Jump to content

I must be missing something. A little help please?


Battleweary

Recommended Posts

I just upgraded my system. I bought the DFI Lanparty nF4 Ultra-D motherboard with an Athlon 64 3500 Venice core. I'm using the Dual Channel OCZ EL Platinum Series PC3200 Ram that I bought a while back. I followed the OCing guide to the letter, maxed out the processor with all the tests and then clocked it down to crank up the RAM. Did all the tests on the RAM, found the speed I liked, and put the two together and tried to run it. Didn't work so good, so I relaxed the timings on the RAM a bit more and got it into Windows. Tried running SuperPi and it failed after about a minute. Same thing happened with Prime, so I put it all back to stock. Here are the specs I tried, and if I'm doing something wrong, please point it out to me. Thanks.

 

Processor: Got it to 2618MHz(11X238)@ 1.58V

Ran all benchmarks fine, OCCT, SuperPi no problem.

Passed Prime for over ten hours with no errors.

This speed is fine with me, now on to the RAM.

 

Since I was looking to run the RAM at 1:1 ratio, I just started it at 238 MHz.

I set the timings at 2.5-3-3-10 and ran it at 2.93V. It passed memtest for a few hours no problem, so I went on to the other tests. It passed SuperPi, OCCT and the three benchmarks fine. I put it on Prime and it passed for 4 hours, so I figured it was okay.

 

I went back into the bios, set the CPU multiplier back to 11, and put the LTD/FSB divider at 4. This should have given me an HTT of 952, which is almost perfect. I tried booting, and it posts fine, but would reboot instead of going into windows. I relaxed the timings on the RAM to 2.5-4-4-10, and it booted right into windows. I tried running SuperPi at that time and it failed within a few minutes, as did prime.

 

The processor never went over 50C the whole time I was stressing the CPU, and my voltages in everest are rock stable. Am I doing something wrong????:confused:

Share this post


Link to post
Share on other sites

I would try 3-4-4-8 for the ram, and see if that works. That guide is very good and well respected, even in other forums. One thing I have noticed, is that when you put your max cpu speed and max ram speed together, they sometimes do not play friendly. If the relaxed timings don't work, you may have to sacrifice a little somewhere to achieve stability. Don't be afraid to use a memory divider because "CPU mhz is king". You will get better performance in actual games, by switching to a memory divider and keeping your CPU maxed out.

Share this post


Link to post
Share on other sites

Guest Darkslayer
I would try 3-4-4-8 for the ram, and see if that works.

 

Damnit, lol. U stole my line :D j/k

 

I would personally try working towards this:

 

CPU vcore @ 1.475v

HTT link multi @ 4x

CPU multi @ 11x

FSB @ 250mhz

Ratio @ 1:1

Mem timings @ 3-4-4-8 2T (try 1T later)

Mem voltage @ 2.9v

 

I just got finished builing my mom's computer, and was able to get her 3500 to 2.75ghz with little problem. Only difference between the 3500 and 3700 as far as I can tell is that the 3700 can do the same overclock @ 1.4v, where the 3500 needs 1.475v (at least that is how it worked for me ;))

 

Now, don't just go and plug all of that in and call it quits. Thats how you nuke your stuff. That is just what I would strive for while being careful.

 

As far as timings vs speed with the memory, FOR ME, DDR memory speeds makes a bigger difference then low timings. I have found that while testing my 3700 at 2.75ghz, DDR500 3-4-4-8 1T outperforms DDR400 2-3-2-6 1T. It's not a huge difference, but in some of World of Warcraft's large dungeons (with tons of stuff going on) and in Call of Duty 2's big gunfights, having the faster ram can give between 5 and 8 frames per second more then lower latency ram. In 3Dmark2005, I lost 75 points with the slower ram, and in Aquamark, I lost 300 points off my CPU score. Also in first person shooters, mouse movement seemed a tiny little bit smoother with faster ram. But the differences between the two settings are so small, I doubt most people would notice.

 

Sadly I have no screenshots, and do not have the time to run hours of benchmarks in the name of proof. So believe me or not, what I say is the truth from my perspective.

 

(sorry if I have offended followers of the DFI overclocking bible :()

Share this post


Link to post
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
×
×
  • Create New...