ToddLC

Member
  • Content Count

    5
  • Joined

  • Last visited

Community Reputation

0 Neutral

About ToddLC

  • Rank
    Member
  1. ToddLC

    Performance of RAID5 mixed array

    What's your stripe size on the RAID 5 array? 64KB
  2. ToddLC

    Performance of RAID5 mixed array

    No kidding. The question is what is "off". The WB cache is of course on. In fact, turning it on is what really bumps the write performance. What I'm wondering is whether the mixed nature of the array is causing the problem. That is why I'd like to see some benches from others with RAID5 mixed arrays.
  3. My 7200.11 also shows a 0 byte cache with HDTune. Others have mentioned this also. This pretty much confirms that the problem is with HDTune and not the drive. I have another issue with the 7200.11 500GB drive that I'm not sure is specific to my disk. This issue is described in http://forums.storagereview.net/index.php?...rt=#entry247586. I'd greatly appreciate it if you would perform the benchmark I cited and post your results.
  4. In benchmarking an Intel ICH9R RAID5 array consisting of three 500GB Seagate disks two of which are 7200.10s and the other a 7200.11, my results are substandard: The same drives perform much better in a RAID0 or RAID1 array. Have any of you tried a mixed geometry RAID5 array built with Intel ICH8/9R and if so, how does your performance compare with mine? Gigabyte GA-P35-DS3P; Version F4 BIOS Intel E6600 2GB OCZ2P800R21G 5-5-5-15 @ 400MHz Slots 1,3 2GB Crucial Ballistix PC2-6400 4-4-4-12 @ 400MHZ Slots 2,4 XFX GeForce 8600GTS One 400GB Seagate ST3400620AS Two 500GB Seagate ST3500630AS One 500GB Seagate ST3500320AS LiteOn LH-20A1L-06 Samsung 245BW All stock timings, HSF Corsair HX520W Antec P180 Windows Vista x64 Home Premium
  5. ToddLC

    Seagate ST3500320AS

    I also recently bought one of these. My 7200.11 also has the old firmware, but I think I'm OK with the 32MB buffer as per the Seagate website info. HD Tune does, however, report a size of zero, which is a bit disconcerting. I have another issue with the ST3500320AS. Here's a copy of a post posted on ocforums that describes the situation: I'm in the process of setting up an Intel ICH9R-based RAID array consisting of three (or possibly four) Seagate 500GB drives. This array will be mixed in that two of the drives are 7200.10s (ST3500630AS) and the other is a 7200.11 (ST3500320AS). I've been benchmarking the three drive array in RAID5 and my results are distinctly subpar (see http://techreport.com/forums/viewtop...r=asc&start=30 and look for user ToddLC). This could be caused by a number of things such as the fact that the array is mixed, a buggy Intel driver for an ICH9R running on Vista x64, or individual drive issues. In this thread I want to restrict the discussion to an individual drive issue I encountered while trying to isolate the problem. In general, the 7200.11 is substantially faster than the 7200.10s. This is what one would expect given that the 7200.11 has a higher bit density than the 7200.10. For example, sequential read for the 7200.11 is around 100MB/s versus 70 for the 7200.10. But for a particular scenario, the 7200.11 bogs down and I would like to know if others have experienced this problem or can replicate it. This benchmark was done using IOMeter on raw disks. The non-default parameter values are: 16KB transfer size, 50% read/50% write, 100% sequential, outstanding I/Os between 1 and 10. What's strange with the results is that the transfer rate goes down as the number of outstanding I/Os goes up. Here are a few data points: #I/Os, Transfer Rate (MB/s) 7200.11 1 9.2 5 3.6 10 3.2 7200.10 1 8.8 5 8.5 10 17 Notice that the 7200.11 transfers at around 1/3 the rate at 10 outstanding I/Os compared to one I/O. The 7200.10 exhibits more or less what you would expect - faster at 10 than one I/O. This unexpected 7200.11 behavior pretty much happens with any small transfer size combined with mixed reads and writes. It gets worse the smaller the transfer size. I can't see how this could be caused by anything other than the drive as the 7200.10 behaves as expected. You would think the more overlapped I/Os, the better the drive could schedule the head placement. I thought this was what NCQ was all about. I'm wondering whether there is a problem with the scheduling algorithms in the 7200.11, i.e., a firmware problem. My drive has the SD04 firmware. Please try this test if you get a chance. I'd really like to know if my drive is at fault or this is a general issue. ---------------------------------------------------------- Gigabyte GA-P35-DS3P; Version F4 BIOS Intel E6600 2GB OCZ2P800R21G 5-5-5-15 @ 400MHz Slots 1,3 2GB Crucial Ballistix PC2-6400 4-4-4-12 @ 400MHZ Slots 2,4 XFX GeForce 8600GTS One 400GB Seagate ST3400620AS Two 500GB Seagate ST3500630AS One 500GB Seagate ST3500320AS LiteOn LH-20A1L-06 All stock timings, HSF Corsair HX520W Antec P180 Windows Vista x64 Home Premium Could you run this benchmark on your ST3500320AS? As I said in my ocforums msg, I'd like to know if my drive is peculiar. If yours is faster, I might upgrade the firmware, but don't really want to if it is a futile effort.