ScYcS Posted March 21, 2005 Posted March 21, 2005 real life applications like 3d mark That actually made me laugh out loud..... Share this post Link to post Share on other sites More sharing options...
RGone Posted March 21, 2005 Posted March 21, 2005 Tref calculate. Address bits Refresh Cycles tREF Auto-refresh interval11 row 2^11 = 2048 = 2K 32ms 32ms / 2048 = 15.6 us 12 2^12 = 4096 = 4K 64ms 64ms / 4096 = 15.6 us Now sharp go delete that other post to unclog the waves. RGone... :confused: as usual. Share this post Link to post Share on other sites More sharing options...
Sharp Posted March 21, 2005 Posted March 21, 2005 But how does that relate to the frequency? I have deleted my other post. Share this post Link to post Share on other sites More sharing options...
Sharp Posted March 22, 2005 Posted March 22, 2005 I have been playing around with the settings and this is what I am able to run at. Everything at 200Mhz(400DDR) Command per clock...............Enabled (performance setting) Cas latency...........................2 RAS to CAS delay..................2 Min RAS active time...............6 Row precharge time..............2 Row cycle time......................7 (lowest value possible in the BIOS) Row refresh cycle time..........9 (lowest value possible in the BIOS) Row to Row delay.................0 (lowest value possible in the BIOS) Write recovery time...............2 (lowest value possible in the BIOS) Write to read delay...............1 (lowest value possible in the BIOS) read to write delay...............1 (lowest value possible in the BIOS) Refresh period.......................Auto Wrtie CAS latency..................1 (lowest value possible in the BIOS) DRAM bank interleave.......... .Enabled (performance setting) Everything I said works, I am able to run at the lowest settings possible. And by using these settings my scores have increased by quite a bit. EVEREST (they seem to vary a bit lower/higher sometimes) 5830Mb/s 2455Mb/s 43.7us Usually ~44us Memtest86+ 2437Mb/s I am very pleased with the results Share this post Link to post Share on other sites More sharing options...
sesiom Posted March 29, 2005 Posted March 29, 2005 I have been playing around with the settings and this is what I am able to run at. Everything at 200Mhz(400DDR) Command per clock...............Enabled (performance setting) Cas latency...........................2 RAS to CAS delay..................2 Min RAS active time...............6 Row precharge time..............2 Row cycle time......................7 (lowest value possible in the BIOS) Row refresh cycle time..........9 (lowest value possible in the BIOS) Row to Row delay.................0 (lowest value possible in the BIOS) Write recovery time...............2 (lowest value possible in the BIOS) Write to read delay...............1 (lowest value possible in the BIOS) read to write delay...............1 (lowest value possible in the BIOS) Refresh period.......................Auto Wrtie CAS latency..................1 (lowest value possible in the BIOS) DRAM bank interleave.......... .Enabled (performance setting) Everything I said works, I am able to run at the lowest settings possible. And by using these settings my scores have increased by quite a bit. EVEREST (they seem to vary a bit lower/higher sometimes) 5830Mb/s 2455Mb/s 43.7us Usually ~44us Memtest86+ 2437Mb/s I am very pleased with the results Hi Sharp!! thanks! those settings increased a lot my Ballistix's PC3200 performance @stock speeds!. I´m really happy with them EVEREST 5908 2538 45.7 memtest and prime95 with 0 errors in 3 hours. On everest "SPD information" and on Sandra´s memory info I´ve noticed that my memory is listed as PC4000 Micron. Strange! It happens the same with you? Share this post Link to post Share on other sites More sharing options...
Sharp Posted March 29, 2005 Posted March 29, 2005 Hello, I didn't think anyone would look at this post. In Everest/sandra my says PC3200. You know that some PC4000 modules that dont pass the test for PC4000 get put down as PC3200. Or you got sent PC4000 modules, what does the sticker say on it? I am still testing the bottom settings just incase you wondered why I did not post it. What sisoft sandra mem scores do you get? Share this post Link to post Share on other sites More sharing options...
sesiom Posted March 29, 2005 Posted March 29, 2005 You know that some PC4000 modules that dont pass the test for PC4000 get put down as PC3200. Or you got sent PC4000 modules, what does the sticker say on it? I am still testing the bottom settings just incase you wondered why I did not post it. What sisoft sandra mem scores do you get? Hello. The sticks were sent as PC3200 (I pay'ed them as PC3200)!! They came in 2 brand new littke boxes that were sealed from Crucial - very shining 2 sticks with the same reference: CL1114J.79 BL6464Z402.16TG Didn´t OC yet... but with those settings I've got 5940/5880 in sandra mem score. Maybe my lucky buy or bugs with SPD's info on everest & sandra lastest versions. :confused: Share this post Link to post Share on other sites More sharing options...
Sharp Posted March 29, 2005 Posted March 29, 2005 With my CPU I get 5400/5320 the advantage of 400Mhz seems to be quite impressive. (my CPU 1.8Ghz yours is 2.2Ghz). I am testing these settings, they seem to offer a tiny bit more performance. DQS Skew Control.............................Decrease Skew DQS Skew Value................................255 DRAM Drive Strength..........................Level 8 DRAM Data Drive Strength..................Level 1 Max Async Latency.............................5ns Read Preamble Time...........................5ns Idle Cycle Limit...................................16 clocks Dynamic Counter................................Disable R/W Queue Bypass.............................AUTO Bypass Max.........................................AUTO 32 Byte Granularity.............................Disable(8 Bursts) Thanks for posting your results Also make sure you specify your CPUs FSB Ratio (x11). If you do not your hypertransport will be running at 800Mhz instead of 1000Mhz. http://www.amd.com/us-en/Processors/Techni...71_9706,00.html Use that to see it. Share this post Link to post Share on other sites More sharing options...
sesiom Posted March 29, 2005 Posted March 29, 2005 I'll google another software program that shows SPD info. Thanks for the reply and for posting your results. I'll see the hypertransport issue. I think CPUs FSB Ratio setting is AUTO. Thanks again! Share this post Link to post Share on other sites More sharing options...
Radical_53 Posted March 29, 2005 Posted March 29, 2005 If you ask for such tight timings, may I ask why you want it that way? With the FPS a 3000+ gives you on stock, it ain't even worth thinking about it. A friend of mine tested the difference between TCCD 2.5-3-3-7 and BH-5 2-2-2-5. Guess what? With his FX-55 at 3GHz and his 6800Ultra-SLI (running at 450/1200), the difference was roughly 1-2%! So, do you really think that with your setup, there'll be any difference to be measured? If it's 2-2-2-5 anyhow? Share this post Link to post Share on other sites More sharing options...
Sharp Posted March 29, 2005 Posted March 29, 2005 Hello, All settings on Auto and 2-2-2-6 5684Mb/s 1783Mb/s 49.5ns memtest 2018Mb/s Now compare the results. So why would I want to leave it on AUTO? Share this post Link to post Share on other sites More sharing options...
Radical_53 Posted March 29, 2005 Posted March 29, 2005 EVEREST (they seem to vary a bit lower/higher sometimes)5830Mb/s 2455Mb/s 43.7us Usually ~44us Memtest86+ 2437Mb/s All settings on Auto and 2-2-2-6 5684Mb/s - 2.6% 1783Mb/s - 37.7% 49.5ns + 13.3% memtest 2018Mb/s So? An Increase on memory benchies, wow... it won't give you any more performance on any other bench, so why bother? With some moderately good TCCD you'll get ~7500-8000 mb/s on Sandra... Share this post Link to post Share on other sites More sharing options...
Recommended Posts
Please sign in to comment
You will be able to leave a comment after signing in
Sign In Now