interesting founding of my kingston pc4000 2x512.

[text]

interesting finding of my kingston pc4000 2x512.

For some reason it doesn't like me to increase the voltage into the ram. I am currently running the rams @263mhz 2.5-4-4-7. At 2.6v it will run for hours without any problems but when I first try to increased to 2.8v in the beginning memtest give me errors. Another interesting finding to me is the latency. I cannot make my ram even to go at 264 with a 3-4-4-8 timing. So I guess the beginning and the end number doesn't matter much? Any comments or ideas why this would happens?

 

[disk11]

What ICs do you have? It should say right on the stick.

Generally, most of the pc4000+ rated chips don't like too much extra voltage.

Have you done any modding to the board? I believe that Abit's Intel boards have a problem with providing the correct northbridge voltage or vtt. Check this out and see if that is the issue and not your RAM.

 

[Reefa_Madness]

If these have the Hynix chips then that would be consistent with some of the stuff guys have been posting. It seems that the newer bt-d43 just doesn't respond to increased voltage. At 263, I believe that you're right in the range of what that ram will do.

 

[text]

I have not yet to do the voltage modification due to the reasons that I don't know much about it and maybe afraid that I cannot do it myself. My sticks have heatspreader on them and removing them will void the warranty and beside it is hard to put it back on. So what about the change in latency? How is 2.5-4-4-7 better than 3-4-4-8? I ran sandra memory bandwidth test and it seems the same to me.

 

[Reefa_Madness]

With an Intel rig running 250 plus, CAS 2.5 vs 3.0 will make a performance difference, however dropping those "4"s down to "3"s and "2"s, will have more impact (see OCZ's whitepaper on the Enhanced Bandwidth memory or AnandTech's review of the EB ram for a detail explanation), as well as the "8" closer to the "5-6" range. You will be hard pressed to get them that low, but that would be your "optimal" setting. For example, that is why bh-5, running 2-2-2-5/6 at 250 with a DDR Booster, is so awsome (see eva2000's thread on this). With these faster timings you should not only experience increased bandwidth, but the feeling of a quicker response from your system, because, that is exactly what is happening, the memory is taking less time to perform its operations. The link below is to the AnandTech article, should you like to do a little more reading on the topic.

http://www.anandtech.com/memory/showdoc.aspx?i=2019

 

[flapperhead]

probably 99% chance of hynix. although usually it responds up to 2.8/3.0 volts.. my old simpletech nitro 4000 is rated for 2.6v.. looked just like adata tho

 

[text]

thank you Reefa_Madness for the interesting article. Look like my hynix is in the 1%. I tried lowering one of my 4s to a 3 but it is not stable so I just keep it at 4. I noticed that lowering one of the 4s to a 3 the speed in memtest increase from 2988/s to 3198/s so that's why it is unstable.

 

[Quattro]

I think these are what I have.
So they can handle high frequencies then...!

Good. But those timings are crap.
But I shouldn't think that matters.