The Tesla K20 offers 3x double precision performance (baseline not stated).
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.
The evidence is the person's or their organisation's own records.
Said for an organisation they lead or work for; it counts on both records.
What would count as true or false
True: NVIDIA's official Tesla K20 materials from around May 2012 claim 3x double-precision performance relative to the prior (Fermi) generation, and the published peak DP specifications support a ratio of about 3x (roughly 2.7x to 3.3x versus the comparable Fermi Tesla part). Mostly true: NVIDIA made the 3x claim, but the actual spec ratio is somewhat different (about 2.4x to 2.7x, or 3.3x to 3.6x), or the 3x applied to a closely related baseline. Mixed: the 3x figure applies only to a different metric, such as performance per watt, while raw DP gain is materially lower. Mostly false: the DP improvement was around 2x or less, and the 3x claim distorts a different figure. False: there was no such claim, or the DP performance did not improve substantially.
Set before the research began.
Evidence (4)
-
E1 [PDF] nvidia® tesla® kepler - gpu computing accelerators NVIDIA · May 2012 Own record
“Designed for double precision applications and the broader supercomputing market, the Tesla K20 delivers 3x the double precision performance compared to the previous generation Fermi-based Tesla M2090, in the same power envelope.”
-
E2 [PDF] nvidia® tesla® kepler - gpu computing accelerators NVIDIA · May 2012 Own record
“1Based on DGEMM performance: Tesla M2090 (Fermi) = 330 gigaflops, Tesla K20 (expected) > 1000 gigaflops”
-
E3 About NVIDIA NVIDIA · 15 May 2012 Own record
“Tesla K20にはGK110 Kepler GPUが採用されています。このGPUはFermiアーキテクチャーをベースとしたTesla製品に対して倍精度能力が3倍に強化されている”
-
E4 [PDF] nvidia® tesla® kepler - gpu computing accelerators NVIDIA · May 2012 Own record
“With more than one teraflop peak double precision performance, the Tesla K20 is ideal for a wide range of high performance computing workloads”
Reasoning
NVIDIA's May 2012 Tesla Kepler datasheet says the K20 delivers 3x the double-precision performance of the Fermi-based Tesla M2090 in the same power envelope (E1). The May 2012 Japanese press release makes the same 3x claim (E3). The datasheet's footnote bases this on DGEMM: 330 gigaflops for the M2090 against an expected figure above 1000 gigaflops for the K20, a ratio of about 3x (E2). That figure is a projection, and while the datasheet gives the K20's peak double precision as more than one teraflop (E4), the evidence never states the M2090's peak, so the peak-to-peak ratio the criteria ask for cannot be confirmed; all of these sources are NVIDIA's own.
Cite this claim
Jensen Huang, “The Tesla K20 offers 3x double precision performance (baseline not stated).” (said May 2012). Verdict: mostly true. Reliability API, claim c_928751c7b6b5, method 0.1.0+03d23c0e, dataset release v1.1.2. https://epistacklabs.com/claims/c_928751c7b6b5
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.