Details of the performance results of our work with Dell and Qualcomm for MLPerf Inference v3.0. In the datacenter category, the Dell R7515 server with four Qualcomm Cloud AI 100 Pro accelerators improves performance by up to 109 percent and power efficiency by up to 62 percent compared with the v2.1 submission.

Can acceleration contribute towards sustainability goals? In this article, Natalie Durzynski and Dr Anton Lokhmotov (KRAI) explain the role that accelerated computing can play in efficiency improvements.

Description: KRAI's Saheli Bhattacharjee and Anton Lokhmotov present our first submission to MLPerf Tiny benchmarks including our unique findings.

Our team proudly delivered over a quarter of all submissions to MLPerf Tiny v1.1.

With another successful round of submissions to MLCommons' MLPerf Inference done and dusted, our team at KRAI reflected on one of our most exciting achievements in this round, affectionately known as KILT.

Powered by our software technologies , Qualcomm's MLPerf Inference v3.0 submissions continue to demonstrate performance leadership and achieve highest power efficiency yet.

Using MLPerf results that we obtained on behalf of Qualcomm, Forbes zoom in on how the Qualcomm Cloud AI 100 accelerator outshines NVIDIA GPUs for Computer Vision.

Lenovo demonstrate a diverse portfolio of compute platforms for AI workloads, including results powered by our technologies.

Optimized and delivered by KRAI, industry-leading MLPerf benchmarking results underpin the strong collaboration between HPE and Qualcomm.

Our team, as MLPerf inference veterans, submitted a ResNet training score for a system with two Nvidia RTX A5000 GPUs of 284.038 min. This entry–level option may be compared with one of Nvidia’s results for a two A30 system, which managed the training in 235.574 min, while the A5000s consumed 39% more power and were 20% slower, they are also 2–3× cheaper.

Our MLPerf submissions on behalf of HPE deliver high performance and reduce latency associated with complex artificial intelligence and machine learning models.

We present how we solved the challenges we faced in achieving linear scaling for MLPerf™ Inference vision benchmarks on Datacenter and Edge servers equipped with Qualcomm Cloud AI 100 accelerators.

The article highlights exceptional MLPerf Inference v2.0 results that we obtained on behalf of Qualcomm in the Datacenter and Edge categories.

Our MLPerf Inference v2.0 submissions are noted as being the most numerous and some of the best performing in MLPerf.

Dr Anton Lokhmotov (KRAI CEO) explains the importance and benefits of MLPerf Inference benchmarks.

Dr Anton Lokhmotov (KRAI CEO) makes the case for industrial-grade benchmarking.