Page 3 of 9 FirstFirst 1234567 ... LastLast
Results 31 to 45 of 130
  1. #31
    Joined
    Jan 2004
    Posts
    225
    Originally posted by Maximus[X-D] on 02-17-2004 at 02:52 PM
    OK first your not running an Athlon64 your running an FX so your results are irrelevant since this cpu was left open to encourage overclocking. Its not an apples to apples comparision....i mean its a completely different motherboard.


    2nd. it looks more and more like the agp/pci speed is tied to the memory divider as in your running at less than 1:1. Now i know what your thinking "dude!!! your ram wont handle the high fbs speed at 1:1" to that i say naeee!!!! it will because all i have to do is swap in my IDE drive and it will do 240's with my best stick at 1:1 so there could definately be a link between lowered memory divider and the pci/agp speed. Which is just as bad as not having a lock because once you lower the divider in my testing ( air cooled) you hardly ever make the performance back in games because you have to raise the fbs so high that you sacrifice your HT link speed and the law of diminishing returns kicks in. Now in the case of Phase change or water cooling where you exceed the orginal curve im sure there are more returns because your able to attain a overall speed past the curve at which lowering the HT and memory settings hamper overall performance. Now if there was a true pci/agp lock assuming the HT bus would hold the best performance would be at 1:1 as high as your memory would run.


    3rd yet again a screenshot of high fbs is provided without any indication or mention of the hard drive used. I.E. SATA Raptors ( i keep harping on them because they are applicable to my situation but are also very finicky to pci speed making them great indicators of whether or not a pci lock exist)

    How about you loosen your memory timings and provide me with the highest stable speed your raptors will operate at 1:1


    Thx

    M*Xi
    Bro, may i know which raid controller are you using for yr raid 0 of your raptors?? Is it onboard or standalone PCI raid controller card??

  2. #32
    Joined
    May 2003
    Posts
    182
    ok you silly LITTLE FELLA

    on my main system which is in my sig i use 2 raptors with the onboard controller i also use 2 addition drives in a raid 1 array on the other onboard controller.


    M*XI
    Abit Fatal1ty AN8-SLI w XP90c
    2x1gig Corsair PC3200
    2 Raptor 74gig SATA 10krpm(raid 0)
    2 Maxtor 300 gig 7200 rpm 16Meg(raid 1)
    2 200 7200rpm 7200rpm 8Meg (HotSwap)
    1 200 gig Maxtor External
    2xGeForce 6800 GT 400/1100
    Audigy2
    Pioneer Slot loader DVD 120s
    Plextor 708a DVD burner
    NEC 3500 DvD burner
    Logitech Z 680 5.1
    Windows XP SP1

  3. #33
    Joined
    May 2002
    Location
    Cumberland, MD
    Age
    48
    Posts
    265
    I understand what you are saying. I have read the posts and I do agree that what you are saying seems correct. Perhaps I did not see it in this post but why is all my software saying I am sitting at just over 66MhZ AGP. Or further yet, I am able to run 220FSB then using Clockgen bump up my AGP speed to 70MhZ with no problems. Wouldn't bumping it up manualy via clockgen be just like increasing the FSB? Being both result in increasing the AGP/PCI bus speed to the point where the Raptors go nuts?

    I am using the onboard controller as well.
    Intel Pentium D 805 Smithfield Dual Core,EM64T Processor

    OCZ Gold 2GB (2 x 1GB) 240-Pin DDR2 SDRAM DDR2 800 (PC2 6400)

    XFX PV-T71G-UCE7 Geforce 7900GT

    Antec SUPER LANBOY

    Asus P5N32-SLI Deluxe nForce4 SLI
    Ultra ULT31852 X2 X-Connect 550W Power Supply (Titanium w/UV Blue) Retail

    Plextor PX-750A-BP/BL 16X Internal Dual Layer DVD±R/RW

    2X Western Digital Caviar SE16 WD2500KS 250GB

    Zalman CNPS9500 LED CPU Cooler Retail

  4. #34
    Joined
    Jul 2001
    Location
    The Goat Pen
    Posts
    4,088
    Originally posted by Maximus[X-D] on 02-17-2004 at 09:52 AM
    OK first your not running an Athlon64 your running an FX so your results are irrelevant since this cpu was left open to encourage overclocking. Its not an apples to apples comparision....i mean its a completely different motherboard.


    2nd. it looks more and more like the agp/pci speed is tied to the memory divider as in your running at less than 1:1. Now i know what your thinking "dude!!! your ram wont handle the high fbs speed at 1:1" to that i say naeee!!!! it will because all i have to do is swap in my IDE drive and it will do 240's with my best stick at 1:1 so there could definately be a link between lowered memory divider and the pci/agp speed. Which is just as bad as not having a lock because once you lower the divider in my testing ( air cooled) you hardly ever make the performance back in games because you have to raise the fbs so high that you sacrifice your HT link speed and the law of diminishing returns kicks in. Now in the case of Phase change or water cooling where you exceed the orginal curve im sure there are more returns because your able to attain a overall speed past the curve at which lowering the HT and memory settings hamper overall performance. Now if there was a true pci/agp lock assuming the HT bus would hold the best performance would be at 1:1 as high as your memory would run.


    3rd yet again a screenshot of high fbs is provided without any indication or mention of the hard drive used. I.E. SATA Raptors ( i keep harping on them because they are applicable to my situation but are also very finicky to pci speed making them great indicators of whether or not a pci lock exist)

    How about you loosen your memory timings and provide me with the highest stable speed your raptors will operate at 1:1


    Thx

    M*Xi
    Alright, so your saying,

    1) Some FX boards have a working lock/divider. That would be all well and good except the K8NNXP, AN50R, and K8N Pro are all socket 754 boards.

    2)In my testing so far, there doen't seem to be any performance hit running your CPU/mem async. If your RAM is running 240Mhz is running 240Mhz irregardless of what the "FSB" says. Using the mem divider you can max your CPU and the RAM at the same time. The GFX cards can't saturate the AGP tunnel as it is, so lowering the HTT link shows little to no performance loss in real world. I've only tested UT2003 so far, I'll test a few more games to see if my theory holds true.

    3)Irregardless of what HD your using there is no way a IDE harddrive will take a 60Mhz PCI bus. The Raptors are a good way to test though because they will give up before your RAM or HTT link if no lock/PCI is present. ATM my RAM gives out @230-235 or so 1:1, also gives out at 230-235 using the memory dividers. I should have the Vdimm mod figured out today some time. And I'll test some 1:1 action

    Edit:
    DannyTeets,
    Your GXF card should be able to handle 70Mhz. I don't think that your problem. Did you try raising the Vddq voltage?
    Last edited by Caprid; 02-17-2004 at 10:50 AM.


  5. #35
    Joined
    May 2002
    Location
    Cumberland, MD
    Age
    48
    Posts
    265
    Agreed but is the PCI bus speed running seperate from the AGP bus? If it is not wouldn't increasing the AGP bus in turn increase the PCI bus?

    Danny
    Intel Pentium D 805 Smithfield Dual Core,EM64T Processor

    OCZ Gold 2GB (2 x 1GB) 240-Pin DDR2 SDRAM DDR2 800 (PC2 6400)

    XFX PV-T71G-UCE7 Geforce 7900GT

    Antec SUPER LANBOY

    Asus P5N32-SLI Deluxe nForce4 SLI
    Ultra ULT31852 X2 X-Connect 550W Power Supply (Titanium w/UV Blue) Retail

    Plextor PX-750A-BP/BL 16X Internal Dual Layer DVD±R/RW

    2X Western Digital Caviar SE16 WD2500KS 250GB

    Zalman CNPS9500 LED CPU Cooler Retail

  6. #36
    Joined
    May 2003
    Posts
    182
    yea im saying that no board not even the NF3 has a working pci lock at 1:1 memory.

    so your saying your ram give out at 230 ish 1:1 and that you have tested your raptors stable at 230+ with your memory setting at 1:1? if so can you provide some documentation as to this feat? and tell me exactly which board you did it on please so that i may purchase it immediately.


    in my testing when you lower the memory divider to get past the 220 fbs barrier of no pci lock then you have to run the fbs up to a point that the HT link becomes unstable at its highest setting forcing you to lower it and although in theory it should still note be bus limited in the real world benches ive done if you lower the HT speed to achieve stability because your raising the fbs to get back memory bandwidth you lost because you had to set the memory divider away from 1:1 due to the lack of implemented PCI lock at 1:1. In all that juggling and tweaking i usually end up right back at the same speeds once everything is maxed out as i was at 220 1:1 default multi.

    M*XI
    Abit Fatal1ty AN8-SLI w XP90c
    2x1gig Corsair PC3200
    2 Raptor 74gig SATA 10krpm(raid 0)
    2 Maxtor 300 gig 7200 rpm 16Meg(raid 1)
    2 200 7200rpm 7200rpm 8Meg (HotSwap)
    1 200 gig Maxtor External
    2xGeForce 6800 GT 400/1100
    Audigy2
    Pioneer Slot loader DVD 120s
    Plextor 708a DVD burner
    NEC 3500 DvD burner
    Logitech Z 680 5.1
    Windows XP SP1

  7. #37
    Joined
    Jan 2004
    Posts
    225
    Originally posted by Maximus[X-D] on 02-17-2004 at 03:33 PM
    ok you silly LITTLE FELLA

    on my main system which is in my sig i use 2 raptors with the onboard controller i also use 2 addition drives in a raid 1 array on the other onboard controller.


    M*XI
    Ohh.. thanks!!
    So the raptors on the VIA raid or the Sil image raid??

  8. #38
    Joined
    May 2002
    Location
    Cumberland, MD
    Age
    48
    Posts
    265
    OOOOOOOOkkkkkkkk! Thank you! I understand! Bling! I see the light!
    Intel Pentium D 805 Smithfield Dual Core,EM64T Processor

    OCZ Gold 2GB (2 x 1GB) 240-Pin DDR2 SDRAM DDR2 800 (PC2 6400)

    XFX PV-T71G-UCE7 Geforce 7900GT

    Antec SUPER LANBOY

    Asus P5N32-SLI Deluxe nForce4 SLI
    Ultra ULT31852 X2 X-Connect 550W Power Supply (Titanium w/UV Blue) Retail

    Plextor PX-750A-BP/BL 16X Internal Dual Layer DVD±R/RW

    2X Western Digital Caviar SE16 WD2500KS 250GB

    Zalman CNPS9500 LED CPU Cooler Retail

  9. #39
    Joined
    May 2003
    Posts
    182
    hmmmm been a while i believe that i found the raptors benched better and with lower cpu utiliiztion on the sili controller so thats where they reside and the Raid 1 array ( backups and important files ) are on the via array.


    M*XI
    Abit Fatal1ty AN8-SLI w XP90c
    2x1gig Corsair PC3200
    2 Raptor 74gig SATA 10krpm(raid 0)
    2 Maxtor 300 gig 7200 rpm 16Meg(raid 1)
    2 200 7200rpm 7200rpm 8Meg (HotSwap)
    1 200 gig Maxtor External
    2xGeForce 6800 GT 400/1100
    Audigy2
    Pioneer Slot loader DVD 120s
    Plextor 708a DVD burner
    NEC 3500 DvD burner
    Logitech Z 680 5.1
    Windows XP SP1

  10. #40
    Joined
    Jul 2001
    Location
    The Goat Pen
    Posts
    4,088
    Originally posted by Maximus[X-D] on 02-17-2004 at 10:57 AM
    yea im saying that no board not even the NF3 has a working pci lock at 1:1 memory.

    so your saying your ram give out at 230 ish 1:1 and that you have tested your raptors stable at 230+ with your memory setting at 1:1? if so can you provide some documentation as to this feat? and tell me exactly which board you did it on please so that i may purchase it immediately.


    in my testing when you lower the memory divider to get past the 220 fbs barrier of no pci lock then you have to run the fbs up to a point that the HT link becomes unstable at its highest setting forcing you to lower it and although in theory it should still note be bus limited in the real world benches ive done if you lower the HT speed to achieve stability because your raising the fbs to get back memory bandwidth you lost because you had to set the memory divider away from 1:1 due to the lack of implemented PCI lock at 1:1. In all that juggling and tweaking i usually end up right back at the same speeds once everything is maxed out as i was at 220 1:1 default multi.

    M*XI
    LOL your going to have me hook up the Raptors again No problem, I guess that would show if it's a true 'Lock' or just a divider. I'll get back to you tonight In case anyones interested here's what I tested so far. Same CPU speed, same memory speed. The only thing that was changed is the HTT,

    Dividers
    1:1=200
    6:5=166
    4:3=150
    3:2=133
    2:1=100


    345 X 7 HTT=X1 Mem=220.1 3:2Divider CPU=2421Mhz
    SSmem-6350/6233
    PiFast-50.51
    3Dmark-18911
    UT2003
    antalus-bot-103.111649
    citadel-bot-116.803772
    citadel-fly-261.940460
    SuperPi-35s

    220 X 11 HTT=X1 Mem=219.7 1:1Divider CPU=2416.6Mhz
    SSmem-6250/6191
    PiFast-50.87
    3Dmark-18626
    UT2003
    antalus-bot-101.952034
    citadel-bot-114.942245
    citadel-fly-259.282135
    SuperPi-36s

    220 X 11 HTT=X2 Mem=219.7 1:1Divider CPU=2416.6Mhz
    SSmem-6255-6188
    PiFast-50.87
    3Dmark-18841
    UT2003
    antalus-bot-101.970528
    citadel-bot-115.010040
    citadel-fly-259.162323
    SuperPi-36s

    220 X 11 HTT=X3 Mem=219.7 1:1Divider CPU=2416.6Mhz
    SSmem-6247-6175
    PiFast-50.85
    3Dmark-18877
    UT2003
    antalus-bot-101.920593
    citadel-bot-115.860733
    citadel-fly-261.002716
    SuperPi-36s

    No difference


  11. #41
    Joined
    Feb 2003
    Location
    Boston, MA
    Posts
    793
    To anyone having HD problems when OCing DO NOT use the via controller, instead use the secondary controller(promise 378 on most). The VIA is somehow on a different PCI level and is not affected by dropping the memory in DDR333 or lower. The limit will be around 220fsb no matter what if you use VIA

  12. #42
    Joined
    Jan 2004
    Posts
    225
    Originally posted by Dr.Nick on 02-17-2004 at 04:15 PM
    To anyone having HD problems when OCing DO NOT use the via controller, instead use the secondary controller(promise 378 on most). The VIA is somehow on a different PCI level and is not affected by dropping the memory in DDR333 or lower. The limit will be around 220fsb no matter what if you use VIA
    That is wat i get when i try overclocking with my raptors on the VIA raid.

  13. #43
    Joined
    May 2003
    Posts
    182
    Ahhh good work well laid out results i see why we disagree on the relevance of HT is because you were looking at the UT2003 results and i have been focusing on 3dmark results. Notice how with each jump in HT setting although your UT scores didn’t change your 3dmark scores did increase which although synthetic probably provides a good picture of overall system impact.

    I notice your best benches where performed with just a little more memory bandwidth and this comes back to my original belief that assuming your cpu has the ceiling to do so that you would be better off running whatever speed your memory will max at 1:1 assuming that you had a pci lock and/or did not have to lower the HT bus speed.

    Btw I have noticed even more decrease in speed from Via based Chipsets when lowering HT speed.


    Man you have done a lot of testing and been real gracious. Please test the maximum stable overclock of your memory at 1:1 if that turns out to be say 230 or greater leave it set at that speed and swap your raptors in changing nothing. That would show a lot and thx for all your efforts.


    M*XI
    Abit Fatal1ty AN8-SLI w XP90c
    2x1gig Corsair PC3200
    2 Raptor 74gig SATA 10krpm(raid 0)
    2 Maxtor 300 gig 7200 rpm 16Meg(raid 1)
    2 200 7200rpm 7200rpm 8Meg (HotSwap)
    1 200 gig Maxtor External
    2xGeForce 6800 GT 400/1100
    Audigy2
    Pioneer Slot loader DVD 120s
    Plextor 708a DVD burner
    NEC 3500 DvD burner
    Logitech Z 680 5.1
    Windows XP SP1

  14. #44
    Joined
    Sep 2002
    Location
    JAX, FL
    Age
    46
    Posts
    2,193
    "btw if you have to lower your ddr timing to achieve better dividers ( which also proves a lock is possible) then we are still screwed because thats a significant hit"

    not really, based on my own system by lowering my ram it didnt hurt me due to the overall smooth bandwidth..

    @ 215 , ht@ 200, ddr @ 400 =
    3dmark 17,900
    pcmark cpu 6869 , memory 8905

    @ 225, ht @ 800, ram lowered to 333 =
    3dmark 18,331
    pcmark cpu 7270 , memory 9106

    Though I will try it with 1st setting and try HT @ 800 along with the ram @ 400 but im getting better OVERALL performance with my current setup which is 2nd listed setup.

    I think honestly we're all still learning, so things change and such..at least for me it is.
    heat: the-elusive-dragon

    Dell XPS desktop
    19" monitor
    Intel core duo @ 2.2ghz
    2gb ram
    250gb hdd
    nvidia 8600 gts video with 256 ram
    saitek backlit kb
    razer diamondback mouse (red led)
    windows vista

  15. #45
    Joined
    May 2003
    Posts
    182
    haha ok man did you leave the cpu divider the same? i bet you did thats where you getting your increased peformance by raw cpu speed. Now lower the divider till the cpu is at the same speed as it was at 215 and check it. Theres no way around the fact that your results will be best if you could achieve your max memory speed at 1:1 without having to lower your HT speed or hitting your cpu ceiling.


    M*XI
    Abit Fatal1ty AN8-SLI w XP90c
    2x1gig Corsair PC3200
    2 Raptor 74gig SATA 10krpm(raid 0)
    2 Maxtor 300 gig 7200 rpm 16Meg(raid 1)
    2 200 7200rpm 7200rpm 8Meg (HotSwap)
    1 200 gig Maxtor External
    2xGeForce 6800 GT 400/1100
    Audigy2
    Pioneer Slot loader DVD 120s
    Plextor 708a DVD burner
    NEC 3500 DvD burner
    Logitech Z 680 5.1
    Windows XP SP1

Posting Permissions

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