Jump to content

Jimbo Mahoney

Members
  • Posts

    143
  • Joined

  • Last visited

Jimbo Mahoney's Achievements

Newbie

Newbie (1/14)

  1. Added a yellow graph for my new set of 2GB OCZ, although I'm using a set of G.Skill ZX right now.
  2. Another vote for OCZ Powerstream. I'm using mine to overvolt the 3.3v line and drive my BH5 @ 3.4v using the jumper trick.
  3. Quoted for truth. IMO, the only place you want to worry about is the pump suction side. You want that immediate area as free as possible, ideally a res.
  4. I guess everyone is sick to death of tREF, huh? :shake: Man, I did graphs and everything! :sad:
  5. Agreed. I've often been running BH5 with a 140 divider. It's often the CPU's memory controller that gives up first when clocking. Don't be afraid to drop the memory so it's around 200 Mhz. Dropping memory clock won't really hurt, but dropping CPU speed for memory speed will...
  6. Looking astaris' post here: http://www.dfi-street.com/forum/showpost.p...38&postcount=27 It seems that my best tREFs all fall in his 1.95us category. Namely: 1,95 us 1536= 100mhz -> 195 Mem Clock cycles 2048= 133mhz -> 259 2560= 166mhz -> 324 3072= 200mhz -> 390 1.95 us 3684= 120mhz -> 234 4196= 140mhz -> 273 4708= 150mhz -> 292
  7. Agreed. I've tuned my Powerstream using a cheap MM and a molex for the 5v and 12v. I'm overvolting my 3.3v to 3.6v and running my RAM off the 3.3v line.
  8. Hey guys, I know tREF gets done to death, but I've been doing some tests after reading the numerous posts, including astaris', Sharp's and Dracula's. Here are my results with my BH5 @ 250 Mhz, 200 divider. At this speed it gives enough errors to give an idea, but not so many that it's ridiculous: Blue line is best-fit of the number of errors. Pink line is best-fit of bandwidth (just from memtest86) My questions are: If tREF in the BIOS is divided into groups of refresh rate (e.g. 100/120/133/140/150/166/180/200 3.9us), why is there such a clear correlation between tREF and # errors? Does each set of memory have its own ideal tREF, or will the ideal tREF vary with clockspeed and / or divider? I'm still struggling with Dracula's tables, and I've always been fond of the theory of: best tREF = refresh period / (1/clockspeed)
  9. The only thing I've had as a result of a low tRAS was on my old Abit NF7 (NF2) board. I corrupted the FAT of the hard disk because I was using high FSBs + low tRAS. I always use the (correct I believe) following: tRAS = CAS + tRCD + 2 so with my 2-2-2 timings, I use tRAS = 6 I haven't found tRAS to affect stability, so I treat it as a timing which should be set correctly, rather than lower = better or higher=better.
  10. Things which have worked for me in the past: Swapping RAM to / from yellow / orange slots (currently in orange). Using a lower LDT Multi (like 2x) Turning DEP off in XP SP2 (change boot.ini from /noexecute=optin to /execute) Has your Corsair been OK before? I've only heard negative stuff about DFI + Corsair. Which BIOS are you using?
  11. I've removed the MOSFET sinks on my DFI ULtra D and whacked some watercooling on there.
  12. Just for people's info, I believe the lower slot loses a couple of hundred points in 3DMark '05. So if you're benching, DON'T use the lower slot, but for games (as someone has already mentioned) it will be fine.
  13. Hmm, well if it's an early chip, you might be in with a shot @ 2.8 Ghz on air then. Good luck! Keep us posted!
  14. You will be REALLY lucky if you hit 2.8 Ghz on air dude. What cooling does your friend have to get 3.1 Ghz out of a Venice 3000 and have you seen it with your own eyes? I am just this minute working on clocking a 3000 venice, but it is an exceptional chip and is doing 3Ghz on water. It did 3.2 Ghz on phase. This is an early chip (week 12) and I've had three modern chips (like week 24 - 38) and they have all clocked to about 2.5 Ghz, so I would aim around there. You MIGHT have an exceptional chip, but they tended to be the early ones. Hope that helps?
  15. Hi guys, I've been running BH5 in the yellow slots for a while now and have been having poor success with a CPU I KNOW is excellent. (Week 12 Venice 3000 DPBW - I had one before and it did 2.9 Ghz prime stable). I've just changed the RAM to the orange slots, and what do you know, I can boot and Prime @ 2.7 Ghz default volts.... (I couldn't even get into windows with the RAM in the yellow slots). Therfore, my question is, what's the difference between the yellow and orange slots? Trace length? That's the only thing I can think of. Any other theories? Thanks!
×
×
  • Create New...