Poor Benchmarks - Help me work out why

Archived from groups: alt.comp.hardware.overclocking (More info?)

Hello

I've been trying to work out why my system gives me low benchmarks

on my FX 5600 256ram on 3DMark2001SE I get a score of 5074
on my 6600 GT 128ram on 3DMark2005 the most i've got is 2892

Below are my specs - but I have one question, my Intel P4 3.2gigHT is the
mobile version, could that make a big diffrence to a same spec standard P4?
I've done all things like defrag, spyware, Anti-virus etc etc and even done
a clean install of WinXP and tried lots of diffrent drives (offical, beta
and over-clocked)

Heres my specs

500 Watt PSU
P4 3.2gig HT
6600 GT running 66.93 driver
Maxtor 160gig DiamondMax Plus 9 IDE Hard-Drive (system drive)
Matxor 180gig IDE Hard Drive (storage)
WinXP with SP2 and all current updates
Pionner 106 DVD
Liton DVD
C-Media 5.1 onboard sound
Very sexy case with 7 fans http://www.casetek.com.tw/midi/ck-1020-2a.htm
CPU temp 83f - Graphic card temp 94f
512/133Mhz DDR266
256/133Mhz DDR266
ARock P4v88 http://www.asrock.com/product/product_p4v88.htm

Also remember, this is a new build system any bios setting maybe a help.....

Prim Graphics Adapter = AGP
AGP Aperture Size = 128
AGP Mode 8x
AGP Fast Write - Off
AGP Drive Strenght - Auto
AGP Gadstb Output Delay - Auto (any ideas wot this should be set too?)
8 answers Last reply
More about poor benchmarks work
  1. Archived from groups: alt.comp.hardware.overclocking (More info?)

    "Aerodeck" <tedblight@yahoo.com> skrev i en meddelelse
    news:H_2dncGqBdA0kY_fSa8jmA@karoo.co.uk...
    > Hello
    >
    > I've been trying to work out why my system gives me low benchmarks
    >
    > on my FX 5600 256ram on 3DMark2001SE I get a score of 5074
    > on my 6600 GT 128ram on 3DMark2005 the most i've got is 2892

    Pls. post your futuremark compare links.
  2. Archived from groups: alt.comp.hardware.overclocking (More info?)

    Sorry, I'm new to Benchmarking and Overclocking -
    Here's the link
    http://service.futuremark.com/servlet/Index?pageid=/orb/projectdetails&projectType=6&projectId=8430113


    > Pls. post your futuremark compare links.
    >
    >
    >
  3. Archived from groups: alt.comp.hardware.overclocking (More info?)

    "Aerodeck" <tedblight@yahoo.com> wrote in message
    news:0sucnc1_3ZjThY_fSa8jmw@karoo.co.uk...
    > Sorry, I'm new to Benchmarking and Overclocking -
    > Here's the link
    > http://service.futuremark.com/servlet/Index?pageid=/orb/projectdetails&projectType=6&projectId=8430113

    That's not the right link.You need to publish the project(s), if you haven't
    done already. Then under each one in Project Manager you'll see the Compare
    URL's.

    > I've been trying to work out why my system gives me low benchmarks

    Have you checked that the drivers are configured optimally for the new card?
    VSync disabled, filtering and quality set for speed etc.?

    > on my FX 5600 256ram on 3DMark2001SE I get a score of 5074

    That seems bloody low.

    > on my 6600 GT 128ram on 3DMark2005 the most i've got is 2892

    No real experience with 3DMark 2005, as it's largely useless as a
    diagnostic/testing tool. I take it you've compared this score with other
    systems and know that yours is deficient?

    > Below are my specs - but I have one question, my Intel P4 3.2gigHT
    > is the mobile version,

    Why are you using a mobile processor? These are a bad choice for a desktop
    system as you lose the 800MHz system bus, and henceforth a lot of
    performance, while there can also be other issues. Is it a 512K or 1MB L2
    cache part, and are you certain that it is working at its full rated speed?
    IIRC using a mobile P4 in a desktop motherboard can produce odd results, in
    particular the processor running a lot slower than it should due to the
    board not knowing how to manage the SpeedStep system.

    > could that make a big diffrence to a same spec standard P4?

    It could in certain scenarios, as explained above. However, we can't guess
    this stuff for you, you need to find out for yourself as you're the only one
    with the access to your hardware.

    > I've done all things like defrag, spyware, Anti-virus etc etc and even
    > done Very sexy case with 7 fans
    > http://www.casetek.com.tw/midi/ck-1020-2a.htm

    Whoa. Computer cases are only sexy if you have no life, or issues
    interacting with real people. Take this as a warning sign! ;)

    > CPU temp 83f - Graphic card temp 94f

    Is this at idle, under full and sustained load, what? Have you also checked
    to see whether the processor is throttling at all when you load it up?

    > 512/133Mhz DDR266
    > 256/133Mhz DDR266

    Again this memory isn't the perfect choice for this system. Without knowing
    how the VIA chipset handles dual channel memory, and without knowing how
    you've arranged the memory in the slots it's impossible to know whether
    you're getting optimal memory performance. More work and experimentation for
    you there.

    > Prim Graphics Adapter = AGP
    > AGP Aperture Size = 128
    > AGP Mode 8x
    > AGP Fast Write - Off

    They're as they should be. The Fast Write setting doesn't make any
    difference when the AGP interface is running in 8x mode.

    > AGP Drive Strenght - Auto
    > AGP Gadstb Output Delay - Auto (any ideas wot this should be set too?)

    No idea regarding the latter. Auto's probably the best bet, although there's
    no harm in trying alternative settings and seeing what, if any, difference
    it makes to system stability and performance.
    --


    Richard Hopkins
    Cardiff, Wales, United Kingdom
    (replace nospam with pipex in reply address)

    The UK's leading technology reseller www.dabs.com
  4. Archived from groups: alt.comp.hardware.overclocking (More info?)

    "Aerodeck" <tedblight@yahoo.com> skrev i en meddelelse
    news:0sucnc1_3ZjThY_fSa8jmw@karoo.co.uk...
    > Sorry, I'm new to Benchmarking and Overclocking -
    > Here's the link
    >
    http://service.futuremark.com/servlet/Index?pageid=/orb/projectdetails&projectType=6&projectId=8430113


    Go into your project mangager. Every project should have a compare link.
    Looks like this ---->Compare URL:
    http://service.futuremark.com/compare?3dm05=1234567
  5. Archived from groups: alt.comp.hardware.overclocking (More info?)

    Hey thanks for all that, its given me a lot to think about, I think I'll
    swap my mobile P4 for a normal P4 with the 800fsb

    I've been messing around with diffrent RAM and Graphics cards for the last
    few hours.

    With 3DMark03

    The best result I could get was 6506 on my 6600GT Card with VStudioExtremeG
    71.20 Driver - 2x 256RAM DDR 133 Duel Channel Memory - Temp goes up about 8c
    when benchmarking

    This fell too 5359 with 2x 512 DDR 400 - strange uh?

    More tweaking to come.... Your comment are welcome.... Cheers
  6. Archived from groups: alt.comp.hardware.overclocking (More info?)

    "Aerodeck" <tedblight@yahoo.com> wrote in message...
    > Hey thanks for all that, its given me a lot to think about, I think
    > I'll swap my mobile P4 for a normal P4 with the 800fsb

    Yeah, that's a good move. Eliminates a load of potential trouble and will
    give you a deal more performance. If you can find one, you'd be better off
    with a Northwood 512K cache CPU than the equivalent "Prescott" 1MB cache
    part.

    > I've been messing around with diffrent RAM and Graphics cards for
    > the last few hours.
    >
    > With 3DMark03
    >
    > The best result I could get was 6506 on my 6600GT Card with
    > VStudioExtremeG 71.20 Driver - 2x 256RAM DDR 133 Duel Channel Memory -
    > Temp goes up\
    > about 8c when benchmarking

    8 celsius difference between idle and full load? That sounds a little low.
    It's largely academic if you're going to swap the processor for a desktop
    part, but I'd lay reasonable odds that the processor is either clock
    throttling under load, or isn't running at full speed in the first place.

    >
    > This fell too 5359 with 2x 512 DDR 400 - strange uh?

    Maybe, maybe not. You've left us assuming far too much there. If these
    results are consistent, and fell when you swapped the memory, the only
    explanation is that the system obviously wasn't configured properly for the
    DDR400 (or, putting it another way, the system was configured better for the
    DDR266 than it was for the "quicker" memory).

    Either you didn't use the correct memory slots (and hence turned off the
    dual channel), didn't have the correct memory bus divider, or timings set.
    If you left the motherboard to sort all this out, its a given that its
    auto-setup functions don't work as well as they could.
    --


    Richard Hopkins
    Cardiff, Wales, United Kingdom
    (replace nospam with pipex in reply address)

    The UK's leading technology reseller www.dabs.com
  7. Archived from groups: alt.comp.hardware.overclocking (More info?)

    The biggest score increases to be found by bios tweaking can be found
    by tweaking CAS Timings. Usually found in bios under Advanced Chipset
    Features.

    Changing CAS Rate from 3 to 2 can significantly improve 3DMark scores,
    especially in 2001SE which is based on the whole CPU, Memory Subsystem
    and GPU.

    2003 benches mainly GPU as does 2005.

    The lower the cas timings the better is the general rule of thumb,
    however there are specific rules and formulae which should be adhered
    to to maintain data integrity.

    See here for an indepth explanation of Conformal Cas Timings (no
    registration req'd):
    http://www.over-clock.com/ivb/index.php?showtopic=438&view=findpost&p=42165

    From there it's down to driver level tweaking of the GPU... google for
    rivatuner and 3dmark tweaking guides for rivatuner.

    From there it's overclocking time - increase speeds of GPU Core and
    Memory, increase FSB of CPU. All details covered in above thread,
    altho for Intel ignore multiplier related points.
  8. Archived from groups: alt.comp.hardware.overclocking (More info?)

    ANd you can't compare the score of one card in 3DMark 2001 vs the
    score of another in 2003 etc.

    Run a bench on the 5600 in 2001SE and compare to a bench on the 6x00
    series card in 2001SE.

    The score weighting for each years' benching software is completely
    different.
Ask a new question

Read More

Overclocking