Comparing Shuttle AN35N-Ultra & AN35N-400

Discussion in 'Shuttle' started by yuki, Dec 23, 2003.

  1. yuki

    yuki Guest

    Is the difference between the boards strictly the dual vs single channel
    memory controller?

    How does that actually translate to performance?
    I read somewhere it is 1-2%.

    Is the AN35N-Ultra worth $15 more than AN35N-400 ($70 vs $55)?

    Thanks,

    yuki
     
    yuki, Dec 23, 2003
    #1
    1. Advertisements

  2. yuki

    Wes Newell Guest

    Actually the single channel nf2 400 chipset is faster in most test. Only
    in apps that require huge bandwidth does the dual channel really help
    out. There are comparisons on the web. You might want to check them.
    I guess it is if you want dual channel. I think that's the only difference
    between the 2.
     
    Wes Newell, Dec 23, 2003
    #2
    1. Advertisements

  3. yuki

    Frank Hagan Guest

    For a difference of $25 get the 400. I thought the price was closer than
    that :-( When I bought my AN35N Ultra's they were $62 delivered.

    FRH
     
    Frank Hagan, Dec 23, 2003
    #3
  4. yuki

    Frank Hagan Guest

    Too darn early in AM here! Can't do simple math :-( Even $15 is a big
    spread for the somewhat problematic difference in speed. Get the 400!

    FRH
     
    Frank Hagan, Dec 23, 2003
    #4
  5. yuki

    yuki Guest

    Turned out the saving was more than $25.
    I just picked up an AN35N-400 for $40 at a local retailer (Fry's).

    I am happy that I managed to come under my budget of $200 for a MB & 512MBx2
    combo.

    Thanks for playing along.
    Merry Xmas,

    yuki
     
    yuki, Dec 23, 2003
    #5
  6. Unfortunately, this does not end in a happy ending yet. So the thread
    continues.

    The good news was the RAM (Kingmax PC3500 512x2) from Newegg arrived
    yesterday (24th) instead of Fri. I suppose the extra $3 for quick
    processing was worth the money.

    The bad news is the system with the new Shuttle AN35N-400 MB is not
    stable at 200 FSB when I put both sticks in (tried both 1-2 & 1-3
    combinations)

    With a single stick, it will do 202x11.5 fine (did not try going
    higher since my target is 200x11), but with both sticks in, memtest86
    will fail on test #5 & #7 even at 200x8.

    So I have 4 choices:
    1. RMA the RAM, get Corsair or Geil Value PC3200.
    2. Return the MB, get AN35N ultra 400
    3. Do both 1 & 2
    4. Return both MB & RAM, go back to my old Asus A7V133 running XP2100+
    at 100x20

    Suggestions, opinions?

    yuki
     
    Yuki Takizawa, Dec 26, 2003
    #6
  7. yuki

    Frank Hagan Guest

    Did you Memtest each stick individually? If they both will pass at 202x11.5
    try "Optimal" on memory and go for 210x10.5. That will give you 2200MHz with
    a higher FSB for memory bandwidth. Of course work your way up to 210. You
    may have to give the PC3500 2.6 or 2.7 volts and you may have to up the
    voltage on the XP2100. This is a trial and error process. It usually takes
    time.

    Of course you may have a problem with the memory. I'm not familiar with
    Kingmax. I just run single sticks of PC3200 (512MB). I set CPU to
    "Aggressive" and memory to"Optimal" at 100%.

    I've got 2 sticks of PC3500 Kingston HyperX on order to test out dual
    channel. Probably a waste, but I have to try ;-)

    Good luck. Let us know how it works out.

    FRH
     
    Frank Hagan, Dec 26, 2003
    #7
  8. yuki

    Kristi Guest

    Frank, I forgot what brand of memory you have been using. Also
    wanted to say that if you were going to throw that old xp2100
    out, throw it in my direction!!! heeheehee
    Kristi (the original one)(still on xp1700 333 11
     
    Kristi, Dec 26, 2003
    #8
  9. Yes, I tested the stick individually. Each worked fine at 200x11.5.
    This is one of the reasons I am willing to swap the MB for ultra 400.
    With both sticks in, I can't even get a clean test run with Memtest86
    at 133x10. There seems to be a known problem with AMD & Memtest86 #5
    test, but I was able to run clean tests with the same CPU on an old
    Asus A7V133 MB.

    It may not encounter any errors in the actual usage, but I figured I
    should be able to build a value system without paying $100 per 512MB
    of RAM.
     
    Yuki Takizawa, Dec 26, 2003
    #9
  10. yuki

    Frank Hagan Guest

    That's strange! Since you don't have dual channel capability it shouldn't
    matter much that the sticks aren't "hand picked " for dual channel. They
    test well separately, but dont't play well together :-(
    I agree! Both 512MB sticks of PC3200 "mystery memory" that I'm using cost
    less than $65 each. I tried them once in the same machine as dual channel
    when I was using an XP2100, but they were too different to go much over
    200MHz. The AIDA32 memory bandwidth results did go higher, but not enough to
    justify using two mismatched sticks together as dual channel. I've got two
    512MB sticks of PC3500 in transit. I will test dual channel and then non
    dual when they arrive.

    I hesitate to give you advice because it may be the memory, the board, the
    lack of dual channel (though I doubt it), or just an unfortunate
    combination. The easiest thing would be to swap the memory. I'm not saying
    that's the problem for sure, but it would be easiest and the results would
    tell you more than you know now.

    Again good luck.

    FRH

    P.S. I forgot Merry Christmas!
     
    Frank Hagan, Dec 26, 2003
    #10
  11. yuki

    Frank Hagan Guest

    Ooops! My bad! Sold on eBay after I got Barton XP2500 and T'bred XP2600.
    Both unlocked (bought 'em on eBay :) Have no clue what my memory is, but
    one stick runs FSB210x10.5 on the Barton and the other (chips mfg. by
    Legacy) runs 204x10.5 on the XP2600. Prime95 stable overnight (8 hrs.
    minimum)

    I've got some PC3500 in transit, tied up in the Xmas rush :-( Will try dual
    channel feature on the Barton and depending on results may have 1GB dual on
    the Barton and 1GB single on the XP2600.

    Sorry.

    FRH
     
    Frank Hagan, Dec 26, 2003
    #11
  12. yuki

    Kristi Guest

    LOL not a prob! You refer to the Barton 2500 and the 2600 as
    being unlocked... I thought they came from the factory that
    way... yes?
    Thanks.
    Kristi
     
    Kristi, Dec 26, 2003
    #12
  13. yuki

    Frank Hagan Guest

    From what I understand all Barton's manufactured after week 39 are
    multiplier locked. I went the eBay route to get a week 31 Barton. Rumor has
    it that the T'bred new production is being locked also so I picked up a nice
    2600 for my backup system.

    The PC3500 (Kingston HyperX) arrived and I'm running tests in dual channel
    mode. So far I can boot and run XP at FSB220x10.5,but it won't pass Prime95.
    So far it's FSB214x10.5 solid. I'm testing 214x10.5 as I write this ;-) When
    I max out in dual channel I'll see what I can do in single channel.

    I tried my "mystery memory" in the XP2600 and found one stick dragging the
    other down. So I'm using the "good" stick by itself and have FSB210x10.5
    Prime95 stable. Been running all afternoon. I'll declare victory tomorrow
    morning.

    FRH
     
    Frank Hagan, Dec 27, 2003
    #13
  14. So I took a new tactic, flashed the bios with the latest version
    AN3S500L (the original was dated from May). All of the sudden, the
    system is much more stable wrt Memtest.

    So I am back to FSB around 200 (mem freq @ 100%) again.
    I think I can live with this now:)

    happy holidays,

    yuki
     
    Yuki Takizawa, Dec 27, 2003
    #14
  15. yuki

    Frank Hagan Guest

    Very interesting. Thanks for the update.

    FRH
     
    Frank Hagan, Dec 27, 2003
    #15
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.