Quote:
Originally Posted by Ubermann
corrupt is using an old C0 chip at 300+ mhz.
Printable View
Quote:
Originally Posted by Ubermann
corrupt is using an old C0 chip at 300+ mhz.
Shadco, have you tried running cas2-2-2-5 with your OCZ PC3700 EL with your OCZ booster installed yet? If you get a chance do so and let us know your results. I installed my booster again last nite and tried throwing 3.45v thru them and did nothing for me in terms of running tight timings above 250htt.Quote:
Originally Posted by shadco
Gonna throw my 3200+ back in and see if it will allow my to run those timings compared to my 3200+DTR.
Just a update seems my 431 chips do react to voltage, though not real tight I'm able to run cas2-3-3-5-1t 1:1 280htt at 3.4v but only sandra proof at 270htt.
Not great results but compared to only cas2 at 250htt with anything below 3v isn't bad. Maybe the new Powerbrain PCB is the real difference why some might run 2-2-2-? above 250htt with voltage.
http://home.comcast.net/~phippy1/voltage.bmp
http://home.comcast.net/~phippy1/voltage2.bmp
so a gig of this stuff will do well in a UT?
Timeline of Events:
11/10/04 (1:30am - 4:00am)
---------------------------
- Flashed BIOS to 10/15 Beta 4V DIMM
- Increased voltage to 3.3V (3.3V rail has not yet been modified)
- Boot, test, boot some more, test some more
- Highest memory frequency with one 512MB module: 310MHz @ 2.5-3-3-10 (Drive Strength=1)
- Fine-tune using A64 Tweaker
- Implement final settings displayed on A64 Tweaker in BIOS
- Memtest Test #5 looped for 10 runs without errors
- Sandra memory benchmark reveals 5050MB/s average on several runs
- Desire to push for more kicks in
- Adjusted CPU voltage to 1.9V
- Boot, test, boot some more, test some more
- Highest memory frequency with one 512MB module: 315MHz @ 2.5-3-3-10 (Drive Strength=1)
- Greed kicks in, risks are ignored
- Adjusted CPU voltage to 2.1 placing complete confidence in 92mm Vantec Tornado
- Highest memory frequency with one 512MB module: 330MHz @ 2.5-3-3-10 (Drive Strength=2) :slobber:
- Run Prime95 for several hours
11/10/04 (9:00am - 1:30pm)
---------------------------
- Blue screen of death
- Make adjustments in BIOS
- No luck
- Back to the basics in BIOS with cleared CMOS, Auto settings, and 10x200MHz CPU
- No luck
- Install memory modules onto wife's rig. All systems are go! Thank God it's not the RAM!
- Replace original DFI board with brand new DFI board received last week
- No luck
- Some more fiddling coupled with a few prayers and profane words in between
- No luck
Conclusion: CPU murdered with 2.1V on high-CFM air cooling.
Summary: I have a gut feeling that my OCZ modules will continue to scale well with increased voltage, granted the 3.3V rail will supply the juice and the memory controller is up to handling the task. I'm ordering another 754 on Friday from Newegg and will exercise a little bit more caution next week. Hopefully, the new chip will have a much better memory controller than the one I just killed.
In closing, I'm sure there are still a good number of people that haven't taken the plunge into the 64-bit world of processors. If and when you decide to, it would be a much wiser choice to overlook the 754 and move on to the 939 if you are looking to push your memory to its limits. I've said it in another post and I'll say it again: the memory controller on the 754 is a crippling force that will limit your memory overclocks, especially when running more than one DIMM. Some people get lucky with their C0 chips, others rave about the CG revisions. My CG couldn't put out the juice, even with 2x256MB G.Skill modules. We all know that overclocking and luck go hand in hand...more so when it comes to 754.
Wish my next CPU luck. Either way, 939 is already in my immediate future.
Cheers! :toast:
Just got this ram about 20 mins ago, and although I'm not nearly finished tweaking it yet, you can see that this ram is phenominal. My insanely CO core is no doubt screwing me, but as I'll be upgrading soon, I'm not too worried about that. I can get 290, but it's really not stable at all. I'll have results from the second DIMM after I'm done playing with this one.
More vDimm would help loads - I'm running official 10/15 bios (3.1vDimm with unmodded rail)
http://www.imgbucket.com/4/285.jpg
instead of sandra which doesn't do much for comparison of timings, can we see superpi or 3dm?
Sandra doesn't do much for comparison of timings? I find it works well for me and it's probably a lot better than SuperPi or 3DMark. SuperPi times are rounded to the nearest second. If you tweaked up another 10MHz, there's no way you're going to see it in SuperPi unless you do one of the big runs. 3DMark2001 scores vary from one run to another, even with the same exact settings.Quote:
Originally Posted by Mag Master 21
When it comes to memory benchmarks and comparing one tweak to the next, I think Sandra and Everest take the cake.
DaGoof
Flash to the 4V BIOS and push that rail to 3.3V. ;)
I'm gonna flash right now, just hope my crappy PSU doesn't fail on me :thumbsup:
I doubt this thing can do 3dmark, and I was getting better superpi scores when i had a 9x multi x 272 fsb with slightly tighter timings.
Last night, I was able to run in the 270MHz neighborhood at 2-3-3-7. If CAS2.5 gives you grief in Memtest in the mid-ranges, try CAS2.
Good luck with your testing...and try not to fry anything. ;)
I'm gonna try single stick first because I think I'm up against the memory controller in the CPU.Quote:
Originally Posted by RacerX
The ram boosters are prolly gonna find a home in the P4 boxes which are using PC3700 Gold Rev 2 chips and I know they can benefit from more vdimm.
My max cpu speed is 2450mhz, so I had to limit myself to 272fsbx9... And to be honest, that was a lucky SuperPi run - I usually get 35s, Time to push her back to 290fbs, and see how stable she is with the new bios.
http://www.imgbucket.com/83/2721.jpg
http://www.imgbucket.com/69/2722.jpg
DaGoof
What are your timings at for that 272MHz run?
If the data on my hard drive isn't corrupted from my suicide run, I'll post up my last Sandra screenshot. It was very :slobber::slobber::slobber:, to say the least.
FSB: 272
Multi: 9x
HTT: 3x
VDimm: 3.1v
Chipset: 1.8v
ratio: 1:1
cas: 2.5
trcd: 3
tras: 5
trp: 3
trc: 7
trfc: 14
trrd: 00
twr: 2
twtr: 1
trwt: 2
tref: 4708
twcl: 1
DQS: Increase Skew
Skew Value: 64
DRAM Drive: Level 2
all the rest are auto
Thanks for the tip on the 939. I'm still on an intel system and will be switching to AMD sometime next year, DFI 939 Nforce 4 SLI :slobber: I enjoy reading every ones oc'ing on the 754 I just wouldn't want to only have 512 mb of ram no matter how fast it is. Then again I like gaming and 2x512 is a must.Benchmarking is fun but whats the point if it dosn't help in real "computer world" applications. Just my two sense ;)Quote:
Originally Posted by conrad.maranan
After I finish mapping for CSS tonight, I will be testing my second dimm, and then both because I dont plan on only using 512 either. No school tomorrow, so a whole day of tweaking for me :banana3:
No work tomorrow, so a whole day of looking at my dead Clawhammer for me. :(Quote:
Originally Posted by DaGoof
I know I said I was going to buy another 754 on Friday. On second thought, I'm not going to anymore. I'm jumping on the 939 bandwagon even though the DFI nForce 4 isn't out yet. I just placed an order for a 3000+ Winchester and an MSI K8N Neo2 Platinum with mWave and will be picking it up from them on Friday or Saturday morning.
Good move. I'm waiting for January or something to roll around before I move up, but I really can't wait to be off this damn 3200+
Oh... just great... got my Plat rev2 with the new looking heatsink. Guess what.
wouldn't pass memtest stock setting no matter what! 2-2-2-5 @ 2.5-2.85v as well as 4 different bios.
I get curious and start testing sticks individually.
1 stick fails @ 3-3-3-8 in no time... ahahaha.
the other one doesn't make it 2-2-2-5 but is scaling up on 2.5-3-3-7 setting decently. (past 260 now)
Now before someone plays "foul" I had perfectly well working EB (still do have it as I plan to ship it out to someone else tomorrow) running happily @ 2.5-2-2-8 @ 2.85v @ 230 or 3-2-2-8 @ 240 on this particular board. Also samsung TCCC I had works as well as on this board as any other.
I am starting to wonder quality control of OCZ now. I KNOW they have awesome tech support as I had to RMA my Powerstream before (another quality issue?) Out of 4 product I have had chance with from them, 2 of them didn't have good first impression so far. I definitely will be contacting OCZ tomorrow first thing in the morning as my good sticks of EB are getting shipped out and I will need something to play with soon...
Well, this is the first product I've had from OCZ, and I could not be more impressed. The fact that I max out at the same speed no matter whether I'm using 1 DIMM or 2 DIMMS, leads me to believe that not only is this ram being screwed by my memory controller, but this ram has TONS more headroom. That damn CO core that I got in January is surely screwing me.
285 = 512x2, impressively stable @3.3vDimm
http://www.imgbucket.com/39/285x2.jpg
That was what I am hoping for but I definitely god dud of sticks. I've had powerstream that died first day on me with spark too :P
Previous 2 memories I got from OCZ was good to okay at least (3700gold when it was hot and 3700eb). This one just blows though. Even one of this stick I have only does 260 2.5-3-3-7. The other one is just not stable in any speed whatsoever (okay I didn't try anything below 200 obviously... but tried timing up to 3-3-3-8 on that one)
I seriously doubt if it is memory controller on the dud stick considering my EB is running perfectly fine in dual channel on mobo that is reputed not to like EB...
Hopefully something will be resolved quickly with OCZ (which I have little to doubt for considering their track record with me on this kind of issue) without me having to go without memory for time as I am shipping my sold 3700eb out tomorrow -_-;
What are you using for htt value in bios? have you tried 3?Quote:
Originally Posted by jinu117
htt? You mean LDT or so called HT Frequency on this one?Quote:
Originally Posted by shadco
http://www.home.no/jimramse/Pictures...10%202.55v.JPG
Have to confess. It's not stabile there. The highest I've benched is about 285. But I really love this RAM. You have to consider that this is Dual Channel and 1t.
which woman is on your desktop ? :D