-
Notifications
You must be signed in to change notification settings - Fork 15
Home
Frank Haverkamp edited this page Mar 12, 2018
·
54 revisions
- CAPI accelerated GZIP Compression Adapter User’s guide
- Generic Work Queue Engine (GenWQE) Application Programming Guide
- Generic Work Queue Engine (GenWQE) introduction
- IBM® SDK, Java™ Technology Edition, Version 8: Enabling hardware compression acceleration on Power Systems
- Ubuntu GenWQE Packages
- RHEL GenWQE Packages
- [SuSE](FIXME ... put in link)
- EJ1A full-height: CAPI Compression Accelerator Adapter (FC EJ1A and EJ1B; CCIN 2CF0)
- EJ1B low-profile: CAPI Compression Accelerator Adapter (FC EJ1A and EJ1B; CCIN 2CF0)
- EJ12 full-height: PCIe3 FPGA Accelerator Adapter (FC EJ12; CCIN 59AB)
- EJ13 low-profile: PCIe3 LP FPGA Accelerator Adapter (FC EJ13; CCIN 59AB)
- Center for Genome Research and Biocomputing (CGRB): A better way to compress big data
- GenWQE Linux Device Driver required for the PCIe accelerator
- CXL Linux Kernel Support required for the CAPI accelerator
- Initial posting of GenWQE Linux device driver on LKML
- IBM LinuxONE™ and Open Source Demo
- POWER8 Scale-Out: CAPI & Hardware Acceleration
- POWER8 Coherent Accelerator Processor Interface (CAPI)
- Nallatech Collaborates with OpenPOWER Foundation
- POWER8 CAPI Flash: In-memory expansion to speed data access
- Convey CAPI Developer Kit for Xilinx FPGAs
- Alpha-Data: Coherent Accelerator Processor Interface (CAPI) for POWER8
- CAPI Developers Community
- IBM z Systems Data Compression (zEDC) for z/OS and zEDC Express
- New zEDC Express Adapter Compresses Data Faster While Using Less CPU
- Nice animation of GenWQE card
- Addressing Is The Secret Of Power8 CAPI
- Sam2bam: High-Performance Framework for NGS Data Preprocessing Tools
The following information is only needed for the PCIe version of the card.
Example how to setup the hard_limit parameter in your KVM-host XML config file:
<memory unit='KiB'>33554432</memory>
<currentMemory unit='KiB'>33554432</currentMemory>
<memtune>
<hard_limit unit='KiB'>134217728</hard_limit>
</memtune>
Not setting the hard_limit parameter correctly can result in severe performance reduction.