hey bro, you forgot to mention that I am the first one who told you this. Pay me!!:D
Printable View
hey bro, you forgot to mention that I am the first one who told you this. Pay me!!:D
A friend of mine said that he could do this with his R.AD0.
http://www.geocities.jp/oc_community...html#20060113a
Look at the Trtw.
R.AD0 is as good as Expert for memory.
But not as good as it for CPU.
Expert uses 4 phases, and so it has advantage for it.
His review translated by Google.
http://translate.google.com/translat...language_tools
I can't see the pic... :(Quote:
Originally Posted by Enz0
Sorry, I've fixed the post.Quote:
Originally Posted by g0dM@n
You would be able to see the image by clicking the link.
i think it's often a mis-set MAL/RP that makes some dividers not work ;)Quote:
Originally Posted by g0dM@n
It showed me the white box with the red "X" in it... thx tho. :DQuote:
Originally Posted by Enz0
Then, does this link work?Quote:
Originally Posted by g0dM@n
http://www.geocities.jp/oc_community...html#20060113a
Yes :)Quote:
Originally Posted by Enz0
guys, i know the Ultra-D dosnt say its got SLI...but can u actaully run in SLI on it? beacuse if i get a Ultra rev R.AD0..its going to be better than the SLI-D for overclocking memory and £30 cheaper, just for no sli sticker?
And no SLI bridge. And no rights after doing the SLI mod ;)Quote:
Originally Posted by Vex
the ultra-d comes with a sli bridge tho?Quote:
Originally Posted by SlackeR
http://www.overclockers.co.uk/acatalog/Socket_939.html
whats envoloved in the mod? just jumper swapping?
http://xtremesystems.org/forums/show...110#post844110Quote:
Originally Posted by Vex
http://xtremesystems.org/forums/show...658#post856658
Ultra-D does NOT come with the bridge.. You can email Donna @ dfi and get a bridge from her.. I dont have her email handy but ill update this if i find it..
now here's an interesting development. on a whim, i looked at the layout of stuff on my board:
http://eclipseoc.com/phpwcms_filestorage/1/1_155.jpg
huh.. that's weird. i got this thing back in the beginning of october. it's an A.C0:
http://eclipseoc.com/phpwcms_filestorage/1/1_156.jpg
but it has the "new" layout. i think there's more of these out there than we thought before ;)
I got my AD0 Ultra D in yesterday and will drop in tonight. I'll compare the layouts when I get my old board out. Your AC0 is the first I've heard with the revised layout (other than AD0 in the Ultra D). I've got all my settings wrote down from the bios, so it'll be interesteing to see if there is a difference. The AD0 will probably have a different bios since mine is using the (6/25)?
i think all A.C0's have the same basic layout as my board :toast:
One thing water users may like is, the socket on the revised board is very nice, the high part of the socket now sits lower than your waterblock/cpu, there is no longer a problem of bad contact due to 1 side of the socket being too high. :) Kudos DFI.
My socket is the same way but it isn't the new revision.
and mine isn't the same way, i had to shave a bit off when i went lidless ;)
mine was. so that's good for me, and I am lidless ;)
Hmmmm, not quite sure what you're talking about :confused: I just installed some water on my lidless FX55 on an AB0 SLI-D and didn't have any issues......Quote:
Originally Posted by fareastgq
http://img17.imageshack.us/img17/9268/sidewc8cr.jpg
http://img468.imageshack.us/img468/4...tertemp6pa.jpg
I swapped in the AD0 last night (replacing my AB0). Stability-wise, it went from a stable 305HTT to unstable 310HTT to a stable 320HTT and unstable 325HTT. Otherwise, like some of the others, my temps now average about 5-7F higher and now fluctuate more then before.
Voltages appear to be more stable across the entire board. I see zero fluctuation across the core now.
Overclocks are the same for the same voltages.
The temps are probably higher b/c of difference in temp sensor. I doubt your CPU is actually running hotter. Also, the same voltage settings on the AD0 board may be pushing more volts than the other.Quote:
Originally Posted by Rapier
I doubt it directly affected the voltage:temp ratio. It's probably just offset figures... :D
--------------------
s7e9h3n, I have the same block. I was using it on a no IHS Opt148, and since the IHS was missing, I had to remove the mobo bracket, so that I can make contact with the die. I think b/c of it I warped the metal bracket on the block. Now I have to use the block without a mobo bracket everytime, even with my Opt170 that still has the IHS in tact.
I love the block, though... with and without the mobo bracket... :)
see the top of your block in that pic? that side of the Socket sits a bit higher, on the very first nf4 boards out, that little side of the socket was up too high, so when you took off your IHS, the block came down and hit the socket on that side, thus bad contact unless you really clamped down your cpu, and even then it would be uneven. I don't know if the oem socket maker is manufac. them diff now or what to fix the problem, but it's all good on mine now. Now that I think about it, it was a common problem with all nf4 based boards, not just DFI. Seems like the problem has lessened now though, not many are talking about it anymore.Quote:
Originally Posted by s7e9h3n
I never even knew of an issue... :DQuote:
Originally Posted by fareastgq
I'm not sure which one of my boards that the bracket I'm currently using came from (I've got 5 DFI's), but all I had to do was take the little spacers out which sit between the mobo bracket and the w/b bracket. No problem making contact with the core that way ;)Quote:
Originally Posted by g0dM@n
Or you could do that :), but you did that where your CPU has no IHS? I didn't even try that (stupid me).Quote:
Originally Posted by s7e9h3n
It's the FX56 (the one in my sig) - pictured in the middle ;)Quote:
Originally Posted by g0dM@n
http://img435.imageshack.us/img435/7261/trio6mn.jpg
i was under the impression that the temp sensor for the cpu was inside the cpu, so any change in temperature would be a direct result of more heat from the cpu?Quote:
Originally Posted by g0dM@n
there's a thermal resistor in the cpu who's resistance is converted into a temperature by a sensor chip on the mobo, if my understanding is correct ;)Quote:
Originally Posted by aiya_au
Yes you are - ITE IT8712F-A handles all the temperature monitoring on the boards ;)Quote:
Originally Posted by ozzimark
^^^ sorry for the OT... I recieved the CPU safe and fine today. :)
-Edit..
And back on topic, I did the same when I was mounting my storm on my old naked 3700+, but alas mine had idiot pads, so I couldnt really screw anything up, When the new cpu I didnt even bother putting the troublesome lotes bracket (The one with the lugs for stock hsf..) because it isnt really needed.
Good to hear....make sure to keep me up to date on how it does. ;) Drop me a PM if you have any further questions.....Quote:
Originally Posted by JasonDTM
Where is the black gooey stuff on the other edges of the CPU? Did you get it off somehow? Mine has the black glue as a square around the die... not that it affects anything, but just wondering where yours went... :DQuote:
Originally Posted by s7e9h3n
I cleaned my cpu :D I thought it was ugly so I used a bit of NAPTHA and a q-tip to take if off.....Quote:
Originally Posted by g0dM@n
Well, how about that? I think I now understand why OPB prefers his Rev. AD0 over his Expert. I finally got a chance to mount up the AD0 on the autocascade last night. A few hours tweaking the board bested my few weeks worth of work with the Expert on Spi1m. I have a feeling a better insulation job helped matters, but regardless, I was able to run at a higher frequency and netted a better time using LESS VOLTAGE with my Ultra-D:
Here's my best 1m with my Expert:
http://www.cryo-laboratory.com/uploa...h3n/14x265.jpg
And here's the Ultra-D:
http://www.cryo-laboratory.com/uploa...7/14x2681m.jpg
Good stuff s7e9h3n, AD0 Ultra-D looking really strong. :thumbsup:
Stephen, since i'm too lazy to read through all the posts...can you confirm that it is in fact a better clocker...and that these simple changes are now supplying a better/cleaner power almost?
And, are all the boards that people purchasing now, getting a Rev. AD0? or is it something different? And...one more thing...is the SLI mod, still do-able with these new revisions?
It seems from the last post that it clocks better. Other than Stephen and opb, i don't think anyone has spoken out.. Oh yes, one guy said that he had a LOT of boards in his hands and at som point they got better. It was this layout.
No one has really measured anything, at least not scoped.
Most Ultra-d's should be AD0.
SLI-mod is doable with all DFI boards.
A mobo (or any hardware for that matter) is only as capable as the person's hands it's in ;)Quote:
Originally Posted by Reinvented
In terms of better/cleaner power - I will say this: The board doesn't allow me to boot quite as high a Vcore as I'm able to on the Expert. I'd say a good .02V difference between them.....
Steven, don't you have an alternative vtt mod than what's public?
THis board isn't modded ;) In fact, I hadn't needed the mod on my last 5 or so DFI's......Quote:
Originally Posted by Mr. Tinker
hmm... I thought vtt tracks pretty poorly. EMC2 hasn't been around to finish off his vtt mod thingy. I read that you and Tony had an alternative vtt mod that was under NDA somehow.
well i've got an AD0, lots of bh-5 and i think the vtt may still be an issue cause my bh-5 sucks on this mobo. (2x256 is cool but 2x512 or 4x256single sided is forced to use 2T for s'pi)
i can barely hit 225fsb with 1T timings on the shipping 623 bios. while i'm used to the ram going well over 250 on anything else. (i tried all drive strenght too)
should i just flash this thing to tony's 702 bios or what?
i'm avoiding flashing till an expert ;) on this mobo/ram combo thinks its worth it.
i hesitate cause i once flashed a mobo and it never posted again.
my 3700+ kab3e has already 1M'd at over 3ghs using stock past and 'sink but with a fast 80mm fan.
i would love to get this running properly so if someone has some good words for me please speak up.
I don't know if anybody's aware of this, but ALL BIOSES ARE BASICALLY THE SAME THING only in different packaging/default settings. Albeit, some options are enbled in certain bioses which aren't enabled in others, but when it comes down to basic memory settings/timings, you really should get to know ALL the settings. Once you do, it's possible to use ANY bios for WHATEVER ram you choose (as long as that bios has all the necessary options enabled for you). For example, bios 510-1, 510-2, and bios 510-3 are the exact same bios. The only difference is what the ram response time defaults to. 510-3 has a table which is better suited for TCCD, while 510-2 was set for LL/BH type memory/ 510-1 is somewhere in between. If one knew their memory timings well, they could use any of the bioses and have them perform equally. The bios itself on the most part does not help performance, it's all up to the user ;)Quote:
Originally Posted by flytek
thanks. i pretty much new that and have a64tweaker'd my bh-5 before proper testing. (to the bench settings you and the other pro's use)
so that means the vtt is the damn issue and only EMC2 has a proper solution to my problem.
unless you know something we don't.
i'm starting to feel i need to learn enough about circuits so i can construct a vtt bypass circiut to fix this dfi engineering foulup
anyway i'm getting rid of this board tommorrow and i might just wait for a yonah board before trying to put together a half decent rig again. or maybe an expert has better vtt.
thanks for taking the time :)
http://i20.photobucket.com/albums/b242/flytek/27.jpg
Oh, one last thing I forgot to add.....I talked to OPB last night and he verified this - The rev. AD0 ultra-d is somehow able run the cpu @ a cooler temp than older revisions and even the expert. I say it's just a borked sensor, but he insists that the temps are for real. Here's what I booted at on my board:
http://img392.imageshack.us/img392/6553/455pb.jpg
Have you taken vtt measurements on the board? Test the voltages @ idle and load and report back if you can.....Quote:
Originally Posted by flytek
Last thing....I promise....For those of you who are looking for Tweaker settings, I DEFINATELY SUGGEST NOT TO USE A TREF OF 200, 7.8u. I think I just set the WR for the slowest time ever @ 3.7Ghz..lol:
http://www.cryo-laboratory.com/uploa...3n/3700sux.jpg
hehe... I was wondering about that also... but I can see faint "swab" marks on the top edge, so I figured you had used some sort of safe solvent (say that 10 times fast! some sort of safe solvent ...lol) to wipe it off... good times, y0. :)Quote:
Originally Posted by s7e9h3n
wow thats awsome, we shoudl have a competition for the slowest times at the highest clocks :DQuote:
Originally Posted by s7e9h3n
:banana: :banana: :banana:Quote:
Originally Posted by s7e9h3n
i'll go out on a limb and that it is indeed the worst time i've ever seen at 3.7ghz.. on the amd side of things at least ;)
ok here's the results.Quote:
Originally Posted by s7e9h3n
at idle with 2x512mb bh-5 @ ddr466: 2.0-2-6-2, 2T (1T will crash no mattrer what i try so far)
vdimm is 3.04 solid
vtt is 1.528
loaded 32M super-pi
vdimm is 3.04 solid (no finer measurement available)
vtt fluctuates rapidly from 1.529 to 1.521 and all figures inbetween
(not looking too good is it?)
thanks agaain for the help so far :cool:
Isn't that pretty good though? Only .008v change
BTW 200MHz 7.8us TREF is 0016 cycles, gets about 1300mb/s bandwidth and 175ns latency for me. And your time with it is still better than mine with normal settings :(
Your VTT is absolutely dead on. FYI - VTT = 1/2 of VMem so looks like your board is holding your VTT levels perfect. The problem is somehwere else......;)Quote:
Originally Posted by flytek
lol...but I'm sure if your were @ 3.7Ghz, a 28sec 1m would be just a bit off....For the record, I can go >1sec faster than that on air @ 3G.....Quote:
Originally Posted by Repoman
you sure? what if i'm getting the vtt spikes scoped out by EMC2 when using 4x256 or 2x512 at 1T? he was noticing severe problems clocking bh-5 in test 8 of memtest.
eveything seems fine at 2T timings but just not at 1T :( (also 2x256 works fine at 1T )
anyway i know this is an age old problem but i just have no further ideas on how to solve it.
any further bright ideas would be most welcome
this mobo is really nice for cpu clocking
right now i'm undervolted to 1.25vcore down from 1.4 while still overclocked from 2.2 to 2.56ghz and all on the bog stock cooler. so that part is great but i would obvoiusly like to get the gig of ram to 1T.
but it really isn't the end of the world as i wont notice the 5% performance improvement anyway.
The same temperature drops when changing from old Ultra-D to rev. AD0.Quote:
Originally Posted by 4X4N
There is a temp offset between Smartguardian and the BIOS of 6C. Did you notice that too.
Marc
didn't notice much of a temp drop on my end.. went from 33/40c to 31/38c.. probably due to a better mount this time around
though one thing to note, this is especially directed towards steve.. my board has brown caps just like my .AC0 did. they're actually all brown except for the ones by the pwmic area
The largest variance i have seen on my board for vtt is 30mv. Only once though. Othre than that, pretty solid results, no more than 7mv-10mv.
IIRC EMC2 was getting 50mv-100(?) droop at some stage throughout the test 8. I think he also got 200mv at one instance.
Read this entire thread and I'm still stuck on this question:
If cost is not an issue and you were trying to get the absolute most out of a cpu and ram, like an Opty 170 @ 300HT 1:1 for example; would you recommend the Ultra-D AD0 or a new Expert?
Quote:
Originally Posted by krylon
Ultra-D, No doubt and you'd better use 1114 bios;)
that bios for dual core or will single core benefit tooQuote:
Originally Posted by Onepagebook
it's not bad hey....just realised my one is AD0 too :)
http://img108.imageshack.us/img108/2...htt32msup1.jpg
How much did you gain in overclock changing to the AD0 rev.?
Made in Taiwan better?
Cheers Monge :toast:
-;) -Quote:
Originally Posted by Monge
I have a NF4 Ultra-D rev. AC0. This board is the same that rev. AD0.
Pics.:
http://geocities.yahoo.com.br/thigob.../board_rev.jpg
http://geocities.yahoo.com.br/thigob...u/mem_reg3.jpg
Got an AD0 here, bought it a couple of weeks ago. Never owned an earlier revision so I can't really say if it's better or not. Are AD0's supposed to be made in Taiwan, because mine specificly says "china" on the box... hmm
They're made in both china and Taiwan, the latter being harder to find....Quote:
Originally Posted by masterofpuppets
Looks good.
With my Ultra-D A02 i can do 375MHz. I took this shot since few months ago, actually i made that just for fun :D . Of course, that was in single mode (with the strong dimm in a Patriot TCCD set), but i think is good HTT.
http://img493.imageshack.us/img493/3...esca7pk.th.png
A few days ago, i took the same stick, i could do a 400 MHz suicide shot, i just could boot with that frequency, no PI o other test.
http://valid.x86-secret.com/show_oc?id=74144
All that with R.A02 and 623-2 bios.
Are the Taiwan-built boards any better?Quote:
Originally Posted by s7e9h3n
not sure i never compared to earlier ones..........also i never really tried to find out the absoute max HTT.........but roughly in terms of htt i tested this one last night up to 425MHz HTT stable on 6xmulti..........i haven't tried loosening MAL any more CBFQuote:
Originally Posted by Monge
my first ultra d did 350MHz i think
my SLI-D did 375MHz
Expert does 400MHz easy i didn't even try more
and this one going past 400 no probs
i never tested RAM OC difference but stephen probably knows
one thing that is important when OCing is stable voltage regulation so i'd say you should be getting max clocks with boards like AD0 or Expert
imo, the max HTT of a board is a useless measurement. my old .AC0 couldn't get any ram over 306mhz, although i could easily break 425mhz htt... ;)
Vtt gross measurement is NOT the issue. The question is, are there any "glitches" or voltage spikes on the Vtt as there are on pre-AD0 boards.
Since the circuit looks the same, the only change that could be made to actually CURE the Vtt issue would be to use DIFFERENT value resistors as the terminator resistors.... that long line of resistors along the DIMMs. Instead of using say 47ohm, they could use 68ohm, and this would lower the MAX current the Vtt regulator would have to produce and eliminate the over-current spike the regulator was putting out under heavy load on pre-AD0 boards.
So, what is the answer. Can someone with a AD0 look at the resistor packs along the DIMMs and tell use what value they are using, and compare it to the pre-AD0 boards?
i can only compare an AD0 to an AC0, which kinda have the same layout, but they also have the same resistor values. the resistors are marked 470 along the top 2 slots, and 390 for the bottom two.Quote:
Originally Posted by uwackme
Does anybody know this?Quote:
Originally Posted by masterofpuppets
i doubt there's too much of a difference. i've got a chinese AD0 that's pretty solid.Quote:
Originally Posted by Ferry82
Got me a new Ultra-d AD0 today gonna build it in this weekend.
My old one is a A02 so hope it give my system a little more power/stability
I will place results when i have some:woot:
About sensors on NF4...
http://www2.shrani.si/files/128438046.jpg
tnx