Jump to content

System Testing and Benchmarks

NoelC

By NoelC
in Windows 8

 Share

Recommended Posts

TELVM

TELVM

Posted
Posted (edited)

You've got significant variation only in the 4K-64th read. The reason total score varies so much is because 4K read and 4K-64th read 'weigh the heaviest' in the calculations for total score that AS SSD does.

 

For typical use the really important thing is the 4K random read/write (though SSD makers love to mesmerize us with the sequential numbers  :rolleyes:  ). I see no significant variations in Seq, 4K, or Access Time (all within the normal fluctuations).

 

Always before running AS SSD benchmark, shut down as many applications and processes as possible, manually send a TRIM command to the SSD, and then leave the system idling logged-off for an hour or so (this allows the controller to perform its garbage collection and housekeeping to bring back the SSD to optimal condition for the benchmark  :yes:  ).

 

 

EDIT - So much for the latest version delivering better results:

 

Z1ffdVKf.gif

 

sia19sk6.gif

 

In the second test the SSD hadn't been allowed time for housekeeping after a previous run, and all numbers went south.

Edited by TELVM
Link to comment
Share on other sites

NoelC

NoelC

Posted
  • Author
Posted

I found that the ATTO numbers don't vary much for me depending on the recent load on the SSD array.  I'm not sure what it does differently, but again for the purposes of checking for developing system problems perhaps it's been the better choice.

 

By the way, I've seen that the 4K AS-SSD numbers are (not surprisingly) greatly affected by RAM caching between the low level I/O and the hardware.  For example, the Intel RST drivers integrate the Windows file system cache into their low level write operations, where my driver (from HighPoint) does not.  You've also shown how a caching hardware controller affects them, TELVM.  But that's just for low level I/O - applications in general use higher-level file system I/O and in that case the file system WILL use the RAM as a write-back no-wait cache for write operations if you configure it, thus speeding everything up.

 

TurnOffWriteBufferFlushing.png

 

The best benchmark I've found for comparing real-world I/O performance is the Advanced Disk benchmark in PassMark PerformanceTest, which will create a threaded mix of read and write operations of different sizes in an attempt to simulate a real-world I/O load. The difference between uncached and cached I/O (the latter of which most application file I/O uses) is monstrous.

 

PassmarkAdvancedDiskTest.png

 

PassmarkUncached.png

 

PassmarkUncachedResults.png

 

By contrast to the above, selecting "normal" I/O type operations...

 

PassmarkCached.png

 

PassmarkCachedResults.png

 

-Noel

 

Link to comment
Share on other sites
NoelC

NoelC

Posted
  • Author
Posted

Thanks DosProbie.

 

At first blush those seem more like "fixit" type tools more than pure benchmarking software.

 

Bart's Stuff also looks a bit antiquated (latest date I can see on the page is 11 years ago).

 

The latter (BootRacer) looks somewhat interesting, though I would tend to use Andre Ziegler's advice on tracing bootup time issues if I had them.  I can time actual bootup time with a clock or watch, and as it is, boot time issues are really a non-issue for me since I only boot about once every few weeks.

 

-Noel

Link to comment
Share on other sites
DosProbie

DosProbie

Posted
Posted (edited)

Thanks DosProbie.

 

At first blush those seem more like "fixit" type tools more than pure benchmarking software.

 

Bart's Stuff also looks a bit antiquated (latest date I can see on the page is 11 years ago).

 

The latter (BootRacer) looks somewhat interesting, though I would tend to use Andre Ziegler's advice on tracing bootup time issues if I had them.  I can time actual bootup time with a clock or watch, and as it is, boot time issues are really a non-issue for me since I only boot about once every few weeks.

 

-Noel

Your right Barts tool is older but still works with Windows 8 , plus need no installation since its a standalone and use little resources,

the Bootracer is a good tool for tweakers to test boot times accurately as small adjustments are made, I will say it works best in Win 7

as it boots straight to the desktop..DP

Edited by DosProbie
Link to comment
Share on other sites
jaclaz

jaclaz

Posted
Posted (edited)

For no apparent reason, this machine:

http://www.lancashirelife.co.uk/people/kendal_brown_house_continuing_the_art_of_snuff_production_1_1569201

produces snuff tobacco since 1792 (but is was re-cycled and was actually built originally around 1750).

 

Sometimes "old" does not mean "outdated" or "no good", sometimes it simply means "thoroughly tested and working" ;).

 

jaclaz

Edited by jaclaz
Link to comment
Share on other sites
DosProbie

DosProbie

Posted
Posted (edited)

## H2benchw.exe v3.16 HD Benchmark Utility ##

 

h2benchw.exe was written by Harald Bögeholz, senior editor for the German magazine C't

This is a great little third party utility that runs from the commandline and performs various
hard drive benchmark tests then saves to a designated .txt file, h2benchw.exe has been
out for several years and is used by various ssd review sites.
 
Here is a batch file that I use for my Hard Drive testing with Windows 8.1, just copy to a .txt file then
rename to Run_Tests.cmd and right-click as admin to run.
Hint: Place `H2benchw.exe` in the same folder as the batch file or run both from your desktop.
Enjoy..DP
Download ==>  http://www.heise.de/ct/c-t-Systeminfo-473388.html
 
:: INFO::: Runas Admin!:: Batch - Run_Tests.cmd:: Uses the third-party utility h2benchw.exe  :: d/l @ http://www.heise.de/ct/c-t-Systeminfo-473388.html@echo off&color 4f&title, ## Hard Disk Benchmark Utility ##cd %~dp0 :: Sequential Read Speed Test (saved to Results file)h2benchw -english -s 0 -w Results.txtdel *.ps>nulecho.&echo.& echo DONE! - Press any Key to Exit..pause>nulexit:: MISC: (a)ll measurements -  h2benchw -english -a 0: (s)eek measurements - h2benchw -english -s 0: (z)one measurements - h2benchw -english -z 0: View switch options - h2benchw -english -s 0 -w
 
 
All_Options.pngAccess_Time_Test.png
Edited by DosProbie
Link to comment
Share on other sites
DosProbie

DosProbie

Posted
Posted

For no apparent reason, this machine:

http://www.lancashirelife.co.uk/people/kendal_brown_house_continuing_the_art_of_snuff_production_1_1569201

produces snuff tobacco since 1792 (but is was re-cycled and was actually built originally around 1750).

 

Sometimes "old" does not mean "outdated" or "no good", sometimes it simply means "thoroughly tested and working" ;).

 

jaclaz

Good point Jaclaz!.. :yes:

Link to comment
Share on other sites
TELVM

TELVM

Posted
Posted

... Sometimes "old" does not mean "outdated" or "no good", sometimes it simply means "thoroughly tested and working" ;) ...

 

If this isn't old enough I've got them older in my Jurassic Park  :D  :

 

N0IaoMZx.jpg

 

... The difference between uncached and cached I/O (the latter of which most application file I/O uses) is monstrous....

 

I like how this table illustrates the pecking order in a more human-friendly time scale. On the rightmost column, if we assume 1 CPU cycle = 1 second, then ...

 

AmmTZONp.png

 

^ RAM is miles faster than anything else outside the processor die, SSDs included.

Link to comment
Share on other sites
DosProbie

DosProbie

Posted
Posted

I am working on a App installer and need a good standalone (non-installer) Benchmark or System tester, If anyone knows of something

I would appreciate it, btw Heavyload is all I have found so far..

Thanks, DP :whistle:

Link to comment
Share on other sites
NoelC

NoelC

Posted
  • Author
Posted

Behold. The Windows Experience Index has returned.

 

http://winaero.com/comment.php?comment.news.220

 

Neat.  I love it when people bring back things Microsoft deletes.  Only thing is I'm not sure I want a WEI score badly enough to trust it not to have a malware payload hidden somewhere inside.  I have stored copies of winsat output and can easily compare them looking for evidence of problems.

 

Microsoft probably figures the majority of people don't need to know about the assessment, since knowing causes some of them to try to game it for bragging rights, which could lead to maintenance problems.  Parts of the system do alter their behavior by the stored winsat results.

 

-Noel

Link to comment
Share on other sites
NoelC

NoelC

Posted
  • Author
Posted

I could believe that.

 

And let's not forget that the exact same hardware got higher scores with Windows 8.0's WEI measurement than with that of Windows 7.

 

-Noel

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share
Go to topic listing

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...