search for: gicv2

...to the fact how our Fast Models are implemented, I mostly us block > devices and 9p protocol over virtio and I get enough performance from > them :-). > >> As arm doesn't support msi-x now, > > To be precise: "ARM" does "support" MSI-X :-) (google for GICv2m) Sorry, I mean ARM with GICv2. > > The correct statement would be: "normal memory mapped devices have no > interface for message signalled interrupts (like MSI-X)" > Yes, that's right. >> we use GSI for multiple irq. > > I'm not sure what GSI stands...

...to the fact how our Fast Models are implemented, I mostly us block > devices and 9p protocol over virtio and I get enough performance from > them :-). > >> As arm doesn't support msi-x now, > > To be precise: "ARM" does "support" MSI-X :-) (google for GICv2m) Sorry, I mean ARM with GICv2. > > The correct statement would be: "normal memory mapped devices have no > interface for message signalled interrupts (like MSI-X)" > Yes, that's right. >> we use GSI for multiple irq. > > I'm not sure what GSI stands...

Hi I'm running Openstack which is installed by using devstack. But it is not launching VMs. >From command line with gic_version=3 option it is running. But openstack glance doesn't have any privilege to specify gic version. On my ARM64 board gicv2 is not supported, so i want to make gicv3 as default one to pass to qemu. Kindly suggest any specific version of vibvirt or patch such that libvirt should pass gicv3 as default one. Thanks in Advance -Vishnu.

On 11/05/2014 03:12 AM, Shannon Zhao wrote: > Hi R?my, > > On 2014/11/5 16:26, GAUGUEY R?my 228890 wrote: >> Hi Shannon, >> >>> Type of backend bandwith(GBytes/sec) >>> virtio-net 0.66 >>> vhost-net 1.49 >>> vhost-net with irqfd 2.01 >>> >>> Test cmd: ./iperf -c 192.168.0.2 -P 1 -i 10

On 11/05/2014 03:12 AM, Shannon Zhao wrote: > Hi R?my, > > On 2014/11/5 16:26, GAUGUEY R?my 228890 wrote: >> Hi Shannon, >> >>> Type of backend bandwith(GBytes/sec) >>> virtio-net 0.66 >>> vhost-net 1.49 >>> vhost-net with irqfd 2.01 >>> >>> Test cmd: ./iperf -c 192.168.0.2 -P 1 -i 10

...at all, sorry. > > > The approach with a shared memory region is dropped as you can see from the mailing list. Glad to hear that :-) > The approach of this patch get a net performance improvement about 30%. > This maybe makes sense to the paltform without MSI support(e.g ARM with GICv2). I appreciate the improvement. I'm just cautions when it comes to significant changes to the standard so late in the process. Pawel

...>> The approach with a shared memory region is dropped as you can see from the mailing list. > > Glad to hear that :-) > >> The approach of this patch get a net performance improvement about 30%. >> This maybe makes sense to the paltform without MSI support(e.g ARM with GICv2). > > I appreciate the improvement. I'm just cautions when it comes to > significant changes to the standard so late in the process. > Sorry, I didn't notice it's at the final phase of the process. It's the first time in my life to make a change for virtio-spec like y...

...>> The approach with a shared memory region is dropped as you can see from the mailing list. > > Glad to hear that :-) > >> The approach of this patch get a net performance improvement about 30%. >> This maybe makes sense to the paltform without MSI support(e.g ARM with GICv2). > > I appreciate the improvement. I'm just cautions when it comes to > significant changes to the standard so late in the process. > Sorry, I didn't notice it's at the final phase of the process. It's the first time in my life to make a change for virtio-spec like y...

As the current virtio-mmio only support single irq, so some advanced features such as vhost-net with irqfd are not supported. And the net performance is not the best without vhost-net and irqfd supporting. This patch support virtio-mmio to request multiple irqs like virtio-pci. With this patch and qemu assigning multiple irqs for virtio-mmio device, it's ok to use vhost-net with irqfd on

As the current virtio-mmio only support single irq, so some advanced features such as vhost-net with irqfd are not supported. And the net performance is not the best without vhost-net and irqfd supporting. This patch support virtio-mmio to request multiple irqs like virtio-pci. With this patch and qemu assigning multiple irqs for virtio-mmio device, it's ok to use vhost-net with irqfd on

...ation, load-balancing, hardware offload and other advanced features. A basic patch set has already been committed to head including: * PCIe Root Complex support * Cache Coherency Unit driver * North Bridge Service driver * Updated Alpine HAL * Extended MSI support in GICv2 and GICv3 code Additional work, such as an MSI-X driver and full Ethernet support, is currently undergoing community review on Phabricator. The multi-user SMP system is stable and fully working, along with the 1G and 10G Ethernet links. The interrupt management code has been adjus...

...ation, load-balancing, hardware offload and other advanced features. A basic patch set has already been committed to head including: * PCIe Root Complex support * Cache Coherency Unit driver * North Bridge Service driver * Updated Alpine HAL * Extended MSI support in GICv2 and GICv3 code Additional work, such as an MSI-X driver and full Ethernet support, is currently undergoing community review on Phabricator. The multi-user SMP system is stable and fully working, along with the 1G and 10G Ethernet links. The interrupt management code has been adjus...