Faster in what apps, 3D??Quote:
Originally posted by Hallowed
OPP, 10-2-2-2.0 or 11-2-2-2.0 should be noticeably faster than 5-2-2-2.0.
Give it a shot. :)
OPP
Printable View
Faster in what apps, 3D??Quote:
Originally posted by Hallowed
OPP, 10-2-2-2.0 or 11-2-2-2.0 should be noticeably faster than 5-2-2-2.0.
Give it a shot. :)
OPP
Quote:
Originally posted by OPPAINTER
With 1.7V vdd I got to 236 stable at 49c load
With 1.8V vdd I got up to big old 238 stable with 51c load.
Now I'm at 1.9V vdd at 240.
It's looking to me that good cooling on the NB will help get higher FSB/Mem. And this was point. After taking it as high as possable at 1.9 vdd I'll cool the NB with better and better cooling and see if I can go up.
This for agressive timings by the way.
Well, 5-2-2 anyway. Things may differ at non agressive timings for all I know.
OPP
so with more VDD voltage, you're able to pass higher fsb tested 3dmark?Quote:
Originally posted by OPPAINTER
I'm useing 3D2001 for stress testing.
I only made it to 239 3D proofed, NB temps get very hot at 1.9V vdd.
Which i think will prove that the higher the VDD the further you'll go but only if cool the bridge, we'll see.
Heres somthing cool, 3D has the seven tests. The tests that stressed the NB the most were the Lobbys (Bandwidth) :D during those 2 lobby tests the temps were peaking at 55.5c, when I went into nature the temps quickly dropped 1c to 54.3c.
Gunna slap on a swifty 159 cooler and strat again.
OPP
i'm not testing my fsb with 3dmark, i'm just seeing how high i can do sandra right now. interesting, you got a NF7 rev2 board? what color are the 4 capacitors by the 6 vcore mosfets?
max i can get in sandra is this:
http://marcin.fastlanehw.com/sandra_nf7_249fsb.gif
that's the 1.7Vdd through the bios. I've tried 1.74, 1.8, 1.84, 1.9, 1.94, and i couldnt' get a higher fsb for sandra test.
i'm wondering if the higher vdd will allow me to pass higher fsb in 3dmark2001. gonna try soon.
Faster in everything..... I've tested it myself, it's true.Quote:
Originally posted by OPPAINTER
Faster in what apps, 3D??
OPP
At 1.7 volts temps droped 9c under load with the swiftech 159 air cooler.
OPP
OKAY, i just tested 3DMark2001,Quote:
Originally posted by OPPAINTER
With 1.7V vdd I got to 236 stable at 49c load
With 1.8V vdd I got up to big old 238 stable with 51c load.
Now I'm at 1.9V vdd at 240.
and i'm about seeing the same increases...
1.8V = 242 max
1.84V = 244 max
1.9V = 246 max
haven't tried higher... the southbridge is getting pretty warm, i will probably add a heatsink to it soon.
At 1.71 vdd and new cooling I've already been able to run 2 mhz higher, 236 to a now stable 238. Still going.
Who ever thought this new Ultra 400 chipset doesn't need to be cooled needs to do more research;)
OPP
i wish there were other nb coolers out there that would fit with the SP-97. the vantec iceberq , i dont feel like playing with, as it's just the same as stock.
OPP, what version NF7 do you have, and what color are the 4 capacitors next to the 6 vcore mosfets on your board? also, what bios are you using??
mine is a rev2 with black capacitors, using equito's d10 bios.
OPP, what is your temps like with NB undermax load with the MCX? Mine are like 45c and I got a bloody stock amd cooler on there with a delta! :D The temps are horrid though cause my room is ambient 80 degrees farenheit.
1.9VDD for me.
Mine is a NF7-S Rev2, I have bluish purple Caps.Quote:
Originally posted by Stang_Man
OPP, what version NF7 do you have, and what color are the 4 capacitors next to the 6 vcore mosfets on your board? also, what bios are you using??
Sidebinder,
I'm testing 1.9V now but at 1.8V I got 44c max temps at 242 FSB.
I'm measuring the back of the chipset with a Monogram Omega temp gauge, very acurate.
Room temps at 21.5c.
OPP
Good, then I know my temps are pretty normal then, have you tested you 2400 on air cause I got the same cpu and the thing will only do 2.4ghz :( same steppings and week.
Time to order 5 more and cherry pick. :toast:
Did you lap the northbridge chip itself? The thing is about as flat as a soup bowl...
lapping your northbridge will gain you at least 5Mhz, I lapped my NF7-S and got a 10Mhz gain. I decided to lap it after finding out that my waterblock was only making contact with the four corners. After lapping I could do 245fsb, ram limited.
Theres an article on this here
Testing is coming right along:)
I saw that going from Stock to Swifty at 1.9 vdd there was a gain of 5Mhz and 9c in max temps. Now for water block cooling. A mild cooling rig, useing a waterchill NB block.
Then all I have left is watercooled pelt cooling and I'll be done testing and then I'll see what this sucker can really do.
OPP
Good luck can't wait to see you hit 300FSB on an NF2! Don't know if possible, but hey, you are OPPAINTER. hehe.
lap the nb chip itself? wouldnt that wreck it???
300 on a DFI maybe... LastViking was at 280 so thats almost there. Xtreme cooling of NB on a good DFI board might do the trick to 300 may be. But not on an NF7 that is doing what his doing.
Opp, did you try lower Vdd for high FSB?
Mine bluescreened with anything over 1.65 and FSB @240, backing off Vdd to 1.55V just did fine!
I know it sounds weird, but i wasn't alone here, more ppl had exactly the same.
I have to add, this was all with bios d10.
Not if use fine sand paper and sand it just a little.Quote:
Originally posted by r3b0rN
lap the nb chip itself? wouldnt that wreck it???
but then again that very possibly won't give you an additional increase. i had a friend who did it.. didn't give him anything.
so it's either a watercooled pelt or all the way phase changed, which would be overkill at any rate and won't be as much help as with p4 or older rev nf2 boards.
/edit typos
If you lap too far, yes it will. But I went at my nf7-s 2.0's nb for a solid half hour with 100 grit sand paper before switching to fine grit just to get it flat. Then higher grits to finish it off and make it smooth.Quote:
Originally posted by r3b0rN
lap the nb chip itself? wouldnt that wreck it???
I only did 400 and 600 grit. A bit of a difference in stability not a lot of extra MHz, but then again I also need a good NB cooler that fits under an SP-97.
theweasel, did you notice any improvements over the non-lapped nb. what are / were those results and what cooling did you use to get them?
OPP, 1.76V vdd, 240 FSB 1:1 CPU interface enabled prime stable :DQuote:
Originally posted by OPPAINTER
At 1.71 vdd and new cooling I've already been able to run 2 mhz higher, 236 to a now stable 238. Still going.
Who ever thought this new Ultra 400 chipset doesn't need to be cooled needs to do more research;)
OPP
Stock nb cooler with arctic silver 3 tim ;)
Oh and please give bios d_14beta a try ;)
http://service.futuremark.com/compare?2k1=7463753
^ Abit NF7-S V2.0 @242 MHz FSB x 10 @1.76V || 2x256MB Muskin DDR3500 Level II 1:1 Dual cas2/2/2/11 @3.20V|| AXP2500+ "Barton" AQZFA - 2420MHz @2.0v || Sapphire 9800 PRO @438/384 || *ALL AIR COOLED* ||
If you dont use a big glob of thermal paste in the middle of the nb, it wont make contact at all, except at the four corners. You just need to sand the corners to make it flat.
Another tip OPP, best timings are 2-2-2-11 yep tras 11 :D
*NF7-S @240FSBx10 (480MHz DDR) 6/6 divider and Dual Ch. Sisoft sandra max 3 bandwidth test*
2.0-2-2-3:
Integer: 3661(95% eff.) / Float: 3431(89% eff.)
2.0-2-2-4:
Integer: 3665(95% eff.) / Float: 3430(89% eff.)
2.0-2-2-5:
Integer: 3666(95% eff.) / Float: 3443(89% eff.)
2.0-2-2-6:
Integer: 3675(95% eff.) / Float: 3440(89% eff.)
2.0-2-2-7:
Integer: 3679(95% eff.) / Float: 3446(89% eff.)
2.0-2-2-8:
Integer: 3674(95% eff.) / Float: 3445(89% eff.)
2.0-2-2-9:
Integer: 3686(96% eff.) / Float: 3448(89% eff.)
2.0-2-2-10:
Integer: 3691(96% eff.) / Float: 3452(90% eff.)
2.0-2-2-11:
Integer: 3702(96% eff.) / Float: 3457(90% eff.)
2.0-2-2-12:
Integer: 3698(96% eff.) / Float: 3450(89% eff.)
Oh and for voltage FSB on the fly (excl vdd):
http://www.hasw.net/8rdavcore/index.html