Results 1 to 8 of 8

Thread: Render time comparisons - very odd results!

  1. #1
    Certified AUGI Addict patricks's Avatar
    Join Date
    2004-06
    Location
    Memphis TN area
    Posts
    7,047
    Login to Give a bone
    0

    Default Render time comparisons - very odd results!

    I ran some render time comparisons on our 4 most powerful machines in the office, and came up with very odd/interesting results. To run the test I opened a local copy of a particular project on each machine, went to the same view, and simply clicked Render.

    I started with my machine and rendered 3 times: first with other programs open, 2nd with no other programs open, and 3rd immediately after a reboot. Times varied only by a few seconds so I just took that average, and then ran the other tests with no other programs open on the other machines.

    The results (all machines run Windows 7 x64):

    My new machine - Dell Studio XPS 8100
    - Core i7 Quad 8700 2.93 GHz
    - 12 GB (2x 4GB and 2x 2GB) DDR3-1333 SDRAM
    - ATI Radeon HD 5770 1 GB graphics

    Render time = 7:58

    My old machine - Dell Precision 490 (purchased 12/2007, upgraded 10/2009)
    - 2x Xeon dual-core 5150 2.67 GHz
    - 16 GB (8x 2GB) DDR2-667 fully buffered SDRAM
    - Quadro FX 580 512 MB graphics

    Render time = 13:29 (surprised it took this long)

    Dell Precision T3500 (purchased 5/2010)
    - Xeon Quad W3530 2.8 GHz
    - 12 GB (6x 2GB) DDR3-1066 SDRAM
    - Quadro FX 580 512 MB graphics

    Render time = 14:49

    Dell Precision T3500 (purchased 10/2009)
    - Xeon Quad E5520 2.27 GHz
    - 12 GB (3x 4GB) DDR3-1066 SDRAM
    - Quadro FX 580 512 MB graphics

    Render time = 11:56

    Well, as you can see my new machine is a good bit faster than all the others at rendering. I figured it would with the newer, faster processor and faster memory clock speed.

    What gets me is the last two. Those were the 2 newest machines in the office before the XPS, and yet the 1-year old T3500 with slower processor renders faster than the newer T3500, and even faster than my old machine.

    This link shows a direct comparison between the W3530 and the E5520 Xeons: http://ark.intel.com/Compare.aspx?ids=40200,41313, On paper it looks like the W3530 should perform a bit better, but in reality the E5520 machine rendered nearly 25% faster. Why would this be the case? It seems pretty sad that the 2nd newest machine in the office actually renders the slowest.
    Intern Architect, BIM Manager/Coordinator
    AERC, PLLC
    Hernando, Mississippi

    Revit - all up in your voxel space

  2. #2
    AUGI Addict iru69's Avatar
    Join Date
    2004-10
    Location
    usa
    Posts
    1,591
    Login to Give a bone
    0

    Default Re: Render time comparisons - very odd results!

    Interesting... I do find the results surprising... and I have to conclude that there's something unusual going on there with your rendering tests.

    I would expect the following order: i7-870, W3530, E5520, 5150x2... but more than the order, it's the time differences.

    As an example, on paper, I would expect the W3530 to be 5-10% slower than the i7-870 (on a rendering test like this), not 45% slower - there's no explanation for that discrepancy other than there's something wrong with the computer or your test is somehow flawed.

    Also, there's no way the E5520 should be able to keep up with the W3530... again, maybe the simple explanation is that there's something wrong with that W3530 machine... actually the more I think about it, the more everything would fall into place if that W3530 was actually in the 8 - 10 min range.

    The 5150 is a Core 2-based dual core CPU - two of them only gets you to a Core 2-based quad-core level of performance.

  3. #3
    Certified AUGI Addict patricks's Avatar
    Join Date
    2004-06
    Location
    Memphis TN area
    Posts
    7,047
    Login to Give a bone
    0

    Default Re: Render time comparisons - very odd results!

    Well I performed the tests on all the Precision machines in the same way: close all programs besides Revit, open local copy of file, go to 3D view, close any other views, click the render dialog button, and hit Render.

    Perhaps I should do it again and focus more on the exact process, including rebooting the machine and launching Revit immediately after.

    I should note that the W3530 user has always reported that my old 5150 machine rendered much faster than hers. She has also complained from time to time of the computer generally being "slow" ever since the computer was brand new, despite coming up from a Precision laptop with XP x86, Core2 T2500 2.0 GHz, and 2GB RAM.
    Intern Architect, BIM Manager/Coordinator
    AERC, PLLC
    Hernando, Mississippi

    Revit - all up in your voxel space

  4. #4
    AUGI Addict iru69's Avatar
    Join Date
    2004-10
    Location
    usa
    Posts
    1,591
    Login to Give a bone
    0

    Default Re: Render time comparisons - very odd results!

    Sounds like something's wrong with the W3530 machine. The W3530 is essentially a re-badged i7-930... certainly no slouch.

    On the software side, there could be some configuration problem, anti-virus software issue (the cause of many problems), bad drivers, etc.

    On the hardware side, it could be old BIOS (first thing you should check!), overheating, misconfigured or bad RAM, etc. That's still a really nice computer spec - would be good to figure out what's going on.

  5. #5
    AUGI Addict
    Join Date
    2001-12
    Posts
    1,714
    Login to Give a bone
    0

    Default Re: Render time comparisons - very odd results!

    Quote Originally Posted by iru69 View Post
    Sounds like something's wrong with the W3530 machine. The W3530 is essentially a re-badged i7-930... certainly no slouch.

    On the software side, there could be some configuration problem, anti-virus software issue (the cause of many problems), bad drivers, etc.

    On the hardware side, it could be old BIOS (first thing you should check!), overheating, misconfigured or bad RAM, etc. That's still a really nice computer spec - would be good to figure out what's going on.
    I am thinking a good check would be to run a quick synthetic benchmark like GeekBench on all machines. If those numbers generally match up with the render tests, then likely the issue is in hardware or windows. And maybe Geekbench even sheds some light on what, perhaps with the one machine showing a crazy out of kilter RAM benchmark or something. If Geekbench ranks things more like hardware specs would lead one to expect, then perhaps the issue is with the Revit install or config on that machine.

    Gordon

  6. #6
    AUGI Addict iru69's Avatar
    Join Date
    2004-10
    Location
    usa
    Posts
    1,591
    Login to Give a bone
    0

    Default Re: Render time comparisons - very odd results!

    Sounds like a great idea. Have you used GeekBench? After a superficial glance at their website, it's not clear whether you need to buy it, and if so, do you need to buy it for each computer you run it on? ($20 a pop).

  7. #7
    AUGI Addict
    Join Date
    2001-12
    Posts
    1,714
    Login to Give a bone
    0

    Default Re: Render time comparisons - very odd results!

    Quote Originally Posted by iru69 View Post
    Sounds like a great idea. Have you used GeekBench? After a superficial glance at their website, it's not clear whether you need to buy it, and if so, do you need to buy it for each computer you run it on? ($20 a pop).
    Yeah, I currently use GeekBench. Mostly because it is one of the few cross platform/multi core/64 bit benchmark tools. Pretty sure the $20 gets you access to all the comparison data. If all you want is to compare to your own machines I don't believe there is any need to pay. At least that was the case when I downloaded it a year or more ago. I downloaded the windows version more recently, to compare old machines at work to a new test configuration. No problems at all with the free download.

    Gordon

  8. #8
    Certified AUGI Addict patricks's Avatar
    Join Date
    2004-06
    Location
    Memphis TN area
    Posts
    7,047
    Login to Give a bone
    0

    Default Re: Render time comparisons - very odd results!

    I have a 32-bit version of GB downloaded in 2006... guess I need to look for something new!

    That would be awesome if we could find something wrong on that machine and fix it to make it run better. Currently I don't think we're getting our money's worth out of it.

    There shouldn't be any AV issues. We run the Trend Micro Security Agent client on all machines in the office (recommended to us by our IT consultant) and I made sure to have no other AV software on any of the machines. I even removed the free 1-year McAfee that came with my new machine so that I could install Trend onto it.
    Intern Architect, BIM Manager/Coordinator
    AERC, PLLC
    Hernando, Mississippi

    Revit - all up in your voxel space

Similar Threads

  1. Results processing time - any hints?
    By bauskas in forum Robot Structural Analysis
    Replies: 0
    Last Post: 2011-02-23, 10:03 PM
  2. Render Time
    By kristina.bollin596153 in forum 3ds Max - General
    Replies: 7
    Last Post: 2008-06-13, 02:42 AM
  3. Render Benchmark Results
    By IanR in forum Revit - Hardware & Operating Systems
    Replies: 255
    Last Post: 2008-05-16, 11:24 PM
  4. Render Comparisons ( was Courtyard Comparisons )
    By SkiSouth in forum Revit - Rendering
    Replies: 30
    Last Post: 2007-02-04, 06:43 PM
  5. Render time
    By IanR in forum Revit - Rendering
    Replies: 2
    Last Post: 2004-11-24, 07:44 PM

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •