• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

David's Learning and Rambling Thread

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

David

Forums Super Moderator
Joined
Feb 20, 2001
This is just a thread to keep track of my tweaking, playing and benching. I'm a complete S775 newb, so this is a learning curve for me. I've also never had a board with as many tweaking options :). I just find it easier to collect my notes and thoughts in a thread, on the offchance that I run into either an issue someone knows their way around, or I manage to overcome an issue someone else has.

So, I've got my benching setup together and booting:
  • Asus Rampage Formula Motherboard (I'll check the Rev.). The BIOS is quite old I think (2008 IIRC), I might flash it to something newer in case it helps.
  • Geil or Micron PC6400 DDR2 RAM, 2 GB in each case, one stick. I also have a pair of 1 GB PC8500 DDR2 sticks on the way.
  • Celeron D 2.8 GHz, Pentium 4 3.0 GHz and Core 2 Duo E6300 (1.86 GHz) CPUs. These are effectively disposable, I'm happy to burn them out and move on. I hope that I can effectively fund my entertainment with the occasional cheap chip from Ebay
  • Corsair CX600 PSU
  • Corsair A70 cooling (if this works out, I'll get a pot this summer)
  • BFG GTX260 graphics (going to replace this with a cheap Quadro NVS 285 I got from Ebay, so I can return the GTX260 to the gaming rig)
  • Seagate 320 GB SATA HDD (old, but can't imagine it'll matter for 2D; I don't plan on tackling PCMark05)

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Sunday 15th January

I had issues with the Celeron. It would boot, but if I changed *any* BIOS setting it crapped it and wouldn't boot until I reset the CMOS again ... it's a 533 MHz FSB chip which may have been the issue.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Tuesday 17th January

The E6300 works. Windows 7 (32 bit) installed. I had a quick go at some tweaking (set FSB strap to 400 MHz). I've submitted CPU Frequencies all the way up to 3.325 GHz (http://hwbot.org/submission/2242780_) with only a little voltage nudge. The RAM is running well out of spec (950 MHz vs. 800 MHz) which I reckon is why I can't get past 475 (480 no boot, even with DDR RAM voltage at 2.1 V and CPU at 1.5 V, and timings loosened as far as they'll go (IIRC). I'm satisfied having hit the average overclock, so I'm hoping that the faster RAM will give me enough headroom to hit closer to 4 GHz and maybe take some boints (air overclocks have hit around 4 GHz on the bot). The 3.325 GHz run validates, but isn't stable enough for UCBench.

I plan to push this as far as I can, validate as high as possible, then work back from there to get the benchmarks to run. The ebay seller informed me that he has a family crisis (his wife has had to go into hospital, I do hope she's OK) so the RAM will be delayed (not a problem for me, this is just me killing evenings). I might pop the P4 in tomorrow :).

There's probably also some room to compare the Micron and Geil, in case I can get a little more out of one than the other.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
Last edited:
Looks good David.

It looks like you may want to get your hands on some better RAM, decent DDR2 can be found for less than $50 in the classies, including some D9 memory which scales really well with voltage (although depending on the motherboard requires a low-voltage stick to be put in there to set the higher voltage, then shut down and put the D9 in).

I highly recommend setting up a google spreadsheet for your benching, I use it for all of my hardware, it allows me to keep track of pretty much everything pretty easily.
 
Think I've hit the limit of the P4 530 - 3.75 GHz boots, validates, runs UCBench. 4.0 GHz struggles to get into Win7 - the RAM isn't what's holding me back, either. 1.65 V doesn't work, 1.7 V gives me an error.
 
Managed to validate at 4GHz, but it took 1.65 v on the cpu and 1.55 on the north bridge. The temps are too high to bench - 80 C. I suspect heat is holding the overclock back.
 
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Saturday 21st January

Some new toys have arrived: a Pentium D 925, and a pair of 1 GB PC8500/DDR1066 DDR2 sticks that should hopefully allow me to push the E6300.

The Pentium 4 is being held back by heat (>80C at 4 GHz) so I also have a small pot on the way, courtesy of a fellow benching team member ... :D

I won't get to play with things until Tuesday, unfortunately.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
You're progressing nicely! One of the quickest transitions to extreme cooling so far :D

Keep up the good work!
 
Nice, David. Having an evaporator makes a world of difference.
 
I'm in it for fun and boints and it looks like I need to go cold to score the later :)
 
Looking at the HWBot CPU Frequency rankings, to score >= 0.5 boints I need:

- 4.7 GHz on the P4-530 (I had 4.05 on air; fastest on air is 5.3, but the next is 4.2)
- 3.6 GHz on the E6300 (I had 3.3 on air; fastest on air is 3.9) - the RAM might net me some boints on air.
- 5.2 on the P4-D-925 (untried on air; fastest on air is 5.1)
 
Yeah, I maybe need to buy easier chips :D

I'll see what I can get - at worst, UCBench is often worth a few boints due to it's new-ness.
 
Haha yeah, I even had a go at reference clock today just for the hell of it and gained a few boints "bonus" lol
 
Nice :) I'm aiming for the best valid. I can get, then back down to highest possible stable, then try and knock through UCBench/wPrime/SuperPI.
 
I found that a clock speed that passed spi,would not pass wprime, and I would spend another few hours trying to find my max wprime clocks, verry frustrating (but good)
Sorry if there's any spelling/grammar mistakes, I'm on my phone at work, in you're neck of the woods (Glasgow) getting my truck unloaded lol
 
I found that a clock speed that passed spi,would not pass wprime, and I would spend another few hours trying to find my max wprime clocks, verry frustrating (but good)
Sorry if there's any spelling/grammar mistakes, I'm on my phone at work, in you're neck of the woods (Glasgow) getting my truck unloaded lol

No problem :) Forums on a phone aren't always easy!

(I used to live in Dennistoun until recently, staying with a friend in Dennyloanhead until I finish my PhD at the end of March and move to St. Andrews. Definitely going to miss Glasgow)
 
It makes sense for a clock to pass spi but not wprime as wprime is multi-threaded and spi is not, thus a lot more stress :)


David, incase you didn't know F7 is the keyboard shortcut in CPU-Z for a quick-validation (that you can then upload the file to the cpuid website). it helps a lot when you are creeping on the edge of stability.
 
Managed to get the C2D E6300 to valid. at 3.38 GHz, which is a slight improvement. I can't for the life of me get past 483 MHz FSB - I *once* got Windows to boot to desktop then it BSOD'd. Then wouldn't even boot. I've tried upping Vcore, CPU PLL, NB and SB voltages (to the highest 'green' setting, except for vcore which I've had up at 1.6 V!) and it wont budge. I suspect I might need the DICE to push it further?

The 925 is currently at 4200 MHz and stable, running benchmarks as we don't have any for the 925 currently. Then I'll try and push as close to 4.5 GHz as I can get before heat kills my game.
 
Back