G.Skill DDR500 showing incorrectly

[NLight95]

I recently purchased 2x1GB of G.Skill Extreme Series DDR500 (PC4000) memory; model F1-4000USU2-2GBHZ.

 

http://www.newegg.com/Product/Product.asp?...N82E16820231021

 

Though it easily runs at DDR500 @ 3-4-4-8, it shows up as 3-5-5-10 in CPU-Z v1.34 or the latest, 1.35. Has anyone else experienced this? I've already seend several reviews of this exact same memory with CPU-Z showing 3-4-4-8.

 

Does this perhaps mean that G.Skill is speed binning DDR400 memory as DDR500, as long as it passes their testing?

 

Here are the screen shots:

 

 

 

[SeanDude05]

If you click on the Memory tab instead of the SPD timings tab, you should see that your settings are actually 3-4-4-8. SPD is just generic factory programmed timings that are used to ensure compatibility in a wide range of systems.

 

And this set of RAM is incredible once you push it.

[Guest GripS]

I remember a g.skill tech saying that the SPD timings were changed for plug and play compatibility. It seems that some motherboards would set the timings too tight by default when the SPD was set to 3-4-4-8. So they switched it to 3-5-5-10 to avoid this issue seeing as how most overclockers set their timings manually they wanted to make sure anyone who bought the modules would have their system boot without issue from the get go. I could be mistaken but i believe i am correct on this. My modules show the same SPD in CPU-Z as well by the way.

[Guest mars]

I recently purchased 2x1GB of G.Skill Extreme Series DDR500 (PC4000) memory; model F1-4000USU2-2GBHZ.

 

http://www.newegg.com/Product/Product.asp?...N82E16820231021

 

Though it easily runs at DDR500 @ 3-4-4-8, it shows up as 3-5-5-10 in CPU-Z v1.34 or the latest, 1.35. Has anyone else experienced this? I've already seend several reviews of this exact same memory with CPU-Z showing 3-4-4-8.

 

Does this perhaps mean that G.Skill is speed binning DDR400 memory as DDR500, as long as it passes their testing?

 

Here are the screen shots:

 

 

 

As other people noted, they want to ensure people can just pop the RAM in and OC to 250MHz HTT 1:1 with CPU without any tweaking. Not a big deal if you don't like these settings, just override in the CMOS...

[NLight95]

I remember a g.skill tech saying that the SPD timings were changed for plug and play compatibility. It seems that some motherboards would set the timings too tight by default when the SPD was set to 3-4-4-8. So they switched it to 3-5-5-10 to avoid this issue seeing as how most overclockers set their timings manually they wanted to make sure anyone who bought the modules would have their system boot without issue from the get go. I could be mistaken but i believe i am correct on this. My modules show the same SPD in CPU-Z as well by the way.

Thanks for the tip. Currently I have the G.Skill @ 3-4-4-8 at 271.9MHz, 1T. I posted my 2748MHz results in the 939 Overclocking Database. It was stable at the settings listed but for some reason now it isn't. Dunno why. The only thing I changed was the NB cooler. I now have an Evercool VC-RE cooler in place of the stock DFI unit. Even used Zalman ZM-STG1 thermal grease. As a result the NB temps are down 5C.

 

Another thing I noticed is that when I set both MaxAsync & ReadPre to AUTO the numbers come up as 11 & 8.5 respectively. How are they calculated? Am I better using the AUTO numbers for stability? I know it takes a hit on latency.

[steelballzz]

T

Another thing I noticed is that when I set both MaxAsync & ReadPre to AUTO the numbers come up as 11 & 8.5 respectively. How are they calculated? Am I better using the AUTO numbers for stability? I know it takes a hit on latency.

you should try whats the best MAL and RP for your rig

testing stability to notice any diference ....pi 32 is a good measurement

[NLight95]

you should try whats the best MAL and RP for your rig

testing stability to notice any diference ....pi 32 is a good measurement

I find SP2004 or Prime95 to be much more thorough, even if it does take much longer to run. There's been more than one time when PI 32 passed easily (even multiple passes) but Prime crashed. I only use PI 32 for quick testing cycles but not for a stability test.

 

Praz brought up a very interesting point--testing with Prime or SP2004 alone doesn't always indicate stability, even if it's run at priority 10. He encourages running other apps while running Prime and I fully agree with him. That is how I achieved my first absolute stable oc of 2700 MHz. I ran Prime at priority 3 for 24hr and at the same time an 8hr DVD encode, Firefox, Nero, Everest, etc. No hiccups at all and the DVD output was flawless. Now that's stability. The LP Ultra-D is truly an amazing piece of engineering.

 

In any regard, I lowered my clock to 2718 for the time being and so far SP2004 has been running for 1hr 22min while I'm running other apps. I'm certainly happy with 2700 but I always like to push just a little more to see where I can go. All this out of an A64 3000.

[SeanDude05]

2700 on a 3000+ is pretty good!

[BadWillHunting]

2700 on a 3000+ is pretty good!

 

Yup... some of the 3000+ venice cores are awesome. I had mine over 2700 with the stock AMD cheapo HSF.

[NLight95]

2700 on a 3000+ is pretty good!

Thanks--it was a lot of bench work to achieve a solid, stable 2700; let alone 2748. I don't think any other mobo could have done as well with the possible exception of an Epox (I've had good luck with them too). The wonderful thing about the Ultra-D is that it provides a 9/10 CPU divider. This allowed me to keep the G.Skill running very fast despite the lack of multipliers for the A64 3000+.

 

I've read here and in other forums where many have had difficulty getting past 2600. AnadTech reached 2700 on an Infinity but just barely, and that was setting the divider to 5/6 (DDR333). I didn't want to do that since I knew the GS could go much higher than its stock DDR500. Once you get past the 2700 "threshold" there's very little return unless you really up the vcore. Then you're talking about employing a phase-change or water cooling solution. Without spending the money on a dual core processor and/or exotic cooling I was able to finally purchase something I've long wanted and needed--a decent widescreen LCD display. Just set up my BenQ FP202W and it's absolutely wonderful. Never, ever thought I'd like an LCD display.

 

Though I've been building PCs for close to 20 years and overclocking for some time--since the Celeron 300A. I profess to be an amateur-intermediate when it comes to the science of it all. People like Happy_Games, Kakaroto, and Praz wow me all the time with their results. We can really be thankful for their (and many others) contributions. Couldn't have done it without their great advice.

[SeanDude05]

Once you have a widescreen LCD, its hard to go back to CRT.