Sign in with
Sign up | Sign in

CPU Utilization And Conclusion

Southbridge Battle: 780a, ICH10 and SB750, Compared
By

If motherboard-down storage is so great, then how are vendors like LSI able to eke out a business selling add-in cards, such as the 6 Gb/s SAS controller used here? A combination of onboard processing and plenty of memory help offload the intensive parity calculations, which otherwise have to be handled by a host processor. In the chart above, you can clearly see the impact on CPU utilization, LSI's controller doing a great job of offloading storage-based calculations and Nvidia's 780a SLI taking the most significant hit.

Conclusion

The only conclusion that can be easily drawn is that you need to have a professional storage controller such as the LSI MegaRAID SAS 9260-8i, which we used for performance comparison, if you want to get the most from your complex storage arrays. The SAS/SATA controller was the only one to deliver high (or at least balanced) performance in all benchmark disciplines with some exceptions for Intel's ICH10R. The LSI still was the only controller able to deliver up to 1 GB/s throughput using the six Intel X25-E SSDs we had for testing, deliver high performance at writes and even when an array is degraded, and support a battery backup unit and professional RAID management features. Hence this card or similar professional products can be the only choice for truly mission-critical applications. The fact that the integrated storage controllers typically beat it in the I/O performance benchmarks may be due to the SSDs, which we found can be CPU-bound (see article Does Power-Saving Technology Kill SSD Performance?).

I/O Performance

All integrated solutions deliver amazingly high I/O performance, but results differ depending on the I/O benchmark pattern, which varies in access type and block sizes. Intel is superior in most of our I/O tests. Running high I/O loads on a degraded RAID 5 array results in a performance drop of around 30% to 60%, which seems acceptable.

Throughput

If you need a RAID array to deliver maximum read throughput, then you might want to go for a fast system using the ICH10R, since Intel’s controller delivers more than 650 MB/s in RAID 5 and 600+ MB/s in RAID 0. LSI stays tightly in the 460 to 480 MB/s range, while AMD and Nvidia limp along, struggling to catch even 200 MB/s.

Keep in mind that this is valid only for reads. If you switch to writes, AMD and Nvidia look even worse, yielding performance numbers not worth mentioning. Intel’s ICH10R still does well in write throughput and can be considered a competitor to LSI’s professional card—until you look at performance in degraded RAID 5. The only solution capable of maintaining at least 390 MB/s and still reaching 680 MB/s is LSI’s MegaRAID 9260-8i with its RAID-on-chip architecture.

Integrated RAID Good for Temporary High Performance Storage

The southbridges we looked at were designed to power desktop PCs, a purpose in line with our results. These chips aren’t built to be server/workstation solutions, so we don’t recommend creating complex RAID arrays on desktop systems. RAID 0 and RAID 5 performance work well for reads, but results are less than acceptable for writes with the exception of Intel’s ICH10. The AMD and Nvidia RAID offerings only are satisfactory, provided you don’t need high write performance…at all. We recommend using integrated RAID 0 or 1 options for temporary high performance storage, but RAID 5 should be left to professional products.

Display all 48 comments.
This thread is closed for comments
Top Comments
  • 19 Hide
    lonewarder , August 11, 2009 7:05 AM
    Please get that Gate.com pop-up ad window OUT OF MY FACE!!!! Every damned article has that freaking thing immediately jumping down in my way from page two on and I'm sick of it. I'll be reading Fudzilla first if you keep this crap up.
  • 11 Hide
    gkay09 , August 11, 2009 7:08 AM
    Why there are not many reviews about the nvidia chipset mobos for AMD ???
    It would be nice if TOM gives full reviews of those boards too...Many people show the desire to go SLI with AMD but are unwilling because of the problems and issues that are floating around the net about them...So it would help them much...
Other Comments
  • -1 Hide
    Netherwind , August 11, 2009 6:34 AM
    First! But anyway, it's nice to know that the Intel southbridge still stands up as a competitor to the LSI card, meaning I don't need to drop a ton of money on a raid controller for a desktop raid setup.. I just ordered 2 1TB Samsung Spinpoint F1's and plan to Raid 0 them. It's not exactly 'high performance' but it's high enough performing for the needs of the normal desktop user.
  • 19 Hide
    lonewarder , August 11, 2009 7:05 AM
    Please get that Gate.com pop-up ad window OUT OF MY FACE!!!! Every damned article has that freaking thing immediately jumping down in my way from page two on and I'm sick of it. I'll be reading Fudzilla first if you keep this crap up.
  • 11 Hide
    gkay09 , August 11, 2009 7:08 AM
    Why there are not many reviews about the nvidia chipset mobos for AMD ???
    It would be nice if TOM gives full reviews of those boards too...Many people show the desire to go SLI with AMD but are unwilling because of the problems and issues that are floating around the net about them...So it would help them much...
  • 0 Hide
    amnotanoobie , August 11, 2009 7:28 AM
    Interesting article, though I may not setup a RAID 5 on such a board (due to financial constraints), at least I have a reference when choosing a board to do a RAID 1 or 0.
  • 2 Hide
    dangerous_23 , August 11, 2009 8:09 AM
    at last, i've always wanted to see this comparison! well done intel!

    what do the transfer diagrams look like? can these raid controllers flatten out the transfer diagrams of conventional drives i.e. make max and min read about the same?
  • 6 Hide
    goose man , August 11, 2009 8:44 AM
    What I do not understand is why bother with the degraded writing performance.

    When you build a RAID array, you expect that array run normally most of the time, don't you ?

    My RAID 5 server, in last 3 years, only 1 time has one disk failed, and only took 1 day to rebuild the array and run normal again.
    1 day out of 3 years (> 1000 days) means less than 1 per mil of its life time it has to run in degraded write. And the quality of HD is much improved since 3 years ago (when I bought my HD to build my array)

    Let's be conservative, let's say when you build a RAID 5, you expect it failed twice a year (very bad expectation), to rebuild it still take less than two days, mean less than 1% of time per year.

    I will choose solution that has more performance that run more than 99% of time (i.e. ICH10 on RAID5) over one that has more performance that run less than 1% of its time (i.e MegaRAID on RAID 5) any day. ^__^

    So if degrade write performance only contribute less than 1%, why bother ?
  • 3 Hide
    haplo602 , August 11, 2009 9:06 AM
    how about a software raid in the mix ? like the linux implementation, just to compare is the chipsets have any benefit compared to a fully software colution ? you did throw in a HW card so a SW solution should complete the picture ...
  • 3 Hide
    Anonymous , August 11, 2009 9:20 AM
    1000 microseconds = 1 milisecond
    The access time is between 110 mircoseconds and 360 microseconds, so between 0.11 ms and 0.36 ms
  • 4 Hide
    dirtmountain , August 11, 2009 9:29 AM
    Since Schmid and Roos refuse to post any retail prices for anything, the LSI’s MegaRAID SAS9260-8i is $600.
    http://www.wwpi.com/storage/storage/7415-lsi-debuts-next-generation-6-gbs-sas-raid-controller-cards-
  • -2 Hide
    cangelini , August 11, 2009 9:40 AM
    adrianbalan1000 microseconds = 1 milisecondThe access time is between 110 mircoseconds and 360 microseconds, so between 0.11 ms and 0.36 ms


    Corrected on page 10. Thanks Adrian.
  • 2 Hide
    Anonymous , August 11, 2009 11:06 AM
    Good article but really strange choice of hardware. Surely the people who want these results are people looking to use SSD's in servers as well as the very small enthusiast community?
    In which case why havent Nvidia based server boards for Opterons or Intel based servers been submitted?
    Also raid 0 is great for a theory but realistically raid 10 is the most used raid solution, especially with solid state drives. It would have been good to have an update with this configuration.
    Hope you do a follow up article.
  • -1 Hide
    MartenKL , August 11, 2009 11:07 AM
    All of a sudden "Intel seems to have optimized its ICH10R for RAID 1+0"
    On page "Benchmark Results: I/O Performance"
    Nowhere else do I see any mention of 1+0 benchmarks nor any graphs.

    Does Nvidia and AMD solutions mean you have to have a working OS installed first? The descripcion of the raid setup tools is a bit unclear.

    To me the conclusion should be don't spend $600 on a new bottleneck for your raid performance. Who cares about degraded performance? No one in their right mind keeps a system running on degraded performance as you use your raid to secure important data you are not willing to risk it being destroyed by a second drive failure.

    Also too little focus is spent on the processor difference. Todays multicore cpus lend them self very well to these integrated solutions and processor overhead is almost a non-issue. Lastly, is it possible intel has some advantage as the SSDs are intel? $600 for LSI's card with the not included battery backup option as the only selling point, is better invested in a UPS for the entire system.
  • -1 Hide
    Anonymous , August 11, 2009 11:43 AM
    there is a big oops in the charts on the i/o section
    amd sb750 raid5 is exactly the same as intels ich 10r raid0 in every chart
  • -1 Hide
    qwertymac93 , August 11, 2009 11:46 AM
    um, you put one too many zeros on one of the access time charts. amd sb750 is listed as 600, i think it should be 60...
  • 0 Hide
    midnightgun , August 11, 2009 12:30 PM
    Should probably note that raid 5 is not supported on all 6 of the intel chipset ports (only the first 4).

    I'm still living with that error.
  • 0 Hide
    bk420 , August 11, 2009 12:53 PM
    I'm curious if the INTEL SSD are a case for the Intel chipset to have an advantage? I don't know, but they should have used another brand like Corsair or WD for this test. The LSI card I guess very expensive and out of reach for most home sever builds as are the Intel SSD's. It was a good article though seems like a lot of work! Good job TH!
  • -1 Hide
    Pei-chen , August 11, 2009 1:17 PM
    I moved from JMicron Raid on a Gigabyte board to an Intel ICH7R Asus board for the same reason Tom’s pointed out. My Raid 0 setup with two 7200.10 drives tops out around 120mb/sec read while the Intel let me have 140mb/sec. I have since replaced the 7200.10 with 7200.11 1.5TB and have 200mb/sec read speed from the Intel chipset. Equivalent offer from AMD and Nvidia 3 years ago couldn’t push more than 120mb/sec like the JMicron.
  • 1 Hide
    altoidman85 , August 11, 2009 2:44 PM
    "Please get that Gate.com pop-up ad window OUT OF MY FACE!!!!" I agree lonewarder. I have been reading toms for years now and frankly the ads are getting really annoying. I know they have to pay their bills too but how about keeping them less obtrusive.

    Good article tho. I was curious as to how ich10 stacked up against the offerings from amd and nvidia.
  • 3 Hide
    TidalWaveOne , August 11, 2009 3:08 PM
    lonewarderPlease get that Gate.com pop-up ad window OUT OF MY FACE!!!! Every damned article has that freaking thing immediately jumping down in my way from page two on and I'm sick of it. I'll be reading Fudzilla first if you keep this crap up.


    I don't see it... use Firefox with AdBlock add-on.
  • -1 Hide
    nemi_PC , August 11, 2009 3:47 PM
    would have been great to see what a Windows software managed raid 5 performance is like on any of the platforms as a reference point.
Display more comments