Results 1 to 14 of 14

Thread: Lightwave and P4 Hyper Threading

  1. #1

    Lightwave and P4 Hyper Threading

    Does Lightwave 8.0 take advantage of HT Pentium 4 Processors. Thanks

  2. #2
    obfuscated SDK hacker Lightwolf's Avatar
    Join Date
    Feb 2003
    Location
    Stuttgart, Germany
    Posts
    13,711
    Yes, during rendering, if you set the renderer to use more than one thread (in the render panel). You will probably get an extra 5%-10% of performance.

    Btw, this even works if you don't boldface your question

    Cheers,
    Mike
    Last edited by Lightwolf; 10-20-2004 at 09:16 AM.

  3. #3
    CORE 5718 mattclary's Avatar
    Join Date
    Feb 2003
    Location
    Tallahassee, Florida
    Posts
    5,228
    Any multi-threaded app will utilize hyperthreading. All it is, is a virtual second CPU.

    Yeah, the boldface comes across as kind of rude.

  4. #4
    Amused
    Join Date
    Mar 2003
    Location
    Keeping Austin weird
    Posts
    1,322
    -Hyper- threading works best when using Screamernet to render. I have dual Xeons so Screamernet sees and uses four render nodes. The render time -per node- is a little slower than a straight out render (not using Screamernet) but with four nodes going it ends up being a lot faster overall. My scenes are relatively small, and I've discovered that setting -multi- threading to only one works best. Experiment with the settings to get the best performance for your system/scenes.

    TW

  5. #5
    Sorry I didn't know

  6. #6
    So I would only see 5-10% faster rendering performance if I went with HT Pentium 4 over an Athlon 64 Skt 939. I'll probably just go with a Athlon 64 3500+.

  7. #7
    obfuscated SDK hacker Lightwolf's Avatar
    Join Date
    Feb 2003
    Location
    Stuttgart, Germany
    Posts
    13,711
    Hi mlmiller,
    no probs, no offense taken, I was just pointing it out

    What I meant is, you will get a 5%-10% performance increase on the _same_ processor if you utilize HT.
    Comparing against another processor (especially one with a different architecture like an AMD) has no impact on that, and isn't easy to do, because processors behave differently on different scenes due to different optimizations.
    Have a look at the benchmarks at www.blanos.com to check out how a couple of standard scenes behave with different CPUs.

    Generally, with less optimized code, Athlons run faster in rendering, while P4s are faster with heavily P4 optimized code.
    If you look at LW, you will see that P4s are faster when rendering global illumination, while Athlons tend to shine rendering volumetrics.
    It also seems that the lastest patch has increased performance on Athlons a bit.
    If you get an Athlon, get one of the 64 range, because they have the SSE2 instruction set, which can speed up LW a bit as well.

    You also can't really compare clockspeeds. I tend to compare the prices of the complete system to find the best price/performance ratio.

    Cheers,
    Mike

  8. #8
    I am probably just going to stick with an Athon 64 3500+, 1GB Corsair XMS PC3200(Dual Mode), 256MB Nvidia GeForce 6800 GT,etc. I believe this setup is good for a Lightwave beginner/Gamer. Thanks for the info.

    Cheers

  9. #9
    Newbie Member
    Join Date
    Apr 2004
    Location
    Ontario, Canada
    Posts
    21

    p4 vs amd in my tests

    just thought i'd add my observation to this discussion.
    I have two systems one is:
    P4 HT 2.6 GHz with 1Gig DDR
    the other:
    AMD 3700+ with 1 Gig DDR Ram
    both systems are configured roughly the same ie: XP SP2 etc.

    in marketing theory a 3700+ AMD is equivelant to a 3.7Ghz P4...
    thus should run 1.42 times faster than my P4.

    however actual render times the AMD is about 1.96 times faster standalone and about 2.2 times faster via screamernet (i dont know why)
    considering bang for busk, I recommend AMD.

    maybe its the SSE2 and 3 extensions i dont know...
    but thats just my observation.
    regards,
    Daniel Young
    Daniel Young
    CrusaderX
    [email protected]

  10. #10
    Quote Originally Posted by crusaderx
    just thought i'd add my observation to this discussion.
    I have two systems one is:
    P4 HT 2.6 GHz with 1Gig DDR
    the other:
    AMD 3700+ with 1 Gig DDR Ram
    both systems are configured roughly the same ie: XP SP2 etc.

    in marketing theory a 3700+ AMD is equivelant to a 3.7Ghz P4...
    thus should run 1.42 times faster than my P4.

    however actual render times the AMD is about 1.96 times faster standalone and about 2.2 times faster via screamernet (i dont know why)
    considering bang for busk, I recommend AMD.

    maybe its the SSE2 and 3 extensions i dont know...
    but thats just my observation.
    regards,
    Daniel Young
    I'm have debating whether to go with AMD or Intel. The benchmarks I've seen haven't shown their is that much of an increase in rendering with Hyper threading and AMD Athlon 64 are alot better with games. Since I am novice and don't need professional level rendering I'll probably just stick with AMD 63 Skt939. Just gotta wait for Nvidia NForce 4 chipset to come out. According to an email I got from Nvidia it should be out by the end of this month.

  11. #11
    Registered User
    Join Date
    Mar 2003
    Location
    Los Angeles
    Posts
    183
    The 'heavy P4 optimizations' that make radiosity render faster in LW are mostly SSE2 optimizations. An Athlon64 nullifies most of that. For LW use, an Athlon64 is pretty much the best of both worlds betwwen an Athlon XP's floating point power and a P4's SSE2 power. Just make sure you get the socket 939 Athlon64 so that you get dual-channel DDR memory...

  12. #12
    Registered User
    Join Date
    Mar 2003
    Location
    Los Angeles
    Posts
    183
    Also, as far as render nodes go, the socket 754 Semprons are a good deal. Though you lose the future 64-bit capability, they ar cheap and have SSE2... the skt 754 Semprons are essentially skt 754 Athlon64s with the 64-bit section disabled.

  13. #13
    Registered User
    Join Date
    Mar 2003
    Location
    Los Angeles
    Posts
    183
    Hi there, me again...

    my experience shows that HT can either be a blessing or a curse with LW rendering... it varies from rendering to rendering. Some renders, you'll gain 5-10%, but others, you can lose that same margin. My perception is it depends how cache-heavy your scene is... if youe scene causes alot of cache hits (large image maps do this, AFAIK) then BOTH threads will need the data in the cache and be hitting it alot... there's only so much cache to store the data these processes need. Maybe that's why the P4s (Xeon and EE) with larger cache are less susceptible to the HT performance hit.

    Ask me later when i can check from home and i can give you examples of scenes on the LW content CD that perform worse with HT.

  14. #14
    I just recently purchased a 3.4C Pentium 4. I've decided to stick with the Pentium 4 cause my Matrox RT.X100 Xtreme Pro works best with Pentium 4 CPUs.

Bookmarks

Posting Permissions

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