Sponsored Content
The Lounge War Stories Linus Torvalds reply about Meltdown and Spectre. Post 303012275 by Corona688 on Thursday 1st of February 2018 04:07:36 PM
Old 02-01-2018
Quote:
Originally Posted by dodona
I read that with the 4.15 Kernel with build-in patches for the hypothetical problem comes with 15-30% performance loss.
That's way overblown, because the performance loss happens during the switch between userspace and kernelspace. Sensible programs don't spend most of their time doing thousands of tiny system calls.

A 30% reduction in performance there is bad, but not the same as a 30% slower computer.
 

5 More Discussions You Might Find Interesting

1. Filesystems, Disks and Memory

What the best HD format for Unix & Linus

Hello people, Im just about to install linux and unix but would like to know what the best Partition type format is for both of them. Can anyone tell me Thankyou in advanced (1 Reply)
Discussion started by: jeffersno1
1 Replies

2. Filesystems, Disks and Memory

hard disk meltdown

I had an issue with a second hard disk in my machine. I have a sparc station running solaris 7. It was working fine but now it wont mount on boot up and when you try to mount it manually it gives an I/O error. I tried a different disk as a control which was fine. What I want to know is if my... (3 Replies)
Discussion started by: Henrik
3 Replies

3. News, Links, Events and Announcements

A Couple of Short Linus Torvalds Videos

This link is to a page that has two short videos that feature Linus Torvalds discussing the Linux kernel. After you watch the first video, scroll down some more to reach the second video. (0 Replies)
Discussion started by: Perderabo
0 Replies

4. What is on Your Mind?

Meltdown and Spectre CPU bugs

This seems a bit serious: Meltdown, Spectre: The password theft bugs at the heart of Intel CPUs • The Register Vulnerability Note VU#584653 - CPU hardware vulnerable to side-channel attacks Project Zero: Reading privileged memory with a side-channel (8 Replies)
Discussion started by: Scrutinizer
8 Replies

5. AIX

AIX 6.1, POWER5 and Spectre/Meltdown

Apologies for this newbie question. We have inherited an IBM p5 520 (9111-520) running AIX 6.1.0.0 which seems to be the base install and no further patches installed. Is this vulnerable to the Spectre/Meltdown threat? Are patches available? Looks like AIX 6.1.0.0 went 'end of support' in... (8 Replies)
Discussion started by: the_garbage
8 Replies
fcache_fb_policy(5)						File Formats Manual					       fcache_fb_policy(5)

NAME
fcache_fb_policy - policy for flushing dirty file data to disk VALUES
Failsafe Default Allowed values Minimum: Maximum: DESCRIPTION
HP-UX asynchronously flushes dirty data to disk to manage the risk of data loss and avoid burst mode I/O. However, excessive flushing can cause extra I/O and reduce performance. The tunable is used to determine the trade-off between write performance and data availability. The tunable has three settings: This is the default setting. It is a high-performance setting. It minimizes the performance impact of optional flushing. For example, in case of mapped access to a file, the dirty pages are only flushed out asynchronously when they are unmapped by a call (see munmap(2)). The asynchronous flushing allows the call to complete quicker than if the flushing was syncrhonous. For file system access, the dirty data is flushed every 30 seconds or so by the syncer daemon. This setting is a high-performance setting. It disables the VxFS Flush Behind feature. Refer to the next paragraph, for more details about VxFS Flush Behind feature. Please note that the semantics of default behavior are liable to change as HP-UX identifies more optimal settings. This setting is a lower performance setting, but it provides more frequent flushing to disk. It increases the amount of flushing for file system access by enabling the Flush Behind feature of VxFS File System. Dirty pages are flushed when the number of dirty pages in memory for a file reaches a certain threshold. These flushes are controlled by a variety of VxFS tunables, such as and See the vxtunefs(1M) manpage for more details on the VxFS tunables. This setting corresponds to the HP-UX 11i v2 default behavior. It does the same amount of flushing as the tunable setting 1, but the I/O requests for flushing data are scheduled immedi- ately (synchronously). Who Is Expected to Change This Tunable? System administrators who run applications that do large file writes may change this tunable. Restrictions on Changing Changes to this tunable will take effect immediately. When Should the Value of This Tunable Be Changed to 1? This tunable should be changed from 0 to 1 if applications running on the system require more data integrity than the HP-UX syncer daemon provides. In addition, should be changed to 1 if the VxFS tunable, is set. The tunable lets an administrator lower the number of dirty buffers per file that the file system will generate before writing them to disk. The default value of is zero. The default value places no limit on the number of dirty buffers per file. With the default setting, there is no problem bypassing the Flush Behind feature. However, if is enabled, the application performing writes may encounter longer delays as it waits for to flush some of the dirty buffers before the appli- cation can continue. When Should the Value of This Tunable Be Changed to 2? This tunable should be changed to 2 if applications running on the system have been highly tailored for HP-UX 11i v2 behavior. It provides the most backwards compatible behavior. What Are the Side Effects of Changing the Value? Changing the value from 0 to 1 will cause increased I/O activity. Changing the value from 0 to 2 will cause lower response time for and system calls. It may result in loss of overall write performance. What Other Tunables Should Be Changed at the Same Time? None. WARNINGS
All HP-UX kernel tunable parameters are release specific. This parameter may be removed or have its meaning changed in future releases of HP-UX. Installation of optional kernel software, from HP or other vendors, may cause changes to tunable parameter values. After installation, some tunable parameters may no longer be at the default or recommended values. For information about the effects of installation on tun- able values, consult the documentation for the kernel software being installed. For information about optional kernel software that was factory installed on your system, see the at AUTHOR
was developed by HP. SEE ALSO
syncer(1M), vxtunefs(1M), fsync(2), munmap(2), write(2). Tunable Kernel Parameters fcache_fb_policy(5)
All times are GMT -4. The time now is 07:17 PM.
Unix & Linux Forums Content Copyright 1993-2022. All Rights Reserved.
Privacy Policy