UhClem

Members
  • Content Count

    265
  • Joined

  • Last visited

Community Reputation

3 Neutral

About UhClem

  • Rank
    Member

Converted

  • Gender
    Undisclosed
  • Location
    NH, US

Recent Profile Visitors

The recent visitors block is disabled and is not being shown to other users.

  1. As for the 7GB/s "estimate", note that it assumes that the 9300-8i itself actually has the muscle. It probably does, but it is likely that it will be PCIe limited to slightly less than 7.0 GB/s (ref: your x4 test which got 3477; [i.e. < 7000/2]) Interesting. The under-performance of my script suggests another deficiency of the PMC implementation (vs LSI) : My script uses hdparm -t which does read()'s of 2MB (which the kernel deconstructs to multiple 512KB max to the device/controller). Recall that LSI graphic you included which quantified the Databolt R/W throughput for differe
  2. Try the B option. It might help, or not ... Devices (and buses) can act strange when you push their limits. The nvmx script uses a home-brew prog instead of hdparm. Though I haven't used it myself, you can check out fio for doing all kinds of testing of storage. I completely agree with you. I do not completely agree with this. I'll send you a PM.
  3. Certainly ... but as an old-school hardcore hacker, I wonder if it could have been (at least a few %) better. I have to wonder if any very large, and very competent, potential customer (e.g., GOOG, AMZN, MSFT), did a head-to-head comparison between LSI & PMC before placing their 1000+ unit chip order. That lays the whole story out--with good quantitative details. I commend LSI. And extra credit for "underplaying" their hand. Note how they used "jumps from 4100 MB/s to 5200 MB/s" when their own graph plot clearly shows ~5600. (and that is =~ your own 5520) I suspect tha
  4. You're getting there ... 😀 Maybe try a different testing procedure:. See the attached script. I use variations of it for SAS/SATA testing. Usage: ~/bin [ 1298 ] # ndk a b c d e /dev/sda: 225.06 MB/s /dev/sdb: 219.35 MB/s /dev/sdc: 219.68 MB/s /dev/sdd: 194.17 MB/s /dev/sde: 402.01 MB/s Total = 1260.27 MB/s ndk_sh.txt Speaking of testing (different script though) ... ~/bin [ 1269 ] # nvmx 0 1 2 3 4 /dev/nvme0n1: 2909.2 MB/sec /dev/nvme1n1: 2907.0 MB/sec /dev/nvme2n1: 2751.0 MB/sec /dev/nvme3n1: 2738.8 MB/sec /dev/nvme4n1: 2898.5 MB/sec Total = 14204.5 MB
  5. Excellent evidence! But, to me, very disappointing that the implementations (both LSI & PMC, apparently)] of this feature are this sub-optimal.. Probably a result of cost/benefit analysis with regard to SATA users (the peasant class--"Let them eat cake."). Also surprising that this hadn't come to light previously. Speaking of the LSI/PMC thing ... Intel's SAS3 expanders (such as the OP's) are documented, by Intel, to use PMC expander chips. How did you verify that your SM backplane actually uses a LSI expander chip (I could not find anything from Supermicro themself; and I'm no
  6. Please keep things in context. OP wrote: Since the OP seemed to think that an x16 card was necessary, I replied: And then you conflated the limitations of particular/"typical" PCIe3 SAS/SATA HBAs with the limits of the PCIe3 bus itself. In order to design/configure an optimal storage subsystem, one needs to understand, and differentiate, the limitations of the PCIe bus, from the designs, and shortcomings, of the various HBA (& expander) options. If I had a single PCIe3 x8 slot and 32 (fast enough) SATA HDDs, I could get 210-220 MB/sec on each d
  7. In my direct, first-hand, experience, it is 7100+ MB/sec. (I also measured 14,200+ MB/sec on PCIe3 x16). I used a PCIe3 x16 card supporting multiple (NVMe) devices. [In a x8 slot for the first measurement.] [Consider: a decent PCIe3 x4 NVMe SSD can attain 3400-3500 MB/sec.] That table's "Typical" #s are factoring in an excessive amount of transport layer overhead. I'm pretty certain that the spec for SAS3 expanders eliminates the (SAS2) "binding" of link speed to device speed. I.e., Databolt is just Marketing. Well, that's two tests of the 9300, with differen
  8. It looks to me like you are not limited by PCIe bandwidth. PCIe gen3 @ x8 is good for (real-world) ~7000 MB/sec. If you are getting a little over 200 MB/sec each for 28 drives, that's ~6000 MB/sec. (You are obviously using a Dual-link connection HBA<==>Expander which is good for >> 7000 [9000].) Either your 9300 does not have the muscle to exceed 6000, or you have one (or more) drives that are stragglers, handicapping the (parallel) parity operation. (I'm assuming you are not CPU-limited--I don't use unraid.)
  9. OK. I'd still suggest 24 hrs of MPrime (aka Prime95) Torture-Blend on the one in play here.
  10. JB, do you have ECC memory? (I know it's not a guarantee, but it gets you 90-95% of the way there.)
  11. The whole time, or just post-read verify ?? (I don't use Unraid, but I vaguely recollect the details of Joe's preclear.) No, it will not affect the CPU usage. It does (effectively) eliminate the I/O bottleneck of the on-board (chipset) Sata sub-system. That CPU usage you saw during pre-clear (x2) should not guide any (re-configure) decision you make.
  12. I've looked into "staggered spinup" for a DIY DAS. The key search term you want to research is "disk PUIS" ... Power Up In Standby. It looks a little tricky, but quite doable. (That isn't a solution for me because my drives are connected via a SAS expander.) [I don't use Unraid.]
  13. Before recommending a seller, I'd like to make sure we are seeking the best solution. Based on your system specs (mobo/cpu), you actually have 5 PCIe g3 x8 slots -- OR only 3 slots if one (or two) need to supply x16 lanes (each). If it is the latter case (you need to "free up" an x16 slot, then I'd suggest considering a SAS(/SATA) expander, which can use one of the soon-to-be x0 slots (expanders only use a PCIe slot for power (no signals/lanes) and a place to live). If it's the former case (you want to repurpose an x8 slot now used by an H200, then, yes, probably a 16 (
  14. Thanks! Good points. I think that spec'd endurance (e.g., 600 TBW for 860 EVO 1TB) won't be an issue, for all but extreme use cases. (For a data point, I used a 860 EVO 500GB in a DVR (DirecTV HR24) for the last year. It had ~8000 hours and ~30 TBW when I secure-erased it. Sadly, I didn't think to do any write-performance tests before the erase.) An "extreme use case" might be an array for multi HD security cameras [e.g. 4 feeds @ 10GB/hr each (24/7) =~ 350 TBW/year]. Note, though, that you need a near-server-level NVME to exceed 1 PBW rating (for 1 TB device). As you said, t