Questions tagged [rdma]

RDMA refers to "Remote Direct Memory Access," which is a set of networking technologies typically used for high performance, low latency communication.

RDMA refers to "Remote Direct Memory Access," which is a set of networking technologies typically used for high performance, low latency communication.

RDMA networks have three main attributes:

  1. Asynchronous queueing: network operations are submitted by adding requests "work queues" which are asynchronously executed by hardware; when complete, the result is returned to software in a "completion queue."
  2. Kernel bypass: userspace processes can submit operations directly to the network adapter hardware without any system calls.
  3. One-sided operations: one system can read from or write into the memory of a remote system without involving any software on the remote system.
177 questions
1
vote
1 answer

To create multiple Queue Pairs in RDMA

I am a new newbie in RDMA programming. My workflow is as follows. struct rdma_cm_id *cm_client_id = NULL; struct rdma_cm_event *cm_event = NULL; ret = process_rdma_cm_event(cm_event_channel, RDMA_CM_EVENT_CONNECT_REQUEST, &cm_event); cm_client_id…
HuangJie
  • 1,488
  • 1
  • 16
  • 33
1
vote
1 answer

Is it possible to use RDMA Mellanox libraries from within a kernel module?

I want to develop a kernel module that is able to send/receive RDMA messages. I am wondering if the Mellanox libraries can be called from kernel space. Can I call Mellanox RDMA functions from a kernel module? Answer: I have some working code here:…
JC1
  • 657
  • 6
  • 21
1
vote
2 answers

What is the difference between RDMA and DMA?

I'm concern about hardware component that are involve in Direct Memory Access (DMA). I read that for DMA to work there is the need of DMA-controller, however I did not found if the same component is involve in the RDMA (Remote DMA) operation such as…
Fopa Léon Constantin
  • 11,863
  • 8
  • 48
  • 82
1
vote
0 answers

what are the purpose of `MLX4_EVENT_TYPE_COMP` and `MLX4_EVENT_TYPE_CMD` in mlnx-ofed-kernel-3.1 interrupt handler?

I'm using a mellanox Infiniband card MT26428 [ConnectX VPI PCIe 2.0 5GT/s - IB QDR / 10GigE] with mlnx-ofed-kernel-3.1 on a linux kernel version 3.13.0. When the card is connected to another one and configured (ibstat said both cards are active in…
Fopa Léon Constantin
  • 11,863
  • 8
  • 48
  • 82
1
vote
2 answers

How RDMA WRITE/READ behaves when the remote memory buffer is unpinned by the OS?

let two infiniband process execute on two machines. When one process terminates, all its resources (allocated file, buffers, ...) are freed even the memory that the process registered for RDMA WRITE/READ operations. Can someone has references on…
Fopa Léon Constantin
  • 11,863
  • 8
  • 48
  • 82
1
vote
1 answer

Why disabling IRQ on linux causes rdma_read and rdma_write to fail?

I have two host machines connected by Mellanox infiniband HCA. I'm executing a simple RDMA application to perform RDMA write and RDMA read operation from one machine (client) on the other machine (server). To know which interrupts are related to HCA…
Fopa Léon Constantin
  • 11,863
  • 8
  • 48
  • 82
1
vote
1 answer

How do I get error information from Infiniband verbs functions?

In particular, I'd like to know why ibv_create_qp fails. When it returns NULL, indicating that it failed, is there a way to find out exactly what the problem was?
1
vote
0 answers

rping server crashes in ported kernel

I have installed Soft-RoCE and the rdma connection between two nodes are ensured successfully using "rping" in Kernel 3.0.0+. I need to do the same in the kernel 4.0.4. I have copied "rxe driver (/driver/infiniband/hw/rxe)" from 3.0.0+ to kernel…
Dinesh Kb
  • 31
  • 2
1
vote
1 answer

Choices for shared memory system, MPI library, original RDMA or ULP over RDMA?

I am new on High Performance Computing (HPC), but I am going to have a HPC project, so I need some help to solve some fundamental problems. The application scenario is simple: Several servers connected by the InfiniBand (IB) network, one server for…
foool
  • 1,462
  • 1
  • 15
  • 29
1
vote
1 answer

RDMA atomic operation's implementation

I heard about that RDMA reads and writes are implemented like, when requests come the nic find the physical page and then using DMA to get the data to nic and then to the target. This is straightforward for reads and write, but it is odd when it…
1
vote
1 answer

RDMA CM for Windows

Does anyone know if there is a Windows equivalent for the RDMA Communication Manager found on Linux (http://linux.die.net/man/7/rdma_cm -- librdmacm), and also ib verbs (libibverbs)? Both these libraries are available when installing drivers for…
1
vote
0 answers

Error in RDMA Atomic operations

I have a problem with executing RDMA atomic operations (FETCH_ADD and CMP_AND_SWAP). When I try to submit an atomic RDMA request, the ibv_post_send() function fails, with Errno set to "Invalid argument". I have no such problems with RDMA READ/WRITE.…
narengi
  • 1,345
  • 3
  • 17
  • 38
1
vote
1 answer

Multi-node connection management

I'm designing an application involving multi-node communications using Infiniband (ibv_*). What is the standard way to keep connections between nodes? I'm thinking of O(N^2) connections for all pairs of node as the easiest one, but it's kind of…
w00d
  • 5,416
  • 12
  • 53
  • 85
1
vote
2 answers

Has someone tried SoftiWARP?

I'm trying to use SoftiWARP for a project. I installed it using instructions from here, which I followed to the word, except I'm using Ubuntu 10.04 server. To test it out, I used the client-server programs from the-geek-in-the-corner…
Anuj Kalia
  • 803
  • 8
  • 16
1
vote
4 answers

RDMA over ethernet

I have never used RDMA. Though on single PCs, I have run commands that uses DMA for memory transfer. Can I use RDMA over ethernet ( I am using switch that support 10BASE-T, 100BASE-TX). I will be using MPI to access data from two seperate machines…
gpuguy
  • 4,607
  • 17
  • 67
  • 125