Page 2 of 6 FirstFirst 123456 LastLast
Results 16 to 30 of 84
  1. #16
    Joined
    Nov 2002
    Location
    Ohio , USA
    Age
    43
    Posts
    602
    Thus a Tbed 1800 and a Tbred 2600 both at 133 should produce the same RAM bandwidth.
    The faster the CPU is . the faster the memory will run...

    if your runnin CPU with multi. of 11.0 @ 166
    and then you run the same processor with a multi of 12.0 @ 166 , the 12.0 will run the memory much faster even tho you didnt changed the mem or FSB frequency...
    an 1800+ and a 2600+ ,that use the same memory,will not produce the same results... 2600+ will run the mem faster

    in other words.. increasing your CPU speed alone , WILL increase mem speeds , without actually increasing the timings or the frequency of the Memory
    **New Build**
    AMD Phenom 9850 BE Quad Pumper /w Zalman CNPS9700 HSF
    MSI K9A2 Platinum Motherboard
    2x2GB = 4GB PC6400 OCZ REAPER memory
    500GB SATA WD Caviar SE16 7200RPM HD
    Pioneer 20x20 SATA CD/DVD ROM
    19" Viewsonic LCD display
    ATI/MSI Radeon HD 2x 4850 512MB DDR3 in Crossfirex
    Soundblaster X-FI Xtremegamer
    Antec true power trio 650w Powersupply
    Antec 1200 Case
    WinXP Pro

  2. #17
    Joined
    Aug 2000
    Location
    Vvardenfell
    Age
    58
    Posts
    10,886

    Cool

    XP2700+ on Shuttle AK37GTR (KT400):

    197 x 8.5 = 2783/2473
    197 x 9.5 = 2828/2563
    197 x 10.5 = 2924/2659
    197 x 11.5 = 2880/2672
    197 x 12.5 = 2789/2636


    OK, I have no idea what is going on there, beyond the fact that this setup appears to have a sweet spot at about x10.5! But certainly there is no "faster CPU = faster Sandra mem score".

    M

  3. #18
    Joined
    May 2002
    Posts
    75
    SiSoft just sucks as a mem bench.Totally unreliable.CPU speed does influence the score.

  4. #19
    Joined
    May 2002
    Posts
    75
    Originally posted by HauZDauG
    The faster the CPU is . the faster the memory will run...

    if your runnin CPU with multi. of 11.0 @ 166
    and then you run the same processor with a multi of 12.0 @ 166 , the 12.0 will run the memory much faster even tho you didnt changed the mem or FSB frequency...
    an 1800+ and a 2600+ ,that use the same memory,will not produce the same results... 2600+ will run the mem faster

    in other words.. increasing your CPU speed alone , WILL increase mem speeds , without actually increasing the timings or the frequency of the Memory
    Only according to SiSoft.

  5. #20
    Joined
    Sep 2002
    Location
    London
    Posts
    44
    o.k. how the hell did you do that with an 8RDA+

    I have a xp 2400+ and corsair 3200, set at SPD (i.e. mem @ 400mhz) fsb at 133. if i set the timings to 6,2,2,2 the memory comes unstable. 7,3,3,2 is the best it'll do there and i don't get it.

    i can up the fsb and take the spd to manual and BEEBAWBEEBAW, no POST.

    clear cmos start again

    dimmit!
    Nick

    Windows ME
    Asus ADSL PCI modem
    IBM & Maxtor Hard drives
    Epox 8RDA+, T'bred 2400+ (12*182)
    512 M Corsair 3200 (1 stick so far)
    Radeon 9700 (317,304)
    430W Antec Truepower
    Creative Inspire 6.1 speakers
    A case with an UV lamp and lots of diodes that will light as and when...

  6. #21
    Joined
    Dec 2002
    Posts
    1,178
    Originally posted by nbhamilt
    o.k. how the hell did you do that with an 8RDA+

    I have a xp 2400+ and corsair 3200, set at SPD (i.e. mem @ 400mhz) fsb at 133. if i set the timings to 6,2,2,2 the memory comes unstable. 7,3,3,2 is the best it'll do there and i don't get it.

    i can up the fsb and take the spd to manual and BEEBAWBEEBAW, no POST.

    clear cmos start again

    dimmit!
    Set sys perf to EXPERT, memory to 100%. Hit F10.

    problem resolved. Don't overclock if you're clueless.

  7. #22
    Joined
    Sep 2002
    Location
    London
    Posts
    44
    I actuqally meant how did you get the fsb to run that high with that chip. NOt how you fiddle with the bios! I currently have the fsb at 146 (mem @100%) and stable on 15x multiplier have tried out numerous memory configurations and thats the best 3dmark score i can get. bizarrely enough, i can no get the mem to go to 5,2,2,2, but the score is (just a touch) lower than at 7,2,2,2 - all other things being equal.

    mind you, tere is alot more to play with in this bios than my old ASUS a7V - i don't know much about the AGP aperture and speed setting, so piut it at 128M and Auto.
    Nick

    Windows ME
    Asus ADSL PCI modem
    IBM & Maxtor Hard drives
    Epox 8RDA+, T'bred 2400+ (12*182)
    512 M Corsair 3200 (1 stick so far)
    Radeon 9700 (317,304)
    430W Antec Truepower
    Creative Inspire 6.1 speakers
    A case with an UV lamp and lots of diodes that will light as and when...

  8. #23
    Joined
    Aug 2000
    Location
    Vvardenfell
    Age
    58
    Posts
    10,886

    Cool

    OK, more KT400 figures.

    First of all, the Shuttle AK37GTR again, with new BIOS (00S) and a new install:

    150 = 2275/2127
    160 = 2341/2193
    170 = 2500/2339
    180 = 2643/2470
    190 = 2798/2610
    200 = 2841/2748

    - a slight improvement.


    Now an Epox 8K9A2+:

    150 = 2310/2188
    160 = 2451/2315
    170 = 2597/2437
    180 = 2743/2558
    190 = 2895/2699
    200 = 3051/2839
    210 = 3201/2989


    But I've noticed that neither board is happy as high bus speeds, and anything over 185 is very twitchy. All the gear worked just fine at 200 in my 8K3A+, so I have to assume it's a chipset thing, or else I've damaged something.


    M

  9. #24
    Joined
    Feb 2003
    Location
    Raleigh, NC
    Posts
    85
    Originally posted by HauZDauG
    The faster the CPU is . the faster the memory will run...

    if your runnin CPU with multi. of 11.0 @ 166
    and then you run the same processor with a multi of 12.0 @ 166 , the 12.0 will run the memory much faster even tho you didnt changed the mem or FSB frequency...
    an 1800+ and a 2600+ ,that use the same memory,will not produce the same results... 2600+ will run the mem faster

    in other words.. increasing your CPU speed alone , WILL increase mem speeds , without actually increasing the timings or the frequency of the Memory
    Absolutely Right. We have seen the numbers here, and I have done alot of testing myself, thought not for THIS purpose exactly, but have been tracking the results, and increasing the CPU speed alone WILL in fact net higher mem speeds.

    Though I question the accuracy or reliability of Sandra.

    But here's one for you to ponder...

    You can see in my sig what my system specs are. I was originally running my memory at 8-4-4-2.0 according to the old kingston hyperX tech standards. But they revised them and I have been running my memory at the new recommended specs of 7-3-3-2.0. Recently, I tried it at 3-3-2-2.0 as advised by k1114's site, and retested. I found virtually NO improvement in Sandra scores whatsoever between the three settings.

    EVERYTHING being equal as per my sig, EXCEPT the change in memory timings, here are the results:

    @ 8-4-4-2.0
    Si Sandra Test 1 = 3156 / 2987
    Test 2 = 3160 / 2975

    @ 7-3-3-2.0
    Si Sandra Test 1 = 3157 / 2975
    Test 2 = 3156 / 2966
    Test 3 = 3151 / 2981

    @ 3-3-2-2.0
    Si Sandra Test 1 = 3157 / 2981
    Test 2 = 3150 / 2986

    Now, bear in mind that if you test and then re-test under the SAME timings, you will get variations (as seen above), the scores rarely come up with the exact same numbers, but they are often very close. Try it and you will see, as I am sure you have already, that there are variations, but they are slight.

    But, I would have assumed that with the very different timings above that there would have been a somewhat noticeable improvement, but there is not. In fact, the two highest numbers are with the slowest settings. These are just the ones I had written down, but you get the picture.

    Conclusion (at least within my system), NO effect due to ram timing changes alone. Same went for the Multimedia tests, same type of slight variations, but NO discernable differences due to the timings.

    Any educated guesses, or similar experiences?

    PS: Drew, my best scores were at 11.5 @ 205 like you are running, but would NOT run stable under P95. By using an 11.0x I am stable to 209/210. Anything over that, and it wont pass P95. Where I am at now can chew on P95 for hours on end, with NO errors, even with running other APPs during. I guess 11x210 is my peak.
    Last edited by jconforti; 03-12-2003 at 10:45 AM.

  10. #25
    Joined
    Aug 2000
    Location
    Vvardenfell
    Age
    58
    Posts
    10,886

    Cool

    But do the RAM settings make any different to other benchmarks? The problem may not be Sandra, just that RAM settings don't make much difference. My experience of using 3DMark2001SE to test this was that RAM settings really do NOT make a lot of difference - maybe 100 points out of 12k.

    M

  11. #26
    Joined
    Sep 2002
    Location
    London
    Posts
    44
    cas latency definitely changes 3dmark, other timings, pretty much nothing.
    Nick

    Windows ME
    Asus ADSL PCI modem
    IBM & Maxtor Hard drives
    Epox 8RDA+, T'bred 2400+ (12*182)
    512 M Corsair 3200 (1 stick so far)
    Radeon 9700 (317,304)
    430W Antec Truepower
    Creative Inspire 6.1 speakers
    A case with an UV lamp and lots of diodes that will light as and when...

  12. #27
    Joined
    Jan 2003
    Location
    Langley, Virginia
    Posts
    109
    Originally posted by nbhamilt
    I actuqally meant how did you get the fsb to run that high with that chip. NOt how you fiddle with the bios! I currently have the fsb at 146 (mem @100%) and stable on 15x multiplier have tried out numerous memory configurations and thats the best 3dmark score i can get. bizarrely enough, i can no get the mem to go to 5,2,2,2, but the score is (just a touch) lower than at 7,2,2,2 - all other things being equal.

    mind you, tere is alot more to play with in this bios than my old ASUS a7V - i don't know much about the AGP aperture and speed setting, so piut it at 128M and Auto.
    I've ran my 2400+ on a 210 FSB at 2415Mhz. How you ask???

    WIRETRICK!!!

  13. #28
    Joined
    Sep 2002
    Location
    London
    Posts
    44
    I'm impressed

    I've got my 2400+ now running at 12 x 186 and stable with the exception of real player when watching news 24 on broadband.

    i'm starting to think that its the software, not my overclocking.

    scared to up the voltage to get it much higher.

    thing is even 11.5x higher fsb is unstable, so much for my corsair 3200!
    Nick

    Windows ME
    Asus ADSL PCI modem
    IBM & Maxtor Hard drives
    Epox 8RDA+, T'bred 2400+ (12*182)
    512 M Corsair 3200 (1 stick so far)
    Radeon 9700 (317,304)
    430W Antec Truepower
    Creative Inspire 6.1 speakers
    A case with an UV lamp and lots of diodes that will light as and when...

  14. #29
    Joined
    Aug 2000
    Location
    Vvardenfell
    Age
    58
    Posts
    10,886

    Cool

    Please stay on topic - this is NOT the "how do I optomise my RAM" thread.

    M

  15. #30
    Joined
    Sep 2001
    Location
    Patras - Greece
    Age
    53
    Posts
    531
    Good thread here

    Dimitris....
    AMD ATHLON64 3200+ @ 2358(262X9) 1.6 Vcore
    Cooling BY AMD STOCK HEATSINK

    ABIT KV8 PRO rev. 1.1 @ Beta Bios 16
    512x2 ADATA VITESTA DDR500 at 2.6 V
    Sound Blaster AUDIGY 2
    Sound Blaster Megaworks Speakers 2.1 250D THX / 300 WRMS
    LEADTEK FX5900 128MB My ViVo
    System Powered By ANTEC True Power 480W
    =============


Posting Permissions

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