Abit TH7-II RAID overclocking benchmarks

Hi guys,

Just want to share my overclocking result with you.

Take a look here:
23 answers Last reply
More about abit raid overclocking benchmarks
  1. kewl, thogh on the HD, the fans make more nois, so get a faster HD then a quiet one..

    <font color=green>
    *(k)eep (I)t (S)imple (S)tupid*
    </font color=green>
  2. Awesome,
    Good work there!
    Nice bench scores wow! Question noticed that you got a Matrox G550 just wondering what kind scores do you get on 3dmark2001 default?
  3. LoveGuRu, you may have a point there. Had the Western Digital Caviar WD1000BB been available at the time, I would probably have chosen that instead. With everything assembled, I must say that the power-supply and the CPU fan are most dominant in the noise-picture, but still very quiet compared to the AMD 1.4 GHz based machines I have heard.

    By the way, the one-platter Barracuda isn't really that slow, and it is absolutely quiet.
  4. Have not tried that, but the G550 is absolutely not for 3D-gaming. I think it performs more or less like a GeForce2 MX200. Go for a GeForce3 Ti 200, Ti 500 or maybe even a ATI Radeon 8500 if you believe ATI are able to optimize their drivers in time.
  5. i use IBM 60GXP 40 gig drive and its quite quiet and the fastest in its catagory, i really developed a dislike for WD products- they just catch bad sectors like a bad cold!

    i just stopped buying them 1/2 years ago..trust me and folow..

    <font color=green>
    *(k)eep (I)t (S)imple (S)tupid*
    </font color=green>
  6. I figured that, Matrox make great 2d cards and i like the DualHead feature. Anyway I will probably get GeForce3 Ti 200 or ATI radeon 8500 but i will wait until januari (Northwood) before my next major upgrade(Ati should have there drivers optimized by then).
  7. You said the core voltage dropped slightly under load. I found that when I got rid of my cheap power supply (Antec) and put in a generic heavy duty OEM power supply of the same size (250w on a loaded PIII) I was able to get rid of the problem and stabley set my core voltage lower by .1v! So you may want to consider a better power supply, if you already have a really good one, maybe a larger size.
    Also, have you cansidered setting the memory to PC600 in BIOS? At 133MHz (QDR533) FSB, your memory would be back up to 800MHz that way, this could give you the ability to reach even higher clocks (if the motherboard can supply the lower PCI/AGP dividers).

    Back to you Tom...
  8. Yes, as I mentioned memory performance was not consistent with FSB set to 122 because the memory is most likely running on the edge at PC976. The BIOS offers to run memory corresponding to PC600 bringing it back to normal PC800 with FSB set to 133 as you mention. The motherboard even provides fixed PCI/AGP dividers (33/66) so I guess it would be possible to set the FSB to at least 133, meaning that when the Northwood B with 532 FSB arrives next year I could actually use it with the same motherboard provided it got hold on some PC1064. When I get to work on monday, I will try to crank up the FSB and see what SiSoft Sandra scores I can reach.

    I really don't think it's a big issue that the core voltage drops a bit under load, so I will hold on to the relatively silent high quality Antec power-supply.
  9. The memory is likely not having any problem. It is the Direct Rambus Clock Generators (DRCGs) on your motherboard that cannot handle the high speeds. If you lower the multiplier to 3x then the DRCGs will no longer give you any trouble.


    = The views stated herein are my personal views, and not necessarily the views of my employer. =
  10. A Danish test-site has obtained a FSB of 168MHz ! on the Abit TH7-II RAID with fixed PCI/AGP (33/66) and the memory multiplier set to 3 corresponding to PC1008. If it's realiable information, I guess it corresponds to a 3360 MHz P4 CPU as they are using a 2 GHz unit. They don't mention anything about the cooling used.

    If you understand Danish you will be able to read it here:

  11. I did not see any mention of any overclocking at that URL. It seemed more like a review of the motherboard. Do you have a more specific URL?


    = The views stated herein are my personal views, and not necessarily the views of my employer. =
  12. Try page 4. If you examine the text you see the number 168MHz FSB. There is no typical benchmark figures in the article. If the PCI/AGP is 33/66 and memory is run with a multipler of 3, I am not sure if it's right to just say 1.66 * 2000 Mhz = 3320 MHz, they don't claim this hihh number in the article, it is something I have concluded, which may be wrong.

  13. They likely had an unlocked Pentium 4 processor; an engineering sample. I highly doubt they got a Willamette Pentium 4 that high.


    = The views stated herein are my personal views, and not necessarily the views of my employer. =
  14. No, I don't think they had an unlocked Pentium 4 processor, after all they overclock by changing the FSB, indicating that the clock multiplier is locked.
  15. They were increasing the memory bandwidth by overclocking the FSB. Noone said their target by increasing the FSB was to overclock the CPU.


    = The views stated herein are my personal views, and not necessarily the views of my employer. =
  16. Hmm...first you say you did it, now you say you didn't?

    <font color=green>I post so you don't have to!
    9/11 - RIP</font color=green>
  17. Hello Guys,

    I have been doing some further experiments with the Abit TH7-II RAID equiped with a 1700 MHz P4 Socket 478.

    In short, I was not able to exceed my previous record of 2074MHz on a locked P4 1700 MHz with stock cooling.

    Using a fixed PCI/AGP clock of 33MHz/66MHz and lowering the RDRAM Bus Frequency to 300 MHz didn't allow me to increase the External clock above 122. I have a feeling that more cooling will be needed to go any higher. The Intel designed cooling solution is surely adequate for running at 1700 MHz, but it's nothing near the best air-cooling available. For one, the current 80 mm stock fan only rotates between 2400-2750 rpm and could be substituted by a screaming 7000 rpm Delta fan. Also the heat-sink itsef is made completely in ALU. Something in the league of the Switftec MC-462A would most likely perform better.

    Maybe one day in the near future, I will try to do some more experiments with improved cooling, just to see how far the P4 can be taken. I plan to put the complete PC in a very cool and dry place.
  18. I will get a TH7-II very soon to check it out.

    I been playing with the P4B for passed 2 day, Im not setting any performance records, but I can post some decent speeds.

    P4B has the ability to change the multiplier, I am able to use any multiplier between 15~24, im able to fire up at 24x100 and 15x133 (example).

    Do you have the ability to change the multiplier on the TH7-II? whats the highest core voltage you can set too?

    Some thing that might help you post a higher clock speed.

    Remove all PCI cards, run with just AGP video or PCI video (seems like this is popular with highest overclocks), break down your ram to a single "best" chip, kick up your core voltage, unplug CDROM drives and other removeable drives (power and data), create a wind tunnel out of your case for smooth flowing air form back to front, check the direction of air from from your PSU and have it blow outside the case (heat rises and PSU is usually at top of case), be aware of FSB/AGP/PCI ratio's you might not be able to post at 132FSB but you might be able to post at 134FSB (example @ 120FSB your AGP is at 80Mhz (and goes higher till you reach 133FSB), thats only 14Mhz overclock, but at 133FSB it drops back to 66Mhz FSB.
  19. Yes, the Abit TH7-II allows you the change the multiplier, but it doesn't help much, as the CPU's you can buy are all locked units. Only test-sites seems to get those nice un-locked P4's, to make the mobos look good in the reviews.

    The highest core voltage you can set in the BIOS is 2.20v; most P4 boards can't go this high. It has to be said that the actual core voltage is somewhat lower (2.05v - 2.10v, according to Hardware Doctor) and dropping 0.05v - 0.10v under load.

    My benchmarking (take a look at www.sitecenter.dk/hjv) was done with the core voltage set to 2.20v.

    You mention a lot of tricks to obtain some more overclocking. Initially I actually just wanted to see what could be reached in a completely normal and silent setup, as this is the only way I can make a daily use of it, but for pure "scientific" reasons (my own curiousity) I plan to give the system a serious go the upcoming weekend. Some of your hints will be used also.
  20. interesting that you said that your core voltage dropped allitle under full load...

    i would say thats a sign that your reaching your PSU's power output limit.

    Religious wars are 2 groups of people fighting over who has the best imaginary friend.
  21. Cool, higher core voltages possible than 1.85v thats a good thing.

    I will have a TH7-II to play with tomorrow. I picked up a P4X266 board last night for $89 fresh off the boat from tiawan. very cheap board supposed to support northwood. I gotta run down a DDR2100 real fast and I will swap it out with the P4B (only problem I had was a driver for XP. I extracted the .inf with -s option) other than that its a great overclockers board.

    This P4x266 is super cheap in quality and design. there is a front panel for audio and USB (kinda nice) but everything else is inferior quality (shine a flashlight thru the manual). At least this board doesnt have "Warranty void if broken" on it. what am I looking for in the P4X266? stability, easy instalation, overclocking, knowledge of product.

    CompUSA now sells the power supply tester plug, only $12.95 near the cases and cooling products.
  22. What brand of P4X266 did you get?

    <font color=green>I post so you don't have to!
    9/11 - RIP</font color=green>
  23. Generic box, plain maual, made by Azza after checking with fcc.gov FCC ID search.


    <P ID="edit"><FONT SIZE=-1><EM>Edited by FUGGER on 10/25/01 02:39 PM.</EM></FONT></P>
Ask a new question

Read More

CPUs Overclocking NAS / RAID