I'll try with more vcore... then i'll let you to know. :)
for now 400x10 it need 1.23 in BIOS to be solid like a rock :p
Printable View
I'll try with more vcore... then i'll let you to know. :)
for now 400x10 it need 1.23 in BIOS to be solid like a rock :p
No... i can't bench with 600 fsb neither 550 :s
With.. any vcore :s i don't know.. what's going on :s
haveyou tried 1.7 vnb?
that gave me the stability for 8x500 quadfsb.
I haven't read the hole thread, but are there still issues with micron ram (2gb kits) and high freq. with tight timings?
Did someone spotted a3 rev of p45 on p5q3?
there was talk of one but no one has said much about it yet. the guys that would have any info about it sold their p5q3's and no longer post any info here.
im hoping to see some development soon. i hate the fact that my 2x1gig of crucial ballistix 1600mhz worn run 1600mhz or faster on this board but on a x38/x48/p35 board they do well over 2000mhz.
i ran a 2x1gb OCZ reaper 1333 6-6-6-18 fine on this. I don't think OCZ used micron ICs for those kits, but it worked fine even on the 0605 BIOS that came with the board.
I'm currently running Mushkin 2x2 1600 7-7-6-18 at 7-7-7-20. I haven't done much tweaking as i've been focusing on getting maxing out my cpu OC before I move on, but the sticks are running solid. Honestly I haven't experienced any of the problems I read about before I bought the board.
I will say this though: the board is very specific with timings and voltages. I tried some preliminary tweaking (just set this and see if it works) and even just going down to 7-7-6-18 was unstable.
really now? heh. well i can't say he's wrong, but i'd like his source on that.
I guess I've never really tried lower vnb, so I'll give it a shot...
but at the same time I'll bet any money you'll get stable with more vnb--if you're willing to try.
hell I've been running at 1.6 with no problems...
Hi Ace80,
my new hardware arrived yesterday. However they gave me the same board again that I reutrned last week. I could tell from the missing stickers on the RAM and CPU socket....so that had me suspicious.
Ace80's screenshots have cleared the mist up now, actually. The key to stability is the tRFC setting just as with DDR2 memory. Too tight timings give you a slew of weird crashes and other sideeffects.
My system would only get stable at a tRFC setting of 82 for two Corsair XMS3-1600 modules, and 110 for two pairs of 1 GB each. But this is all clear if you are peeking at Ace80's XMP profile in his screenshot. Corsair programs the XMP settings to a tRFC of 41, two modules require 82 then. I have seen people setting their tRFC to 60, it would be the best to do a thorough test if that's really sufficient.
I am still in the process of figuring all the overclocking details of my current hardware, but here is my current stable settings, very briefly and out of the top of my head:
CPU: Q9300 M1/C1, 4x1 GB Corsair XMS3-1600C9DHX w/o XMP
FSB: 460 (The Maximus Formula bailed out at 436!)
RAM: 1472 (1:1 clocking doesn't work, which means DDR3-1840)
Timings: 8-8-8-24-5-110-10-5
CPU Volt: 1.2750
FSB Volt: 1.36
NB Volt: 1.40
Hi, anyone with 2x2gb Corsair XMS3 1600 9-9-9-24 stable @1600. Some mate can´t do it,just 1333, and we would need some help with the config. Do u use the template ? Thx
it seems there is a 1402 bios for the P5Q DLX..... any word of one for us?
im about to call it quits with this board.
This is stable for me with 1 GB modules. 82 or more. And set the Performance Level to 11. I cannot exactly tell fro 2 GB modules but it's worth a shot.
2 modules: 82, 4 modules: 110 (tRFC)
What makes me wonder, stability at 110 and 110 being the max. possible selectable option...shouldn't ASUS give the P5Q3 users some more options higher than that?
So far my system has proven to be stable at FSB 460....
for some reason if i set my performance level to anything other than auto my board wont post.
I'll say it again: this board is very picky about timings and, in general, really likes it's auto settings.
My system is stable at 8x488fsb (Q9450) with 4gb of mixed IC RAM at 1464mhz. Really, I'm surprised at the amount of stability i've gotten expecially with the amount of complaints I've read on this thread.
Slov,
I have had extreme troubles with both my OCZ and initially also the Corsair XMS3-1600 RAMs. 2x2 GB and 4x1 GB. The defaults were definetly not stable. Mainly because it seems that the board is deliberately ignoring some SPD programmed timings. What helped me was my experience with DDR2 timings on the Maximus board. Like, the P5Q3 Del setting the tRFC wrong. What was also very confusing to me was that the commom performance level has been around 4-7 seemingly randomly until I set it to 11 manually, to exclude sources of instabilities.
I wonder, what board revisions have been found so far? I have a 1.02G that was made in China. To my experiences, 1.02G boards are usually the first production releases followed by the large mass of boards being 1.03G soon afterwards.
Also I head of issues of some of the Maximus boards that were made in China, and not in Taiwan. Does anybody here have a different revision than 1.02G or a board made in a different place than China?
actually i have the same problem with my OCZ 2x1 1333 kit, it doesn't recognize the SPD settings.
So I set it to the same exact timings as my Mushkin's XMP and just upped to 8-8-8-20 to OC to 1464 and now I've got the 2x1 and 1 stick of the mushkin running on those timings. and yeah i think i did set PL to 9.
Partially. I set all timings to 250 ps advanced with no success, it was all unstable except on auto. I reduced it step by step towards normal, but no stablity.
The 82 does make sense. I saw a screenshot of the XMP version of my Corsair modules and the XMP data said clearly that tRFC for one module would be 41. Two modules do require twice the amount of clocks. Not ALWAYS but it is no surprise if they do. Theoretically four modules would double that value again, but 110 works for me, besides I can't go any higher. It is entirely possible that ASUS does not offer higher tRFC settings than 110 for a good reason. They tried it, and four modules usually work with it, I guess.
As for compatibility, what in the world could be better than Corsair RAMs?
Granted, OCZ is prolly not the best quality. All that I had so far were troublesome to me. Mainly the DDR2-1066 modules, two pairs died on me on an earlier system after only three weeks of usage.
I tried with my P5Q Deluxe and no luck with skews and pull-ins. Just relaxind 3 informations I pushed to almost 1100, I have to play more for getting at least 1100 stability.
When I know abouth this guy I will comment u for reference.
And yea, OCZ is sucking with lasts RAMS, Corsair is still great. I´ve known Mushkin and they´re very good too.
I'm dumping this board.
I think this is #3 that I've RMA'd through newegg for no good reason.
It's a fine board, i'm just bored.
i sold mine, it pissedme off, I think im done with ddr3 ingeneral, any one wanna buy 2gb ocz gold ddr3 1333mhz!
i have found that it doesnt like D9GTR IC's but seems to be fine with D9GTS IC's.
my GTR's cant get 1600mhz stable but GTS's do 2000mhz no prob.
Hi all i am new here !
I came from German sorry for my bad english.
3 weeks i must wait before i can write here after my Registration.
I have problems to became a high memory Frequency with high fsb.
My System how kills my nerve at firstime but now its running a month
P5q3 1201 BIOS
Q9450 @3,6 GHz 480Fsb 1,385 Volt Watercooled
Ocz Intel Edition Xmp 2*2GB
DDR3 1600 Cl7 and 1900 Cl9 @1444 Mhz Cl766 Performance Level 8
VFSB1,48Volt
Nb 1,5Volt
24hours no Probs
but no way to 1500Mhz or 1550 Mhz ram frequency 1550mhz Cl7 was my favorit but the system shootdown from allone.
[IMG][img=http://img237.imageshack.us/img237/7959/480fsboh5.th.png]
I hope anyone have an idear
@visualist
read through the thread from the beginning. I think you'll find some...uh...interesting tidbits about 2x2gb RAM kits at 1600mhz+...
should be fun reading....
Beta mate :
Link 1402 here : P5Q3-DELUXE-1402.ZIP
Pmp made a thread where all the new biosses will beupdated regularly instead in the thread itself look here for updates :
Asus Socket 775 bios thread
someone tell me if it's worth flashing this before i pack this board away.
I flashed 1402, but I have problems still with dram 1600 and spi 32m.
I-m with 1402, but i don-t have too much time to test now... this afternoon i-ll same test with stability
Hi,
been away from the thread cause my rig was at assistance. I gave up after 1 month of unsuccessful tryings.:(
The final response from the lab after extensive tests has been: mobo is faulty :mad:
I'm gonna RMA it now.
Well here's the question: I asked them for another mobo (quite same specs and price range) to replace the "epic failure one".
I'll be using this pc mainly as a media center (High definition stuff basically), hd home video encoding, BD enconding, torrenting :D...no gaming. No particular need to OC it.
What do you guys suggest me to go for?
I just want to use my system...it could be even go for a "quite old" chipset (P35?). Or does X38/X48 worth it?
System spec:
cpu: q9450
mobo: p5q3 deluxe wifi @n rev 1.02G (to be replaced)
gpu: hd4870
psu: enermax modu 82+
Case: silverstone cw02
cpu cooler: thermalright hr01 plus
fan: noctua s12
ram: corsair TW3x4g1333c9dhx
hdu: seagate 500gb (sata II)
Thanks a lot fro your help
;)
Hi there,
I am still trying to get my system completely stable. I had my P5Q3 Del. replaced today, but now that I am homme again, I found something absolutely weird:
My supposedly identical Sets of Corsair DDR3-1600 CL9 modules are NOT identical.
The packaging is the same, the stickers on the modules, there is absolutely no difference in the looks.
However, one set appears to have XMP programming, and the other doesn't! :shocked:
The part number of module in DDR socket 2 and 3 reads (I am using CPU-Z):
CM3X1G1600C9DHXIN
Part Number of modules in socket 1+4
CM3X160C9DHX
Is eem to ahve messed the sets up, but that shouldn't matter.
What do you think, does this cause poblems? The modules should still be the same, and we are usually setting the BIOS to not use XMP anyway. This shouldn't hurt then....
I'll make sure to only use XMP modules in the slots A1 and A2, and the ther modules in B1 and B2.
This is D9GTR:
http://www.xtremesystems.org/forums/...&postcount=557
With 1306 I was able to run RAMs ok with E8500. However, with QX9650 I was not able to run mems as high as with E8500. Also, after playing for a while with QX9650 I managed to break both a SATA-disk and an IDE-disk and loos my BIOS :D It CAN be that my QX9650 is halfbad though.
I will try 1402 together with E8600 a bit today.
@Slovnaft
I have my Board July at this time i read from beginning this thread.
I dosent can find help here.
Or my english is to bad to understand all.
Have anyone here 2*2gb OCZ XMP with quadcore and high fsb over 1600+mhz.
I thanks for all idears but dont post me konfig for Dualcores Dualcore Oc is to easy thats not the sam.
And 1024mb*2 is not interrestet for me.
Thanks at first i want be so happy when anyone can help me .
In german forum all yousers are sell his Boards all says dumb board.
And i hope and hope that will be better.
it's not so much a dumb board as a really weird and puzzling board.
but not so weird and puzzling as blackops, thank god.
Visualist, basically the jist of the last 20 or so pages of this thread has been that 2x2gb modules are rarely able to operate at speeds of 1600mhz or above on this or any other P45 board because of inherent instabilities in the memory controller of the P45 chipset itself. Tony (the OCZ tech support guy who posted in this thread a lot) has tested and tested again 2x2gb modules on this and many other P45 boards and has concluded that your best bet is to drop your RAM speeds, tighten your timings, and be happy.
Now on the quad core overclocking side of things, i may be able to help you there. This board has been great to my Q9450, and I've been able to get it to 3.9ghz stable with 488fsb, and 4ghz benchable with 500fsb, and 4040mhz validate at 505fsb. This chipset seems to be a great option for quad OCing as it allows for higher fsb than many and seems to stay stable. Are you having problems OCing as well?
I have no problem with the OC itself with quadcore. However, I can not at the same RAM-speed as with dual-core. Same for you?
I had no problem with 2000MHz on the RAMs together with dual-core. With quad-core I started to get trouble below 1900MHz (or even lower, don't remember)
I haven't really made any passes at RAM overclocking with this board. It's sodamn particular about timings and ratios that I haven't bothered. Plus it didn't help that one stick on my 2x2 kit is possibly broken so i'm using a 2x1 1333 cl6 kit plus one stick of the 2x2 1600 cl7 kit at 1464 cl8, so i'm just happy it works.
It seems this board is not the greatest for RAM tweaking, and you just need to get a stable setting and be satisfied.
Now your whole dual v quad thing could probably be a function of the chip, and more likely of the NB. I mean, to get 1900mhz on the RAM, assuming you're running a 1:4 ratio you'd need an fsb in the high 400's, something that may be either a function of your chip or vNB. Try bringing vNB up to 1.6 and dont be afraid to go as high as 1.7, just make sure yyou've got a high cfm fan on it.
IF you've already done this though, i'm not quite sure.
@Slovnaft
My Q9450 running fine of this board but i need high voltage for 3,6 Ghz 1,385Volt on 480 FSb 7,5 Multi.
And 1,365Volt for 450 at 3,6 8Multi.
I think i understand you i will check it again at weekend.
Thank you at first for the fast answher .
Hi People,
you might remember my hardware specs from my previous posts.
Despite the Corsair DDR-3 1600 CL9 DHX modules being certified to WORK on the P5Q3 Del., neither OCCT nor Orthos 2004 do run successfully using a blend or RAM test. App. runtime is 25 minutes. Only Prime 95 does pass 1 hour running four instances, each one with an affinity to one core.
Oh, one detail I am not even trying to do 1600 on the modules. I am running those modulesa roughly 1333, give or take a bit depending on the FSB clock.
Using four of those modules makes it all harder, and I am aware that only TWO of those modules were specified to run stable with this board.
However, all my other software appears to work stably. 3DMark 06, 3DMark Vantage, Secondlife (running stable for HOURS!), WinRAR system bench/test...
I cannot make sense of this at all. I am at the verge of hurting my old rule to not disregard those tools....
What do you people think? How are you testing the stability of you systems?
I am quite frantically looking for someone who can give me damn clue about this board.
I am at the end of my damn patience. I have had this board replaced, I am trying the SECOND type of RAM now, and still I am screwed....:mad:
Any one with e8600 try to make 600 fsb? i cant reach on mine :s
I have just started to try my E8600 and it looks good after 30-40 minutes of testing. 590MHz is SP32M stable. It is almost SP32M stable at 595MHz so I think 600MHz should be possible.
EDIT: 600MHz FSB SP1M is no problem.
I will try to post tomorrow. It is in the middle of the night here in sweden so it is time for bed. :p: CPU-REF is 0.615v for me, that I remember.
Just thrown bios 1402 on and for me atleast its by far the best.
I'm not testing memory at the mo, probably in the next few evenings, just cpu for now.
http://img.techpowerup.org/080910/Capture007611.jpghttp://img.techpowerup.org/080910/Capture008536.jpg
With last bios (1306) this speed needed around 1.38-1.4v
4Ghz now needs only 1.28v (1.3v set bios).
My board is rev 1.02G and made in China.
Well, I have not been able to run Sp32M @+600MHz FSB with this E8600 yet. 590MHz FSB work though and this is the settings I use when I boot at 585MHz FSB:
CPU GTL Ref=0.630v (both of them)
PLL=1.54v
FSB Term.Volt.=1.50v
NB=1.50v
NB GTL Ref=0.550v
CPU clock skew=400 (200)
NB clock skew=200 (normal)
SP1M works above 610MHz FSB.
I use 1402 atm. With 1306 I could go higher but that was with a different CPU so... The settings are rather simulair between the BIOSes anyway.
For CPU and NB clock skew, 400/200 seems to work the same as 200/normal.
I gave up on mine as far as OCing it is concerned
gonna build my dad a nice rig with it
I am really frustrated about it... ARGH!
hi guy
I (re)buy :D a new ram kit...wait for mushkin pc14400.
stay tuned for tests :up:
i have found that my gskill HZ's work great on this board, just how they should. however, a good friend of mine has the same board and same ram and he cant even get the stock 1600mhz to run stable. theres something fishy going on here.
Hi Guy,
I use bios 1402, with this setting for 621Mhz
VNB:1.4
NB GTL:0.54
Vcpu:1.38
Pll:1.54
VFSB::1.32
Vdim:2.00
E8600
2x1Gb CellShock Pc12800
P5Q3 deluxe
Galaxy 1Kw
http://www.pctunerup.com/up/results/...185707_621.JPG
http://valid.x86-secret.com/show_oc.php?id=416737
It is working - and really well now
Hi people,
I have finally gotten my system stable, and after some thorough research I am coming to the conclusion that the board is perfectly fine, except for a stupid BIOS bug.
We all know it, when you're setting the memory timings manually for the first time, the AUTO entries are being changed to display the default memory timings. What's extremely irking and a spoiler to me, too, is that the defaults of the P5Q3 Deluxe board are 5-5-5-15. This is ridiculous tho, as we all know, the crucial timings are roughly doubling for DDR3 RAM. So we are setting our timings to something like 9-9-9-24. All fine and well, the problem in my opinion is the ACK# to ACK# Delay (tRRD) setting. It is not stored in the SPD, hence there's only ASUS faulty default of 5 clocks. And this is exactly what did not allow me to get prime stable, and potentially was driving lots of others up the wall.
Setting ACK to ACK delayed to AUTO will always put it at 5 (DDR2 typical!), while 8-10 is required for DDR3. You can find that, for example, in the forums of the OCZ support or at Corsair. So I think I have a good professional backing up for my thesis.
Another step to getting our board stable is tRFC. 60 is often not enough either. Corsair for example programs their SPDs to 41 in the XMP profile, which is usually ignored by the BIOS. Dual Channel requires twice the SPD's programmed value, so I recommend 82 or more for stability.
You can roughly determine your instability cause by the kind of issue you have. Too tight tRFC usually leads to sudden bluescreens and system resets, while the too tight tRRD is giving you errors in Prime 95, Orthos 2004 or OCCT.
Another factor is the Common Performance Level. The BIOS has a tendency to recommend lower values than 9, but at least 9 should be set manually. I am using 11 myself right now.
I hope this helps. Here's a lil screenshot of my memory settings....if you need to see more, let me know.
Good to see that... i reach estable 400-410-420-430-440 right now, but i-m always testing for me
But i want to joing to the 600 FSB club :P
hi guy, today I've corrupted the bios :D but the die hard bios work very well on this mobo :up:
When this happens, you may be asked for a special BIOS-file to load. To save some time you can take what ever BIOS-version you want, rename it to what-file-you-have-been-asked-for, put it on an USB-stick and it load from there.
@faina77: Can you test how high you can run SP32M?
Dam i-m tryin with a old bios to reach 600 fsb but i can-t.. what i must believe? cpu or mother?
it's possible that you wrong the memory divider.
What's your PL?
I can set 10 but is not stable...PL11 more stable ;)
http://img526.imageshack.us/img526/3...lo16666mu7.png
sav, BND_MOD:
Try setting tRFC to 82 and tRRD (Act to Act delay) to 8-10
And read post 661 on this matter, further up on this page (page 27).
Should help....
I cant believe people are still buying this junk. Stop buying this board. LOL j/k how is this board coming along, i had it for 1 week, i wanted to kill myself, it drove me crazy :/
I have test PL AUTO and 11 and 7 and all :p nothing work...
I-ll read that post :) thanks... i-ll try all again and i let you to know..
well i have try all this
nothing work :s into OS with setFSB i get 5Ghz validation and 580 max FSB :s no more :sQuote:
i Overclocker Tuner: Manual
CPU Ratio Setting: 07.0
FSB Frequency: 600
PCIE Frequency: 100
FSB Strap To North Bridge: AUTO
DRAM Frequency: All Speed
DRAM CLK Skew on Channel A1: Auto
DRAM CLK Skew on Channel A2: Auto
DRAM CLK Skew on Channel B1: Auto
DRAM CLK Skew on Channel B2: Auto
DRAM Timing Control: Manual
CAS Latency: 7-8
DRAM RAS to CAS Delay: 7-8
DRAM RAS Precharge: 7-8
DRAM RAS Activate to Precharge: 18-24
RAS TO RAS Delay: Auto
Row Refresh Cycle Time: AUTO-82
Write Recovery Time: Auto
Read To Precharge Time: Auto
DRAM Static Read Control: Disabled
DRAM Dynamic Write Control: Disabled
DRAM Read Training: Disabled
DRAM Write Training: Disabled
MEM OC Charger: Auto-Disable
Ai Clock Twister: AUTO
AI Transaction Booster: AUTO-PL 11-7
CPU Voltage: 1.3v-1.4
CPU GTL Voltage Reference (0/2): Auto
CPU GTL Voltage Reference (1/3): Auto
CPU PLL Voltage: 1.54-1.6
FSB Termination Voltage: 1.4-1.6
DRAM Voltage: 1.9-1.96
NB Voltage: 1.4-1.55
NB GTL Voltage: Auto or 0.545
SB Voltage: AUTO
PCIE SATA Voltage: Auto
Load-Line Calibration: Enabled
CPU Spread Spectrum: Disabled
PCIE Spread Spectrum: Disabled
CPU Clock Skew: AUTO-200-400
NB Clock Skew: AUTO-200
Try setting RAS to RAS delay to 8-10. It is the same setting that's called ACK to ACK delay in memset, I had that confused.Quote:
nothing work :s into OS with setFSB i get 5Ghz validation and 580 max FSB :s no more :s
A default value of 5 should almost be impossible. These timings usually DOUBLE when going from DDR2 to DDR3 and it seems ASUS is using DDR2 defaults here....:shocked:
Another lil change, up the Write to Pre delay prolly to 24 and increase tRFC beyond 82. Upt o 110 is possible.
If that doesn't help your OCing capabilities are prolly exhausted.
I am running my Q9300 at FSB470, but seriously, I am happy with it. Four cores are a lot of calculation power.
Let me know how it works with the new settings. My best guess is the RAS to RAS delay.
I confirm, with tRFC to 82, tRRD 8-10 and PL11 I have more stability :D but I had to change also memory divider and lose in memory read....
http://img82.imageshack.us/img82/6551/thrshtmw9.png
http://img177.imageshack.us/img177/2663/ththhwa6.png
I think that with 2x1 gb (gtr) Pll 10 is possible if I set high memory divider.
@faina77
Validate is not a good "stabiliy" test :D
[QUOTE=Sav;3287639]I confirm, with tRFC to 82, tRRD 8-10 and PL11 I have more stability :D but I had to change also memory divider and lose in memory read....
Sav:
Depends. One pair of my Corsairs works at 1600 when using the XMP profile.
But four modules...bear in mind that the P45 is only specified by intel to run four modules at 1333. ASUS' 1600 specification almost ONLY means one pair of modules...using four 1600s makes you lose overclocking ability anyway. it is a law of physics. If you are putting more load on the same lines to the chipset it will diminish the signal quality. My CL9 modules max out around at 1440, maybe some CL7 modules might make it that far.
yes, but I use 2x2Gb ;)
I think we must find a correct memory divider according with fsb frequency.
Now with high tRFC and tRDD I can do this:
http://img300.imageshack.us/img300/5917/3d3d3d3dcx5.png
http://img213.imageshack.us/img213/2...lo13434zl2.png
With old setting I can't do spi or everest because I had freezing :)
now i probe
For 600 FSB doesn-t work :sQuote:
i Overclocker Tuner: Manual
CPU Ratio Setting: 07.0
FSB Frequency: 600
PCIE Frequency: 100
FSB Strap To North Bridge: AUTO
DRAM Frequency: All Speed
DRAM CLK Skew on Channel A1: Auto
DRAM CLK Skew on Channel A2: Auto
DRAM CLK Skew on Channel B1: Auto
DRAM CLK Skew on Channel B2: Auto
DRAM Timing Control: Manual
CAS Latency: 7-8
DRAM RAS to CAS Delay: 7-8
DRAM RAS Precharge: 7-8
DRAM RAS Activate to Precharge: 18-24
RAS TO RAS Delay: Auto-8-9-10
Row Refresh Cycle Time: AUTO-82
Write Recovery Time: Auto
Read To Precharge Time: Auto
DRAM Static Read Control: Disabled
DRAM Dynamic Write Control: Disabled
DRAM Read Training: Disabled
DRAM Write Training: Disabled
MEM OC Charger: Auto-Disable
Ai Clock Twister: AUTO-Ligther
AI Transaction Booster: AUTO-PL 11-7
CPU Voltage: 1.3v-1.4
CPU GTL Voltage Reference (0/2): Auto
CPU GTL Voltage Reference (1/3): Auto
CPU PLL Voltage: 1.54-1.6
FSB Termination Voltage: 1.4-1.6
DRAM Voltage: 1.9-1.96
NB Voltage: 1.4-1.55
NB GTL Voltage: Auto or 0.545
SB Voltage: AUTO
PCIE SATA Voltage: Auto
Load-Line Calibration: Enabled
CPU Spread Spectrum: Disabled
PCIE Spread Spectrum: Disabled
CPU Clock Skew: AUTO-200-400
NB Clock Skew: AUTO-200
But for 4Ghz 4.1-4.2-4.3-4.4-4.5 Ghz this setting work
Quote:
i Overclocker Tuner: Manual
CPU Ratio Setting: 10.0
FSB Frequency: 400-410-420-430-450
PCIE Frequency: 100
FSB Strap To North Bridge: AUTO
DRAM Frequency: All Speed
DRAM CLK Skew on Channel A1: Auto
DRAM CLK Skew on Channel A2: Auto
DRAM CLK Skew on Channel B1: Auto
DRAM CLK Skew on Channel B2: Auto
DRAM Timing Control: Manual
CAS Latency: 7-8
DRAM RAS to CAS Delay: 7-8
DRAM RAS Precharge: 7-8
DRAM RAS Activate to Precharge: 18-24
RAS TO RAS Delay: Auto
Row Refresh Cycle Time: AUTO
Write Recovery Time: Auto
Read To Precharge Time: Auto
DRAM Static Read Control: Disabled
DRAM Dynamic Write Control: Disabled
DRAM Read Training: Disabled
DRAM Write Training: Disabled
MEM OC Charger: Auto-Disable
Ai Clock Twister: AUTO-Ligther
AI Transaction Booster: AUTO-PL 11-7
CPU Voltage: 1.3v-1.4
CPU GTL Voltage Reference (0/2): Auto
CPU GTL Voltage Reference (1/3): Auto
CPU PLL Voltage: 1.54-1.6
FSB Termination Voltage: 1.4-1.6
DRAM Voltage: 1.9-1.96
NB Voltage: 1.4-1.55
NB GTL Voltage: Auto or 0.545
SB Voltage: AUTO
PCIE SATA Voltage: Auto
Load-Line Calibration: Enabled
CPU Spread Spectrum: Disabled
PCIE Spread Spectrum: Disabled
CPU Clock Skew: AUTO-200-400
NB Clock Skew: AUTO-200
Hi Guys/Amurtigress,
Have you any recommendations on what I should set in the bios for my specs?
RAM:
Corsair XMS3 DHX DDR3 TWIN KIT 2X2048-1333 (1GBx2) X 2 (4 sticks of 1GB each, 2 dual channel kits) - model CM3X1024-1333C9DHX
Processor:
Intel Q9550
Motherboard
P5Q3 Deluxe Rev 1.02G, BIOS 1402
I have tried your settings without any CPU overclocking paramaters (e.g. CPU voltage was set to auto, and CPU ratio setting was 8.5, and memory speed was 1333 etc)
I have been trying to get things stable (e.g. memtest) with these parameters. I have even tried trfc set to 110 but still get errors within the first 10 minutes of testing.
One thing to note is that legacy usb is set to on in my bios , since I am using a bootable usb and running memtest off it. In the corsair forums they recommend to disable legacy usb support (but when I do this, I lose USB functionality). Should I test with a floppy instead ?
I have also read in the corsair forums that if I am to use 4 sticks then I should lower the memory speed to 1066 - if I do this, would I lose a lot of performance? Could I then have tighter trfc timings (I think a loose trfc can also lead to bad memory performance). Should I get 2 X 2GB instead?
I haven't tried with default bios settings btw - is that something I should do first? Lets say the system is stable at 'stock' settings (e.g. if I use 1GB X 2 only, or even 1GB X 4), does that mean the board will automatically downclock everything to a very slow speed for compatiblity (e.g. reset bios to default, then use memtest)?
Would appreciate any help, thanks
Well, frankly, I think it's not worth the hassle to try and get 1600 stable. I've seen posts stating that people are getting better throughput at 1333.
2 GB modules are a bit more of a load for the NB, but not as much as 4x1 GB. It is possible that you cannot get 1600 at all.
Other than that, Sav, I don't know your full timing settings, hence I can't tell you more than that.
BND: It is likely that the board is changing the FSB Strap once you hit 600. I'd recommend to see which FSB strap is active and stable below 600 and then set this strap manually. Just guessing...I remember on the P5B Deluxe when people were getting to FSB 400 it all got stable but from 366-399 there was a stability hole...
Hi,
first off it would be helpful if you posted your settings like BND does.
1) Legacy USB. I read it, and I know about issues of some older mainboards with it enabled, but I can't tell if this is affecting stability for you. A floppy might be worth a shot...but it's not my first guess.
2) Memory at 1066. RAM that is not being specified to run in two pairs on this board is likely not specified because it cannot hit the memory clock it was sold as. But usually the next lower step (1066) should work one way or the other. As for losing performance, I'd say you try and get it stable at 1066 first and benchmark with Everest. Then work your way up, with somewhat looser timings. Eventually benchmark with Everest on the unstable but stable enough system to see what you're losing.
In my case 1600 does only work with one pair of modules and only with the XMP programmed set. Weird enough I got two packages of the same modules and one set had XMP programming....
As for defaults, they won't work due to the falsely set tRRD timing (5 instead of 8-10). You can certainly leave all at auto and only set tRRD (RAS to RAS delay) to 8-10...
Hope that helps
hi guys...this is my first post @ xtremesystems.org...:D
nice to know you all...hahahaha...I think I have the same probem with you BND_MOD...:p:
I can't go with FSB 600...I don't know for sure, this is because of my motherboard chipset limitation or my processor FSB wall, because I've tried to pull the voltage for CPU to 1,6v...Still can't reach FSB 600...:confused:
I've also tried to push vNB to 1,56v and vFSB to 1,48v...still can't make this motherboard to reach FSB 600...:(
I just can reach FSB 575 with this motherboard...
http://img92.imageshack.us/img92/871...75onairwc1.jpg
This Board starts annyoing me.
I got LLC enabled. I cant put up any VCore between 1,240 and 1,248.
Sames goes for VCore between 1,248 and 1,256.
Though smaller steps are available in the Bios-Setup.
Under certain circumstances its overvolting under Load. :shakes:
(NO I AM NOT TALKING ABOUT LLC - ITS THE OPPOSITE OF VDROOP.)
And it got a VDrop of 0,02.
Anybody can help with CPU margin Enhancement ? What will it Change ?
And this is my daily setting...
http://img227.imageshack.us/img227/5417/dailyuseun4.jpg
After tired to "play" with FSB 600...:(
I try to push my RAM...:)
and the result is not bad with this motherboard...
I can reach 2000Mhz with timing 7-6-5-18 PL 7 tRFC 60 :D
http://img134.imageshack.us/img134/1...zcl7pl7dw5.jpg
I am liking thsi board for regular use - VERY challenging though that is fun to me - getting a PCI POST reader to make life easier - plus I got it new for cheap so I cannot complain - worth it IMHO:D
Did you mean my BIOS setting?
CMIIW...
for my daily setting I have 2 setting (saved @ OC Profile)...
1. CPU @4050Mhz (450x9) RAM @1800 7-6-5-18
Vcore 1,32v
Vdimm 2,04v
Vpll 1,54v
Vfsb 1,28v
Vnb 1,36v
2. CPU @4275Mhz (475x9) RAM @1900 8-7-6-21
Vcore 1,42v
Vdimm 1,82v
Vpll 1,54v
Vfsb 1,32v
Vnb 1,38v
but for 2000Mhz RAM...I used this setting...
Vcore 1,35v
Vdimm 2,2v
Vpll 1,54v
Vfsb 1,36v
Vnb 1,56v
Nbgtlef 0.570
Nb skew delay 200ps
BTW...anybody could help me..?Why I can't reach FSB 600..?:(
Hi Amurtigress,
Thanks for your quick response - I have now used the CD iso version of memtest (straight from memtest website) so its booting from the cd room and straight into test - for some reason it is very slow as compared to booting off a USB and doing memtest - if I use a usb to boot from dos and do memtest, it goes to 7% in about 7 minutes, whereas with with the iso version, it does 7% in about 1/2 hour.
I have two more questions:
you said
"Memory at 1066. RAM that is not being specified to run in two pairs on this board is likely not specified because it cannot hit the memory clock it was sold as."
What does this mean? I am using two pairs of DD3 1333MHz dual channel memory kit that was on the QVL of this motherboard - and also tested in all 4 slots:
http://216.239.59.104/search?q=cache...ient=firefox-a
search for "CM3X1024-1333C9DHX" on the above page - in the "DIMM socket support (Optional)" column you will see that the C configuration has a star under it , which means it has been tested by asus in the following manner:
"C* : Supports 4 modules inserted into both the orange and black slots as two pairs of Dual-channel memory configuration"
Also you said:
"Then work your way up, with somewhat looser timings."
How do I do this in increments? Do I change the memory speed if I find out that memtest is stable (and what are the increments?)? If I up the memory speed, what other parameters do I need to tweak. How long do I test memtest for (how many passes?) before it is safe to say the ram is stable?
You also said:
"Eventually benchmark with Everest on the unstable but stable enough system to see what you're losing.""
I am not sure how something can be "unstable but stable" at the same time. Do you mean to say that after I keep on tweaking the memory speed on my system/retesting with memtest, the first time I get memtest errors, I should revert to previous good memory speed, and install XP, go back to the same memory speed again (the one that gave first signs of problems with memtest), go into windows and retest with everest?
The board is not "bad" when using it as a "normal" computer. This is for instance more than enough for daily use regarding memspeed and FSB:
http://loppmarknader.se/blossas_albu.../10001/605.JPG
My "problem" with this board is that it is not strong enough to be able to "compete" with boards that does way over 630MHz FSB benchable.
This board is kindof a cheapout for someone looking for a cursory P45+DDR3 board.
That's why I bought it, and why I inevitably returned it.
1) the memory at 1066. Okay, I thought they were not certified for four DIMMs. My fault but it still means what I said. <g>
2) Unstable. Unstable can mean either your system doesn't even boot. It boots and crashes after a while. Or it works well overall but makes calculation errors under extreme load.
In most cases you can STILL run software like Everest and see how much you get under 1333 to compare to 1066, even if it's not entirely stable.
3) Working your way up: Find a stable setting even if you are underclocking certain parts. A stable FSB clock, etc. And then increase it and adjust other timings to get further up.
4) losen timings. Use more clocks on the memory timings if your present settings don't work stable. Give the chipset and RAMs more TIME to complete their tasks.
Seriously, how many threads about stabilizing systems and mainboards have you read so far? Man you are dealing with three screen pages worth of settings only about overclocking in your BIOS. Most of them cannot just be set and "work", a well part would have to be figured out for your individual combination of RAM, your particular board and processor. No system is like another in all regards when you're getting to the very limits of what you can squeeze out of your hardware....
Not a lot actually, all I am trying to find out is that if the problems are being caused by the ram occupying all 4 slots, so will I have a better chance to stabilise my system with only a dual channel kit (but still 4GB) - I don't need to overclock, and never will. I just need a stock, trouble free system, in which I do not need to go into the bios at all. My last computer upgrade was about 4 years ago, and I got an asus p4c800-e deluxe with kingston hyperx dual channel kit - added rest of the components - et voila; never had to go into the bios. The reason I have to go into the bios NOW is that the ram is not being run at 1333MHZ if I set everything to auto, so what's the point of buying expensive memory and having to downclock it?
The thing is, I need guidance on only a few settings as I am not going to overclock my system. If it's just a little overclock (5% changes, ) AND if I don't have to install an aftermarket cooler after this, I will probably start tweaking a few settings. Otherwise I just need a plug and play kind of scenario. If I do have to setup numerous parameters for ram on this board no matter which brand I install, then probably this motherboard is not for me.
my board is in a good mood tonight so far......
4.6ghz with memory 1:1 @ 1840mhz on air. (gskill HZ 1600mhz 2x1gig)
http://forums.techpowerup.com/attach...1&d=1221628640
I'm tired of running my Kingston 2 x 2GB 1600MHz memory at 1066,
If i make the changes listed within the last page or so(Ras to Ras, etc.. ) should that help with stabilizing at 1600?
Here are my bios settings: Is my NB voltage ok, or should I raise that?
Kingston recommends 9-9-9-27 @ 1.9V but I BSOD when I attempt this.Code:COnfigure Advanced CPU Settings
-----------------------------------------
Manufacturer: Intel
Brand String: Intel (R) Core(TM) 2 Quad CPU Q9550 @ 2.83GHz
Frequency: 3.63Ghz
FSB Speed: 1708Mhz
Cache L1: 128 KB
Cache L2: 12288 KB
Ratio Status: Unlocked (Min:06, Max:8.5)
Ratio Actual Value: 8.5
CPUID: 10677
-------------------------------------------
CPU Ratio Setting [8.5]
C1E Support [Disabled]
Max CPUID Value Limit: [Disabled]
Intel(R) Virtualizatopn Tech [Disabled]
CPU TM Function [Enabled]
Execute Disable Bit [Enabled]
Configure System Performance Settings
--------------------------------------------
Ai OVerclocking Tuner : [Manual]
CPU Ratio Setting : [08.5]
FSB Frequency : [425]
PCIE Frequency : [100]
FSB Strap to North Bridge: [Auto]
DRAM Frequency : [DDR3-1066MHz]
Dram CLK Skew on Channel A1 : [Auto]
Dram CLK Skew on Channel A2 : [Auto]
Dram CLK Skew on Channel B1 : [Auto]
Dram CLK Skew on Channel B2 : [Auto]
Dram TIming COntrol : [Auto]
Pri. Information:
CAS# Latency : [7]
DRAM RAS# to CAS #Delay : [7]
DRAM RAS# Precharge : [7]
DRAM RAS# Activate to Precha : [20]
RAS# to RAS# Delay : [Auto]
Row Refresh Cycle Time: [Auto]
Write Recovery Time : [Auto]
Read to Precharge Time : [Auto]
Sec. Information:
Read to Write Delay (S/D) : [Auto]
Write to Read Delay (S) : [Auto]
Write to Read Delay (D) : [Auto]
Read to Read Delay (S) : [Auto]
Read to Read Delay (D) : [Auto]
Write to Write Delay (S) : [Auto]
Write to Write Delay (D) : [Auto]
third Information:
WRITE to PRE delay : [Auto]
READ to PRE delay : [Auto]
PRE to PRE delay : [Auto]
ALL PRE to ACT delay : [Auto]
ALL PRE to RED Delay : [Auto]
DRAM Static Read Control : [DISABLED]
DRAM Dynamic Write Control: [DISABLED]
DRAM Read Training: [DISABLED]
DRAM Write Training: [DISABLED]
MEM. OC Charger [ENABLED]
Ai Clock Twister: [Auto]
Ai Transaction Booster: [Auto]
**** Please key in voltage numbers directly! ****
CPU Voltage [1.36250]
CPU GTL Voltage Reference(0/2) [Auto]
CPU GTL Voltage Reference(1/3) [Auto]
CPU PLL Voltage [1.54]
FSB Termination Voltage [1.22]
DRAM Voltage [1.70]
NB Voltage [1.28]
NB GTL Reference [Auto]
SB Voltage [1.40]
PCIE SATA Voltage [Auto]
**************************************************
Load-Line Calibration [Disabled]
CPU Spread Spectrum [Disabled]
PCIE Spread Spectrum [Disabled]
CPU Clock Skew [Auto]
NB Clock Skew [Auto]
CPU Margin Enhancement [Optimized]
Any suggestions?
Which Vdimm do you use??? and other volts set?
I only can do this.. in a fast test
CPU with Multi at 8x
The ram is the sam like yours DDR3-1600 HZ 7-7-7-18 1.9volts
940Mhz - 9-9-9-24-Auto-82
http://img82.imageshack.us/img82/811...9924qb6.th.jpg
960Mhz 9-9-9-24-Auto-82
http://img82.imageshack.us/img82/684...9924zp1.th.jpg
980Mhz 9-9-9-24-Auto-82
http://img171.imageshack.us/img171/8...9924ic6.th.jpg
Hi guy :)
finally I break 1000 mhz :D
I received Mushkin PC3 14400, very nice memory :up:
http://img88.imageshack.us/img88/5459/1000vf4.png
http://img89.imageshack.us/img89/8594/cachememjsy3.png
I've understood what is my problem.
1) the mobo (mine) undervolt on memory of 0,06V
2) the nb want more volt of 1,44/1,46
tRDD and tRFC are not the porblems with 2GB kit.