Overclock.net › Forums › Components › Hard Drives & Storage › RAID Controllers and Software › PERC 5/i RAID Card: Tips and Benchmarks
New Posts  All Forums:Forum Nav:

PERC 5/i RAID Card: Tips and Benchmarks - Page 316

post #3151 of 7150
Here is my Raid0 V-Raptor setup. It seems around the same as yours. Minus the two dips that i really dont like. Is that normal?



Quote:
Originally Posted by dc5rsx View Post
My benchmark for 3x 300GB VelociRaptors RAID 5/64k/Adaptive/Write Back:
236.7MB/sec Maximum
191.1MB/sec Average
7.5 Access

These numbers seem really disappointing to me, considering there 7200RPM drives here getting ~300MB/sec average. What do you guys think?

Edited by zerotrace992 - 8/28/09 at 6:22am
Main PC
(13 items)
 
  
CPUMotherboardGraphicsRAM
Q9550 @ 3.4 DFI LP LT X48-T2R 8800GT 512MB 8GB OCZ Reaper HPC 
Hard DriveOptical DriveOSPower
2x VelociRaptor 300GB - Raid0 LG BluRay SATA & Pioneer DVD IDE Windows Vista x64 Extreme Edition R2 CoolerMaster 850W Real Power Pro PSU 
Case
CoolerMaster Stacker 810 
  hide details  
Reply
Main PC
(13 items)
 
  
CPUMotherboardGraphicsRAM
Q9550 @ 3.4 DFI LP LT X48-T2R 8800GT 512MB 8GB OCZ Reaper HPC 
Hard DriveOptical DriveOSPower
2x VelociRaptor 300GB - Raid0 LG BluRay SATA & Pioneer DVD IDE Windows Vista x64 Extreme Edition R2 CoolerMaster 850W Real Power Pro PSU 
Case
CoolerMaster Stacker 810 
  hide details  
Reply
post #3152 of 7150
Quote:
Originally Posted by dc5rsx View Post
My benchmark for 3x 300GB VelociRaptors RAID 5/64k/Adaptive/Write Back:
236.7MB/sec Maximum
191.1MB/sec Average
7.5 Access
One more thing to Keep in mind is that in RAID-5 you only get the performance of n-1 drives writes where n is the number of drives you have. This is because the drives have to write a drives worth of parity data. On reads you also take some penalty because the drives are skipping the parity data.

So your 3 drives act like two for writing, and slightly less than 3 for reading. Per the reviews of the 300GB Velociraptor it averages about 100MB/sec read and 100MB/sec write. Your benchmark's seem to be about what you should expect.

Some of the 7200 RPM drives have much higher areal densities which gives them comparable or even better STR, which is why you see people with similar or better speeds from 3 of the latest gen 7200RPM drives.

Just as a point of reference, here's a the STR of single 5400RPM drive with 3 500GB platters.



Even it isn't that far from a single 300GB Velociraptor in terms of STR.
Edited by Stereodude - 8/28/09 at 8:33am
post #3153 of 7150
Quote:
Originally Posted by Stereodude View Post
What's your stripe size and what application did you use to arrive at those numbers?
64k stripe size / Benched using HDTune

I reformatted and put the 3 VRaptors in RAID0. My max went up to 270Mb/sec and my average went up to 224MB/sec. So about a 24MB/sec increase in speed from RAID5 to RAID0.

The RAID0 speeds even seems a little low to me. My friend has 3 7200RPM drives in RAID 0 and he's getting 280MB/sec average.
post #3154 of 7150
Quote:
Originally Posted by zerotrace992 View Post
Here is my Raid0 V-Raptor setup. It seems around the same as yours. Minus the two dips that i really dont like. Is that normal?
I got similar dips and speeds, so I guess it's normal. I expected the V-Raptor's to perform a bit faster, especially because it's on the Perc card and all, no?
post #3155 of 7150
Quote:
Originally Posted by dc5rsx View Post
I got similar dips and speeds, so I guess it's normal.
Do you have Windows installed and running on the array that you're testing?
post #3156 of 7150
How would one go about changing a Raid 5 array with 7x 1TB drives into a Raid 5 array with 7x1.5TB drives?
post #3157 of 7150
Quote:
Originally Posted by ara View Post
How would one go about changing a Raid 5 array with 7x 1TB drives into a Raid 5 array with 7x1.5TB drives?
I'm pretty sure you will need to reinitialize the drives which is destructive
    
CPUMotherboardGraphicsRAM
AMD Phenom 9950 Black Edition @ 3.0Ghz Gigabyte mATX 785G? HD3200 IGP *Passively cooled* 2x1GB WINTEC Crap, 2x2GB Patriot | DDR2 800 1.9v 
Hard DriveOSMonitorKeyboard
Western Digital AAKS Series 500GB Windows 7 Enterprise x64 Custom Cooked for SPEED Samsung 58" B650 1080p Plasma Logitech DiNovo Mini 
PowerCaseMouse
Silverstone 350Watt Silent PSU Cooler Master 590 Logitech DiNovo Mini 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
AMD Phenom 9950 Black Edition @ 3.0Ghz Gigabyte mATX 785G? HD3200 IGP *Passively cooled* 2x1GB WINTEC Crap, 2x2GB Patriot | DDR2 800 1.9v 
Hard DriveOSMonitorKeyboard
Western Digital AAKS Series 500GB Windows 7 Enterprise x64 Custom Cooked for SPEED Samsung 58" B650 1080p Plasma Logitech DiNovo Mini 
PowerCaseMouse
Silverstone 350Watt Silent PSU Cooler Master 590 Logitech DiNovo Mini 
  hide details  
Reply
post #3158 of 7150
Quote:
Originally Posted by ara View Post
How would one go about changing a Raid 5 array with 7x 1TB drives into a Raid 5 array with 7x1.5TB drives?
I'm not sure you can with the Perc cards in the way that you're thinking.

I read a procedure on how to do it on some 3ware cards once, and it sounded like something I definitely wouldn't want to attempt. Basically you would fault out the smaller drives one at a time and rebuild with the larger drives. Once you had replaced all the drives there was some trick (don't recall what it was) that you did before starting their online capacity expansion routine which would then let you expand the array into the extra space on all the drives.

IMHO, the best way to change from 7x1TB to 7x1.5TB is to get a 2nd Perc card, build a new array, and copy all the data over.
post #3159 of 7150
Quote:
Originally Posted by dc5rsx View Post
I got similar dips and speeds, so I guess it's normal. I expected the V-Raptor's to perform a bit faster, especially because it's on the Perc card and all, no?
defrag your array and run the test again, sometimes that causes dips.
post #3160 of 7150
Quote:
Originally Posted by Stereodude View Post
Here's are the early results from my RAID-5 setup.

Gear:
  • 6x Samsung Spinpoint F2EG HD154UI 1.5TB 5400 RPM 32MB cache
  • Dell Perc 5/i PCI-E x8 SAS RAID controller with 256MB cache & battery with latest LSI firmware.
  • 2x 32pin SAS to 4 SATA cables
  • Gigabyte GA-EP45-UD3P rev 1.0 motherboard
  • Intel E5200 CPU
  • Windows XP x64 SP2

The array is a single RAID-5 volume with a 512k stripe. The Perc 5i is in the 2nd PCI-E x16 slot (farthest from the processor). It required taping the SMbus pins to boot.
Here's the final benchmarks from my RAID-5 array.

I used IOmeter to duplicate the style of type of benchmark ATTO produces, but without all the inconsistency and errors*.



*ATTO produces inconsistent and inaccurate results because the 256MB data length written during the benchmark is too small when considering in the amount of cache on the Perc 5/i card (256MB). In my above IOmeter test the smallest data size written was ~7.36GB which takes prevents the cache from skewing the STR results.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: RAID Controllers and Software
Overclock.net › Forums › Components › Hard Drives & Storage › RAID Controllers and Software › PERC 5/i RAID Card: Tips and Benchmarks