MLPerf retweetou
We've just released @MLPerf Mobile version 5.0.2 on the Google Play Store and GitHub. This version adds support for devices based on the Samsung Exynos 2500 SoC.
play.google.com/store/apps/det…
English
MLPerf
97 posts
MLPerf
@MLPerf
Building fair, useful, industry-standard benchmarks for ML training and inference performance of hardware, software, and services from TinyML to supercomputers
Entrou em Ocak 2019
2 Seguindo704 Seguidores
MLPerf retweetou
@MLCommons & @AVCConsortium announce @MLPerf Automotive v0.5 benchmark results—a major step for transparent, reproducible automotive AI performance data. mlcommons.org/2025/08/mlperf…
English
@zhentan We prefer Tekken, @TheKanter happens to be quite fond of it from his college days!
English
@MLPerf please add to your rankings! 😬
Matthew Berman@MatthewBerman
LLM Fighter - A new way to benchmark LLMs (not April Fools) What if you let the top AI models take over Street Fighter? Which model would win? What does it tell you about the performance of the model? Here's a full breakdown and tutorial:
English
@sandyasm @JeffDean @matei_zaharia @NaveenGRao @HazyResearch @DrJimFan You should reach out to @TheKanter if you are interested!
English
are there end to end system level benchmarks for AI than just model level (@mlperf)? TPCx-AI is purely classical ML, Any open industry effort on end to end AI benchmarks including data proc.. eg RAG, Agent workflow? @JeffDean @matei_zaharia @NaveenGRao @HazyResearch @DrJimFan
English
@SwayStar123 @EMostaque @intel @HabanaLabs The dataset for MLPerf Stable Diffusion is ~6M images, and is likely smaller than what @EMostaque uses in production training.
English
@EMostaque @MLPerf @intel @HabanaLabs Wow, so how much would that cost to rent. I'm guessing an hour is the minimum at the least. So Probably more efficient to use 16 gaudi 2s instead
English
Latest @MLPerf number show how @intel @HabanaLabs #Gaudi2 AI chip is top on price/perf even according to NVIDIA's own numbers on stable diffusion (!)
Now it has fp8 it beats A100 easily w/ 96Gb HBM, 2.4 TB/s interconnect great work by $INTC
AMA on them
servethehome.com/nvidia-shows-i…
English
MLPerf retweetou
MLPerf retweetou
On a personal level, I loved seeing .@MLPerf being mentioned so much! I can’t wait to see more submissions from across the industry as we push for the next 1000x in machine learning performance.
English
MLPerf retweetou
@MLCommons @MLPerf The Perf Storage Benchmark Suite will be a critical resource in improving machine learning and we’re honored to have participated alongside these industry leaders and academic organizations!
English
MLPerf retweetou
Inference keeps getting faster ... and more demanding. @MLCommons
MLPerf Inference 3.1 adds large language model benchmarks for inference venturebeat.com/ai/mlperf-3-1-… via @VentureBeat
"“We’re evolving the benchmark suite to reflect what’s going on,” @TheKanter said.
English
MLPerf retweetou
@MLCommons, we just released! New @MLPerf Inference and Storage results. Record participation in MLPerf Inference v3.1 and first-time MLPerf Storage v0.5 results highlight the growing importance of GenAI and storage. See all the results and learn more mlcommons.org/en/news/mlperf…
English
MLPerf retweetou
AI semiconductor company @Syntiantcorp announced its Syntiant Core 2™ programmable #deeplearning #architecture ranked as utilizing the lowest power #energy across three categories in @MLCommons @MLPerf Tiny v1.1 benchmark suite. Read more about energy performance below:
Syntiant Corp.@Syntiantcorp
Great news! Our second-generation architecture achieved the lowest power energy performance across three categories in the most recent @MLCommons @MLPerf Tiny v1.1 benchmark suite. Click bit.ly/3PJGd8t to read the full results. #edgeAI #tinyML
English
MLPerf retweetou
If you are at the @databricks Data+AI Summit this week be sure to check out @MLCommons Exec Director @TheKanter's session - Advancing ML Through Benchmarks & Data, Thursday, June 29 at 2:30pm
databricks.com/dataaisummit/s…
English
@braindersnn Maybe you should move the party over to .@MLCommons…I hear they are pretty nice!
English
It’s been great to see so many neuromorphs come together for #NeuroBench. I am hopeful that it will do for #neuromorphics what @MLPerf did for traditional ML accelerators.
Get involved here:
neurobench.ai
Check out the whitepaper here:
arxiv.org/abs/2304.04640
English
MLPerf retweetou
Calling all ISC attendees, stop by the @MLPerf roundtable 5/24 at 9am, Hall E - 2nd floor, to learn more about the MLPerf HPC benchmark suite V3.0 from the experts who created it. #mlcommons #MachineLearning #ISC23
English
Driving ML Forward in Automotive - David Kanter - 2023 CASPA Spring Symp... youtu.be/-TxJbUSv7EA via @YouTube
YouTube
English
MLPerf retweetou
MLCommons: MLPerf Inference Delivers Power Efficiency and Performance Gains
ow.ly/5vE250NBs04 #HPC @MLCommons @MLPerf
English
MLPerf retweetou
Learn about the business implications and check out the technical details of our @MLPerf submission in our blog.
neuralmagic.com/blog/neural-ma…
English
MLPerf retweetou
MLPerf retweetou
New @MLPerf inference results from us at @neuralmagic are now up! We show 1,000x performance improvements and 92% reduction in energy consumption. Dive into the blog to learn more: neuralmagic.com/blog/neural-ma…
#mlops #mlperf #ml
English