Tag: Graphcore
Google TPUv4 Selectively Joins NVIDIA MLPerf Training 2.0
Google TPUv4 selectively joins in the NVIDIA MLPerf Training 2.0 effort dropping NVIDIA-only benchmark tests from 6 of 8 to 5 of 8
Graphcore BOW IPU Launched
The Graphcore BOW IPU offers solid incremental performance gains by leveraging new packaging technologies for higher clock speeds
Graphcore IPU-POD128 and IPU-POD256 Extend Scale
With the new Graphcore IPU-POD128 and IPU-POD256 the company is demonstrating its capability to scale its solution to larger cluster sizes
Graphcore Colossus Mk2 IPU at Hot Chips 33
We get a deep dive into the Graphcore Colossus Mk2 IPU AI accelerator during the Hot Chips 33 (2021) presentation by the company
Inspur MLPerf Training v1.0 Insights
We get a chance to chat with Inspur about its MLPerf Training v1.0 results and some of the trends happening in the industry
Graphcore Celebrates a Stunning Loss at MLPerf Training v1.0
We dig into the strange case where Graphcore celebrated a stunning performance and price/ performance loss to NVIDIA in MLPerf Training v1.0
MLPerf Training v1.0 Results Still NVIDIA Led but with 500W NVIDIA...
MLPerf Training v1.0 again is primarily a NVIDIA benchmarking exercise but we see a few other interesting results in the mix
Intel IPU is an Exotic Answer to the Industry DPU
The Intel IPU is the company's exotic answer to the industry-wide DPU effort to separate data center infrastructure and application layers
Hot Chips 33 Advance Program Launched
The Hot Chips 33 advance program has been launched and features a number of very interesting talks coming in a few months
Graphcore GC200 IPU Launch Very Cool but Glaring Omission
We have some very cool technology with the Graphcore GC200 IPU launch, however, there was a glaring omission in the launch that seemed shocking