Re: [Qemu-block] [PATCH 2/2] block: align bounce buffers to page

2015-05-12 Thread Denis V. Lunev
On 12/05/15 13:27, Kevin Wolf wrote: Am 12.05.2015 um 07:47 hat Denis V. Lunev geschrieben: The following sequence int fd = open(argv[1], O_RDWR | O_CREAT | O_DIRECT, 0644); for (i = 0; i 10; i++) write(fd, buf, 4096); performs 5% better if buf is aligned to 4096

Re: [Qemu-block] [PATCH 2/2] block: align bounce buffers to page

2015-05-12 Thread Kevin Wolf
Am 12.05.2015 um 12:36 hat Denis V. Lunev geschrieben: On 12/05/15 13:27, Kevin Wolf wrote: Am 12.05.2015 um 07:47 hat Denis V. Lunev geschrieben: The following sequence int fd = open(argv[1], O_RDWR | O_CREAT | O_DIRECT, 0644); for (i = 0; i 10; i++) write(fd,

Re: [Qemu-block] [PATCH 2/2] block: align bounce buffers to page

2015-05-12 Thread Kevin Wolf
Am 12.05.2015 um 15:41 hat Denis V. Lunev geschrieben: The following sequence int fd = open(argv[1], O_RDWR | O_CREAT | O_DIRECT, 0644); for (i = 0; i 10; i++) write(fd, buf, 4096); performs 5% better if buf is aligned to 4096 bytes. The difference is quite

Re: [Qemu-block] [PATCH 2/2] block: align bounce buffers to page

2015-05-12 Thread Denis V. Lunev
On 12/05/15 17:08, Kevin Wolf wrote: Am 12.05.2015 um 15:41 hat Denis V. Lunev geschrieben: The following sequence int fd = open(argv[1], O_RDWR | O_CREAT | O_DIRECT, 0644); for (i = 0; i 10; i++) write(fd, buf, 4096); performs 5% better if buf is aligned to 4096

Re: [Qemu-block] [PATCH 2/2] block: align bounce buffers to page

2015-05-12 Thread Kevin Wolf
Am 12.05.2015 um 16:20 hat Denis V. Lunev geschrieben: On 12/05/15 17:08, Kevin Wolf wrote: Am 12.05.2015 um 15:41 hat Denis V. Lunev geschrieben: The following sequence int fd = open(argv[1], O_RDWR | O_CREAT | O_DIRECT, 0644); for (i = 0; i 10; i++) write(fd,

Re: [Qemu-block] [PATCH 2/2] block: align bounce buffers to page

2015-05-12 Thread Kevin Wolf
Am 12.05.2015 um 07:47 hat Denis V. Lunev geschrieben: The following sequence int fd = open(argv[1], O_RDWR | O_CREAT | O_DIRECT, 0644); for (i = 0; i 10; i++) write(fd, buf, 4096); performs 5% better if buf is aligned to 4096 bytes. The difference is quite

Re: [Qemu-block] [PATCH 2/2] block: align bounce buffers to page

2015-05-12 Thread Denis V. Lunev
On 12/05/15 17:26, Kevin Wolf wrote: Am 12.05.2015 um 16:20 hat Denis V. Lunev geschrieben: On 12/05/15 17:08, Kevin Wolf wrote: Am 12.05.2015 um 15:41 hat Denis V. Lunev geschrieben: The following sequence int fd = open(argv[1], O_RDWR | O_CREAT | O_DIRECT, 0644); for (i = 0; i

[Qemu-block] [PATCH 2/2] block: align bounce buffers to page

2015-05-12 Thread Denis V. Lunev
The following sequence int fd = open(argv[1], O_RDWR | O_CREAT | O_DIRECT, 0644); for (i = 0; i 10; i++) write(fd, buf, 4096); performs 5% better if buf is aligned to 4096 bytes. The difference is quite reliable. On the other hand we do not want at the moment to enforce

Re: [Qemu-block] [PATCH 2/2] block: align bounce buffers to page

2015-05-12 Thread Paolo Bonzini
On 12/05/2015 12:27, Kevin Wolf wrote: I think it would make more sense to keep this specific to the raw-posix driver. After all, it's only the kernel page cache that we optimise here. Other backends probably don't take advantage of page alignment. I don't think it makes sense to keep it

[Qemu-block] [PATCH 2/2] block: align bounce buffers to page

2015-05-11 Thread Denis V. Lunev
The following sequence int fd = open(argv[1], O_RDWR | O_CREAT | O_DIRECT, 0644); for (i = 0; i 10; i++) write(fd, buf, 4096); performs 5% better if buf is aligned to 4096 bytes. The difference is quite reliable. On the other hand we do not want at the moment to enforce

[Qemu-block] [PATCH 2/2] block: align bounce buffers to page

2015-05-04 Thread Denis V. Lunev
The following sequence int fd = open(argv[1], O_RDWR | O_CREAT | O_DIRECT, 0644); for (i = 0; i 10; i++) write(fd, buf, 4096); performs 5% better if buf is aligned to 4096 bytes. The difference is quite reliable. On the other hand we do not want at the moment to enforce