Not yet, I am going to work on tweaking the most of my memory, async right now :DQuote:
Originally Posted by Kin Hell
Printable View
Not yet, I am going to work on tweaking the most of my memory, async right now :DQuote:
Originally Posted by Kin Hell
You do realize that orthos will stress up to 2 cores so you dont need to run a second instance...Quote:
Originally Posted by marauder16
Well, after a lot of meh-larking around, I`ve stopped chasing the fast FSB's. Got to the magic 464 on a 7 Multiplier & was benching Sisoft with 34% Bandwidth efficiency. Anyway, I got there on NB 1.53v & ran Orthos for ten minutes, so I guessed it was some where handy. 3DMark is way slower on such poor Bandwidth efficiency. I'm convinced being able to use an 8 Multiplier would give better bandwidth efficiency even at a slightly lesser FSB than 464 but faster than 370 1:1, ...say 410.
I also notice that by using a 300 Strap, SiSoft is loosing approx 500Mb/sec, so all my results are based around 1:1 on SYNC with 266 Strap. Dropping to a 300 Strap does not make the CPU or Ram any more Overclockable, at least not for me.
I`m now Orthos Stable at 370FSB 1:1 & get a snappier feel to everything and anything I do at any other settings. Benchies are the biggest too. Thinking back to when I was trying the OCZ9200, I`m not entirely sure I tried a 7 Multiplier with 464 board Voltages running Sync on a 266 Strap. Is it possible?
As I go on waffling about this, & maybe even talking myself into an answer which totally eludes me right now, ....I am concerned at to wtf is going on when playing with Multipliers!?
I'm using a 2.4ES here, stepping 5. Orthos seems to like showing the CPU speed as if it`s actually got a 9 multiplier, even tho it's set to 7 X 464fsb on a 266 Strap in the Bios. CPUz reads it as 7 with the correct Mhz-idge of 3248Mhz!
So, Orthos & "rmb Sys Properties Gen TAB" show CPU speed at 4.176Ghz?
Emotionally disturbed by this, "I did a rmb Properties" on "My Comp/Gen TAB", which also reported my CPU Speed as it would if I was on a 9 Multiplier?
Running Orthos or rmb Sys Properties on an 8 Multiplier shows it contrary to CPUz which again shows the CPU at it's correct Mhz-idge. At 9 Multiplier, there are no conflicts between Orthos, "rmb Sys Properties Gen TAB" or CPUz.
Erm.....can someone nudge me nicely about this please? :confused:
Gah m8y..looks like it`s RMA time. Try & sort it with your Supplier on a 28 day affair. Might be your quickest fix. Sorry for ya but good luck as well. :)Quote:
Originally Posted by M3kk
:stick: :stick: :stick:Quote:
Originally Posted by Kin Hell
:rolleyes:
Software, dude, is only as intelligent as the lines of code. Open up Everest and check out the FSB speeds when using a lower multi...
Fact of the matter is that they don't know how to read the bios.
..yeah but come on...wtf is the chipset actually doing if Software that folk use as a bit of a Bench Standard isn't reading right?Quote:
Originally Posted by cadaveca
Its a bit like me looking at the bad Axe 2 Rev 505 Mobo & constantly seeing CPUz NOT reading 9 Multi...but 6 Multi...??????
Is that software too when earlier Revs of Bad Axe with Latest CPUz read fine?
I don't think so! :rolleyes:
*Edit*
Further, it must ALSO be a software thing for WindowsXP Pro to show the Hardware it is on being 4.176Ghz?
This is System Properties on the General TAB.
Again.....I Don't think so!
Anyone else want to nudge me some more?
Well said mate, I bought this motherboard JUST because Tony said it was good, NEVER has he led anyone on and EVERY bit of advice he has given has worked perfectly for me, I feel very lucky to be able to read what he writes, there's not many people you can say that about.Quote:
Originally Posted by EnJoY
And since I've only had it 3 days I haven't done much but what I have I am very pleased with, certainly looks a very stable piece of kit.:toast:
I a gree 100%Quote:
Originally Posted by kimandsally
Kin Hell..are you back on the 12/22(official bios) with your most recent result...Hmm..maybe I should try to fall back to 390 1:1 sync and see how things go
Kin Hell..does bandwidth eff. change only via multiplier? or sync status? or strap all all have an effect
e6600 9x 390....bios 12/22
I tried 9 x390 1:1 sync with 266 strap...I got on Sandra 55% eff(howeverI crash 3dMark05..not really stable)
I tried 390 cpu: 500 mhz ram, async 266 strap..got better memory score(small amount) and same 55% eff.(FWIW..i was 2ohrs+ blend stable when i stopped with 390x 9=3500: async ram @500 with 1.42v on NB...only other increases were cpu vtt 1.23, MB 1.8b PLL 1.93 and convert at 1.97)
The software doesn't recognize the dividers used for independant clock control is all. The independant control, and it's ratios, for all the busses, isn't yet programmed into the app.Quote:
Originally Posted by Kin Hell
The most obvious differences between mobo revisions(especially on the badAxe series) is the clockgen used, and so, yes, it's the software. It's been programmed for a different clockgen, and so when it polls the clockgen, it doesn't get the results it's looking for.:stick:
And, um, did everyone forget that RD580 and RD600 are kinda the same family of chipset(580 is AMD, 600 is Intel)? How long ago did RD580 come out? Forr this board to perform as well as it does...well..completely blows me away.
DFI's fault or AMD's, or perhaps Intel's???Quote:
Originally Posted by cadaveca
Look, I respect your views but cr4p like this just isnt right. It's a bit like the Industy deciding to call EIDE devices SCSI devices. I'm competent, but millions think they have SCSI drives!
FSB Holes in the 680i Chipset;
Bull Sh1t CPU speeds on RD600. :slapass: - gimme a break!
Overclocking since 1989 on various platfroms. I guess there has to be a first-time sometime! :rolleyes:
Whats next, Vista Ultimate on a Creative Muvo 256Mb Mp3 player. - without compression? :D
This kinda leads to another subject which is on the next post as a reply.....
Kin ;)
Hey guys, just started delving into tweaking this board, and figured I'd start with the ram. Currently I have 4Gb's of Ballistix BL2KIT12864AL1005 (2 kits) However, I'm wondering, do the nb voltages affect memory clocks at all? Just because by simply upping mem voltage to rated 2.2V, setting the timings to 5-5-5-18 at 2t and trying to run them at the rated pc2-8000 (500 memory fsb right?) No dice. What else should I be trying to get them up to rated specs so I can go on from there? Thanks.
did try to run memtest86 first to see if the ram works at this speedQuote:
Originally Posted by Suicide King
I have 2 x 1GB 10th anniversary crucial...I am at 500 5-6-5-5-18-24-42-3-6-3-2-3--2-2-5-2t with 2.2v(I know your ram is binned better but try this as a starting point)
My Strap is set to auto
Yes NB voltage and other settings may affect ram stability
to run ram @500 make sure strap is set to auto, sync=async
I would at least push NB to 1.36v
also in the bios my ram is set to 499..in winxp=500
I'll give memtest a shot, but I wasn't aware it could test speeds.... Could you explain abit further, I thought memtest only looked for physical defects.
Hi nealh,Quote:
Originally Posted by nealh
Running Sync on this board seems hard to do over 372 fsb for me. I know I'm on Geil 4300, but the OCZ just wouldnt go any further. Effectively, it's a dead end with my CPU.
Running Async, I could get DDR540 @ 2.4v but with the CPU on 380fsb, I was still faster at 360fsb 1:1 with Geil 4300 @ 2.1V 4-4-4-12 1T.
Also, dropping the Multi to 7 x 464 was nice to see. I had around 8700 Mb/sec
with a CPU flapping around in massive bandwidth that it couldnt fill. Bandwidth efficiency with the 300 or 333 Strap left Bandwidth Efficiency down @ 36%! :(
Benchies were on par, but everything was Volted-to-bits to get there so it was all a bit pointless. I really am not sure if I was right on top of the Strap/Sync combos whilst using the OCZ 9200 so you will have to experiment.
I found that the 300 Strap was about 550mb/sec slower than 266, but the Ram stayed at 1:1. Same with 333 Strap from what I recall, just lower than 300 by about the same margin.
When you select 200 Strap, don't give in easily. You might have to reduce your Multiplier to get a boot, but check out the bandwidth results! I know I saw 76% Efficiency, but gave up trying to get stable. I was a bit of a noob on the PLL overclocking then! ;)
Right now, I'm trying to hang on to 372fsb 1:1 4-4-4-12 1T & had to push up the board voltages slightly overstock. 1:1 is definately the best way for me on this ram. Anything else just sucks! :D
Just remeber that above 266Strap will be kinder to your ram, but will affect Sisoft by around 500 to 550Mb/Sec.
Have a look at the bandwidths at easily obtainable fsb's using 200 Strap. A few hours & you should have your head around what does what - & what goes where! Keep notes...it'll defo help.
I havn't quite yet worked out if the Chipset is lacking in bandwidth as it doesn't do what i'd be expecting it to do when reaching certain FSB curves. I was having a bad time around 400fsb getting the voltages set just right. I noticed my bandwidth had just reduced by 500 mb/sec, yet all the Straps were still set! :confused: I just couldnt get it back. :confused: Dunno if i'd fallen in a hole, but it didn't feel right. Loading a known working saved setting & then a corrupt Windows start forced me to turn the PC off & reset CMOS. :mad: Iv'e seen this @ 372 FSB 1:1 too....about half an hour ago....Turned off Comp at Dos prompt that files are corrupted...turn back on...it boots.
Gunno have to sort this fast!!
Having already done 464 for no gain, I gave up trying for a mid range compromise, largely cos Async is poor for me!
As I left answer to "cadaveca" on my previous post, I have to whole heartedly agree, This board absoutely rocks! I feel it would have been a better product & ROCKED even more if AMD wern't lurking when they were! ..But then I'm a bit of a Cynical old git thesedays.....Age inflicted but hell, I can always pour another JD. :D
I feel the FSB's arn't quite doing what they should be doing at certain curves & wonder if this ties in with Tony Discovering 3T faster than 1T back a while!
He was on crazy fsb's & multi's at the time, - might even be in the first half dozen pages of this thread I think, or maybe the Repository?
Perhaps this is why & where some progs read incorrectly for CPU Mhz'dge when at a particular Multi and/or FSB setting. :confused:
Anyways nealh, always Remember 1:1 will be harder on your NB, CPU & Ram than Asyn can ever be! This will result in a slightly lower clock, but the bandwidth effciency gain compensates for the clk loss. This might also mean voltage ups for stability. Don't forget the notes & compare Efficiency, Mb/Sec, FSB. Try slightly faster FSB's on same strapping/multi combos & see how much quicker the curve UP is, than ASYNC. Do 3DMark01 runs & see what 10Mhz extra on the fsb @ 1:1 does for Dragothic going over the wall. :cool:
I`m dry & fingers are nearly bleeding. Not sure if I just imagined Ice hitting glass tumblers again...gotta go check! :D
Hope this helps & yeah, I'm back on the 12/22. Beta .041 defo has s bit more bandwidth. I'd avise staying on 12/22 to get stable. when you have found your MAX, get .041 onto a floppy for a Dos Boot & flash, reset & save your Bios Config to the Bios Store, ...heh :D, then turn off PC. Press n hold reset , press and hold Power Button holding for 5 seconds..then let go reset & boot up for a DOS flash at default bios settings. You know what to then eh!
Get some Ref figures with 3Dmark/Sisoft bandwidth/Super Pi 1Mbs ASYNC before you try. Make notes! ;)
I got some voltages you can try for 380ish FSBs if you get stuck. Just PM me if you think I can help you at all.
Kin :)
PS Hope u don't have to "Ram Dance" @ default bios ...forgot to say! :p:
Forgot to say above, it'd be cool if you come back on this. I dont know if its the Chipset that cant handle fast 1:1 FSBs it or the CPU. You have a better CPU than me, so if we cant get you better than me at a 372 fsb, then it has to be chipset or a blend of CPU/Strap. I've managed to tweak from getting orthos crashing @ 360 1:1 to ...1 hour stable- "stopped by me" @ 372fsb!
53850 01 marks from GX2.
I'm still testing some settings but I did manage to part bench @ 400 1:1. Dragothic maxxed out a lot of it & never left the 900's.
I might be able to get 400 with GTL tweaks but thats another story for another time later! I`m still working on it & trying to get a listing of my settings that worked together.
L8rz
Great infoQuote:
Originally Posted by Kin Hell
I have no floppy so winflash for me...
do you find running at 3400 cpu and 1:1 gives you equal performance vs say 3500 with async and higherram clocks
I always thought cpu mhz is still king with C2D....I have to look back at some of my bios settings but at one point I was 3400 stable at 1T ram cant recall if I was sync at 1:1 at 380
BTW I get 6800-6900 mb/sec on sandra for ram at 3500(9x 390):500 async 2t
well..I went ahead and dropped down to 3420(8x 380) with 1.4625..24hr blend stable settings and set it to sync 380....well...I got it to post and moved to the second screen..update and hung..soooQuote:
Originally Posted by Kin Hell
I reset and it posted...then went to second screen and said updating DMI....hung
Reset..and nothing...blank screen
turned off...tried EZclear instruction, clear cmos with jumper...removed battery for 10 min with clear cmos jumper
Nothing...I cant get a post...I have removed 1 stick...placed a stick in slot 1,2,3...nothing..I cant get any boot
I get a C1 error......I have no other sticks of DD2
Why, oh why do I not just quit screwing around...never worth the aggravation
So..now I am removing ATX, 8pin, removed battery, clear cmos jumper set...off to bed for me
Hopefully tomorrow..I can get a reboot at least...if not I will need to buy a damn stick of cheap ram and hope I can get a boot...
No place but Best buy, Circuit city or Compusa ...what type of ram should I get
My crucial worked fine on my initial build...I have never seen a bios do this...
I did not push any farther than I did earlier...(I tired 9x 390 1:1..without this)
this time was I went 380x 9..with auto strap, sync....crap
No more for tweaking until they get a better bios.....
Well, I can't seem to get it to go above 487 on the memory... And then it won't survive an overnight run of two prime 95 instances (one per core) and super pi 32m loops, though it runs them without throwing errors, so the reboot might have been due to something the os did. Tried upping voltage to 2.25 on the ram, didn't it allow to budge, 487 is where it sticks... Need to figure out why.
Sorry to hear this bad news. :( Surely, the Crucial is okay for RMA with a lifetime warranty!? That is,......If it has died of course!? .....It could be the Board. Perhaps your best option would be to get your board to a supplier to try another Ram stick or three before writing the board off! I've been here several times so I know how you must be feeling. I know your PSU has some good reviews on not loosing when under load, but that X1950 is a Power Hungry Monster, especially when OC'eed. Are you sure it's upto the demand you are putting on her? Again, just another thought worth considering. Hope you find solution really quickly, best of luck m8y! :)Quote:
Originally Posted by nealh
Also, try both your ram sticks in diff slots - 3 is usually a winner. I have seen Dual kits fail to post with 1 stick, but ok on the other.
I wonder if SYNC at these speeds is just too much to ask of the CPU rather than the Chipset, with the weakest link blowing it's Ass in the process. Does anyone have any information pertaining these circumstances?
With regard to your 6800-6900 mb/sec on sandra for ram at 3500(9x 390):500 async 2t, - I'm at 7400+Mb/sec 63% efficiency at 3347Mhz (9 x 372) SYNC 266Strap, 4-4-4-12 1T.
The DFI RD600 is a nice stable board this is what iam using for 24/7 for mine nuclear reactor QX6700 LOL.
http://img123.imageshack.us/img123/7...quadba9.th.jpg http://img263.imageshack.us/img263/6...tureoe7.th.jpg http://img174.imageshack.us/img174/8...e828py4.th.jpg
my setup was run at 3500/500 with x1900xt at set speeds(but remember in 2D/startup it is at 500/594..ie default..only 675/800 in 3D mode)Quote:
Originally Posted by Kin Hell
I ran 20 hrs of blend at these settings, 3Dmark05,3Dmark06, win memtest for 1000%...so no issues with that
Mind you I set my cpu to 380x8=3420 @ 1.4625 which was known to run 24hrs of blend(3Dmark etc) with ram at cas 4-4-4-4-18-2t 400,async
The only change was sync ram and lowered its cas settings(from 500mhz 5-6-5-5) because when it is at 400 or less mhz it sometimes at cas5 wont boot
The hardware could fail at any time of course but I highly doubt the PSU died at that exact moment
RAM was never on more than 2.2v(verified this is 100% ok with crucial)
The board will not clear or killed my ram or something...seems bizarre to me....I left battery out, clear cmos jumper set, atx/8pin removed, unplugged DVDrw, HDD..only thing left plugged in a few fan..unplugged speakers etc
All I get is a C1 error.....this board is so frustrating..yes in theory I was tinkering but I never had a board that you could not clear the CMOS trying the above stuff
If the board is dead this amazingly fragile crap...I mean 3400 sync 380 1:1 as a test and it wont post
My crucial when first installed would post in slot 1&3 at 1.8v the board defaulted too: but the cmos is not clearing to allow ram to work IMHO its the bios not resetting itself back to defaults....
I am trying to decide how to proceed I am going to get a cheap stick of ram that I hope is rated to 1.8v....and hopefully post..if not RMAing it...and will sell the new one
I think I will go with BadAxe2...seems like you cant get a great fsb but you can get 380/390 with this crap
Look the board is sold to people like us..who tinker...I need nothing really extreme.....
I tried BA2 after RD600. I had a ver 505 of the board & it sucked. Legacy USB issues (BIG TIME) CPUz reporting Multi at 6 when it was Bios LOCKED at 9. ROFL....I have an ES CPU dammit! :mad: ....Board is a right messy farce from what I saw of it. I read elsewhere they are now making over-clocking floppy disks now!?!?!?...I didn`t bother reading anymore cr4p & RMA'd it's ass along with the OCZ 9200 & went back to RD600 & Geil 4300. Bank is better off for £480 anyways; Legacy USB is fine & CPU Mhz-idge is way better; & resetting CMOS is easy compared to needing 3 hands the size of an Aztec Pigmy on BA2. The ONLY plus for me on the BA2 was Matrix Raid. :eek:Quote:
Originally Posted by nealh
Dunno about the 505, but earlier Revs still need V-Mods to go much further.
Just don`t forget, as fragile as DFI LP mobos have been in the past for me, it's a case of when you have a good one, just turn all the right taps on ....& it works without performing Hardware Mods.
*edit* I did read somewhere that anything over around 370fsb ( I can't remember exactly) the Strap will need to be set @ 300 or above for stable Fast FSB's. I'm wondering if the 266 Strap @ 372fsb 1:1 is right on the edge & thats why I get the Odd corrupt boot sometimes!? Like I said earlier, powering off & back on gets rid of this. Someone like Tony or Raja might be able to shed some light on this subject!? ;)
Any ideas fellas?
Best Wishes,
Kin :)
Good points..why I bought RD600Quote:
Originally Posted by Kin Hell
so were RMAing you dead mobos to DFI?
NO issues with BadAxe2 here, except for slot-distance...not enough room for my DD Tyee blocks.
I'll say it again..RD600 is Intel version of RD580. Very old chipset, in the scheme of things.
CAS5 not working right is because you must adjust secondary timings. Board is tweaked for 1200mhz 4-4-4-x. Well, bios is, anyway.