Printable View
Finally, the kit is here ( after several delivery issues ) and I'll start testing it in an hour or so
Let's see what it can do...the usual suspects, Micron D9GTR ( aka ex Z9HWQ ) are "under the hood", let's hope for DDR3-1900+ 7-6-5-15
First tests coming soon!
Take 1:
Fast tests, nothing tweaked ( O/S, Bench, sub-timings, etc )
5-5-5-15 tests
config:
E6420 @ 350MHz FSB x 8 Multiplier
Super Talent W1866UX2G8 2GB Kit
Asus P5K3 Deluxe ( BIOS v0604 )
266MHz Strap, Transaction Booster @ 1, 1T Command Rate
http://valid.x86-secret.com/show_oc.php?id=240175
http://i8.tinypic.com/34nnxue.png
http://i12.tinypic.com/6bejqz4.jpg
http://i1.tinypic.com/6g9dor6.jpg
Max benchable 5-5-5-15 1T @ 2.25V should be around DDR2-1450
[ sorry for the lazy posts, a m8 is sleeping and this keyboard is noisy :( ]
More tests coming soon :D
More of a mobo limitations test than actual D9GTR test...
500MHz FSB on 266 Strap, Performance Level 8 ( 7 looked ok as well ), DDR3-1500 6-6-5-18 1T Command Rate, E6420 @ 3,5GHz, Untweaked ( OS, Subtimings, Bench, etc )
http://i16.tinypic.com/5xxjwqc.jpg
don't waste time
go straight for 7-6-5-18 1T and let's see some magic :D :up:
lol, your benching with your roomate sleeping next to you? :lol:
how many volts to reach 1500 665?
Think I used 2.05V for the 6-6-5 runs.
Perhaps they can run there with lower voltage, haven't tried because I didn't have much time :(
Semi-drunken friend hanging at my house because his is far away and didn't bother dropping him by :p:
push up the speed, with some mem you can get 1600+ 665 2v :D
move the sticks around, black slots and then try stick1 chA stick2 chB and then stick2 chA and stick1 chB.
lol@drunk guy in the background while you bench hahahh :D
Well, looks like my kit is unstable at default settings ( DDR3-1866 8-8-8-24 1.9V ).
It's totally unstable in SuperPi 32M ( not even initial value calculation ) and unstable in Windows Media Player also :D
Tried with 1.95V as well, without success.
Sub-timings doesn't seem to be a factor at all, strict or loose, still the same results and instability level.
Gonna try some more, maybe different BIOS and/or mobo later on.
[ current config = Asus P5K3 Deluxe, BIOS v0704, Black Slots ]
Bench, where did you get that BIOS? Is there any news on it? Could you share it?
Quote:
Originally Posted by BenchZowner
hmmmm not stable at stock speed.... that sucks... :(Quote:
Originally Posted by BenchZowner
Whats the max speed you can get stable with cas8?
Maybe the board is limiting you?
What chipset voltage are you using?
the bios is not public yet i got it from a friend and sent it to a couple of persons ;)Quote:
Originally Posted by jugeen
I'm using it and see no diference with 604 so far
Haven't tested them thoroughly with CAS 8 to say "there's the limit", but will do soon ( gonna start the testing process in 30 mins or so ).Quote:
Originally Posted by saaya
This board is a killer :D Reached DDR3-2100 with the Super Talent DDR3-1600 CL7 at 8-7-7-24, and 1900+ with 7-6-5-15 both at 1T Command Rate, and full stable the latest.
All possibly working settings have been tried, and even with "inflated" voltages nothing seems to happen ( except errors :p: )
Now why did those m*rons had to destroy my decent 1600CL7 kit on trasportation ? :(
Gonna have a chat with my contacts in Super Talent, something's definitely wrong here :D
And I'm not the only one having problems running them at their stock settings... :shrug:
yeah, i heard that many people have problems running their 1866s at stock.
im not sure if its really the mem though, ddr3 is tricky...
try this :)
stick1 in slot2 and stick2 in slot3 and then the other way around.
EDIT: what do you mean with fully stable btw?
how did you test 2100 cas8?
Already done, same results ( instability )Quote:
Originally Posted by saaya
I was talking about my previous DDR3 kit from Super Talent, the 1600CL7 with Z9HWQ chips, that did 2100 CAS 8 Benchable @ 2.35V, and 1900+ 7-6-5-15 stable ( Double Pi32M, Orthos, etc )Quote:
Originally Posted by saaya
I have a feeling that the ES codenamed Z9HWQ meant "Z9 Hardware Websites Quality" samples :p:
Off to start testing now...let's hope for something better, maybe low voltage loving ? ( hope so, else :down: )
Just tried DDR3-1866 8-8-8-24 at 1.7V/1.75V/1.8V/1.85V
1.7V & 1.75V = no POST at all
1.8V = Pi 32M 'till loop 1 or 2
1.85V = Pi 32M 'till loop 6 or 7
1.9V yesterday gave...nothing, not even initial value calculation.
Not going to give up on them easy ( never did, even with the worse HW I've ever touched )...but I don't feel like they'll ever work full stable at DDR3-1866 8-8-8-24
hmmm did you contact stt about this already?
hope they swap the sticks quickly or know some trick :D
and yeah, its a shame the new micron chips arent clocking as well as the earlier ones. the early d9s were nice as well but the recent chips are... meh...
i have a feeling that micron sent out the best chips first and now sends out the rest...
I've e-mailed my contact, but no news yet ( I'm currently in EU not in the states and there's a time difference :p: ).
But...I'm afraid that I have to retract my previous statement!
The sticks proved me wrong this time, they're stable at DDR3-1866 8-8-8-24 as they're supposed to...but...at 2.25V... 1.9V is their rated voltage from Super Talent but they're not stable there... 1.7V/1.75V/1.8V/1.85V/1.9V/1.95V/2V/2.05V/2.1V = instability ( max SuperPi 32M loop = 6 )...haven't tried 2.1V < Vddr < 2.25V yet, but will do.
At least they're stable with 2.25V :D
http://i10.tinypic.com/6f9yxro.jpg
hmmmm and what chipset voltage did you use?
1.7v right?
strange... so you can get the mem stable but need a LOT of vdimm...
hmmmm
Yup, just like you said.
Vnb = 1.7V
I can't complain, I prefer having to supply it with 2.25V to be stable at 1866 8-8-8-24 than a 1866 no go :p:
Had taken a break, I'll start benching it again in a bit :D
yeah bt 2.25v for only 1866 cas8 is a bit high ^^
im currently testing a kit that came back with an rma that does 1975 cas8 with 1.8v :D
but i dont blame the guy, ddr3 is really weird and tricky...
6-5-5-16 stable up to 1620 on the P5K3 Deluxe now, at 2.25V of course ( seems like 2.25V is their daddy :p: )
http://i16.tinypic.com/643lc38.jpg
**::.. exchanges decent DDR2 2GB Kit for a decent DDR3 2GB Kit..::** { LOL }
you can blame me :) ddr3 is an X-file. i am really sry if it sent back a good kit.
Quote:
Originally Posted by saaya
I finished all my tests for this kit just yesterday.
My final verdict it is the same of BenchZowner. These memory kit is not superPI 32M stable at rated specs.
No way I tried a lot of bios settings but nothing to do.
They are stable at 1866 8-8-8-24 with 2.05v (a little bit better of the BenchZowner kit) but this is too much voltage.
I am very confused because DDR3-1600 7-7-7-21 performs better than this kit. DDR3-1600 are superPI 32M stable at 1866 8-8-8-24 with 1.9v :confused:.
With same setings I get out from ram slots DDR3-1866 and get into ram slots DDR3-1600, and the DDR3-1600 performs better than DDR3-1600 with every bios settings, timings and voltages. Very strange :confused:.
I think that in Supertalent made a big mistake with chips selection it is not possible DDR3-1600 performs better than DDR3-1866.
I think mine work stable at 1866mhz but is weird i can boot much higher but not stable
I am a little bit confused because I do not understand if it is a trouble related to kit in my hands or it is a general trouble with these kits.Quote:
Originally Posted by metro.cl
It seems very strange that Supertalent was able to build a kit with default specs DDR3-1600 7-7-7-21 1.8v that is able to work stable at 1866 8-8-8-24 1.9v, and after they build a new kit with default specs DDR3-1866 8-8-8-24 1.9v that are not able to respect default specs.
Then I have to verify if it is a trouble with sample in my hands or it is a generalized trouble with these memories.
I think is a general problem i tried testing at 1.8v and seemed to work better than 1.9v but i have had stability issues with the kit.Quote:
Originally Posted by v_parrello
hey, its fine with me, i just hope your not sad that you swapped the kit since i think your replacement is probabaly not as great as the one you sent back ^^Quote:
Originally Posted by jugeen
i mean 1950 876 with 1.85v 100% stable, thats pretty damn good :D
i dont think its a problem of the chips, i checked several batches again and i made a mistake before, i tested on the crappy old board that only does 1825 max ^^Quote:
Originally Posted by v_parrello
anyways, the newer chips are the same as the older chips...
i think supertalent has a problem with their binning procedure...
vincenzo, did you try the sticks in single channel?
sounds like you have a good and a bad stick paired together, try them in single channel.Quote:
Originally Posted by metro.cl
did you guys contact supertalent?
what did they say?
To whoever has problems....... :D
IF your motherboard has the "DDR3_REF" voltage features then put ALL THREE of them at +30mV.....and feed your Northbridge with ~1.85 - 1.91Volts AND (0.67x).....I believe that you'll get more out of them.... :)
sounds like the blue kit finally arived? :D
george, i suspect the vdimm and vtt to be unstable on the p5k3, i hope we will soon see some nice clean ripple free vdimm and vtt from a maximizer :D
EDIT: btw, whats this x0.67 thing about?
i really wonder why asus adds options like theese and then doesnt explain what they actually do :confused:
Sascha,Quote:
Originally Posted by saaya
I think you made a right question.
I already made these tests and the modules work well in single channel (1866 8-8-8-24 1.9v) but they are not able to work in dual channel.
Edit: Therefore I do not know if you are right when you saw that sounds like we have a good and a bad stick paired together.
I am waiting some reply from my contact in Supertalent.
Yeap they finally came....and already fed them via a DDR Maximizer......:DQuote:
Originally Posted by saaya
http://www.xtremesystems.org/forums/....php?p=2438496
.
:eek:
Started testing the kit for the review today.
Just finished finding the best timings for fully stable operation at DDR3-1333.
Un-tweaked SuperPi 1M run screenshot follows...
http://i14.tinypic.com/4xvo57c.jpg
Edit:
A Everest Cache & Memory Benchmark:
http://i4.tinypic.com/66t02z8.png
Used Performance Level 4 instead of 5 this time.
It's stable :D
Edit #2:
Oh well! Performance Level 3 as well :D
http://i4.tinypic.com/4zeicgi.png
waithing for some more testing at higer clocks :)
read the rest of the thread, i dont think there will be much higher speeds :D
maybe single channel... it seems one of the sticks is bad.
can you post something we can't already do with DDR2 :p: :D wink wink :DQuote:
Originally Posted by BenchZowner
go easy on him, hes handicapped ^^
come on then bill, show us some nice SC then if DC doesnt work well with this kit.
sometimes maybe not.Quote:
Originally Posted by hipro5
leave dimm_vref to auto seems better in most situation.
set mch_vref = +30mv when u wanna stable @ 1900mhz+
Any more info on these 3 things? I haven't found the thread with the P5K voltage points again, if anyone knows where it is post it up plz, want to meter those. :up:Quote:
Originally Posted by saaya
There was some suspicion that P5K3 Vdimm was actually higher than set in BIOS, no? And what are these "reference" multipliers about? Like you saaya, I'm bemused by them.
about the voltage being stable or not, i have no idea... but the weird results some are seeing on the orange slots on the p5k3 dont seem to be voltage related after all.
George, how much improvement did you notice from switching to the maximizer on the blitz? would you say the blitz has good vdimm or is it rather unstable?
how much of 10 points would you give it, 0=can kill mem, 10=maximizer :D
and yeah, what exactly do the x0.67 options do?
DDR3 must still be in the "doesn't quite work correctly" stage since I can almost pull that with standard DDR 400 and cold temps. DDR2 makes those numbers look positively pathetic.
Perhaps they've mixed up IC's or summin' 'cos those bandwidth numbers are very low indeed. Not to mention it's using the same vdimm as DDR2.
DDR3 ARE FAST.....Depents on who's working them correct.....;)Quote:
Originally Posted by SOLDNER-MOFO64
ASUS Vdimm is good enough - 2 phase circuit - .....I haven't placed it onto the osciloscope though to "investigate" it a bit...... :)Quote:
Originally Posted by saaya
BUT I trust my Maximizer much more....Even at cold boot it helps.....;)
The 0.67Volts option - in simple words - it's like GTL_REF Voltage of a Core 2 CPU but it's the Northbridge's......:)
what ddr1 numbers are you talking about?Quote:
Originally Posted by SOLDNER-MOFO64
maybe a64+ddr1 can get similar bandwidth results, but even that would surprise me. a big problem with ddr3 is the fsb limiting the performance... we need csi to see the full bandwidth :(
and about the voltage, dont know what your talking about, ddr1 was used with almost 5v, ddr2 is being used with almost 4v, ddr3 is only used up to 2.3v. and in the end who the hell cares about the volts? what matters is the power consumption, and thats where ddr3 is a lot better than ddr2, the difference is so big that you can feel it with your own hands, you dont even need to meassure the temperatures. getting ddr2 to the point where you burn your fingers touching it is easy, good luck trying to get ddr3 to those temps... :D
thats what raju said as well, but i dont quite understand it.Quote:
Originally Posted by hipro5
if i select that option the the chipset option will be 76% of another voltage, right? but which one?
Hey man,Quote:
Originally Posted by SOLDNER-MOFO64
what are you talking about? :confused:
Looking at DDR3 benchmark results and technology argumentations about DDR, DDR2 and DDR3 RAMs it is very hard to understand what you want to mean. :shrug:
Let me revive this thread :D
Got another P5K3 Deluxe some days ago, just tried the sticks on this one, and they seem to be running fully stable at DDR3-1868 8-8-8-24 1T @ 1.8V
Gotta test them more thoroughly, later on...this is getting weird...the rest DDR3 kits that I've got there work fine on the first P5K3 Deluxe, and reach very similar ( almost the same ) frequencies on both P5K3 Deluxe's :o
Sorry to ask Hipro but isnt 1.85v way to much for X38 chipset?Quote:
Originally Posted by hipro5
I also tested upping all DDR3_REF to +30mV but didnt help, im on maximus extreme.
and 0,67x didnt help either :(
You're asking the man who used 2.00V on the NB & 1.85V ( or so I'm not 100% sure on the SB voltage ) on 680i :D :rofl:Quote:
Originally Posted by metro.cl
1.80V+ should be ok for benching with P35...but you gotta keep the chipset cool.
I know somebody who has even pushed 2.4V on 975X :D
Mobo still alive after all :D
Yes i know that hipro is crazy :DQuote:
Originally Posted by BenchZowner
But im on X38 now, and 1.8v on the NB is well in the "red" zone
Hipro is just stating what is needed to reach certain clocks.. whether it's safe of not - well that is another matter LOL
what clocks? i think he just gave a broad recomendation. I've got more out of my DDR3 that anyone here in this thread and way lower volts than hipros recomendation.Quote:
Originally Posted by eva2000
you just have better sticks or mobo probablyQuote:
Originally Posted by metro.cl
they differ a lot Felipe :( (mobos that is)
probably high clock tight 6-5-5-x or 6-6-5-x with tight sub timings >DDR3-1866 - 1900+ Mhz hehQuote:
Originally Posted by metro.cl
and i think these days hipro running quad core so naturally they need more NB volts to go higher FSB wise to match DDR3 limited mem dividers
i needed 1.63v NB for this DDR3-1820 7-6-5-18 1T
http://fileshosts.com/intel/Asus/Bli...1m48s031ms.png
:DQuote:
Originally Posted by metro.cl
On my P5E3 WS Pro, I have placed a Swiftech heatsink on it with a DELTA fan 4 x 4 10000rpms....There is NO way that it can handle 1.85V IF I was using the DEFAULT heatsink EVEN with the DELTA on it.... ;)
I feed in on every bench I do, 1.81Volts and for the "light" benches like Super-pi 1M, WinRAR, pi-fast, etc, at 1.85Volts...
On my Blitz Extreme I'm with 2.01Volts for ALL benches.....
My other UNMODED P5E3 WS Pro OR the P5E64 WS Pro, I feed them with NO MORE that 1.69Volts.......They can't handle more than that due to their DEFAULT heatsink..... ;)
EDIT: ALSO remember this 32M bench of mine that was done with 2.01V on P53 and 2.42Vdimm..... Not bad if you ALSO see the tight subtimmings here..... ;)
http://www.hipro-tech.com/images/hip...1040rams_1.png
http://www.hipro-tech.com/images/hip...1040rams_2.png
......and this 32M was done with the rams at 1T and 1029MHz (no further tweak like CW or so)....
http://www.hipro-tech.com/images/hip..._NO_MAXMEM.png
there are a lot of variations between boards/nbs ... some need high volts some dont... 0.67x setting didnt help me either, havent tried more than 1.7v chipset voltage.
why do you think 1.85v vmch is too hight though?
i havent heard of anybody killing their mainboard with vmch yet...