I believe the 4-15-05 beta and 5-4-05 official have better "rev.E support". You can find them (4.0 Vdimm modded by TicTac) here http://www.tictac.katrourke.com/modu...ownload&cid=14Quote:
Originally Posted by iddqd
Printable View
I believe the 4-15-05 beta and 5-4-05 official have better "rev.E support". You can find them (4.0 Vdimm modded by TicTac) here http://www.tictac.katrourke.com/modu...ownload&cid=14Quote:
Originally Posted by iddqd
Just so you guys are aware, the 5/04 BIOS with 4.0VDIMM option provided the best stability for my set-up. My SuperPI 32M clocks were in the neighborhood of 60MHz lower with the 4/15. ;)
Have fun, guys...and keep on pushin' those clocks. :up:
well im tryin right now im at 2760 with 230fsb on a 180 devider
for some reason at ddr 200 2-2-5-7-16-2 my copm fails horrible on memtest
I have to have cpc disabled to run it at any speed or hundreds of thousands of errors.
Is there some kinds of timings specifically for speed premium bh-5 utt?
I had it at 3.4 and was failing at ddr 200 1-1 with fsb with those timings
I know this baby can go way higher
right now im at 1.45 volta idle at 32-34 so i know i have room to go with this setup
Someone pla help me out with some timings i want 250 1-1 X12 multi maybe 11 multi well see
until then ima sit right here at 2760 :slapass:
Have you tested your memory on another system to verify that it can reach 250MHz and beyond?
well today i tested each module individually and each can hit 240 at 3.2 which is a good starting point for me thats with cpc on also single sticks
problem is when i put both sticks inthere i have to turn cpc off and set fsb to 235 with multi of 166 which is actually ddr 195 :(
i know there has to be a setting in bios i have th change in order to ge tthese sticks to run together 1:1 with fsb
plz help me out here fellas
i had a LOOOOTTTT of trouble getting both of my ballistix sticks over 210mhz when i first got the newark... forget everything you know about timings, and start over again. especially trc/trfc (8/10 is best for me, higher gives errors, lower gives raid detect) :stick:
also, max async latency and read preamble are very picky for me.. :(
cool ozzi anyway u can post me what u have so i can try them out and go from there
I cant seem to get anything stable because of this memory stuffs
like a screenie of a64 tweaker would be nice
or maybe a screenshot or bios whichever u want
so i can get an idea
right now im 2-2-5-2-8-16-2 drive strength 1
and not workin
even with cpc off
i want 1:1 250 cpc on :0
actualy right now i would be happy with anything with cpc on and over 200 at this point. im getting errors in everntthing i dop
cant game or anything
last night iwads fraggin at 1760 now i cant
wierd
i dunno but i hope i can fix this soon
its killin me
at 2820 fully loaded it goes to 37 mostly sometim4s hits 38 :)
at 1.375 110%
plz help me get this figured out.
atleast i know someone else had same problem and it was resolved
so i know im not gonna be throw to the dogs
as we speak im at 2-2-2-11-13-15 old bh5 timig to see if they work out
so far priming ok
better than 5 8 16
my ballistix like drive strength of 1, and bank interleaving on. i actually get more errors with interleaving off :confused:
and here's my timings for 229mhz on the ballistix.
to get to 245mhz with 2.8v, i need to drop twr and trtw to 3.
i have async latency and read preamble on auto, as it seems to work quite well
also, tref is very picky. mine likes a weird 0776.. when i crank up the speed, somewhere in the 0900's works a bit better, but not much.
i hope this helps some :D
as a side note, i have two single sided bh-5 sticks sitting around, want me to pop them in and take it up to 270mhz or so?
Yes plz :)Quote:
Originally Posted by ozzimark
sorry for the delay, i had a few things to take care of. currently pushing 265mhz with 3.2v, trying to tweak the last bit out without bumping the voltage again. i'll get a screen when i max the stuff out
EDIT!
here we go. ended up needing 3.3v to make 265mhz stable.. the trick is in the trc and trfc again. i could change most of the other things to whatever i wanted, and it works
Finally got my 3700+ running, for now just trying to find out its max at default voltage. Currently feeling out stability at 2.5GHz. :)
i finally switched my bios to 4-15 and now my mem will run higher, just tryin to get errors out of prime now
best ive done so far is 5 mins :( @ 230 with 3.2 volts :(
im at 230X11 2530 at stock volts as well
im tryin 1.5-2-2-5-10-12 and some other various things to try stability :(
not doin to well so far, im sure some of its prob the psu I set it to 3.3 or 3.4 in bios but it never goes higher than 3.16 is that why most ppl have ddr booster? because nf3 wont let it go higher?
I hope it will because if not that blows.
well for now tweaking has to stop because my all in wonmder is starting to record pimp my ride for me :) so prime will have to wait :banana:
you need to being the 3.3v rail up, the vdimm will only go up to ~.15v under what the 3.3v rail is. when i want 3.4v for my ram, i have to bring the 3.3v rail up to 3.6v ;)Quote:
Originally Posted by Philosophy
Are these avalable retail yet?
i believe there is a member here who ordered one from shentech.com and get it in a while ago :D
just got my 3700 from shentech. its CABSE 0515UPMW. ive been following this thread pretty much since it started. i finally just went ahead and ordered one up to see what i might get out of it. its gonna be watercooled so im lookin at some decent OC potential....**crossing fingers**. im a regular reader at ocforums, which is where i first heard about this thread. good job on all the work u guys have done so far, keep it up. :banana:
Let us know how it goes! :cool: Oh BTW, :welcome: to XS!Quote:
Originally Posted by DmatkO
PS: Did you pay the $23 shipping fee at Shentech?! :eek:
yeah, i did pay the shipping, but i didnt have to pay tax. only odd thing about the whole ordeal was where it shipped from. from the tracking directions, the chip came from California :confused: .still, as bad as that seems, i cant imagine paying newegg $10.00 to ship a $12.00 fan. i think i would just end myself at that point lol.
My 3400 from Shentech came out of Pennsylvania, but the company has a NY address. My 24/7 stable OC is 2695 at 1.595v, less than impressive. I can bench and play games at up to 2805 1.65v so far without problems, but can't pass Prime95 above 2695 (23.8 32bit or 24.11 64bit). Temps on air are decent - 38 C peak at 1.595v (SLK-948U + Delta 92mm 3-blade 2800rpm). Any other 3400 results?
Maybe you should give it more volts (1.65V). I’ve followed this thread right form the beginning and I am waiting more results for 3400+ (3700+ is too expensive for me :().
P.S. I already have a crappy 3400+ mobile :slapass: Clawhammer, I don’t need another bad 3400+.
Hey guys, I saw some place that the Sandies (?) need Drive Strength 3 to run right on the DFI 939 boards. Seems to help a bit on my Newark too. Give it a try to see if your RAM likes that better or not.
hey guys
i have been following this thread just for a few hours....
when i posted to get info about 754 mobile....and after that i saw this one
its 4 am here i am sleepy as hell but i would like to ask u guys a couple of things
that "sentech"..i think its correct, ships worldwide......so i was wondering if any of u owners of newark cpu s live in europe and have tryed the services of that company?
r u aware of any dealer in europe?
the costum rates can be very high shipping from us...so i was just trying to do things the fast and cheapper way
i am not changing to 939 right now and i was very well impressed with the results i have seen here
can someone help me find one of these "babys" near my small and distant Portugal
i want to join the 3,00mhz 754 club
:banana: :banana: :banana:
thnxs
i finally have put my system together. now comes the interesting part i need some help with. plz take note of my sig to see system specs. a few posts up i posted my 3700 "signature" 1. my 3700 on startup is running at 800mhz, and .950volts. would a bios update fix this?? 2. according to clockgen, i have a max multi of x12, max volts of 1.4. 3. as of right now, i have the system set at a x7 multi, .975v, and 210 htt. the memory is runing 1:1 at 2226. the chip is runing at 1470mhz. iv done a little oc on my old system, but i was hoping to get more in depth with this one. so far, im not lookin to optimistic.
one last thing i would like to know is what kind of temp monitor people are using. i dont see an option for the dfi board in mbm5. thnx for the help.
i use smartguardian, check dfi's site for it ;)
also, when on auto, the dfi board defaults to 4x and 0.9v.. i think i mentioned this in the first post. simply change it to 12x and 1.35v for stock :p
lol...man, i read this thread from front to back about 4 times already. guess i didnt see that part, or was just to giddy with joy to pay close attention. o yeah, i also d/l mbm5 and did a quick little change to the data folder. i swapped it with soundx98 folder, from dfi-street.com. im pretty happy with the setup he had since my 5v line is actually now staying at 5v. it was usually jumping around and would be reading anywhere from 2.38 to an extreme 6.85. thnx alot ozzi for the reply today.
ouch! :stick:Quote:
Originally Posted by DmatkO
haha, well it's good to hear that you're basically set. good luck overclocking it! :D
edit: just a random thing i did for kicks. nothing under 2.5x ht link multi seems to work right, and greater than 1.7v for the chipset is weirdness, so i topped out here:
http://valid.x86-secret.com/show_oc?id=23871
ozzi,what was your's max suicide SS and spi 1mb clocks on AIR ?
And the ambient temp ?
hehe, i guess this thread is getting a bit too long to read, but the max suicide is definitly in the first post ;)
max 1m stable was somewhere around 2950mhz.. was just trying to find a screen of it in this thread, but couldn't, so here it is:
ambient was around 14-15c
I interrupt this thread to bring you my new cat avatar. w00t!
thanksQuote:
Originally Posted by ozzimark
a good clocker,this newark :)
yes it is.
hey ozzi, i was wondering how your running your setup now. im clocked at 2806mhz; a 12x multi; 1.47v; 25c idle; 1:1 mem at 233mhz 2-3-2-8 and 3.05v. i was just wondering how my upmw compares to your dpmw. i think my limiting factor is my ram, i might try to save up for 1 gig matched sticks of tccd or tcc5 ram in the future. overall, im still happy with the performance of this little chip. thnx
quick question, what exactly is the HTL in central brain identifier?
i normally run 229x12 for 2.75ghz with 1.45v that undervolts to 1.42-1.43v (max stable with 1.45v is 2765, but i like to have a wee bit of headroom just in case :p: )
ram is 1:1 at 2.5-2-2-10 :D
and HTL = hyperstransport link :toast:
seems like our chips are pretty similar
Just got my Newark 3700+ in from ComputerBrain.com, sitting here at stock speeds atm, lol kind of nervous after mounting up the XP-90C, but it's ok.
Mine is stepping CABSE 0515UPMW, anyone know if they are good/bad? I guess we will see...
I really hope it's good :)
*EDIT* Btw it's running pretty cool so far. Ran a SuperPi 1MB at stock 2400mhz 1.35v, fan speed on lowest setting, and temps reached 28C :lol: Though A/C is on in my room ATM.
same as mine!!!!...so far, im gettin pretty much the same results as ozzi. only thing is that im at 1.475v right now. im tryin to burn the chip in and see if i can relax my volts a bit more with equal timings. g/l and post up what you end up with. im still in the process of finding a happy medium with my upmw.
thnx for the info ozzi. im using your example as a barrier for my chip. :D
im hatin this setup ovberall over again, other day after folding for hours on end i come back to pc and its ruhnning sluggish as usual, i restart, and when come to windows same thing. Check some things and see its only showing one 512, wierd shut down and open her up, and I dont get it its in socket and all so i pull it and reseat it and no post, was at 2880 on 4-15 in dimms 1 and 3 now i cant get it to go past nf3ut screen at begining of post it hangs there, right now im on 5-04 200htt with 166 multi just to get into windows to post these problems, i cant get 200 1:1 stable on 5-04 but in 4-15 i was up to 232 at one point 1:1 1.5-2-2-5-2-10-12 etc.
now after folding all night come back and dim isntshowing, now i can pout in 1 and 2 slot on 4-15 and itll post but cant get 200 stable on 5-04 i can do 1-2 slot and 1-3 slot but no 200 stable, only thing i had b4 was 1-3slot on 4-15 bios 232 stable, now it locks at bios start screen, i have already flashed biossavior many times to both bios's is it bad mobo? Either dimm will do 130 solo in slot one so whats the deal here?
Lol, I'm retarded :with: I even quoted you on previous page of thread, but forgot your stepping.Quote:
Originally Posted by DmatkO
I'm pleased to here someone else has one just like mine! :toast: Your's looks good. I'll def post here once I see how mine OC's :)
Btw, anything in particular I should know before I start? (about BIOS settings or whatever) Thnx.
that's really weird. did you try leaving all the settings on auto, but the main ones in the bios?Quote:
Originally Posted by Philosophy
i have noticed that tweaking ram on the newark is... picky at best. :(
from what i have gathered so far, newark no like dividers. as far as bios, i have the 5/4 bios running, and it seems ok so far. i dumbed the chip down today to 2.7ghz:running at x12 multi with 1.45v pretty stable. i also set my ram back to its original settings of 2326 with 2.9v. i pushed the ram as high as 3.1v and 2.5-3-2-10 and some other combos. it all pretty much led to the same end result of 2.7ghz. i figured to drop the volts and just leave it as it is for now. i cant seem to get any more than 225mhz out of my ram. it just wont pass prime no way, no how. i figure im just gonna keep these settings till i can get my hands on some better ram in the future. as far as bios settings, i just tried and did what ozzi has done so far. load optimized defaults, save and exit. reenter and up the volts to whatever tickles your fancy i guess. ive only gone as far as 1.55, but since i didnt see any better results anyway, i dropped it back down. anyone think that 1 gig of dual channel ecc pc4000 ram will yield better results?? in cbid it says that it supports ecc ram. give it shot??? i mean, im building my first dually anyway. if it dont work, i can just pop it into there.
ECC ram won't work in these setups ;) (at least I'm pretty sure it won't)
I'm pretty confident that this statement would be more dependent on what type of memory is being used. I am able to run memory dividers down to 100/200 (100MHz) with my 3700+ and OCZ PC-3200 VX without issue.Quote:
Originally Posted by DmatkO
so then its a question of which ram to get than. any recommendations greatly appreciated.
im guessin this mobo has taken a dump on me, ima call the egg tomorrow and see what can be done about it, right now i took the bh-5 out and put in a spare gig of ch-5 and same results, so i know its not the ram. maybe thats why i have been having all these prblems all this time with this setup, we shall see, sucks to have to wait for rma of mobo but atleast i have 2 spare beasts to use in the meantime
:slapass:
Little update on my 3400. Prime95 is still the only thing I have problems with. This chip is everything-but-Prime stable at 2805 1.61v. I've always used P95 to determine stability before, but since others report having problems passing Prime I've been using other methods. It folds without errors, benches without problems, hasn't crashed on me once at present speeds. There seems to be a lot of debate about Prime95 and A64s, but I think the jury is still out.
I know my vcore is pretty high compared to what others use for 24/7 use, but temps are decent - 38 C peak under load on air. I'm going to see how high I can take her and still bench stable. 2915 is as far as I've gone so far, so there's some more left I think. And Ozzi, mine doesn't seem to be too picky about RAM timings as you said yours is. But I've only tested mine with BH5 and CH-UTT. Still waiting to see more 3400 results, but they're hard to find so far.
i agree. i have weird issues with my ballistix, but none with my bh-5.. i think it's more a relationship between the board and ram though, as my neo wasn't nearly as finicky :stick:Quote:
Originally Posted by conrad.maranan
i had the same problem with my prime. super pi 32m would pass without errors on multiple loops. as soon as i started to prime, it would fail within 4-12 min. i got to the point where i didnt care if it failed prime and just let it fold. end result was me coming into my room to see the computer off. from that point, ive decided not to question prime anymore.
im still blaming the ram and will probably end up switching it. i dont wanna pump too much volts through it so im probably gonna end up ordering some 2x512 ocz gold pc3200 from the egg when it gets in stock.
The wierd thing with mine is it was fine prime stable at 2800 and all with ultra tight timings and 230 1:1 only limited because crapper psu. All sudden one day it wont even post with nothing changed? and now im on bios it pr4viously wouldnt post on but cant pass memtest at loose timings at 200?
I think somewhere my dimm 3 has taken a dump on me somehow, as i said before both sticks will 240 with 3.1 in memtest single but once together cant touch 200 without over 2k errors after playin with timings all d@mn day. It sucks to have to rma the board but i think it is something with dimm 3, now i know it sounds noobish and all but it was workin fine the day b4 and nothing has changes since then, ive done everything in my power to try to get it back to where it previously was with no luck whatsoever, and now i gues sit has came to RMA :(
Is the DFI the only board that can run the Newark? I'm still running an Epox 8KDA3J.
it works on my k8n neo platinum too, but that's one of the boards that is known to support mobiles..
i don't think epox does :(
Is it an actual hardware limitation or is Epox just not supporting the chips in the BIOS?
i want a newark!!! lol. can someone tell me where to purchase one? or are they even out yet? i've heard rumors that they haven't been released yet.... sure looks like a few people have gotten them anyway though o_O
im noticing that this chip is getting better over time. current specs are::2760mhz=1.45v=30c load running folding@home=230x12=2-2-2-5 1:1 mem at 3.1v. next thing im gonna do is replace that stupid chipset heatsink on the dfi...WAYYY TO HOT!!!. other mods will include some new bga ramsinks for the 6600gt. so far, im really happy with the vid card running at stock speeds, so i probably wont be OCing it. other than that, i think im gonna be done with this round of building. gonna start saving up and preparing for my duallie build, and eperimentation with TEC'S. :banana: ((dunno why i put that there, it just cracks me up... :banana: :banana: :banana: :banana: lmao))
nice! that's exactly the same as mine, but 2c cooler :p: :woot:Quote:
Originally Posted by DmatkO
Well I got back from out of town last night, so this morning before I run off to work I had a quick play with this Newark (got the stupid thing a week ago and haven't been able to test it until now :mad: )
Preliminary results look good :) Stepping is CABSE 0515UPMW, cooled by XP-90C. A/C is on ATM so my room is nice and cool. Since I have no AGP card in ATM, I can only test with SuperPi (I prefer 3D), but here is what I got:
http://photobucket.com/albums/v223/t...875mhz155v.jpg
Voltage was set to 1.55v, mem with 166 divider to rule it out. Temps were ~32-34C while running SuperPi 1MB.
When I tried 2886mhz it blue-screened right away when I tried SPi. So I kicked up the volts again, 1.55v x 104% (I had just been running 1.55v with OverVID at AUTO.... so basically I just changed from AUTO to 104%). Tried the same 2886mhz and it ran, but when I tried 2909mhz it hardlocked with a couple seconds of starting SuperPi. Temps were ~36C or so iirc.
Still, I think I have an okay chip :) I hope it 3D's at 2800mhz+ no problems when I put in my brand new X850.
And more importantly, I hope it goes far when I put it under dry ice :cool:
-William
looks good so far. when i first got the chip it would bsod pretty much at anything 2900 and over. it seems like its getting better over time. i couldnt quite stay stable but i did get it to 2980mhz with alot of fiddling around. i opted for a stable system so im gonna just stay at my nice cool 2760 for now. in a month i may start trying to mess around with the settings again. i may end up d/l the dim vmod version of the dfi bios from dfi-street. i got the ocz gold ram a week ago and i want to pump some more juice through it. i wanna see how far i can push this lil guy at 1:1 at 2-2-2-5...
anyone wanna tell me where i can pick of these up in the uk and for how much?
just for kicks, i'm trying to get cool & quiet working.
i downloaded the mobile amd drivers, set minimal power management and set c&q from disabled to auto in the bios
now it changes the cpu multi, but not the voltage. anyone got any ideas on how to get the second half of the "cool" part? :D
What type of air coolign are you using? Can i expect the same overclock virtually if I go newark? How about lower models of it?
Anything over 2.9 Ghz is exciting!
But how on earth only on air?
zalman cnps-7000. the 3.1ghz screen was with ~14-15c air. :D
The DFI nF3 has a few bugs with mobile CPUs; the voltage thing with C'n'Q is one of them. I currently have a 3400+ mobile Clawhammer and no matter what I do, VCore stays “locked”.Quote:
Originally Posted by ozzimark
hmmm.. curious. i don't really remember it working right with my 3200+ clawhammer either :(
oh well.
EDIT!!!! :woot: :woot: :woot:
it's working!
i cleared the bios, left the cpu multi and VID at auto, and changed cool + quiet to auto
get into windows, see that it's at 800mhz, .95v, move a window around a bunch and saw it jump up to 2400mhz and 1.35v :toast:
now to get the % over VID working, and get the thing oc'ed again. :stick:
OZZY you're the man :toast: . You can use RMClock to adjust FID/VID in Windows (http://cpu.rightmark.org/products/rmclock.shtml), very very nice software.
Edit: when using RMClock you need to disable C'n'Q in the BIOS, RMClock takes over...
thanks, have you tried it too?
and i'm using xp-64... i would use rmclock (i tried actually), but it needs 64bit drivers, so i'm tweaking my oc to use stock multi and vid, changing only % over vid and htt/fsb :D
I am using it, but version 1.4. The newest version 1.5 is a little bit strange with P-State transitions...
well, with my testing, i have discovered something very very weird on this board.
if you leave the VID on auto.. it'll just stay where it was when you reboot.. and from a cold boot too. the only way for me to get VID to stock when it's on auto is to kill the psu power for a few seconds so the light on the boards go out, then it'll drop back down to it's normal 0.95v
right now i'm at 236mhz, 12x multi, "auto" VID with 113% (1.52v load, 1.02v idle) and ram 1:1 at 2.5-2-2-10, seems to be doing quite well :toast:
Wait wait wait.. you're using Air cooling with the Zalman 7000 at 15 C?
how is that possible? Where on earth do you keep your computer? In a freezer?
i thought i had a picture of my old cooling setup posted in this thread somewhere. oh well, here goes:Quote:
Originally Posted by D U R A N D A L
:lol: that was a nice setup.
LMAO..........cant think straight after last night, but i :clap: you for the custom setup. watever works i guess.....
Just ordered a 3400+ Newark from Computer Brain. Should be here early next week. Can't wait!
ozzi dude that is totally hilarious bro, keep up the good work on them cold air mods, who knows maybe one day ull be top aftermarket cooler maker in the world, Im thinking of flex tubing some cold air from my central air unit being sucked in by a 92 tornado as well but a 25' run of flex with 5 kids and a cat is gonne be tricky.. Totally awesome swiss cake roll duct mod work bro, im laughing too hard right now. Might be the beer but right now i wanna go tear apart the box of moon pies i have downstairs just drooling over ur cooling duct mod soultion :) :toast:
Keep us posted with your OC-ing results. I still have to decide between a 3400+ or 3700+ (not enough ca$h for a 3700+ :( ).Quote:
Originally Posted by sKiDmArK
very nice results! Where can I get one of these 3700+ Newark chips? :D
-CaT
Vandi had all of the good ones, but I got the last one Cat. :P
Any idea where can i get me a Turion64 chip? :D
Several places sell Newarks, but i need Lancaster...
Quote:
Originally Posted by Korpse
i am in portugal and the answer to this would suit me just fine
anyone in europe who has bought a newark?
come on.....we want to join the newark "club"
i only see it selling at usa and canada....but customs are quite expensive...
thnxs
Hey Eclipse and Conrad - what BIOS do you guys use? Still 4/15 Betas? I have 10/15 but that doesn't seem to work. I'm not sure what's more recent, as the numbering scheme baffles me. Would be nice if it included a year... ^^
are there gunna be dual core S754 chips? if so when?
Quote:
Originally Posted by Korpse
uhm no, the amount of power needed can't be given by 754 pins, it needs 939. And as dual cores 'require' a big memory bandwidth, I think singlechannel just wouldnt do, so a s754 dual core just would be unlogical as it wouldnt perform as it should do (on 939).
correct me if im wrong
5/4 :DQuote:
Originally Posted by iddqd
biohead, s754 won't have a problem powering a dual core cpu, the extra pins are purely for the memory
Has anyone gotten a chance to try 2x 1gig stix on these?
you are right on the no dual core for s754 but wrong as the reasoning. dual cores would run fine on the motherboards and socket. amd just won't implement them because intel hasn't set up a dual core pentium-m mobile chip yet and the market is small for dual core workstation laptops. not because the socket can't support the power and single channel would kill performance.(both of which are not true)
remember the desktop s754 is officially dead but has stayed alive due to the mobile chips.
Quote:
Originally Posted by biohead
Linky plz?
for the bios i would assume? it's the latest official from dfi :p:Quote:
Originally Posted by iddqd
http://www.dfi.com.tw/Support/Downlo...FLAG=A&SITE=NA
Bah, the 4.0 vdimm version.
actually, single channel would kill the performance. There is not nearly enough bandwidth to support the needs of the 2 cores.Quote:
Originally Posted by tristancarton
sabre and/or bio (or even anyone else) any way you can provide a link to backup your statement.(maybe we can start a new thread in amd forum and get someone with a dual core system to post benchmarks in single channel mode?)
i know dual channel with 1 cpu is usually 4-8 percent faster (in most apps) than single channel with 1 cpu. my assumption was that even with a dual core cpu the difference wouldn't increase by that much due to the integrated memory controllers. perhaps it would jump to 8-15 percent but even with those numbers i don't think it is a huge deal and prob not amds reason for not pushing dual core s754 because you will still see a huge difference in going from 1 core to 2 cores.
(this will be my last post on this matter in this thread because i don't want derailment)
btw- anyone else have benchmarks for the 90nm 1mb s754 3400 newark? the 3700 is still too expensive for my liking.
Quote:
Originally Posted by sabrewolf732
Quote:
Originally Posted by tristancarton
trying to have enough banwidth for 2 cores with only 3GB's of bandwith is crazy. Run your 754 system with the ram at 100MHz.
perhaps however i ran a p3 tualatin dual cpu smp system on pc133 (talk about low bandwidth) for a while and it wasn't too big of a deal.
anyway i posted a thread asking for dual core kids to benchmark between single and dual channel. hope someone will respond. here is the thread. all discussion on this may as well move there.
http://www.xtremesystems.org/forums/...d.php?p=953638
Quote:
Originally Posted by sabrewolf732
there's a difference.. with single channel at 200mhz, you still have decent latency.. 100mhz kills itQuote:
Originally Posted by sabrewolf732
edit: seems that dangerden doesn't test as well as they should. popped off the waterblock to put on the clawhammer.. it seems i'm lucky there's no damage from what i can tell...
:slapass: :slapass: :slapass:
that newark looks like it almost got taken down! what are you plans for the meantime? keep using a shiesty block, or go back to air for a little bit.
i dunno, i'll see if i can do a temporary fix and try to get a replacement block.. these things are warranteed against leaks, yes?
dam oz ur lucky u caught it in time bro, get that water problems fixed and get her cookin again man, it woulda sucked to have her taken out.
Mines runnin like sheetz as ity is with only one 512 of speed premium cause other one took a dump on me :(
Damn Ozzi! I hope everything still works. I was going to buy that block for my Opterons but now...... FORGET ABOUT IT! :slapass:
got a question for you fellas, how do i know if my mem controller is gone?
2X512's worked b4 at 240X12 with 3.2 and now i cant get em stable at 200 in any dim combo. Either one of them will do 230 alone but together its no good with alot of different timing configs to go also. Im gonna try rmaing my dimms first as i have already rmad one mobo and if dimms dont work im guessing that mem controller is gone, is there a way to test this? Gimme some advice as to what i can try to rule this out
Right now im on one 512 at 230X12 stable as a rock but if i try other dimm in there crapps out on me.
a quick email with DD's tech support has put me back in good spirits, just make sure you do more thorough testing than i did :p:Quote:
Originally Posted by Vandi423
philo... no clue, did you clear the bios the proper way? not doing so has given me trouble in the past :slapass:
LOL i cant find a newark core for sale anywhere...?
-CaT
It's a good block. Don't send it back, just use some thread seal... the water clearly leaked through the thread on the bolts.Quote:
Originally Posted by Vandi423
Been a while since I've posted in this thread, maybe somebody already beat me to the punch. I was digging around on AMD's site and came by this product revision guide.
http://www.amd.com/us-en/assets/cont...docs/25759.pdf
Check out page 8. According to it, there are no Athlon64 mobiles based on the San Diego single core (SH-E4). Instead, the Athlon64 mobiles are based on the SH-E5 silicon of the Turion64. I had confirmed this (although I didn't realize it), as I measured power consumption of my 3700+ mobile and Turion64 at the same speed and voltage, and power consumption was basically identical. At first I interpretted this to mean that Turion64 was the same core as the San Diego (SH-E4). It wasn't until I came by this tech. document that I had even considered that Lancaster and Newark were the same core, but this appears to be the case.
Doesn't really relate to overclocking or anything, but I thought you guys might find it interesting.
Quote:
Originally Posted by frostedflakes
They are identical. All "tweaks" are transistor-level. High wattage Turion models are basically the same thing as regular mobiles. Low wattage Turions are something different, however.
Well yes, I know the architecture is identical, but it also appears that these have the same transistor tweaks to reduce power consumption as the Turion64. The 90nm 62w mobile A64s and Turions are the only processors on that list that carry the SH-E5 CPU ID.
I guess it just kind of surprised me, because me and a few other members here got in on some Turion action early, and they turned out to be very poor overclockers. Then these 90nm mobile A64s came out OC'g like mad, so we all just assumed they were the same dies as those found in the desktop San Diego's.
Guess between week 7 (what my Turion was) and week 15 (around when the first mobile A64 dies seem to have appeared) AMD worked out the bugs in their tweaked transistors to allow for higher clocks. :woot: