Sorry to hear that. You need to use accetone or nail polish remover with a Q-tip to get the paste off. Be very carefull on the mb and dont hit any of the small resistors beside the nb and sb.Quote:
Originally Posted by Hor$eman
Sorry to hear that. You need to use accetone or nail polish remover with a Q-tip to get the paste off. Be very carefull on the mb and dont hit any of the small resistors beside the nb and sb.Quote:
Originally Posted by Hor$eman
Well I went and bought another board (Wanted a SIIE...oh well), and hooked up the poster to work on the old board. It remained on CPU INIT at post, so I removed the z9700 and the e8400 and found a bent spring pin... anyone know if this can be repaired, or is the board a writeoff because of this?
I removed the heatsink array from the new board, put the previously removed (from other board) fets on, my new thermaltake NB coolers on the NB and SB(impressive 4800+ RPM), lapped the e8400, (added AS5 to all) installed (carefully) the e8400, installed the z9700, and fired it up, checking the temps. Initial NB and SB temps were:
NB less 7c than prior best (bios)
SB less 8c than prior best (bios)
(Also installed my new Corsair RAM cooler)
After 4 hours running at 4.05 Ghz using orthos and Prime95, my CPU max temp has been 59c load, but is coming down as the AS5 works in, and max NB temp has also been 59c (according to temp sensor I mounted), now reduced to a pretty standard 54c after the AS5 is working in.
Once i get the kinks worked out at this frequency, I am going to work on a stable 4.5 Ghz. I had to go with my previous bios settings for over voltage, as trying to run on auto (except for CPU and Mem) kept failing.
New Patriot Viper PC9600 is now flying at 1230 Mhz at 5-5-5-15 @ 2.32v, though will run at 4-4-4-12 at 1200 Mhz @ 2.3v is not passing Prime95 or Orthos at the lower numbers. I might have to buy two more sticks of this awesome RAM since it should be able to run 5-5-5-12 with 4 sticks as easy was with 2. Best bandwidth so far with it (SiSoft Sandra) has been 9883 MB/s
any 4.5ghz stable config 4.1 max here
I have those set of vipers too but i'm having trouble getting just 1000mhz out of them nevermind their rated 1200mhz. also noticed some different timings per stick, check this out:
http://www.xtremesystems.org/forums/...81#post2936381
anybody with Kingston HyperX ram try it with this motherboard?
I wish I'd known Kingston makes PC9600.... I would have went with that.Quote:
Originally Posted by mashakos
Can you post a screen of the SPD tables?
I did lots of testing with HyperX 6400 and it got better timings (though lower overall bandwidth of course) than my Patriot PC9600.
4.05 highest tested stable here. There's a memory hole above 1800 whereby latency skyrockets, if I remember correctly.Quote:
Originally Posted by hmetro24
I've only had this ram for a few days, and have it running stable at 1200, runnable to 1230. I've had stability issues reducing timings too much, so am slowly doing that here and there. I should have my best numbers in a week or so.Quote:
Originally Posted by slugzkea
Have you done any burn in on the ram you have? Try burning it in at 1100 mhz for an overnighter.
I think that would be a strap change??? Same as it did on 680i.Quote:
Originally Posted by Hor$eman
Hor$eman, our ram should go up to 1300mhz although vdimm has to be over 2.6V... I can run 2GB at 1266mhz at 5-3-3-3 stable, but not with 4GB... ((
Can you all say what is the memory bandwith you are getting in everest please?I beleave that the most of the posts won't be higher than 9500MB/s,witch I am getting with my second striker.With the first I had,I could easily get 11000+ MB/s by symply adjusting trd in bios:shrug:
Max I got was 10500...Quote:
Originally Posted by Phil The Greek
but even at 1T, I can only get 10042... ((
is anyone running 4GB (2x2GB) 1066 corsair dominator sticks successfully with this board?
No but I am running some 2x2G A-data pc6400 that does 1080mhz @2V with SIIF and now have it running 1:1 at 475fsb 5-5-5-12 giving 11560mb/s in everest.Quote:
Originally Posted by RAID
will Thermalright Inferno FX-14 fit ?
This shows higher than in SiSoft Sandra.
I noticed something about the gold plated southbridge piece on the southbridge: it's just a cover for the heatpipes. I tried removing it so I can slap on a Thermalright hr-05 heatsink over the copper plate but ended up damaging one of the heatpipes. Now the board won't POST. I'm going to get a new one next week, lap the chipset cooler and try removing that gold cover on the southbridge again. If anyone has good suggestions on safely removing the southbridge covdmer, I'm all ears :D
Quote:
Originally Posted by mashakos
That cover doesn't actually cover the SB. It covers some other smaller unit, which doesn't seem to get even warm to the touch. Removing this and adding a cooler to it is just lipstick. If you want to help yourself, I'd suggest adding something to the NB.
However, if you want to remove the whole assembly:
- Begin by removing the spring pinned "keepers" by pressing the back side of the keepers in with a pair of pliers and pulling on the other side of the pin at the same time, until all of the keepers are through. After two, it will be easy.
- Unscrew the North bridge.Lift the entire unit off. Only the center PCI-E plastic holder is in the way.
To remove an individual unit or plate, use acetone or nail polish remover and high heat (hair dryer style) to soften the glue. I think its like hot glue gun glue.
I took mine completely apart, but screwed the whole unit up in the process of trying to disassemble piece by piece, so ended up using thermalTake NB coolers on the NB and SB, and cut the heat pipes off the fet coolers and put them back on the way they were. max NB temp is now 59, SB temp is now 41 under 100% duo CPU load at 4.05 Ghz (duel instances of Prime95. The other smaller unit (as per above) has air fromthe SB cooler blown straight across it.
Waiting for 8GB to arrive :)Quote:
Originally Posted by RAID
I was actually talking about this one:Quote:
Originally Posted by Hor$eman
http://img166.imageshack.us/img166/1010/sbsq7.jpg
the one you're talking about is the smaller dark orange one correct? I think that's for the SLi bridge chip or something.
how exactly? A hot knife with acetone on it maybe?Quote:
Originally Posted by Hor$eman
I'm not sure. Something sharp to pry with - perhaps if you have a benc and a chisel, it will come apart easy if you heat it up. Getting the heatpipes seperated though, if you want to do that, requires destruction. If not, you may be successful.Quote:
Originally Posted by mashakos
I'd pull the whole unit and replace the thermal compound anyway. My first board, the tc was hard and dry. On the second one, it was better, but one can only tell if they look at it.
Pic. I know the 9700 is oriented differently than normal, but with the large fan at the top, this works better.
I notice that like myself, anyone using a Q6600 on any 780 board, is getting absolutely horrid latency (like 70ns) and terrible bandwidth, no matter what timings.
Using the same timings as my 680i, I get horrible score, where I should be getting 44ns (which is what I was down to :) ) I'm not up in no mans land at 70+ns.
I see alot of people that are getting good latencies are using CPU's with a 333 native FSB, those of us getting suck scores are using CPU's with 266 FSB.
Is it possible that the board is relaxing (radically) the timings that we cant change to people with 266 FSB because of a strap change??
I suggest this because at 450FSB it is 69% overclock for a 266 native CPU, but only 35% for a 333 native CPU.
So what I'm saying is, board thinks its FSB is being overclocked a long way at 69% because it thinks its 266, and for you 45nm users it only thinks its being pushed 35% from 333.
If thats not the reason, then I got no idea what the go is, it just seems to be a trend.
Any evidence to back me up, or shoot me down, or anything?
Thanks for any response.
Very interesting thought:rolleyes:
When I was using my first striker,I was using a QX9650 and with auto trd I was getting this
By simply setting trd 5 I got this
I am almost sure I could get the same bandwith with my QX6850 too(at least after using the 9650 on the board).
Right now with my second board I have only the 6850 and I am getting horrible bandwith no matter what I am setting on trd.
The thing is that both chips have a native fsb of 333,so it can't be something related with fsb:shrug: