• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

What should I run?

D

drizzt81

[H]F Junkie
Joined
Jan 21, 2004
Messages
12,361
Assume that I have access to a PC with a "hardware" raid card (SX4-M), a software raid solution (nforce), four 7200 rpm drives, a 160GB rapor and a 4GB iRAM.

What benchmarks would be interesting to the community, keeping in mind that I'd rather not watch and time a WinXP install 15 times in a row and generally have limited free time.

right now I have nothing important on it, so I can install pretty much any OS/ freeware that you can come up with.
 
I'd like to see a load time test with a game of your choice (tested with a stopwatch) on a single 7200, a raid 0 with 2 7200s, a raid 0+1 with four disks, the Raptor, and the iRAM. That should pretty much cover the bases. Multiple runs (with each individual datum reported, and reboots in between the runs) would be nice, for statistical analysis.

 
sounds doable. Would be pretty easy, since it won't require a Windows reinstall for switching between configs.
 
So here's an initial update, no graphs. Due to a Samsung disk failing on me (before I ran a single benchmark) I cannot do RAID 0+1. Also, the Steam folder that I am using for benchmarking is ~4.5GB in size, so I cannot run any numbers on the iRAM, sorry :(

Software:
WinXPx64 installed from OEM disk, no patches.
nVidia nForce driver 6.86
nVidia Graphics driver 91.47 for WinXP x64
Steam updates current as of Today (10/16/06)
All RAID configurations were done with a 64k stripe size selected in the nVidia storage manager.

Hardware:
AMD Athlon 64 3000+
1024MB of RAM, mixed mushkin, running at 200MHZ, "default" timings (I can dig up the numbers, though I doubt that this will have a nonlinear effect on my benchmark)
Windows installed on 160GB, 8MB cache PATA Maxtor drive
GeForce 6600 PCIe, 128MB of RAM

Network:
100Mbit switched on WRT54g

Methodology:
I rebooted in between the runs, unless otherwise indicated. I have a local area network with a SRCDS running on a Opty 165 @ 2.6GHz with 2GB of RAM. Server is running CS:S, no bots, 24 player max, de_dust, VAC on, LAN. Restarting the server did not appear to affect the loading times on the client, so I did not restart the server.

The time was measured using this stopwatch on monitor 1, connected to the SRCDS computer while WinXP x64 was showing on monitor 2. I opened the server list, where I have manually added the IP of my server. I opened "view server info" and then simultaneously clicked "connect" and "start" using two mice. I stopped the timer as soon as I saw the "MOTD" screen.


3 disk SP1614C RAID-0:
00:34.8
00:35.1
00:35.2

no reboot in between these runs:
00:13.4
00:13.6
00:13.4

Single WD1600ADFD Raptor (dell branded)
Raptor Only
00:29.4
00:29.0
00:28.3

no reboot in between these runs:
00:13.3
00:12.5
00:12.9

I am currently working on single disk (7200rpm) and 2 disk RAID-0 runs.

Should I take and post HDTach screenshots?
Do I need to make a fancy graph so people can see the difference?
Should I run a RAID-1 setup and compare?
 
of course you should take screen shots, how else are we going to believe you? :p

ok, so this is a bit OT, but is it possible to raid iRAM?
 
protias said:
of course you should take screen shots, how else are we going to believe you? :p
Click to expand...
what do you want screenshots of? The stopwatch app?
protias said:
ok, so this is a bit OT, but is it possible to raid iRAM?
Click to expand...
yes. in most cases they behave just like regular disks.
 
OK and here are the other results:

Two disk RAID-0:
00:34.8
00:34.6
00:34.4

no reboot in between these runs:
00:14.1
00:13.9
00:13.9

Single SP1614C:
00:40.6
00:38.3
00:39.1

no reboot in between these runs:
00:15.0
00:14.8
00:14.7

And here's the chart of average startup times for CS:S when rebooting in between runs:
raid.PNG

ps.: How can I create a RAID-5 array with media shield?
 
I can't create or migrate to RAID 5 arrays with my version of Media Shield, but I can create them using the BIOS.

I'd be interested to know IOMeter performance, with the following parameters for example:

10 GB file size. (20 000 000 sectors.)
1 outstanding I/O (This is a tricky variable, but sets a lower-bound, and is used sometimes.)
64k 100% sequential reads
64k 100% sequential writes.

Duration > 1 minute so that the cache doesn't pollute the results.

I'd also be interested in performance with variations in stripe sizes:

a. 64k
b. a / number of data drives (#drives for RAID 0, #drives - 1 for RAID 5)
c. b / 2

I'd also expect fairly poor write performance from NVRAID 5, regardless of stripe size. Don't mean to prejudice you here, just warn you about probable frustration if your time is limited.
 
any screenshots you deem necessary to validate your results. i am kinda interested to see a HDtach run. curiousity has YET to kill me
 
protias said:
any screenshots you deem necessary to validate your results. i am kinda interested to see a HDtach run. curiousity has YET to kill me
Click to expand...
I cannot show you any screenshots that would even remotely come close to valiading my claims:

If I show you a stopwatch with the corresponding time, how would that substantiate my claims? Since the experiment was run on two different machines, I cannot have a "composite" screenshot.
I could make a video of the process, but I do not own a video camera that is fast enough to capture the results.
 
You must log in or register to reply here.
Back
Top