1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Will a 3.4 Prescott reach 4GHz+ ?

Discussion in 'Overclocking & Cooling' started by Toff, 31 May 2006.

  1. Toff

    Associate

    Joined: 11 May 2006

    Posts: 46

    Location: GX Bucks

    I ordered a 3.4 478 Extreme but sadly the box turned up empty! :mad: :mad: :mad: (my first e8ay problem in over 110 transactions :( )

    As it seems almost impossible to get a plain 3.4 northwood (let alone an Extreme) - I'm wondering about a Prescott.

    As previously mentioned, I'd love to reach 4GHz - will a 3.4GHz Prescott over-heat before it gets there?

    My water rig is working well - my 3.0 northwood is running 3.7GHz (1.6v) and the temp is 34 idle, 37.5 under load.

    - Toff.
     
  2. AWPC

    Sgarrista

    Joined: 8 Dec 2005

    Posts: 9,759

    The skt478 will need an extreme cooling solution but it can just about be done although sudden voltage death is a distinct possibility.

    Not sure Intel ever released a skt478 3.4 though. Thought they just paper launched & maybe a few trickled out then went over to the skt775s which can clock well over 4Ghz.

    TBH you are not gonna see much more performance anyway due to latency issues. Probably another 3-5% tops in real world performance.

    Save your money & wait for Conroe.
     
  3. Toff

    Associate

    Joined: 11 May 2006

    Posts: 46

    Location: GX Bucks

    Well that pee's on my bonfire :rolleyes:

    But of course you are right.. even without the latency issues, it'd be at least £150+ for a theoretical 300MHz raise! I guess I just got fixated on reaching 4 GHz!

    I've got my Northwood 3.0 to around 3.7 by raising the voltage to 1.625v that means the FSB is around 255 (any further and it's unstable).

    have I pretty much reached the limits of my processor, or are other people eeking more out of their 3.0 Northwoods?

    Cheers - Toff.
     
  4. AWPC

    Sgarrista

    Joined: 8 Dec 2005

    Posts: 9,759

    If you buy some exotic ram that can reach a native FSB speed of 1066 then it may be possible with what you have as a higher FSB + lower CPU multiplier often gives better ocing results anyway as the mobo is more able to cope with higher FSBs than the CPU being maxed out alone.

    http://www.overclockers.co.uk/acatalog/Online_Catalogue_Corsair_XMS_310.html

    Try and find something which is DDR533 or higher as that may do the trick but not sure if your mobo needs a mod to do a higher FSB than that.
     
  5. Bundles

    Wise Guy

    Joined: 18 Jan 2003

    Posts: 5,384

    Location: Hatton

    a prescott would get you to 4ghz on water np, mine did, and my mobbo was gash!

    [​IMG]
     
  6. Toff

    Associate

    Joined: 11 May 2006

    Posts: 46

    Location: GX Bucks

    Bundles.. that's amazing.. but I thought using a 5:4 ratio on the RAM wasn't as good as having a 1:1 (using suitably quick memory as AWPC suggested).

    I thought raising the FSB (and core voltage - max 1.65v) gradually until the system proves unstable (then back-tracking until stable) was the best way to OC :confused:

    Excuse my ignorance, should I drop the FSB back to standard and raise the multiplier instead?

    - Toff.
     
  7. Bundles

    Wise Guy

    Joined: 18 Jan 2003

    Posts: 5,384

    Location: Hatton

    Cheers. 1:1 is preferable over 5:4, but on older intel systems it makes little difference tbh.

    It is, although i tend to just whack them both up, if it's unstable, give it more voltage, if it's not, give it more fsb until it is, THEN whack up the voltage.

    You would have a hard job upping the multi, it's locked on 99% of P4s.

    :)
     
  8. AlanJ

    Hitman

    Joined: 16 Nov 2005

    Posts: 684

    Location: Torquay

    why isnt the PI better?

    i get 31 @2.9 with a 3700+
     
  9. WJA96

    Capodecina

    Joined: 13 Jul 2005

    Posts: 15,766

    Location: Norfolk, South Scotland

    Probably because the memory is running out of sync with horrible timings. P4's don't do well on SuperPi as it's memory bound.
     
  10. Bundles

    Wise Guy

    Joined: 18 Jan 2003

    Posts: 5,384

    Location: Hatton

    Because P4's are soooo weak at some things, but very strong at others, hmm, just like AMD's.

    It's mainly my pap memory that was to blame for the Spi time, it's only wee Adata PC3200, bless it.
     
  11. AlanJ

    Hitman

    Joined: 16 Nov 2005

    Posts: 684

    Location: Torquay

    ah that'll explain it, thanks
     
  12. Siders77

    Hitman

    Joined: 5 Nov 2005

    Posts: 960

    Location: Yate, Bristol

    [​IMG]

    Drazic did it.

    Okay his timing was shocking but he ran a 5:4 divider and was using AMD RAM as to INTEL ram like he did end up doing.

    So yes it can be done but you have to know what ya doing to do it
     
  13. BISH422

    Gangster

    Joined: 1 Jun 2006

    Posts: 254

    Hi all

    just noticed i have exact same chip as one shown in Bundles post 3ghz p4, and i am interested to know whot board u are using/or could recomend as i am thinkin of trying some overclockin myself .I need to get a new mobo as my chip sits inside a dell pc and as u proberly know dell's dont overclock :( very easy if at all.

    any help would be greatly appreciated
     
  14. Bundles

    Wise Guy

    Joined: 18 Jan 2003

    Posts: 5,384

    Location: Hatton

    That was an MSI 865PE neo2 platinum edition (RIP) but there are much better boards out there, try to get hold of an Asus P4P800/P4S800 or an Abit IC7 of some veriety, these are your best boards for S478 intels.
     
  15. Toff

    Associate

    Joined: 11 May 2006

    Posts: 46

    Location: GX Bucks

    Excuse my ignorance (again) but what does PI stand for please?

    I settled for a 3.2 extreme edition. Without any OC'ing it beats my 3.0 OC'd northwood (3.7GHz) in the benchmark tests :)

    So that's cool, probably due to the extra cache (and lack of errors going on?)

    The extreme edition doesn't seem very happy to overclock though.. it'll only reaches 238 FSB and then it's unstable thereafter. Raising the voltage slightly doesnt seem to help at all.

    Are extreme editions generally less friendly to overclock?

    - Toff.
     
  16. BISH422

    Gangster

    Joined: 1 Jun 2006

    Posts: 254

    thx for the reply , i shall keep my eyes open for 1 of these boards. Then will be back in touch for some advice on overclockin it

    thx again
     
  17. daz

    Capodecina

    Joined: 18 Oct 2002

    Posts: 23,971

    Location: Bucks

    Pi, as in the mathematical constant - the ratio of a circle's circumfrence to its diameter. Pi is an irrational number, which means it has no end - and infinite number of decimal places. This program calculates pi to a specified number of decimal places using a certain algorithm.
     
  18. Siders77

    Hitman

    Joined: 5 Nov 2005

    Posts: 960

    Location: Yate, Bristol

    It has 2Mb catch dude, thats why it'll be tons faster

    It should clock by a milestone being a Galantic core

    What volts you pushing?

    You set your ram to the slackest and hihest vdim and then clocked?

    Oh and what bord too
     
  19. Toff

    Associate

    Joined: 11 May 2006

    Posts: 46

    Location: GX Bucks

    Hi,

    At the standard voltage (1.55v) it will run stably until 238FSB. I've tried pushing the voltage upto 1.625v but it's still unstable past anything over 238FSB (too chicken to raise the voltage any further! - the water cooling is working well, but I don't want to nuke the CPU with too much voltage!)

    The ram is OCZ 4000, so it's actually running below-spec (500MHz native). Should I still slacken off the timings?

    I'm loving overclocking - good job you guys are here to help :)

    - Toff.
     
  20. Toff

    Associate

    Joined: 11 May 2006

    Posts: 46

    Location: GX Bucks

    Sorry, ABIT IC7-MAX3