Video Card, 3D, and high FSB OC

[Gothic Case]

erh... just a dumb question... when your playing for 2hours+, with a 256MB Vid Card and 1GB system memory, depending on how heavy is the game (Doom3:roe in this case), it would already started using the pagefile, right?

 

So it's reading from the HDD... could the nVidia IDE SW driver be my problem?

[NEOAethyr]

Clay.

 

Well when I 1st got my board, I did'nt really notice any breakin stuff.

Except if I hammered it with my vcore jacked up and used the sandra cpu test breakin thing...

That made things worse lol.

 

I did'nt know much back then .

 

It was'nt until I killed my mobo's bios that I noticed breakin.

My board sat for over 2 months before I could get a new chip, I'm pretty poor...

 

 

I used to use 240 with a dividor 24/7 back then btw, no big probs.

 

 

It was when I got some balls and decided to go for it.

I allready knew I could'nt get past 263 with a dividor, it was'nt even close to stable, but I wanted to bench it in sandra, something higher...

 

I found this trick, bad trick, one I'll never use again ever.

Holding the reset button I think it was, either that or the power button, I think it was reset.

I could hold it and get wierd stuff to boot.

I even got into windows at 275 ^^, but it stoped loading at the status bar thingy(the one you'de logon with, that small dialog thing).

It had that little tiny bar that scrolls, it was scrolling over and over without going anywhere.

 

So I reset, I set the fsb to 250.

No boot, stupid bios I said... (619, I was demo'ing it, I liked 506 better).

(I had tons of boot probs with 619...)

So, instead of clearing the cmos liek I shoudl of, I tried the trick again, thinking I'm gonan force the sucker to boot and do what I want from now one, hopefully it'll work heh....

Well, did'nt work, bye bye bios, big time.

Not even a new chip fixed it completely, it took me 2 days I think to get that sucker booting again after I got the new pre-flashed chip.

I had to power on with no cmos jumper, over and over, and put it back on when it was one..., that sort of thing, screwed up stuff.

 

Anyways after I got it booting

 

I could do that, and boot it, and game it :.

And more...

 

I got 280 gaming stable at a pethetic 2:1 ratio :.

Still was pretty cool though, if I could of got higher in the fsb, it would of been worth it back then since that memory did'nt clock very high.

 

Then I got new memory for very cheap, and lucked out, this adata I have.

Did 250 out of the box, allmost gaming stable at the time.

When I actually got it home though, I got it gaming stable in no time.

I even got it 280 gaming stable pretty quickly, at a 1:1 ratio, but only a 5x multi.

 

I had serious probs after 250 with higher multi's.

There was'nt a single bios that helped.

Matter of fact, only one other bios could even get 280 gaming stable, and that was merlin's eb3, the stock bios that could doit was 506.

No other one could.

Even with merlin's bios, the bugiest bios I've ever used for some odd reason, I could'nt use any more multi's, and it was slower.

 

It was'nt until I got my 2nd stick, that I noticed how well merlin's bios really worked.

Hellfire's bios never did it for me either...

I got 263 in dc gaming stable with merlin's bios.

With only a 5x multi though.

 

Eventually I got 9x working at 263, I can't remember when.

I asked merlin if he could get 263x9.5 working and he never answered back.

In tictac's old forum.

 

After that I lost that stability.

I struggled to get it back.

I eventually started getting it back, it took forever.

This was well into where I was making my own modded bios'es.

Eventually I got 263-266 gaming stable again myself bios wise, but not prime stable.

It was'tn until I added a filter to my atx lead on my old enermax that I got it stable, in sc 1st, then dc.

This is where I noticed breakin 1st hand, memory or bus wise.

 

Because at 1st, only sc was halfway prime stable(but not).

And that took a few tried to get it 8hrs stable, and I had to run my system cool.

Then, I worked on the 2nd stick, in sc mode, again took me a while.

Then both sticks, did'nt take to long then, onyl 2 tries I think, I totally forgot now, not important anyways.

 

I showed my buddy, it was pretty impressive...

I then worked on lowering the votlages, then 1.9v and 3.0v I think.

I got it down to 1.6v and 2.9v.

 

Then I encountered s&m.

I needed 1.8v and 3.0v for that...

 

After a bios mod, I needed 2.9v again , but for s&m.

So that breakin probably helped on that one...

 

Now...

I can do about that same setup, at 1.7v and 2.8v, maybe lower, in dc.

I just checked a few hours ago .

 

I re-learned some timings...(tRC and tRFC, then tRCD-R and tRCD-W, I got this new list of all possble good timings(for the main ones, excluding cas and cmd)

That's why I think.

 

 

Somewhere along the way during all that I leanred of the proper way of cpu breakin at a diff forum, after I saw that bh5/ch5 breakin thread.

I then put 2 and 2 together and thats what started my knowledge in breakin.

I know that upp'ing voltages can be bad, and if there's no reason, don't doit...

That and errors are bad.

Cooling helps initially, and after a while, you don't have to run it that cool...

[charlie22911]

@GC, its possible that may be a problem, i hear a lot of people have that problem. but ive used the SW's since i got my board as stock drivers makes a (!) besides my CD-RW in device manager and i cant use it. i used the one from the 4.21 driver package.

 

@neo, hmm, interesting read. ive always severely over volted my parts with some success, but i discovered by mistake a few days before i found blown caps in my psu that im stable at 2.4Ghz 1.625v, used to take 1.75v to get there... i can do 2.3Ghz at 1.5v even, not sure why it takes so much more voltage to get there... mabye it goes back to the caps in my PSU.

 

BTW, whilst we are on the subject of caps neo, what caps do your board have? mine are either chemicon or rubycon, forget which but they are KZG caps, which i hear are quite good . the caps that blew in my PSU were Fuhjyyu caps, same one that seem to be plaguing antecs, and now my vantec.

[NEOAethyr]

Same here, chemi-con kzg I belive, only the cpu ones though.

 

I got mostly teapo's on this board.

It's a china board.

 

 

I forgot to mention I added ferrite beads to most of my stuff.

One in my mouse, one on each drive's power cables.

One on the p4 power cable.

One on my surge protectors power cable...

One on my phoneline(it actually helped a tiny bit, I checked the graphs).

 

The ones on the drives helped out both hd's, in the smart data the power cycle count got lower.

 

 

I've got to add more later on in the future...

One one each end of the cable, from the psu, to the end by the connectors.

To help rid my cables of rfi(radio freq interferance).

 

I will try to get a spool of pure copper wire later, or silver(I know where to get silver too...).

And redo all my cables, the drives love copper cable .

I wonder what my board will do on them.

 

I gotta get 8 80mm fans too later on.

I want these, Oh crap they seem like 92's..., I gotta relook that up :.

 

Delta 92x25mm FFB0912SH $15

92x92x25mm

90.02 CFM @ 4100 RPM

50.0 dBA

8.28 watts, 0.69 amps,

Maximum Air Pressure 10.06 mm H²O

(It's quieter, faster, slimer, and uses less power then my tornado)

 

I need: 4 (Actually I need 8...)

[Gothic Case]

Are they going to blow air directly to the mobo, or are going to make an airduct from your case?? lol

[charlie22911]

12 0C Device Power Cycle Count: This attribute indicates the count of full hard disk power on/off cycles.

 

power cycle count has nothing to do with what you are refering to neo, if that actually went down then somethings up and its not good.

 

here is a page i quoted from that lists the various functions of S.M.A.R.T. and what they are:

http://en.wikipedia.org/wiki/Self-Monitori...ting_Technology

 

tis a good read .

[Clay]

Thx "Neo" when I first got my LP-B it needed 1.7 vcore 1.7chips 2.8 for 252X9.5 now it's 1.675-1.6-2.8 manually entered std timings mains 11-3-3-3-2.5-13-15(Ect) Seems like the more I werk I do the it the better she get's Hehe...As far as psu goes I haven't looked yet. But the psu I have is much better than the 1 I had.That 1 the room would smell of ozone And I could roast wennies off the exauste air.A-power500 (el-cheapie) The 1 I have now runs pretty cool and I have to stick my nose in the exhauste fan to get a hint of ozone. (I may have a small leak on 1(I'm not worried about it though))What is alarming is the amount of vcore my processor seems to want now for 2500mhz's I'll get back to that soon. When I can find time. But right now I'm lucky to find time,to be able to post. But I might be lookin for another pwrstrm520 to eliminate any doubt. Or a bigger approved (Inna' bout 5momths this one has to hold on!)

I haven't even looked at the caps on the boards. The next 1 to get pulled will be the ultra-d

Better chipset cooler,Cpu cooling upgrade.(2months+ anyway) Then I'll check that 1 out.

BRB Got bussiness.

[Clay]

erh... just a dumb question... when your playing for 2hours+, with a 256MB Vid Card and 1GB system memory, depending on how heavy is the game (Doom3:roe in this case), it would already started using the pagefile, right?

 

So it's reading from the HDD... could the nVidia IDE SW driver be my problem?

 

Not if they're not installed thier not. Since I went ahead and backed off to 252X9.5 I haven't noticed anymore problem. But I havent been playing doom3 either. I still think the problem there in my case is-was gpu heat.(Still is) But I've been playin' "NWN" lately which is alot easier on the card. (I still havent had time to fix it.(Yet))

[charlie22911]

okay guys, woth this pwn'in PSU i have rocked away in halflife2 deathmatch for about an hour and a half (man i suck ) @ 260x9.5 1.85v with no hickups. 270 isnt stable enough to game in yet so thats a no go.

[NEOAethyr]

SW ide's from the 5.10 pack are bad news.

The newest ones are ok, should be ok for everyone, but maybe not.

In general though, if the new ones are working, they are working good .

 

 

Charlie22911

Like I said, I don't trust most general user info's off the net.

 

However that one on that one setting is correct.

After looking up info's in forentsics I found the info on it.

Good enough to go by.

Power on cycle count specifically means how many times the drive has been powered on from standby or from being powered off.

 

Smart data is located most of at all times, on the hd platters them selves.

I've done some wacky stuff with my hd's though.

Like full copies, apperently that's not the norm lol...

 

So far I've found these:

 

SMART Specs:

ftp://ftp3.ds.pg.gda.pl/people/macro/S.M.A.R.T./8035R2_0.PDF

ftp://ftp3.ds.pg.gda.pl/people/macro/S.M.A.R.T./8055.PDF

 

8035 is THEE spec.

However it has no mention of what attirb's are commen and such, attib's are diff per disk...

 

NDST Specs:

http://www.seagate.com/docs/pdf/whitepaper...anced_smart.pdf

http://www.seagate.com/docs/pdf/whitepaper..._Tech_Paper.pdf

 

 

You have motivated me .

It appears I need to figuer out what program is correct with SMART data.

I've been going with everest for along time, however recently I noticed huge errors in it's hd info.

Like it reporting 600 bytes per sector lol, which is impossible.

I know because I've tried 256 and 1024 and windows freaks.

 

You can re-write smart, and I'd like to.

To start from scratch.

For one, my 40gig seagate, I give it over a 16 million value on one of it's smart values when I was low leveling it by hand.

:

 

I'm gonna check this out:

http://sourceforge.net/project/showfiles.php?group_id=64297

 

There's some progs out there to redo data too so I gotta figuer that out.

 

 

Btw guys.

My idea of what tRTP was wrong.

It's not just tRAS.

I thaught a tRAS of 11 allows tRTP of 0.

 

However if I slack other main timings I can go lower as well.

So I dn.

 

Still, tRC and tRFC I think I got that down, but I have yet to truly test that still :.

I was able to use for example 11-3-2-3 with tRC of 8 at 275 so I dn, I will figuer it out someday.

 

Anyways I will check you guys later.

Check out my pcr's btw .

The genric data in them is quite good.

I will work on them as time see's fit.

[Sharptooth+]

Just thought I would let everyone know the latest developments with regards to this topic. Originally, I was postulating whether or not AGP-bridged (video card) solutions were having issues with NFII boards at high FSB OCs. I wanted to get feedback from users to see if there was a correlation.

 

I suspected my video card (6600GT) because it has HSI and I was having issues stabilizing at 240-250FSB. Well, I believe that in my case at least, it may not be the card that's at fault.

 

When I was playing Doom3 and ROE (10.5x240), I mentioned in other posts that I was getting an intermittant pausing but no crashing or error message. This would occur randomly. I became suspicious when I observed that I was passing the 3DMarks without issue. That's when I noticed my video card's OC. I had it going at 600/1100 and the Vagp was at 1.8. During one of the times I was testing with Doom 3 ROE, after it paused again, I quit the game to see what the video card temps was. It was 66c!!! That's with an aftermarket cooler and sinks on the ram chips.

 

The same video card needed a Vagp of 1.65v on my AN7 system for 3D stabilty at 10.5x236. I know it did because when I set the Vagp to 1.6 it would crash. That's why I ASSUMED it would need perhaps a higher Vagp setting on the LP-B when testing at 240-250FSB.

 

So, I let the coolbits2 detect an alternative OC and it chose 562/1130 but I also decided to back off the Vagp and set it to 1.5v. I re-tested all the 3DMarks and played Doom 3, ROE and Quake 4 for a couple of hours last night and NO PAUSING OR CRASHING AT ALL!!!!

 

Now, getting to another topic. Earlier in this thread I mentioned how the 'cold-boot' bug had reared it's ugly head as soon as I got into the 10.5 multiplier (240FSB). I also said that it persisted even after I changed my PSU from an older Enermax 431w to a newer Noisetaker 485w. At that point, I tried using my *brand new* PC Power and Cooling SLI 510 and *Bam* the problem hasn't occurred since!

 

So, in summary, I just want to say that I don't think Nvidia AGP-bridged cards are necessarily at fault when pushing NFII systems with higher FSB OCs. At least, that's not in my case. It seems the combination of using too high of a video card OC and/or too high of Vagp was creating this issue for my LP-B. Furthermore, my PCP&C PSU may be helping the situation too. It did with my 'cold-boot' issue.

 

 

--Fernando

[Sharptooth+]

Believing that I may have finally stabilized around 240FSB (10.5x) to the point where I can start testing at other higher FSBs. I tried briefly testing at 250FSB (10x). The first thing I noticed that using the same memory and alpha timings from 240FSB seem to be too aggressive. I couldn't play Doom 3, ROE and Quake 4 for very long before I would get an error message.

 

So, I decided to back off those timings to:

 

8-2-2-2-2.0;14,16;5,4,5,3,3,5,3;e,e,f,e;4-11,4-11,4-11

 

At those settings the games did play further but still eventually produced error messages.

 

Then, I decided to change slightly to:

 

8-2-2-2-2.0;14,16;5,4,6,3,3,4,5;e,e,f,e;4-11,4-11,4-11

 

Again, I played with Doom 3, ROE and Quake 4. I cleared a level in one and quit to play another game and I was going back and forth for about an hour. To my surprise, it didn't produced an error message with any game. Of course it's far from exhaustive testing and more testing is required but it certainly draws interest to the alpha timings. It appears to show tangible benefits with these games.

 

However, even with those laxed timings, it still failed:

 

SuperPI 32m (after a few loops)

Prime95 (5 mins)

 

By the way it's with using a Vdimm of 3.2 and Vdd 1.9. Ironically, when I up the Vdimm to 3.3 it produces errors in the games. Strange isn't it?

 

--Fernando