Couple of questions from novice

CryptoCrypto W.Sussex UK Member
edited January 2004 in Hardware
Hi Chaps,

couple of questions before I start to OC my new rig.

1. Which temperature reading programmes do you generally use when discussing temps. I currently use MBMonitor which gives me 3 readings, currently 25 System, 41 CPU, 54 Diode ( running FAH 100% of course) :thumbup :fold:

2. Which benchmarking programme is the most appropriate for "real world" computer use. Main use for me is Internet browsing and digital photo editing and printing. I understand that I should use Memtest86 to test for memory stability between each change of fsb :confused:

I'm not looking to wring every bit out of this rig, simply to get the best, stable deal for my money.

Thanks in advance

Crypto :ukflag:

Comments

  • muddocktormuddocktor
    edited January 2004
    MBM is a great proggie to use for monitoring the temps on that Asus board because you can directly read the die temps, which Asusprobe doesn't do.

    For benchmark testing for stability, I just run Gromacs WU's and minimize error problems by copying the whole folder that F@H is in to another spot on my hard drive before making a change in speed or fsb and unhooking the cat5 cable for a while so that if the WU errors out it can't send in the error and get another WU to mangle. When I get an error like that, I then shut down the client and delete the entire folder and then copy back the saved folder back to the original location and reboot back to bios to make changes for added stability.
  • Straight_ManStraight_Man Geeky, in my own way Naples, FL Icrontian
    edited January 2004
    And as for benchmarking, for your needs, The Winstone and Winbench CDs, which are created by Veritest, will give you good overall benching. I suggest getting BOTH CDs and the media content (comes with Winstone or Winbench set) bencher from same place if you want SiSoft Sandra is a good detail looker for seeing what is lagging in asystem, but it sometimes likes DMA on for real HD benching. For HDs, there are better benchers, for CPU and RAM effectiveness SiSoft Sandra is good.

    For what you say you do, you do not need to bench 3D Video against gamers, but AquaMark3D should run if you ahve an overall stable system. Most of what you want to do is lower res 3D or (and most will be this) 2D. Bench 2D more, for your purposes.

    Digital photo work can be RAM intensive, and I favor TIFF over jpg if you are going to be putting on web. Idea is this, browsers like jpg or in modern browsers, some will handle png. jpg is lossier, work in tiff and then convert final edit to jpg. Keep the tiff for later use. I like a mix of Adobe Photoshop Elements 2.0 and Paint Shop Pro 8.0 and Corel Draw 11 for intense digital work, but my digicam stores at 1600x1200 TIFF by choice and anything for the web gets editted and resized as tiff(save full size and editted, resized tiff until happy with web looks) and then converted to png or jpg. One of the reasons I do not use PhotoShop is that it can be a memory hog for high-res, true color, digital editting, and my Windows box folds as I edit photos and etc. PhotoShop Elements will suit a novice, but get Version 2.0 or up as it has more and better tools than earlier versions.

    NOTE on RAM, I run 1 GB RAM and a Barton 2500+, can and DO run a heavy accounting program plus Crystal Reports Pro, and digiphoto edit all at once, WHILE folding. I did strip some fo the media stuff out of XP Pro and run it in full Classic Theme, reduces overhead for things I do not need for my own purposes.

    John.
  • csimoncsimon Acadiana Icrontian
    edited January 2004
    You may already know how to OC but I will include my little step-by-step guide that I refined from someplace that I copied it ...hope it helps. Keep in mind that I am an OC newb so don't flame too hard! :thumbsup: Good luck!

    Highest CPU Speed Mhz
    1. Bios ...Softmenu III Setup ...
    2. Change multiplier from auto to 10.5 (10.5x133=1400).
    3. Reboot ...SoftMenu III Setup ...
    4. Increase multiplier one notch.
    5. Reboot into WindowsXP ...test/benchmark.
    6. If BSOD or crash increase vcore voltage one notch and try again.
    7. If no BSOD or crash increase multiplier one notch.
    8. Repeat #4 until highest cpu speed is achieved.

    Highest FSB Speed
    1. Bios ...divide top speed by 166 and set your multiplier to that then set FSB to 166.
    2. Reboot into WindowsXP ...test/benchmark.
    3. If BSOD or crash increase vmem voltage one notch and try again.
    4. If no BSOD or crash ...divide top speed by 176 and set your multiplier to that then set FSB to 176. Repeat #2.
    5. ...divide by 180.
    6. ...divide by 196.
    5. ...divide by 200.

    Lowest Memory Settings
    1. Set FSB back to 133 and multiplier back to auto.
    2. Set vmem as high as possible.
    3. Bios ...Advanced Chipset Features>DRAM Settings.
    4. Set 6-3-3-2 ...reboot ...bench/test.
    5. Set 6-3-2-2 ...reboot ...bench/test.
    6. Set 6-2-2-2 ...reboot ...bench/test.
    7. Set 5-3-2-2 ...reboot ...bench/test.
    8. Set 5-2-2-2 ...reboot ...bench/test.

    Merge Everything
  • CryptoCrypto W.Sussex UK Member
    edited January 2004
    Mudd,
    I agree that MBM is better than ASUS probe, but, when people post their temps ie 35 idle 46 load, do they mean socket temp or die temp. There's about a 10 degree difference between the two. :confused:

    Your "realworld" benchmarking and stability tests seem a bit too much fiddling about for me. Do you really find crunching Gromacs a repeatable stability measure? You've got so many rigs there, you obviously have a lot of experience, I'll give it a try. :smiles:

    Ageek,
    Winstone and Winbench sound as if they are realworld benches, good for testing memory bandwidth settings perhaps?
    Your views on jpg and TIFF are noted and I'll have a look at the difference next time I'm up for some editing. Thanks :D

    Thanks chaps

    Crypto :ukflag:
  • CryptoCrypto W.Sussex UK Member
    edited January 2004
    CSimon,

    thanks for your OC guide. I've never seen one before that sorts out the multiplier before the fsb. Interesting, I'll give it a go :)

    Probably won't get far with improving the memory timings, I was too mean (or skint) to spend on high spec memory. :(

    Cheers

    Crypto :ukflag:
  • csimoncsimon Acadiana Icrontian
    edited January 2004
    Well that OC guide was originally written around the softmenu III nf7 series of bios but I'm sure you can probably manipulate it to suite your needs as well!!
  • GHoosdumGHoosdum Icrontian
    edited January 2004
    Nice guide - it really covers the basic process well, csimon!
  • csimoncsimon Acadiana Icrontian
    edited January 2004
    Thanks GH but I can't take all of the credit ...I think I borrowed most of it from an old madshrimps article.
  • GHoosdumGHoosdum Icrontian
    edited January 2004
    I've never heard of madshrimps. But I have heard of csimon. So I'll give you all the credit. :thumbsup:
  • muddocktormuddocktor
    edited January 2004
    Crypto, most people are quoting the socket thermistor temps with AMD rigs because that is what their mobo supports but the diode temp is much more accurate.

    I find crunching Gro work to be very accurate for non-3d app stability because they will fail with the slightest instability with either the cpu or memory problems. I just disconnect the cat5 cable and save a copy of the folder so as not to flood Stanford with bad WU's if I'm not totally stable at the new setting. Obviously, Gro's don't test stability while gaming as it doesn't tax the system while the video subsystems are heavily loaded though.
  • CryptoCrypto W.Sussex UK Member
    edited January 2004
    muddocktor wrote:
    Crypto, most people are quoting the socket thermistor temps with AMD rigs because that is what their mobo supports but the diode temp is much more accurate.

    I find crunching Gro work to be very accurate for non-3d app stability because they will fail with the slightest instability with either the cpu or memory problems. I just disconnect the cat5 cable and save a copy of the folder so as not to flood Stanford with bad WU's if I'm not totally stable at the new setting. Obviously, Gro's don't test stability while gaming as it doesn't tax the system while the video subsystems are heavily loaded though.
    Thanks for that info Mudd, sort of just what I wanted to hear. :wink:

    I've just found out on the last couple of hours in my OC exploits indeed that Gromac folding goes belly up due to OC instability :-/ , so good call there.

    I'm currently running back at stock speeds and voltages after a couple of nasty frights where I thought I'd lost my hard drive and memory. A nice long CMOS battery out and a walk in the woods got me out of trouble. Maybe I'll have another go later.

    One thing I did notice was that as I increased the FSB between 180 and 190 I noticed little difference in the memory benchmark in Sandra 2004. :-/

    I got as far as 190FSB and 12 multiplier before it all went pear shaped.


    Cheers

    Crypto :wave:
Sign In or Register to comment.