Some Sandy PC benches incl an E5

Discussion in 'Mac Pro' started by xgman, Feb 29, 2012.

  1. xgman, Feb 29, 2012
    Last edited: Mar 1, 2012

    xgman macrumors 601


    Aug 6, 2007
    I saw this on gearslutz for what it is worth:
    From this thread:

    (also see post 9 below "E5 Already in Channel")

    "Its time again with several new products coming out shortly.

    usual place
    Benchmarks for Pro Audio!
    2nd graph down.

    tests are done using Dawbench DSP/RXC/Cubase
    DAW Bench : DAW Performance Benchmarking

    first up the Quad core Sandy E 3820 3.6GHz (already released)
    stock speed.

    256 296
    128 288
    64 270
    32 252

    this is performing very close to the 6 core 3960K (3.33GHz) and 3930k (3.2GHz) stock speeds.
    This certainly makes a case of GHz over core count or in this case where GHz (3.6) catches up with more cores of a slower ghz (3.3)

    overclocked to 4.2GHz (oddly for the first time almost not worth talking about) in comparison to the 3930k/2630k @ 4.5GHz, but it does edge out the 2600k @ 4.5GHz barely.

    256 323
    128 311
    64 293
    32 273

    now for a surprise i was NOT expecting.

    the new Intel E5 Xeon E5 2670 2 .6GHz 8 core
    this is for a SINGLE Xeon 8 core not a pair. I have a pair but no board at the moment that works well) this was dropped into a standard X79 board.
    bear in mind this is a $1600-1700 processor so dont get too excited.

    256 436
    128 411
    64 398
    32 322

    this slow 2.6GHz 8 core 20 meg cache beat everything @ every buffer other than 32 where the 3690k/3630k @ 4.5GHz took a decent lead (353)

    so much for GHz over core count (my mantra)... other than @ 32buffer.
    only thing i can think of is the added cache is really making a difference with effects processing.

    this Xeon is the same silicon as the other Sandy E's so i cant explain the increase with just 2 more cores @ a lower GHz yet i am not surprised its performance drooped @ 32 buffer as this is rather the norm on muliltprocessor rigs. (more cores but lower GHz)

    i also have an Ivy Bridge 3770k and a board coming for the Dual Xeon E5
    so more to come in a week or so

    "we didnt run up any animations tests but for Adobe the 8 core did really bad for the MPE.

    Premiere Pro CS5 Version 5.5 Testing
    Video material - AVCHD 1080P 24 Frame Each Cut to 30 minutes of material
    Export Codec - H264 HDTV 1080P 24 Preset Default
    4 Effects per Layer - Fast Color Corrector, Brightness & Contrast, Video Limiter, Sharpen
    Each Layer Scaled to 50% for 4 frame PinP view.

    I7 2600K 4.7 GHz
    16GB Blackline 1600 CL 9
    4 WD 1Tb Sata 64 Meg Cache 600 Drives in 2 Raid 0 arrays
    3 Layer - 30:46
    4 Layer - 33:36

    X79 3.3 @ 4.5 GHz
    32GB 1333
    580GTX 3GB
    4 1Tb Sata 32 Meg Cache 600 Drives in 2 Raid 0 arrays
    3 Layer - 27:43
    4 Layer - 30:02

    E5 2670 @ 2.6 GHz <--- single not dual
    32GB 1600
    570GTX 2.5GB
    4 1Tb Sata 32 Meg Cache 600 Drives in 2 Raid 0 arrays
    3 Layer -
    4 Layer - 40:41 <--- pretty slow

    but since you own an Apple here it is.

    Mac Pro 2010 Dual 2.93GHz
    24GB 1600 CL 9
    Quadro 4000
    2 WD 1Tb Sata 64 Meg Cache 600 Drives in Raid 0

    3 Layer - 42:24
    4 Layer - 44:05"

  2. goMac macrumors 603

    Apr 15, 2004
    Needs links and an explanation of what the numbers mean.
  3. xgman thread starter macrumors 601


    Aug 6, 2007
  4. derbothaus macrumors 601


    Jul 17, 2010
    Such a sweeping statement after looking at a benchmark based on fully multi-thread aware apps. Now for iTunes test. Whatever has the highest turbo wins. So GHz still wins over cores most of the time. Unless physics has changed.
  5. xgman thread starter macrumors 601


    Aug 6, 2007
    I take the whole thing in stride and with a grain of salt, but somewhat interesting none the less I thought.
  6. derbothaus macrumors 601


    Jul 17, 2010
    Totally. They look nice and the extra thermal headroom should net gains everywhere.
  7. Neodym macrumors 68000


    Jul 5, 2002
    Thanks for posting! For me the most interesting part was that the new Xeons seem to offer features that were thought to be IvyBridge-exclusive (-> Tri-gate).

    Maybe there is more of similar stuff and what is commonly called "Sandy Bridge" may not be that precise a term anymore when it comes to the new Xeons...
  8. Umbongo macrumors 601


    Sep 14, 2006
    Yeah very interesting. Certainly would explain some things about time lines and Intel's attitude. I suppose this will all come out next week at cebit.
  9. xgman thread starter macrumors 601


    Aug 6, 2007
    I also saw this report today. I'll take it as good news so far. Not sure how many model numbers this includes:

    "Intel's Xeon E5 is already in the channel says Boston
    No paper launch after delays
    By Lawrence Latif
    Thu Mar 01 2012, 16:45

    CHIPMAKER Intel is finally set to launch its Xeon E5 chips next week with server and workstation vendors keen to show it isn't going to be a paper launch and that chips are already in the channel.

    The UK distributor for Supermicro, Boston invited The INQUIRER into its labs to take a look at an Intel Xeon E5 workstation. While we were asked not to publish benchmarks until next week, the firm did allow us to share pictures of the working system and the Socket 2011 Xeon E5 chips, known as Sandy Bridge E.

    img-9285 img-9286

    Boston's workstation is based around a dual-socket Supermicro motherboard. The firm said it has many other Supermicro Xeon E5 SKUs and because of the widely publicised delays the firm, like its rivals, has had loads of time to test and validate its kit.

    img-9266 img-9264

    The system has two Sandy Bridge Xeon E5 chips in it, coupled to one 128GB SSD and two 1TB hard drives. Graphics are courtesy of an Nvidia Quadro 4000 board and it doesn't take a genius to work out that it is a fast bit of kit, though you will have to wait a little longer to see how much faster it is than single processor alternatives.

    Although we can't talk about performance figures for a few more days, we can certainly say there were no issues with stability. Heat wasn't particularly noticeable, though we were running the system in a heavily air-conditioned lab, nevertheless the heatsinks could be touched without burning flesh.

    Boston confirmed to The INQUIRER that Xeon E5 chips have been in the channel for a few weeks now and confirmed to us that it won't be a paper launch. We have also been shown Xeon E5 systems from other large vendors that give credence to the notion that Intel's Xeon E5 will arrive, for real, next week. µ

    Source: The Inquirer

Share This Page