FSB ceiling on A7N8X-E, higher than I thought

lemonlimelemonlime Canada Member
edited December 2004 in Hardware
My 3000+ Barton is not much of an overclocker, no matter how I swing it, it never surpasses 2.2GHz. (It is a 13x166 stock chip). I recently purchased some 2x512 OCZ Platinum PC3200 (rev.1 2-3-2-5) that replaced my single stick of Kingston PC3200. I could bump it up as high as 11x205, but any higher and it would not be prime stable. Once I got the OCZ, I took the multi down to 10.5, and was able to crank the FSB to 212MHz. 213 would fail prime after about 3 hours. Again, the 2.2GHz limit.

Since I'm not going to gain anything in the clock department, I decided to drop the multi and take this ram for a spin. I loosened up the timings to 2.5-3-3 to see if I could find the FSB ceiling. To my surprise, it continued to post all the way up to 237MHz! much higher than I expected to see from the A7N8X-E. 237 caused some nasty hangups, but 235 actually appeared to be somewhat stable. Anything higher than 237 would not post.

Anyway, to make a long story short, I spent quite a bit of time trying to find a good balance between clock speed and FSB. Seems like most of the combinations I tried failed prime after several hours, but 228x9.5 (2175MHz) has been prime stable for over 24 hours now. Looks like I found a winner. Sandra scores were actually significantly higher (I'll post those soon). So far, I've been very pleased with this ram. I dont doubt it could go quite a bit higher :thumbsup:

Still waiting to get my antec RMA back, all this was done on a 300W enermax. I saw some nasty rails when upping the vcore. 12V seemed to raise to almost 13V during full cpu load. Vcore is currently at 1.75 (bumped it for testing), but I'm going to try to lower it back to stock, I think it should be okay at stock vcore.
FSB.jpg 54.7K

Comments

  • paroxymparoxym Toronto, Canada
    edited October 2004
    Thats a pretty impressive bus overclock you've got going there. In fact, I'm more than a tad jealous as Im currently running 210x11 on my NF7 despite my best efforts.:scratch: I had even purchased the 2x256 DC OCZ platinum kit myself some time back. I had the FSB at 430 stable but I couldn't get it any further at the time. I ended returning the sticks and getting a cheaper kit that would handle my current OC and fit my budget at the same time.
  • lemonlimelemonlime Canada Member
    edited October 2004
    Well, I dropped my vcore down a bit, which helped my temps significantly. I'm still running an OEM HSF (Watercooling system in the works). To my surprise, lowering the vcore seems to have increased the stability a bit. I've been priming once again at 230MHz FSB :thumbsup: Almost 24 hours strong. At stock voltage, idle temp is at 41 degrees, load is low 50s. It was approaching 60+ degrees at 1.775V. Memory is still at 2.5-3-3. If I take it any higher than 232MHz, it seems to randomly lock up regardless of temperature. I have a feeling the ceiling on the a7n8x-e is somewhere in the range of 231MHz (stock chipset cooling, vchip etc). It certainly has been fun spicing up my aging socket-A a bit :). I'm still curious to see how high this memory will clock, going to try to run async and find the memory ceiling later.
  • muddocktormuddocktor
    edited October 2004
    If you have a spare 40 mm fan lying around from another mobo nb, it's very easy to modify the nb heatsink on the A7N8X boards for active cooling. All you have to do is bend over the 4 outer heatsink fins about 1/2" from the top where the mounting holes for the fan are and then drill holes theough the bent over section under the fan mounting holes and you have an instant (almost) great active cooling heatsink on your northbridge that will keep it nice and cool. I did that with my old rev 1.04 A7N8X-Dlx board and it worked out great.
  • lemonlimelemonlime Canada Member
    edited November 2004
    Do you think that the NB temperature could be holding me back a bit? It is a bit warm to the touch, but by no means is it really hot.

    I've heard that the NF7-S ceiling is somewhere around 240MHz, and that some of the DFI UltraB boards will do 280+ FSB. What seems to be the limiting factor?. The DFI seems to have a similar passive sink to the A7N8X-E. Unfortunatly, I do not have any v-chip settings in the bios. Could these high FSB values be a result of chipset voltage increases? or some kind of improved voltage regulation on the boards?

    I'm certainly not going to be running out to buy an UltraB :) but I'm wondering if there is anything else that can be done to pump it further and get some increased stability. Maybe I should throw an 80W peltier on the NB :eek: (yeah right), but I've seen some people doing voltmods on A7N8X boards. Does anyone have any good info on VChip increases, and what sort of improvements/degradation comes with it?
  • Cygnus_x_1Cygnus_x_1 New Jersey
    edited November 2004
    i would check for modded bios' for the a7n. i am currently running the nf7, and can crank the fsb up to 250 w/o a sweat (the last time i did it was about a year ago ;) ) i had an a7n8x, a long time ago, and remeber the mem timings to be very picky when pushing the fsb.

    cooling the nb as well as the power mosfets along the socket should help, even if its direct cooling with a 92 or 120 fan, as well as getting your antec back.
  • lemonlimelemonlime Canada Member
    edited December 2004
    Well, to my surprise, my FSB ceiling had some more headroom than I originally thought. After getting my watercooling all sorted out, and some active cooling on the northbridge, I was able obtain a ROCK SOLID 235MHz FSB out of my a7n8x-e. I was actually able to post and boot into windows at 240MHz, but it was not prime stable. I didn't think that active cooling the northbridge would make much of a significant gain, but it got me an ~8MHz boost in FSB at 1:1. I tightened up my memory timings a bit too, and managed 2.5-3-2 timings.

    I have been looking around the net, and scoping out some vchip mods for the a7n8x-e. I am pretty confident that my memory has some more headroom at cas 2.5, and I really want to see if I can break 240 on this board. I must have lucked out and got a really good a7n, as most people I've seen online are usually board/chipset limited at much lower frequencies, and most cannot break 220ish.

    My question is: How much more headroom does vchip increases actually get you? I know the NF7s has adjustable chipset voltage, does this help increase your stable FSB frequencies much?

    If I'm only going to get 2MHz out of some soldering, its probably not worth it.
  • edcentricedcentric near Milwaukee, Wisconsin Icrontian
    edited December 2004
    Take the memory to 4-4-4-11 and the multi down to 6 and see where the mobo bails out. Then you could back off a bit and start tighten things.
  • lemonlimelemonlime Canada Member
    edited December 2004
    edcentric wrote:
    Take the memory to 4-4-4-11 and the multi down to 6 and see where the mobo bails out. Then you could back off a bit and start tighten things.

    I gave it a shot with some very loose timings (3-5-5-11), but it appears it is still chipset limited, and not ram limited. 235 is the best I can do. Anything higher than 238 starts becoming unstable. It's not your average failures either, system hardlocks etc :(
  • edcentricedcentric near Milwaukee, Wisconsin Icrontian
    edited December 2004
    9.5x230 sound nice to me.
    My 3000 will only do 2.3.
    My 2500 will do 2.1
  • lemonlimelemonlime Canada Member
    edited December 2004
    edcentric wrote:
    9.5x230 sound nice to me.
    My 3000 will only do 2.3.
    My 2500 will do 2.1

    The watercooling helped me out tremendously. Previously, my 3000+ would not do anything above 2.2. Now at around 1.8V, I can do 230x10.5 with some headroom to spare. (~2.4GHz), or I can push the FSB a bit, and drop the multi a touch at 10x235. In the end, I'm very pleased with the results, and shouldn't complain :thumbsup:
Sign In or Register to comment.