Multiplier vs. FSB... Hmm...
Home | Reviews and Features | Special Reports | Forums |

Results 1 to 9 of 9

Thread: Multiplier vs. FSB... Hmm...

  1. #1
    Join Date
    Feb 2000
    Location
    Deutschland
    Posts
    936

    Multiplier vs. FSB... Hmm...

    OK, so I've decided to play a bit with the FSB and multiplier on my old .25 micron Athlon 650, and see what happens. The CPU itself can go all the way up to 800 (not bad for a .25 micron), but with the stock heatsink it overheats soon, so the max I can use stably is around 750 MHz. So I've tried reaching that frequency in various ways.

    The mobo is an Abit KA7, memory is always at CAS2, set to 4-way interleave. (Though since I have single sided DIMM's, I highly suspect it's only 2-way in practice. I'm not sure, though.) L2 cache is at 1/2. Also bear in mind that the video card is an old G400, so most of the time this is fill-rate limited. (It is overclocked to 160MHz core, 180MHz memory, though, thanks to an old Socket 7 fan I had lying around )

    So let's see what happens in 3DMark2000, default benchmark.

    7.5x100, 133MHz RAM, Normal ... 3007
    6.5x115, 115MHz RAM, Turbo ... 3029
    6.0x124, 124MHz RAM, Turbo ... 3082

    Now the interesting part is that the 7.5x100 combination has the highest core frequency (albeit only by a few MHz), and the fastest memory setting, but it scores the lowest.

    And approximately the same thing happens in games, too. E.g., the cityintro in UnrealTournament gains a good couple of fps average, in 1024x768.

    I'm not entirely sure what to make of this. If memory bandwidth was the issue, then the scores should definitely be ordered differently. AGP overclocking shouldn't make that kind of a difference, or at least not on a card that's already hit the ceiling on fill rate anyway.

    Oh well... I guess it's just useful to know, or something

    ------------------
    Moraelin -- the proud member of the Idiots' Guild

    [This message has been edited by Moraelin (edited 06-03-2000).]
    Moraelin -- the proud member of the Idiots' Guild

  2. #2
    Join Date
    Feb 2000
    Location
    The origin of stupidity - Alabama
    Posts
    735
    The FSB is the speed in which the memory and other and other components operate. So, you aren't just O/Cing the CPU, you are speeding up the RAM to CPU connection, which is the biggest bottleneck in systems with high multipliers. Correct me if I'm wrong people.
    You people who think you know everything really piss people like me off who really do know everything.
    http://geocities.com/r337m0nk3y/net/smileysex5.gif

  3. #3
    Join Date
    Aug 1999
    Location
    Toronto,On,Canada
    Posts
    128
    I aggree. Its what decides when modified what speed the System and all of its devices is run.

    Moraelin I see your examples and I can assume that your 133 bus settings are at normal because you can't successfully boot with the ram in turbo? Wouln't that make your ram PC-166MHz? I don't know the math too well but its up there. The Happy medium here is as high a bus # with high ram settings. My system K7-650/core runs more efficient @8.5x105.5~892 than 9x100~ with ram 4:3. As long as your devices and ram behave and your cpu isin't baking, Keep On Truckin' with a Front Side Bus...


    ------------------

  4. #4
    Join Date
    Oct 1999
    Location
    Hamar, Norway
    Posts
    1,187
    To me, it all makes perfect sense
    You have two options when overclocking athlons:

    1. Increase the multiplier. Increases CPU speed, but nothing else.

    2. Increase the fsb. Increases overall system speed AND CPU speed.

    Just keep the PCI away from speeds over 40MHz and you'll be fine, with the most efficient system

    BTW, I read your posts over in the video forum regarding GF2 vs. V5 - good show I'm thinking of getting a V5 5500 (cause it's $365 while the GF2 is $500 over here, and your arguments definately won me over.

    ------------------
    Ich bin eine wienerschnitzel !

    [This message has been edited by D|sElMiNk (edited 06-03-2000).]

    [This message has been edited by D|sElMiNk (edited 06-03-2000).]
    Carpe noctem!

  5. #5
    Join Date
    Feb 2000
    Location
    Deutschland
    Posts
    936
    Yes, Teebee, that's correct. At 133 MHz I can't set the memory to Turbo. Still, it works at CAS2, so for the cheapest PC133 RAM I could get it's not so bad

    OverclockingAddict, please correct me if I'm wrong, but the whole point is that I'm actually underclocking the memory in the process. At 100 FSB, the memory is set to 133 MHz (FSB + PCI), while in the other two cases it's lower. If the factor was memory speed, then the combination with 133 MHz RAM should be the fastest, followed by 124, and 115 last. But what happens here is that it's the slowest. Hmm...

    ------------------
    Moraelin -- the proud member of the Idiots' Guild
    Moraelin -- the proud member of the Idiots' Guild

  6. #6
    Join Date
    Apr 2000
    Location
    Toronto, Ontario, Canada
    Posts
    141
    hey, could someone explain the concept of turbo to me? i somehow doubt it's the same as the turbo that took your 166 down to 66 so that those old dos games didn't scroll by in a millisecond etc... i didn't have the option for turbo in my old bh6, and i do now in my bp6, so i was just wondering what it did, thanks!
    specs:
    Dual Celly 533
    Abit BP6
    512MB RAM
    Asus v6800 DDR Pure 64MB
    2 * 40gig maxtors in RAID
    2 * 15gig maxtors in RAID
    1 * 20gig maxtor

  7. #7
    Join Date
    Oct 1999
    Location
    Hamar, Norway
    Posts
    1,187
    The turbo on the BP6 is just a 2.5% increase in bus speed - so 100mhz -> 103mhz
    Carpe noctem!

  8. #8
    Join Date
    Nov 1999
    Location
    Markham, Ontario, Canada
    Posts
    3,022
    turbo was used to test for system stability. But I think turbo on the Ka7 might mean something else. From what I can gather, in Morealin's case, it seems that having the fastest FSB or RAM settings with Turbo ON yields the fastest performances.

  9. #9
    Join Date
    Feb 2000
    Location
    Deutschland
    Posts
    936
    Basically the settings for memory are: 10ns, 8ns, normal, medium, fast and turbo. I don't know exactly what they mean. Since the actual memory clock is controlled from somewhere else, those 10ns and 8ns settings are at best misleading. It's not the CAS timing, either, since that's a separate menu entry. The latch delay is also somewhere else, and likewise for the memory interleave. I suspect it has something to do with RAS and the other signals. Like, instead of getting to control each of the timings separately, you have 6 pre-defined configurations to choose from. (But then why give the configurations such cryptic names?)

    Either way, what happens here is that 133 MHz, the memory works ok with "normal" and CAS2. "Fast" and "turbo" are unstable. "Medium"... come to think of it I didn't even try. But if I underclock it to 125MHz, I can get CAS2 and "turbo".

    I don't think it's the "turbo" that's causing the speed increase, though. Methinks the FSB overclock is what's making it fast. I mean, ok, I've set the memory down to "Normal" again, just to see what happens:

    6.0x124, 124MHz RAM, Normal ... 3064

    If the "Turbo" was the cause, the score should have dropped below the 3007 for 133 MHz RAM and "Normal". It didn't.

    ------------------
    Moraelin -- the proud member of the Idiots' Guild

    [This message has been edited by Moraelin (edited 06-03-2000).]
    Moraelin -- the proud member of the Idiots' Guild

Thread Information

Users Browsing this Thread

There are currently 1 users browsing this thread. (0 members and 1 guests)

Posting Permissions

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