Whee!!! I finally did it! Validation Link: http://valid.x86-secret.com/show_oc.php?id=88812 Took me a while but I updated bios to the latest one (NF4LD406.) My goal was to keep it at 2.5-3-3-X So I hope to push this even further or at least try to. Timings are loose and maybe I'll tweak em after I'm done with how far I can get em. I prolly need to drop my multiplier again coz my CPU can kinda only go to 2.75GHz... System: A64 3200+ Venice with Scythe Ninja & 120mm Panaflo High Speed with two 80mm fans blowing over the ram Gskill F1-3200DSU2-1GBLE DDR400 @ 2-2-2-5 DDR550 @ 2.5-3-3-7 (TCCD week 440) DFI NF4 SLI-DR XFX 7900GT (520/1500) Stock OCZ Powerstream 520w XFi Platinum Hitachi Deskstar T7K250 160Gb SATA3.0 Attached below is a shot of the ram. Can't wait to do some 3d benching with this...
nice job. i have had mine stable at 296 but at 300 i would get bosd. that is alot of volts you have going into it.
Thanks. The actual is about 2.8. MBM isn't as accurate as my trusty multimeter. 2.8v should be safe for everyday use. I'm still running prime to see if its good for 24/7 use. Looks good so far... It would boot at 2.7v and run a few benchies. But it wouldn't pass SP32m.
Wow that's sick man. I'm just saving up for some 2x1GB that can do DDR550 to finally satisfy my northwood that can go up to 267FSB.
here is a screenshot of my 296 stable. this was with my gigabyte board and i have not tried it with my abit yet. i just put my money into another 7800gt for a sli setup. http://www.photojunkyard.com/is.php?i=19811&img=2.65_mhz_295_htt.JPG
That damn sweet. Back when I had my Neo2 Plat I could get my 3000+ to 2.8GHz @ 1.7V but not 1:1. Thats about 311HTT. I am not gonna try this now since 1.7V scares me
it is nice to see such a result... but, recommend u try karakato's guide... lowering the voltage of tccd. mayb a memtest windows for 2 days
Wheres this so-called guide by karakato? I wanna do some reading to see if I can get more out of my memory but I probably can't since I don't have a DFI.
From what I've read, different batches of TCCD react differently to voltage. Most of them like it low, around 2.6-2.7 but there are some that give best results even when the VDIMM is cranked up to 3.0V.
So I finally got around to doing some benching with the LEs. I thought it'd be interesting to compare them with my VX. Here's the difference that I saw. All games were run at 1600x1200 Full details. 7900GT is @ stock clocks (520/1500.) VX @ DDR400 & DDR500 = 2-2-2-8 CPU = 2.75GHz (275x10 9:10 divider) LE @ DDR400 = 2-2-2-5 @DDR600 2.5-4-4-8 CPU = 2.7GHz (300x9 1:1 divider) Game benchies are at VX: DDR500 (2-2-2-5) & LE DDR600 (2.5-4-4-8) Everest showed gains in Read and Write performance but the latencies were slower than the UTTs due to timings (duh ). It should be noted however that these are synthetic benchies. There was an increase in FEAR because it is pretty demanding on the memory bandwith. However, there was a slight decrease in Source. Farcry & Doom3 yielded virtually no difference. I didn't bench Oblivion and BF2 coz I can't seem to find timedemos (without using fraps.) However these differences are slight and probably wouldn't be noticeable. However given that they are at high resolutions and not my playing resolutions, it may yield something noticeable at lower resolutions. Basically I get better performance on games that require more bandwith (games like FEAR and BF2.) But it seems latencies do reflect some on Source yet nothing on FarCry, UT2K4, and Doom3. And yes, BF2 feels smoother to me (1280x1024 everything high no AA.)
