3000+ Overclocking Woes!

[Guest Lukano]

I apologize in advance if I forget to mention anything vital in the first post, and on the same note I don't have any screenshots or snapshots of the bios handy (writing this from work right now).

 

My machine specs are as follows;

 

AMD Athlon64 3000+ Venice

OCZ PC3200 Gold VX Rated 2-3-3-8 @ 2.6-2.8v & 2-2-2-8 @ 3.2v

DFI Lanparty Ultra-D

Forton AX500 500w Power Supply.

 

I've been posting on both the Somethingawful forums as well as HardForums in an attempt to track down my issues, and I still can't seem to figure it out.

 

We'll start with the ram, which I had originally assumed was the problem. I had tried it at a variety of settings, variety of dividers, and a variety of voltages. End result was I can do 2-2-2-8 @ 3.2v 1:1 w/ 200HTT and run in Memtest for 12+ hrs without a single error.

 

As of last night (I think a bit of burn-in over the past few days got me to this point, which is nice) I can also do 2-2-2-8 @ 3.2v 2:3 w/ 300 HTT (end result ram running at 192mhz) for 12+ hrs without a single error in Memtest.

 

So I think it's a pretty safe assumption that it's not my ram. It's running at nice and tight timings at reasonable voltage without any memtest hiccups.

 

So I restart and boot into windows. I had my LDT set at 1.5v and my Chipset at 1.8v - both due to previous testing and I had forgotten to change them back. I'll also note here that I've tried every step of Chipset and LDT voltage now, and it does not appear to change the end result (Which I'll get to in a second). But from my reasoning, and the fact that I've tried varying Chipset voltages we can assume it's not the motherboard - correct?

 

So now to the meat of the matter. I'm able to boot into windows fully, with a stable desktop with no problems. I can run through 3dmark2005 without a hitch, but haven't tried any other 3dmark benchmarks (nor have I tried games yet).

 

BUT - When I fire up Prime95 and run it through blended, in-place ftt, or small fft stress tests I'm l lucky if it'll go 2 mins without error.

 

Now at this point, I've got my ram divider still at 2:3 (133) resulting in 192mhz ram @ 2-2-2-8. CPU Multiplier @ 9, HTT Multi @ 3 (to keep it sub 1000mhz as per other peoples recommendations). CPU Voltage has varied between 1.50 and 1.77 depending on what VID special control I apply - and it does not seem to affect it as long as it's enough juice to boot.

 

So I got frustrated at the desktop stability but not prime stability of 300x9 with the above mentioned settings (again, ran 12+ hrs memtest without a hitch) and upgraded prime95 to v24.11. Also no help there, reduced some level of errors (it was erroring faster with the 23.xx version which apprently doesn't play nice with new(er) A64's) but it still errors out within 2mins of in-place ftt or any other test.

 

So I restarted and without touching anything else (same settings as above) dropped HTT to 290. Then to 280. Then to 275. Then finally to 260. Some (I can't recall which as I was doing it rather quickly) wouldn't boot problem without PFI BSOD errors - and others would boot into a stable desktop again with the same problem.

 

So where can I start digging from here? Any idea's on what might be causing it? I know that I may just have an unlucky chip that doesn't like being pushed as hard as I'm pushing it - but it may also be something super silly that I've simply forgotten to change or check.

 

Oh - and on a final note - I've tried these same settings with the 414-1/2 and 510-1/2 bios's (including the Hellfire 510-2 with 310 memory tables).

 

Any idea's?!

 

edit - I've just been reading the stickied thread, especially Angry's post about HTT (and RGone's as well) being fine between 1200 and 1800. I was under the impression it absolutely MUST stay below 1000 - but if that's the case and I have it too low, it very well may be my problem. I'll try this tonight. Still open to any other suggestions.

[Randallw23]

First flash to the 510-2 bios and before you restart the computer make sure you reset the cmos jumper. Then starting from scratch keep the FSB at 200 and put the DRAM voltage jumper on and bump your ram voltage to 3.4V. Then start overclocking with the ram at that voltage. That should solve your issues. If 3.4V gives errors put it at 3.6. Let me know how it goes

[Guest Lukano]

First flash to the 510-2 bios and before you restart the computer make sure you reset the cmos jumper. Then starting from scratch keep the FSB at 200 and put the DRAM voltage jumper on and bump your ram voltage to 3.4V. Then start overclocking with the ram at that voltage. That should solve your issues. If 3.4V gives errors put it at 3.6. Let me know how it goes

 

Ram's currently at 3.2 - but I've had it as high as 3.4 with no gain. Currently running 510-2 Hellfire but I've also tried 510-2 Oskar Wu, 510-1 Oskar Wu & 414-1 as well as 414.2. Each time, I would load default, unplug, hit power a few times, clear cmos, hit power a few more times, reload & load defaults again. Then restart one last time and start configuring.

 

Each bios had varying results, but nothing better than what I've got now with 510-2 (Hellfire or Oskar Wu, doesn't matter).

 

I'm also wary of pushing higher than 3.4-3.5 as OCZ only warranties this ram up to 3.5 - and I haven't a clue what risk I pose to it by over-volting it like that (and if it does kill it, will they be able to read something in/on the chip to see I've pushed higher than 3.5v?)

[Randallw23]

that is odd you can't get above 200 on that ram. I currently have my Bh5 Geil at 3.6v 253mhz 2-2-2-7. I would try having no divider and bumping the ram voltage to 3.6 then seeing if i could overclock it at all above 200. I'm not sure whether or not they can tell, but you could always set it to 3.6v and if something happens say you had your motherboard set at 3.5 ( a small fib never hurt anyone....)

[Guest Lukano]

that is odd you can't get above 200 on that ram. I currently have my Bh5 Geil at 3.6v 253mhz 2-2-2-7. I would try having no divider and bumping the ram voltage to 3.6 then seeing if i could overclock it at all above 200. I'm not sure whether or not they can tell, but you could always set it to 3.6v and if something happens say you had your motherboard set at 3.5 ( a small fib never hurt anyone....)

 

To be honest I haven't really put much effort into pushing the ram past the current 192mhz until I can get it prime stable. I was assuming (perhaps erronously) that slower ram really wasn't going to do much than bottleneck things - but not really cause any errors like what I'm seeing. I have tried different ram dividers (140 and 150 both give PFI error BSOD's on boot into windows) and 166 won't even load windows. Maybe dropping my timing will help - but again, I really can't see it being the ram causing the issues due to speed alone (and it's been put through it's rigors on memtest with no issuess at 2:3 300 HTT).

 

edit - btw - what are your temps at 275 and at 300? I've not seen it push over 52C at 300 HTT.

[Zelig]

You'd do better if instead of starting with the cpu MHz high, you start low.

Raise the HTT up from 200 MHz until Prime starts failing, then give a little more voltage and see if it gets you anywhere.