UPDATE: At 2.7V, passed MemTest86 @ cas2227, 386ddr. However, at normal 2.6V, it can only pass cas 2337 at 384 and below.
Last night I finally got my two 512-megger Corsair XMS3500cas2 sticks. As many of you know, Thugsrook and others had no success running previous (xms3000 and 3200) sticks in pairs with 3:4 ratio, using the popular P4B533 mobo. Even 355mhz (133fsb) was impossible before with 1GB RAM.... well, not any longer!
I have officially passed MemTest86 eleven times in a row (before I ended the test) at 392mhz with cas 2.5-3-3-7 (using Asus default 2.6V ddr voltage). Specs: 2.26B @ 2.5ghz/1.50Vcore, Asus P4B533, 147fsb:392ddr using Alpha PAL8942, AS3, and a near-silent Panaflo L1A @ 7volts. (pic with opened cover) I will test higher, but this is my normal speed.
Unfortunately, stricter timings failed at 392mhz: both cas2336 and cas 2327 caused errors at 65% in MemTest86. My testing (inconclusively) suggests that cas 2327 should be possible in the 380's, because the errors were minimal by the time I got down to 392. I also ran using 1:1 ratio, at 147fsb:294ddr with cas2225 without problems, but I haven't tested it more thoroughly. Also, I tried 133fsb:355ddr with cas 2226 and it seemed to work. However, these are yet unconfirmed--they worked through the first 30 sec of MemTest86 and worked in Windows, but they may not be reliable under further testing.
My conclusion: if you want 1GB of fast DDR memory, this 5ns memory is IT. If I understand previous results correctly, this two-stick pair beats dual-XMS3200 or dual-XMS3000 by a long shot. It won't run as high as my single XMS3200c2, but it comes fairly close. Close enough that I'm only losing maybe 1-3% performance from the higher CAS settings. Once Granite Bay comes around, it will be time to bottom out those CAS settings!
Leo
Last night I finally got my two 512-megger Corsair XMS3500cas2 sticks. As many of you know, Thugsrook and others had no success running previous (xms3000 and 3200) sticks in pairs with 3:4 ratio, using the popular P4B533 mobo. Even 355mhz (133fsb) was impossible before with 1GB RAM.... well, not any longer!
I have officially passed MemTest86 eleven times in a row (before I ended the test) at 392mhz with cas 2.5-3-3-7 (using Asus default 2.6V ddr voltage). Specs: 2.26B @ 2.5ghz/1.50Vcore, Asus P4B533, 147fsb:392ddr using Alpha PAL8942, AS3, and a near-silent Panaflo L1A @ 7volts. (pic with opened cover) I will test higher, but this is my normal speed.
Unfortunately, stricter timings failed at 392mhz: both cas2336 and cas 2327 caused errors at 65% in MemTest86. My testing (inconclusively) suggests that cas 2327 should be possible in the 380's, because the errors were minimal by the time I got down to 392. I also ran using 1:1 ratio, at 147fsb:294ddr with cas2225 without problems, but I haven't tested it more thoroughly. Also, I tried 133fsb:355ddr with cas 2226 and it seemed to work. However, these are yet unconfirmed--they worked through the first 30 sec of MemTest86 and worked in Windows, but they may not be reliable under further testing.
My conclusion: if you want 1GB of fast DDR memory, this 5ns memory is IT. If I understand previous results correctly, this two-stick pair beats dual-XMS3200 or dual-XMS3000 by a long shot. It won't run as high as my single XMS3200c2, but it comes fairly close. Close enough that I'm only losing maybe 1-3% performance from the higher CAS settings. Once Granite Bay comes around, it will be time to bottom out those CAS settings!
Leo
Facebook
Twitter