Lately, the AMI / LSI Logic MegaRAID Elite 1600 and OEM versions from Dell have been very popular amongst SCSI enthusiasts. Unfortunately, many people are complaining about poor write and read performance while in fact their controllers may perform very well. The benchmarks on this page will proof once again there is no direct correllation between sequential transfer rates and actual desktop performance. Despite lower sequential transfer rates, the MegaRAID shows better reald world performance. Thanks to its onboard Intel 960 I/O processor, 128MB cache and intelligent prefetching and write-back caching strategies the MegaRAID Elite 1600 is able to pull ahead of firmware based RAID solutions such as Adaptec's HostRAID and LSI's Integrated Mirroring.
Testbed
The benchmarks were performed on an Asus A7M266-D mainboard with two AMD Athlon XP 2400+ processors, 1GB PC2100 ECC Registered DDR SDRAM and up to three 36.7GB Seagate Cheetah 10K.6 drives. Windows XP was patched with the latest updates from Windows Update. Single drive performance was compared to intelligent RAID 0, 1 and 5 performance on the MegaRAID Elite 1600, HostRAID 0 and 1 performance on Adaptec's 29320LP-R and Integrated Mirroring RAID 1 performance on the Tekram DC-390U4B. The DC-390U4B is based on the LSI Logic 53C1030 Ultra320 SCSI chipset. For some reason the Adaptec 29320LP-R showed disappointing performance on the 760MPX platform. Performance on the nForce2 platform is comparable to or better than the LSI Logic 53C1030 chipset.
The MegaRAID Elite 1600 we used in this article is actually a Dell Perc 3/DC by origin. The card was transformed into a 'real' LSI MegaRAID by updating its firmware to the latest G170 release. 128MB cache and a Battery Backup Unit were present. Alle benchmarks were performed with adaptive read-ahead caching, cached IO, write-back caching and 64K stripe size enabled.
Our desktop benchmarks are based on the Intel IPEAK SPT trace and playback method, which has been thoroughly described by Storage Review. The test methodology is comparable to Storage Review, although we used different wordloads. The Office and Workstation tests are based on recordings of disk accesses during a run of Business Winstone 2001 and Content Creation Winstone 2002. The Virusscan test is based on various recordings of Norton AV full system virusscans. The Software Installation trace was recorded while installing Microsoft Office XP, Adobe Photoshop 7.0, Corel WordPerfect 2000 and several smaller applications. Two variants of the Office and Workstation benchmarks were created; a Light version with almost no file fragmentation and a Heavy version with consirable file fragmentation and some additional background disk usage. Because of time constraints a subset of our complete benchmarksuite was used.
Benchmarks
 | ||||||||
 | | |||||||
| ||||||||
| Adaptec 29320LP-R - HostRAID 0 - 3x 10K.6 |    204000 | | |||||
| ||||||||
| Tekram DC-390U4B - IM RAID 1 - 2x 10K.6 |    82900 | | |||||
| ||||||||
| Tekram DC-390U4B - 1x 10K.6 |    70500 | | |||||
| ||||||||
| LSI MegaRAID Elite 1600 - RAID 5 - 3x 10K.6 |    69100 | | |||||
| ||||||||
| Adaptec 29320LP-R - HostRAID 1 - 2x 10K.6 | 68400 | | |||||
| ||||||||
| Adaptec 29320LP-R - 1x 10K.6 | 68300 | | |||||
| ||||||||
| ||||||||
| | |||||||
| ||||||||
| Adaptec 29320LP-R - HostRAID 0 - 3x 10K.6 | 120000 | | |||||
| ||||||||
| LSI MegaRAID Elite 1600 - RAID 5 - 3x 10K.6 | 69200 | | |||||
| ||||||||
| Tekram DC-390U4B - IM RAID 1 - 2x 10K.6 | 54900 | | |||||
| ||||||||
| Tekram DC-390U4B - 1x 10K.6 | 41400 | | |||||
| ||||||||
| Adaptec 29320LP-R - 1x 10K.6 | 40100 | | |||||
| ||||||||
| Adaptec 29320LP-R - HostRAID 1 - 2x 10K.6 | 40100 | | |||||
| ||||||||
RAID 0 sequential transfer rates on the MegaRAID Elite 1600 are in the 70-80MB/s range.
| ||||||||
| | |||||||
| ||||||||
| LSI MegaRAID Elite 1600 - RAID 0 - 3x 10K.6 | 521 | | |||||
| ||||||||
| Adaptec 29320LP-R - HostRAID 0 - 3x 10K.6 | 493 | | |||||
| ||||||||
| LSI MegaRAID Elite 1600 - RAID 5 - 3x 10K.6 | 426 | | |||||
| ||||||||
| LSI MegaRAID Elite 1600 - RAID 1 - 2x 10K.6 | 403 | | |||||
| ||||||||
| Tekram DC-390U4B - 1x 10K.6 | 302 | | |||||
| ||||||||
| Adaptec 29320LP-R - HostRAID 1 - 2x 10K.6 | 290 | | |||||
| ||||||||
| Tekram DC-390U4B - IM RAID 1 - 2x 10K.6 | 290 | | |||||
| ||||||||
| Adaptec 29320LP-R - 1x 10K.6 | 262 | | |||||
| ||||||||
| ||||||||
| | |||||||
| ||||||||
| LSI MegaRAID Elite 1600 - RAID 0 - 3x 10K.6 | 549 | | |||||
| ||||||||
| LSI MegaRAID Elite 1600 - RAID 5 - 3x 10K.6 | 450 | | |||||
| ||||||||
| Adaptec 29320LP-R - HostRAID 0 - 3x 10K.6 | 435 | | |||||
| ||||||||
| LSI MegaRAID Elite 1600 - RAID 1 - 2x 10K.6 | 415 | | |||||
| ||||||||
| Adaptec 29320LP-R - HostRAID 1 - 2x 10K.6 | 270 | | |||||
| ||||||||
| Tekram DC-390U4B - 1x 10K.6 | 265 | | |||||
| ||||||||
| Tekram DC-390U4B - IM RAID 1 - 2x 10K.6 | 254 | | |||||
| ||||||||
| Adaptec 29320LP-R - 1x 10K.6 | 248 | | |||||
| ||||||||
| ||||||||
| | |||||||
| ||||||||
| LSI MegaRAID Elite 1600 - RAID 1 - 2x 10K.6 | 620 | | |||||
| ||||||||
| LSI MegaRAID Elite 1600 - RAID 5 - 3x 10K.6 | 612 | | |||||
| ||||||||
| LSI MegaRAID Elite 1600 - RAID 0 - 3x 10K.6 | 602 | | |||||
| ||||||||
| Adaptec 29320LP-R - HostRAID 0 - 3x 10K.6 | 428 | | |||||
| ||||||||
| Tekram DC-390U4B - IM RAID 1 - 2x 10K.6 | 392 | | |||||
| ||||||||
| Tekram DC-390U4B - 1x 10K.6 | 356 | | |||||
| ||||||||
| Adaptec 29320LP-R - HostRAID 1 - 2x 10K.6 | 348 | | |||||
| ||||||||
| Adaptec 29320LP-R - 1x 10K.6 | 292 | | |||||
| ||||||||
| ||||||||
| | |||||||
| ||||||||
| LSI MegaRAID Elite 1600 - RAID 1 - 2x 10K.6 | 859 | | |||||
| ||||||||
| LSI MegaRAID Elite 1600 - RAID 0 - 3x 10K.6 | 790 | | |||||
| ||||||||
| LSI MegaRAID Elite 1600 - RAID 5 - 3x 10K.6 | 736 | | |||||
| ||||||||
| Tekram DC-390U4B - IM RAID 1 - 2x 10K.6 | 618 | | |||||
| ||||||||
| Adaptec 29320LP-R - HostRAID 0 - 3x 10K.6 | 528 | | |||||
| ||||||||
| Tekram DC-390U4B - 1x 10K.6 | 507 | | |||||
| ||||||||
| Adaptec 29320LP-R - HostRAID 1 - 2x 10K.6 | 357 | | |||||
| ||||||||
| Adaptec 29320LP-R - 1x 10K.6 | 304 | | |||||
| ||||||||
| ||||||||
| | |||||||
| ||||||||
| LSI MegaRAID Elite 1600 - RAID 0 - 3x 10K.6 | 826 | | |||||
| ||||||||
| LSI MegaRAID Elite 1600 - RAID 5 - 3x 10K.6 | 649 | | |||||
| ||||||||
| LSI MegaRAID Elite 1600 - RAID 1 - 2x 10K.6 | 592 | | |||||
| ||||||||
| Adaptec 29320LP-R - HostRAID 0 - 3x 10K.6 | 524 | | |||||
| ||||||||
| Tekram DC-390U4B - 1x 10K.6 | 361 | | |||||
| ||||||||
| Adaptec 29320LP-R - 1x 10K.6 | 353 | | |||||
| ||||||||
| Adaptec 29320LP-R - HostRAID 1 - 2x 10K.6 | 341 | | |||||
| ||||||||
| Tekram DC-390U4B - IM RAID 1 - 2x 10K.6 | 267 | | |||||
| ||||||||
DIY desktop benchmarking
Intel IPEAK Storage Performance Toolkit has been discontinued by Intel. Luckily, a trial version can still be downloaded at this Sunsite mirror. The trial will run five times. You can perform the Office and Workstation benchmarks on your system by opening the two files below in RankDisk, which is part of IPEAK Storage Performance Toolkit. RankDisk will report results in terms of average service time. IOps (I/O operations per second) can be calculated by dividing 1000 by average service time (in milliseconds). Use Windows Disk Management to look up the number of the physical drive you wish to test. Be cautious: RankDisk will destroy all data on the drive!
Office Heavy trace: office-heavy.RAW (1.98MB)
Office Workstation trace: workstation-heavy.RAW (1.75MB)
Recent storage articles on Tweakers.net
:strip_exif()/i/1046960112.jpg?f=thumbmedium)
/u/1/femme.png?f=community){kind=small}
Wil je het internationaal aanzien van Tweakers.net verhogen of is deze review voor een andere site geschreven?