mnemennth Posted January 7, 2008 Share Posted January 7, 2008 I've been looking around the forum here for an answer, & I've viewed your presentation on memory. It's good, I understood it, and it reiterates everything I've learned in school, but it's pretty general. My interests in RAM timing are a bit more specific, particularly Command Rate. The machine I'm working on will be used as for two purposes: Serving video across my LAN to my two ReplayTV units, & Recoding DVDs for archive. As a result I'm not terribly concerned with frame rates & video bandwidth (I could keep using the 10 yr old MatroxPCI video card I have in it now), but I AM concerned with getting that video onto & off my hard drives & across my home LAN smoothly, even if I happen to be recoding a disk at the time. I've been told that for this kind of data crunching, dropping my Command Rate from its 2T default down to 1T is desirable; but I don't know how to properly test if it functions allright after forcing this mode. My MB is Socket 754/AMD64 (Nice & cheap - all my money went into the drives); as such it will only support RAM in Single-channel mode at best & even then I have to drop the memory speed down to 333MHZ (Limitation of the memory controller). I've been burn testing 48 hours with Memtest86+ with no errors at 2.5-3-3-7-2T - Corsair spec is 2.5-3-3-8-2T by the configurator, but that is where my MB set it at after I forced 333MHZ speed. At this point I believe there is enough leeway thanks to underclocking the RAM a bit that I might be able to tighten up some timing; I've forced Command Rate to 1T & commenced burn testing again, but at this point I have to ask a few questions: Will MemTest86+ actually test the RAM in such a fashion that errors caused by changing the Command Rate will show? Do I have to load the OS (I'll be using XP ProSP2 unless something better suited comes along before it's built) & attempt to abuse the system with video recoding or some other application to properly test it? Is this the best change I can make, or should I try other parameters first, & if so, where do I start for my kind of grinding long-haul application? Any enlightenment you could shed would be greatly appreciated, mnem - Tinkerer-At-Large Link to comment Share on other sites More sharing options...
Corsair Employees RAM GUY Posted January 8, 2008 Corsair Employees Share Posted January 8, 2008 Command Rate is not really a memory setting and would be dependent on the make and model of MB and PSU and the specific CPU you have installed. You might look this setting up on Techarp and get a better explanation. I would suggest setting the command Rate to 2t on that platform, there were only a few CPU's that would run 1t CR as I remember about socket 754 Mb's and for sure if you have more than one module installed 2t Command Rate would be needed as well. Link to comment Share on other sites More sharing options...
mnemennth Posted January 8, 2008 Author Share Posted January 8, 2008 Yeah, I guessed that after reviewing the memory architecture segment on your presentation; how can it execute full command in 1 cycle when it can only access half the RAM in that time? Besides... I kindof got an answer from MemTest86 - while it did run for over 20 hours in 1T setting with no posted errors - the MemTest application itself showed signs of corruption in the form of gibberish characters in the output screen itself. I've reset back to defaults & will update when I have some answers on performance. Thanks, mnem Link to comment Share on other sites More sharing options...
Corsair Employees RAM GUY Posted January 8, 2008 Corsair Employees Share Posted January 8, 2008 NP Please let me know if you have any more questions! Link to comment Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.