The more I thought about using Win 2k8 server as my workstation OS, the more I liked the idea. I did initially have a problem with Battlefield2 locking up and then crashing to the desktop, but I eventually worked through that (details below).
I decided to do some benchmarks for you folks while I was experimenting. I went through three separate OS installs to do these, so there would be nothing to influence the scores (such as where the OS’s partitions would lie on the HDD), and I’m no professional benchmarker, so these are really only benchmarks that are relevant to me, my setup and my applications, but here goes for those who are interested!
First, I’ll recap my hardware:
AMD Athlon 64 x2 4600+ @ 2.4GHz
6 GB DDR2 800 PC6400 RAM (Crucial Ballistix 2×2GB + 2×1GB)
Main HDD: Western Digital WD3200AAKS 320 GB, 16MB Cache SATAII
First Secondary HDD: Seagate ST3120814A (7200.9) 120 GB 8MB Cache EIDE (Data only)
Second Secondary HDD: Western Digital WD2500BB 250GB 2MB Cache EIDE (Data and Virtuals)
- Side note: Even though the WD2500BB by all accounts should be slower than the Seagate, I use it for my virtual storage space since it has ALWAYS benchmarked faster than the Seagate for me (even if by a small margin).
Video Card: PNY Nvidia 7900GS 256MB PCI-E (Overclocked to 500 Core/700 MEM)
Motherboard/Chipset: Biostar TF570SLI/Nvidia Nforce 570SLi
Power Supply: Antec Neopower 650 - 650W
Operating Systems:
- Windows XP Professional x64 Edition SP2
- Windows Vista Ultimate x64 Edition SP1
- Windows Server 2008 Enterprise x64 Edition SP1 (BTW, is this the first time MS has released an OS already at SP1?)
On XP x64, I ran all updates, updated drivers to latest versions, otherwise stock config (XP Themes on).
On Vista, I ran all updates, disabled UAC, Indexing/Search, Media Center, and the Sidebar (Used Vista Aero theme) and updated all drivers to latest versions.
On Server 2008 I installed the Desktop Experience package, Changed Processor Scheduling, Enabled SuperFetch, Enabled Sound, Enabled Themes (Used Vista Aero) and updated drivers to latest version.
Now for the Benchmarks:
3DMark 2006 - Standard setup, just installed and ran.
- Windows XP x64 = 5789 3D Marks
- Windows Vista x64 = 5541 3D Marks
- Windows Server 2008 x64 = 5990 3D Marks
HD Tune on Main HDD (The WD SATA)
- Windows XP x64: Maximum Transfer = 69.8 MB/sec; Burst Rate = 96.2 MB/sec
- Windows Vista x64: Maximum Transfer = 71.1 MB/sec; Burst Rate = 110.7 MB/sec
- Windows Server 2008 x64: Maximum Transfer = 75.6 MB/sec; Burst Rate = 114.3 MB/sec
Cinebench 9.5 (64-Bit Edition); Multiple CPU Rendering:
- Windows XP x64 = 742
- Windows Vista x64 = 745
- Windows Server 2008 x64 = 747
Half-Life2: Lost Cost (Video Stress Test No AA no AF - defaults for my card @ 1280×1024):
- Windows XP x64 = 127.56 Avg. FPS
- Windows Vista x64 = 85.25 Avg. FPS
- Windows Server 2008 = 115.51 Avg. FPS
Counter Strike Source (Video Stress Test - 4xAA no AF - defaults for my card @ 1280×1024):
- Windows XP x64 = 139.65 Avg. FPS
- Windows Vista x64 = 111.42 Avg. FPS
- Windows Server 2008 x64 = 121.94 Avg. FPS
Battlefield 2 (High settings on everything @ 1280×1024):
- Windows XP x64 = 95.2 FPS
- Windows Vista x64 = 80.1 FPS
- Windows Server 2008 x64 = 94.4 FPS
As I reported before, I had problems with Battlefield2 running under Server 2k8 - My sound would stutter, and then the game would lock, and eventually crash to the desktop. I discovered a trick to set up the sound system the same way it is on Vista. Look up this reg key:
HKLM\SOFTWARE\Microsoft\Windows NT\CurrentVersion\Multimedia\SystemProfile
and change the value of SystemResponsiveness from 100 to 20.
VMware Workstation 6 (Vista x64 Ultimate as Guest OS - Guest has 2GB RAM and 1 CPU: WEI Scores):
- Windows XP x64: CPU = 4.4; Mem = 5.6; HDD = 5.2
- Windows Vista x64: CPU = 4.5; Mem = 5.7; HDD = 5.3
- Windows Server 2008 x64: CPU = 4.5; Mem = 5.9; HDD = 5.3
I did not include the Graphics scores as they were obviously 1.0 across the board due to running in VMware. As a point of reference, my machine gets a bare metal score of 5.0 on CPU, 5.9 on RAM and a 5.7 on HDD under Vista x64 (CPU higher mostly due to 2x CPU probably).
VMWare Workstation 6 (Ubuntu 8.04 as Guest OS - Guest has 2GB Ram and 1 CPU - results of hdparm -tT /dev/sda1):
- Windows XP x64 = Timed cache reads = 874.32 MB/sec; Timed buffered disk reads: 38.57 MB/sec
- Windows Vista x64 = Timed cache reads = 894.41 MB/sec; Timed buffered disk reads: 40.21 MB/sec
- Windows Server 2008 x64 = Timed cache reads = 978.39 MB/sec; Timed buffered disk reads: 45.83 MB/sec
For the Ubuntu guest, VMware paravirtual kernel support is enabled as Ubuntu supports it since 7.04 (Feistly).
For all VM’s, I have VMware set to fit all virtual machine memory into reserved host RAM.
Conclusions and Closing Thoughts:
I am not sure why Server 2008 performs so differently than Vista. I have read a lot of blog posts around 2008 as a workstation over the last few days, and I’ve seen people be downright rude and call people liars. Folks are saying that there should be no difference since they are basically the same OS underneath, and if you disable all the “vista” eyecandy it’ll even out. With my tests, I had the “eyecandy” set dead even. I can not explain why there is a difference, but there is a difference, if you don’t believe me, try it for yourself. Even more striking than the benchmarked differences the OS just “feels” so much faster and smoother than Vista, and much more solid - although, that could very well be some sort of placebo effect, and I fully admit that.
For me, however I think the winner for my HDD space is going to be Server 2008 - until something else convinces me otherwise. Again, I stress these were not “scientific benchmarks” in a 100% controlled environment, and I did them originally, strictly as my own point of reference to use in making my decision. Please take the results with a grain of salt, and know that your results may differ dramatically!
If you’ve made it this far: Thanks for reading! 