's GK104 whitepaper gives a lower 2x performance-per-watt figure (E3), which fits "mostly true" rather than fully true.
Cite this claim
Jensen Huang, “NVIDIA's Kepler SMX delivers three times the energy-efficient performance of the Fermi SM.” (said May 2012). Verdict: mostly true. Reliability API, claim c_dd5e4ad52f53, method 0.1.0+03d23c0e, dataset release v1.1.2. https://epistacklabs.com/claims/c_dd5e4ad52f53
This claim is in dataset release v1.1.2, with the same verdict.
Reply to this claim
More factual claims by Jensen Huang
-
Mostly true
“we invented a technology called a GPU, and basically a GPU does parallel processing tasks”
-
Resolved false
“Today, NVIDIA’s token cost is the lowest in the world, not by 10%, by X factors, orders of magnitude, all because we did extreme co-design”
-
Mostly true
“It’s the first one to be PCI Express Gen 6.”
-
Mixed
“We have demonstrated that NVFP4 can achieve inference without loss of precision, while delivering significant performance and energy efficiency improvements, and it is equally suitable for training.”
-
Mixed
“Over the past two years, the computational demands for inference have increased approximately 10,000 times, while usage has increased approximately 100 times.”
Verdicts are made automatically; how verdicts are made.