search for: ofa

...bw = 1.71 GB/sec The second result is bi directional, Gen4 on PCI-E V2 should get quite a lot better. Rob -----Original Message----- From: Jan Ruffing [mailto:ruffing@motama.com] Sent: 11 December 2008 16:18 To: Robert Dunkley Cc: Gilad Shainer; general@lists.openfabrics.org Subject: Re: [ofa-general] Infiniband performance Robert Dunkley wrote: > The newer gen4 20Gb cards get about 14-15Gbit/sec because of their better lower latency processing and support for higher bandwidth PCI-E 2.0 (PCI-E is not 100% efficient so a 20Gbit PCIE connect will hold back a 20Gbit IC). > > T...

...- Performance graph at: > http://mvapich.cse.ohio-state.edu/performance/mvapich/romio.shtml Please feel free to try it out and send your comments/questions to this lustre-discuss list or mvapich-discuss at cse.ohio-state.edu. Thanks, --Weikuan -------- Original Message -------- Subject: [ofa-general] Announcing the release of MVAPICH 1.0 Date: Fri, 29 Feb 2008 00:17:48 -0500 (EST) From: Dhabaleswar Panda <panda at cse.ohio-state.edu> To: ewg at lists.openfabrics.org, <general at lists.openfabrics.org> The MVAPICH team is pleased to announce the availability of MVAPICH 1.0...

This issue was brought up about a year and a half ago. So I''ll bring it up again and see if anything happens. I''ve got an infiniband network and am attempting to pass the infiniband card through the host and give it to the guest. I''m working with standard CentOS 5.2 on both guest and host with their provided xen (3.0.3 ish). I''ve also attempted to install the

This issue was brought up about a year and a half ago. So I''ll bring it up again and see if anything happens. I''ve got an infiniband network and am attempting to pass the infiniband card through the host and give it to the guest. I''m working with standard CentOS 5.2 on both guest and host with their provided xen (3.0.3 ish). I''ve also attempted to install the

Gentlemen: One of the three operating systems on my computer is eComStation 2.0, and it is connected to the outer world by means ofa TCP/IP facility. This is activated by means of "File and Directory Connections (Samba)". I have been through several generations of OS/2, so I know that system, but the Samba is a new field for me. After I fill in the rectangles, Samba takes two minutes to connect (it gets faster later...

I am trying to compile lustre-1.6.6 against ofed-1.4 using following configure options : ./configure --disable-server --with-o2ib=/usr/src/ofa_kernel-1.4 --with-linux=/usr/src/kernels/2.6.18-92.el5-x86_64 I am getting following errors: configure: error: can''t compile with OpenIB gen2 headers under /usr/src/ofa_kernel-1.4 Is it that lustre-1.6.6 will only work with ofed-1.3. or I am doing something wrong. System env is: HP DL5...

...ustre always uses the same NID to talk to a given server from a given node. It choses the NID (a) with the fewest hops (to minimize routing) and (b) appearing first in the "networks" or "ip2nets" LNET configuration strings. Now consider a 2-rail IB cluster running the OFA stack (i.e. OFED) with the following IPoIB address assignments... ib0 ib1 Servers 192.168.0.* 192.168.1.* Clients 192.168.[2-127].* 192.168.[128-253].* ...here are some different configurations you could create... A. I''ve got many more cl...

Hi, today me and a friend of mine spent several hours figuring out why ssh still asked for a password after we set up public key authentication. We have tried to understand the problem by reading 'ssh -vvv ...', but unfortunately the output was not useful. In the end of the day we have found out that sshd actually was logging this problem.... So that's for the context. Now, can you

platform i686-pc-linux-gnu arch i686 os linux-gnu system i686, linux-gnu status major 1 minor 2.1 year 2001 month 01 day 15 language R For years, I used a function in Splus to give me more information than just the object names which

I was looking at the Sparc 64 4.7 or 4.8 port, wondering if its in any state to put on my Solaris v100. Just need the basics, eide, network, kbm,mouse, graphics (x). A bonus would be support for the lom (light out management) features and the flash card. Reason for change is I HOPE that the math (ssl) performance is better on FBSD, as well as I am just tired of keeping track of

It seems the problems running mthca in a Xen domU have uncovered a bug in mthca: mthca uses dma_sync_single in mthca_arbel_write_mtt_seg() and mthca_arbel_map_phys_fmr() to sync the MTTs that get written. However, Documentation/DMA-API.txt says: void dma_sync_single(struct device *dev, dma_addr_t dma_handle, size_t size, enum dma_data_direction direction) synchronise a single

...s either from the host or to the host), so 'cmd' is necessary for the miscq. But the inflateq is exclusively used for inflating pages, so adding a command to it would be redundant and look a little bewildered there. 'flags': We currently use bit 0 of flags to indicate the completion ofa command, this is also useful in the "unused page" usage, and not needed by the "balloon page" usage. >>>> +#define MAX_PAGE_CHUNKS 4096 >>> This is an order-4 allocation. I'd make it 4095 and then it's >>> an order-3 one. >> Sounds goo...

...t; host or to the host), so 'cmd' is necessary for the miscq. But the inflateq is > exclusively used for inflating pages, so adding a command to it would be > redundant and look a little bewildered there. > 'flags': We currently use bit 0 of flags to indicate the completion ofa command, > this is also useful in the "unused page" usage, and not needed by the "balloon > page" usage. > >>>> +#define MAX_PAGE_CHUNKS 4096 > >>> This is an order-4 allocation. I'd make it 4095 and then it's an > >>> order-3...

...t; host or to the host), so 'cmd' is necessary for the miscq. But the inflateq is > exclusively used for inflating pages, so adding a command to it would be > redundant and look a little bewildered there. > 'flags': We currently use bit 0 of flags to indicate the completion ofa command, > this is also useful in the "unused page" usage, and not needed by the "balloon > page" usage. > >>>> +#define MAX_PAGE_CHUNKS 4096 > >>> This is an order-4 allocation. I'd make it 4095 and then it's an > >>> order-3...

On Fri, Apr 14, 2017 at 04:37:52PM +0800, Wei Wang wrote: > On 04/14/2017 12:34 AM, Michael S. Tsirkin wrote: > > On Thu, Apr 13, 2017 at 05:35:05PM +0800, Wei Wang wrote: > > > > So we don't need the bitmap to talk to host, it is just > > a data structure we chose to maintain lists of pages, right? > Right. bitmap is the way to gather pages to chunk. >

On Fri, Apr 14, 2017 at 04:37:52PM +0800, Wei Wang wrote: > On 04/14/2017 12:34 AM, Michael S. Tsirkin wrote: > > On Thu, Apr 13, 2017 at 05:35:05PM +0800, Wei Wang wrote: > > > > So we don't need the bitmap to talk to host, it is just > > a data structure we chose to maintain lists of pages, right? > Right. bitmap is the way to gather pages to chunk. >

...#dA zq`XUx(2U&A$$C`N0_!8w at dkLZ_&IWvX5 at cGUZELzUyxU6M)oJ<O`6f{YqEi58E=v| zXvQVqlM^(f`8VVo&1m@@*~l`rSIFx$qs6ym6H~OOD3)F(eU-dLGcrCT_p%J<TjUhY zX!0p}pJuw=ChyXWtdGfiG$ZvrV)uMAhM7y*X`0dWOY#EE$oz;rM>E{-ke{%q)hOEz zRo%iA!j${O*-3dkmUCFu&(lo$thyClpl-KK-v&nPmFVL$>hm%7S5$GReOFaJR5hD+ zKScSF`Tz=CiF2v)i%K6ve4Z}s1&*=tY!}qgTC}e5u_z96>oomg+2$kQTOX}_ot{BA zXeg_1#Hp#p4vd$c(aEEXHKi~@k1>`)U)IUvjAheTb+VH&N?~nwG3KOi=wvrzI(<tg zdl+j%-_gkv45|RTvX`-33S+X5u{8P?Ax|=9P*{-tj5VWJUk4b=p`Yu~4jyF8MHgx0 z5MvJdl}?^wEQ@}llcyOPgwZ(6Fm<pO&oGuwf6~dbjJYWc#1Y1tQ`m;*7~^ylBhNz# z...