prefect/superfetch performance difference ?

[davexnet]

I have not tried turning Superfetch off to see the impact, but I'm presenting my finding on
one particular app, comparing two different PC's.

DELL Dimension 4600 (2003) Penium 4 2800 GHz single core, 2GB Ram
AMD FX4300 based PC (2014) quad-core processor, 4GB RAM

They both have the same version of Photoshop installed (version 9.02)

On the old Dell, the app screen appears in 2 seconds, ready to use in 10 seconds
On the FX 4300, the app screen appears in 4 - 5 seconds, ready to use in 15

In a task like video encoding the FX4300 is perhaps 5 - 10 times faster that the Dell,
(CPU only, no graphics card assistance)
but for the startup of programs, the DELL is faster.

Any thoughts?

 

[Bree]

for the startup of programs, the DELL is faster.

According to the specs for your Dell Dimension 4600 here on Cnet it has a 7200 rpm HDD. If your other machine has a normal 5400 rpm HDD then that could explain the difference.

15 seconds / 7200 x 5400 = 11.25 seconds, near enough the same as the 10 seconds you measured.

 

[davexnet]

I'm pretty sure the AMD PC has a 7200 RPM; can you recommend a disk utility that may help with some
meaningful disk performance info? Or indeed any system benchmarks that might give an overall picture?
Perhaps there is something in the real world operation that's a factor.
The Dell is running Windows 7 (originally XP; upgraded by the original owner)
AMD FX4300 running Windows 8.1
I took some more timings; I disabled the Superfetch service on the AMD PC and rebooted, there is hit - it takes
approx 25% more seconds for the app to startup.
Tried the same on the DELL, with the Dell Superfetch off, and the AMD on, they app startup is about the same

 

[Bree]

I use CrystalDiskInfo to read a drive's specs, including rotation speed and SMART data. For performance testing I use CrystalDiskMark. Both are available as portable apps.

CrystalDiskInfo Portable (disk health monitoring) | PortableApps.com

CrystalDiskInfo is a HDD/SSD utility which supports S.M.A.R.T. It supports a part of external USB disks, monitoring health status and temperature, graph of S.M.A.R.T. information, control AAM/APM settings and more.

portableapps.com

CrystalDiskMark Portable (disk benchmark utility) | PortableApps.com

CrystalDiskMark can measure sequential reads/writes speed, measure random 512KB, 4KB, 4KB (Queue Depth=32) reads/writes speed, has support for different types of test data (Random, 0Fill, 1Fill), includes basic theme support and has multilingual support (though it will not yet remember your...

portableapps.com

 

[davexnet]

I installed NovaBench 3, and the Crystal BM, I don't see anything unexpected

 

[davexnet]

Both PC's running Windows Defender; I disabled it on the AMD box and I got about a second knocked off the startup time.
I tried some other apps, the old DELL was always slightly faster.
Perhaps it's the Intel chipset on the motherboard that gives it better performance in certain situations?

This AMD bios has an option "load performance defaults" - I'm going to try it and see if it affects this.
(I've got no idea what option actually does)

 

[Bree]

The Dell is running Windows 7....
AMD FX4300 running Windows 8.1

The overheads of the operating system are going to be a factor. Windows 7 almost certainly has lower overheads than Windows 8.1.

I can demonstrate that. I have a test machine with Macrium images I can restore, each with identical software installed but running under a different OS. As this is the same hardware and software for each test, then the only variable will be the OS and the AV it uses.

There is a PowerShell applet called measure-command that you can use to measure the time it takes to run a command. I used that to measure the time it takes to open Excel 2007 with this command:

measure-command {start "C:\Program Files\Microsoft Office\Office12\excel.exe"}

Do not run the app before timing it in PowerShell, else you may well be reading the .exe from a memory cache rather than from the HDD. For consistent results restart the PC before running the test, and make sure the test is the first thing you run.

In Windows 7 with MSE Excel opens in 490ms, in Windows 10 and Defender it takes 643ms. I don't have Windows 8.1 to test, but I would expect it to be closer to the W10 times than to W7 (ie. about 30% slower to open an app on the same hardware).

 

[davexnet]

Hi Bree, thanks for posting info about this tool, it's new to me and I will look into it.

On a positive note I enabled the AMD FX4300 "load performance defaults" bios setting.
After booting the system, and waiting 5 or 6 minutes for all the Superfetch disk activity
to complete, I started Photoshop and saw a 50% startup performance improvement,
it was ready to use in 8 seconds

There is a warning in the BIOS on this option, it says in some cases Windows may not start properly.
I haven't seen any problem. I would love know what it actually does

Thanks for your help and suggestions

 

[Bree]

I enabled the AMD FX4300 "load performance defaults" bios setting....
...I would love know what it actually does

Apparently....

According to this manual on Manualslib.com

Load Performance Defaults​ If you select this item and press Enter a dialog box will appear. If you select [OK], and then Enter, the Setup Utility loads a set of performance default values. These default settings are quite demanding and your system might not function properly if you are using slower CPU, memory, or other low-performance components.

 

[davexnet]

That's interesting. I will certainly keep an eye on it. I'll probably look at CPUZ to see if I can find any memory timing
differences. The old Dell system has such a simple Bios, there are no settings like this.
Perhaps it's some combination of Dell's expertise and Intel chipset