How to find the speed of a non-video card?

[Automata]

I'm running a Perc 5/i in my server. The card fits in a PCI-e slot and is 8x physically and electrically. It is using the top PCI-e slot (16x rated) of my Asus M2N32-SLi motherboard. I'm seeing a trend with the benchmarks, anything that uses a sustained read is being limited to around or less than 150mb/sec. I should be seeing upwards of 500mb/sec and I have enough reason to suspect that it is not really running 8x in that PCI-e slot.

So, bottom line. How do you check the speed of a device in a PCI-e slot that is not a video card?


GPU-z: Doesn't report anything but video cards
CPU-z: Doesn't list PCI-e speeds
Everest: I see it is listed at "16x", which I know isn't right

Now before I get the "But Thideras, you're doing it wrong!" p), here is my Perc 5/i with EIGHT 1tb drives in RAID0 with a 200GB slice at the beginning of the drive; tell me something isn't wrong and I will smack you up-side the head.

Writing is a bit faster, but still nothing quicker than 150mb/sec

 

[Automata]

I did find it is reporting 8x through Everest, the option was well hidden. Now, I need to figure out why this is running so slow...

 

[Dusnoetos]

I have know idea if you have seen this thread from a different forum.... but it looks like it is rather informative - even if it does not help directly with your question

http://www.overclock.net/raid-controllers-software/359025-perc-5-i-raid-card-tips.html

 

[RollingThunder]

Thiddy, I think you're doing it wrong................

 

[Automata]

I have know idea if you have seen this thread from a different forum.... but it looks like it is rather informative - even if it does not help directly with your question

http://www.overclock.net/raid-controllers-software/359025-perc-5-i-raid-card-tips.html

Yes, I read every page of that thread before buying that card

Thiddy, I think you're doing it wrong................

/smack


It turns out that the way that program benchmarks may be incorrect. I'm going to try some "real world" benchmarks to see how it good it does. If this file copy is more than 100mb/sec, I know that I can throw those results out.

 

[bing]

Thiddy, although I'm not sure if this will help in your problem, suggesting you to try this -> HWINFO.

The GUI version is free but I have doubt it is sort of ad-ware.

But you can also try the DOS version, even its in evaluation mode, still fully functional, you just had to wait for few seconds when it started and push some button and continue. Just put it in DOS bootable USB thumb drive, boot and launch it from there.

In the past, the DOS version really helped me with some weird hardware configuration problems.

 

[Automata]

Thiddy, although I'm not sure if this will help in your problem, suggesting you to try this -> HWINFO.

The GUI version is free but I have doubt it is sort of ad-ware.

But you can also try the DOS version, even its in evaluation mode, still fully functional, you just had to wait for few seconds when it started and push some button and continue. Just put it in DOS bootable USB thumb drive, boot and launch it from there.

In the past, the DOS version really helped me with some weird hardware configuration problems.

I have full confidence it is now running in 8x mode, but I'm still seeing weird/mixed results. I'm wondering if there is something in the way these programs test speeds. Parts of the test show normal results like this (this is the exact same run as the ones above):

But some other parts of the same program show the 84mb/sec read speeds. I don't know what to trust...

Bottom line: How do you truly test the speed of a RAID array?

 

[zbo]

Thats really odd....

I tested my WD 640gb Black RAID0 (40gb from each) and got these results, which don't seem to get capped at 150mb like your seeing in yours. Perhaps it's only with a large number of drives, say 4+ that it starts going wrong?

 

[Automata]

That isn't the same RAID controller, I can't compare the two.

Either way, I have two programs giving two completely different results. One portion of the test is saying less than 100mb/sec while another says almost 550mb/sec. I don't know which to believe.

 

[Automata]

Fired up another program, IOMeter. Did 100% sequential, 100% read 50mb file. Says 550MB/s+.

I'm thinking that HDTune just is bad at testing this card.

 

[Dj Spiel]

Off Topic: Is HD Tune a free program?

 

[Automata]

Off Topic: Is HD Tune a free program?

There is a free version that has some basic features. I'm using the "full version" trial in the screenshots.

 

[Neuromancer]

HDtune is no longer used because of compatability issues with SSDs, maybe the same glitch it has with SSDs is also happening with your raid card?? Sounds wrong but who knows?

Try out HDTach or ATTO

 

[Automata]

HDtune is no longer used because of compatability issues with SSDs, maybe the same glitch it has with SSDs is also happening with your raid card?? Sounds wrong but who knows?

Try out HDTach or ATTO

I'm curious to see what the issue is, do you have a link to a thread (etc)? I like to know why stuff fails.

I can try other benchmarks too.

 

[Dusnoetos]

?? Is their a way to determine if the program('s) suport testing your Raid card?

have you given PCwizard 2009 a try?

 

[Automata]

?? Is their a way to determine if the program('s) suport testing your Raid card?

I didn't word that correctly. All the hard drive tests "support" RAID. Meaning, that they are capable of producing some sort of numbers regarding the transfer rate and actually write to or read the disk. My questions stems from "are they producing correct numbers?" or ones that apply to the type of data transfers that I will use.

have you given PCwizard 2009 a try?

I don't see anything that program offers that others don't. I'm currently waiting on the RAID10 to initialize so I can begin benchmarking it.

 

[bbtkd]

ZFS

I didn't word that correctly. All the hard drive tests "support" RAID. Meaning, that they are capable of producing some sort of numbers regarding the transfer rate and actually write to or read the disk. My questions stems from "are they producing correct numbers?" or ones that apply to the type of data transfers that I will use.

I don't see anything that program offers that others don't. I'm currently waiting on the RAID10 to initialize so I can begin benchmarking it.

You could also install Solaris and compare ZFS. We've found that in general software RAID outperforms PERC and other hardware RAID (particularly for RAID 3 or 5) as long as you have enough cores. I've asked our admins to either go to soft RAID 5 on Linux servers, or just switch them to Solaris to use ZFS. Of course we're talking 8, 12, or 16 cores...

 

[Automata]

You could also install Solaris and compare ZFS. We've found that in general software RAID outperforms PERC and other hardware RAID (particularly for RAID 3 or 5) as long as you have enough cores. I've asked our admins to either go to soft RAID 5 on Linux servers, or just switch them to Solaris to use ZFS. Of course we're talking 8, 12, or 16 cores...

I already ran that on the server and I don't like the operating system at all. I'm sticking to RHL/CentOS/Fedora.

 

[LeonX]

Guess im new, but try
Speccy.

TEACHER TEACHER!! I KNOW PLEASE LET ME TAKE THE WORD!!!:.
Ok use Hiren Boot CD; its an DOS interfaz.. so it wont lie, DOS is probably the best OS ever made... also 98SE is better..