-
Notifications
You must be signed in to change notification settings - Fork 16
Home
Jan Eitzinger edited this page Sep 28, 2021
·
56 revisions
The following results are created to the best of our knowledge. The results are created using this script and afterwards reviewed for correctness. Here is documentation how to use this script. In this article we share our take on the memory wall.
The results do not only depend on the architecture, but among other things also on the toolchain, the test case, the number of cores used (if applicable) as well as system and frequency settings. The following bar graphs show the best result for single core and single socket. For details have a look on the specific processor result pages below.

AMD processors:
- Interlagos 2S 2M 16C (2011)
- Naples 2S 4M 24C (2017)
- Rome NPS1 2S 32C (2019)
- Rome NPS2 2S 32C (2019)
- Rome NPS4 2S 32C (2019)
- Rome NPS1 2S 64C (2019)
- Rome NPS2 2S 64C (2019)
- Rome NPS4 2S 64C (2019)
Intel processors:
- SandyBridge 2S 1M 8C (2012)
- IvyBridge 2S 1M 10C (2013)
- Haswell 2S 1M 14C (2014)
- Haswell CoD 2S 2M 14C (2014)
- Broadwell 2S 1M 18C (2016)
- Broadwell CoD 2S 2M 18C (2016)
- SkyLake 2S 1M 20C (2017)
- SkyLake SNC 2S 2M 20C (2017)
- CascadeLake 2S 1M 20C (2019)
- CascadeLake SNC 2S 2M 20C (2019)
- Icelake 2S 1M 36C (2021)
ARM processors:
- Fujitsu FX700 1S 4M 48C (2020)
- Fujitsu FX1000 1S 4M 48C (2020)