Wow, Intel lost it completelly. 1902 BIOS out, maybe it's the final version of the 1900RC.
Printable View
Wow, Intel lost it completelly. 1902 BIOS out, maybe it's the final version of the 1900RC.
Hey guys,
it's long when I post here last time. I was waiting for the hardware RMAs.
As you can read previously I had random reboots on 1600 FSB and I coudn't solve it. After consultation with OCZ guy I decided to RMA memory modules. As my opinion I also decided to buy new PSU, because I just had feeling that previous Enermax Modu82+ 625W is not enough.
I must say RMA on OCZ was very fast, sending modules to Netherlands from Czech republic, exchange and returning to CZ did not took even 2 weeks! Modules were then back to CZ delivered by UPS next day, which is quite expensive service. I'm wery satisfied with OCZ guys.
Then I wait about 3 or 4 months for new PSU, because Enermax Revolution85+ 1250W was at is quite only anounced. Nobody know when it will be available. While this time I not started PC to not again "undervolt" memories.
I switched memories, put new PSU, fired PC. And finaly I'm now on fire :-)
Default settings:
-----------------
! Definitevely new PSU helped a lot ! Previously I needed to increase vMEM even to stabilize the system on default. Othervise reboot. New PSU -> Solved! Default settings OK.
FSB 400, CPU not overclocked:
-----------------------------
This settings was stable. Prime95 Torture test 4 cores run for 30 hours 30 minutes. No errors, then I stoped it.Code:CPU settins: |MEM settings:
vCPU 1,3250V C1E disabled |7-7-6-20
vFSB 1,250V EISST disabled |72-5-10-6-6 1T
vMCH 1,350V |vMEM 1,82V (on memory is label 1,8V)
|1333/333 MEM/REF (1600/400 gave me errors in Memtest as previous modules)
FSB 400 x 6 resulting 2,4GHz (Underclocked from 2,66GHz original)
FSB 400, CPU 3,2 GHz:
---------------------
Finaly when I got stale on 400FSB I started tuning CPU. It was quite easy. Core CPUs are very overclocable :-)
When I put CPU to 3,2GHz I got rounding errors. I increased vFSB, vMCH, vCPU nothing helped. So I decreased memory timings and I got stable. Final settings:
Prime95 Torture test 4 cores run for 26 hours 11 minutes. No errors, then I stoped it.Code:CPU settins: |MEM settings:
vCPU 1,3250V C1E disabled |7-7-7-21
vFSB 1,250V EISST disabled |80-5-10-6-6 1T
vMCH 1,350V |vMEM 1,82V
|1333/333 MEM/REF
FSB 400 x 8 resulting 3,2GHz
Conclusion:
I think it was PSU issue, because according to PSU fan speed it seems that PSU is going on 50-60% of power(in maximum CPU load). Simply mathematics gives 1250*0.5=625W which was total power of Modu85+. And peak power spikes could very easily cause reboots.
Memory modules were possibly OK, because of same configuration behavior as now.
I'm also attaching screenshot. You may notice there is only 3,2GB RAM, it's because of WinXP 32bit. It will be reinstalled with Vista 64bit ;-) I needed to cut images to more parts because of forum file size limitation.
Also there is new BIOS from Intel 1902. I'm a little afraid to update.
Hey Guma, welcome back. Glad your system is up and running. You probably won't need the full 1,82V on your mem, I could lower 1 notch on mine, but I'm running 1333.
Guys, check out the "known errata" on the 1902 release notes: BIOS Auto-Tuning not implemented in this BIOS. Comments anyone?!
Looks strange this new BIOS release, seems so rushed out. Even the release notes file, doc instead of pdf and the other file BIOS update readme missing... something fishy here.
The 1900 BIOS is working very well on my system. No complaints so far. I'm going to watch what happens with the 1902 until it's clear what is happening on the Intel BIOS development front.
However, I'm encouraged that BIOS development still is undergoing for our boards. Although, we are still awaiting the additional memory adjustment settings that had been rumored to be promissed several versions back. Hope this does happen and helps with even further pushing the OC boundaries.
Posting the message above was the prod to my system to exhibit the bad behavior they likely released the new patch for. My SATA drives went into a strange and very loud spasm for lack of a better term. Yikes!
Never heard this before. I'm going to the 1902 and will do extensive testing to ensure my drives are good. Others that have the 1900 may want to roll-back to the prior BIOS or look at the 1902?
Someone else may have already noticed this, but Intel released a new BIOS Glossary by menu for our boards in January:
http://download.intel.com/support/mo...bymenu_v10.pdf
Here is my guess RE 1902 and "auto tuning..." A while back word was that in March IDCC and BIOS would be updated so that we could use IDCC again. see here: email with intel support Also, for sure in the x58 board, and rumored for the x38/48, IDCC was to have an "Auto Tuning" mode for OCing, I am guessing that this is comment in the release notes is simply preparing for new IDCC.
GumaLive, if you are not too afraid of juicing your processor a little (not sure what cooling youre on), you should be able to get FSB up to around 485 or 3.880 Ghz (the highest stable FSB for the 94/9550s that I have ever found from other sites is 520, but the vast majority are in the 480-495 region). You will need to, however, work with the GTL REF settings (dont worry, far less scary than it sounds). I posted some info a while back regarding the use of GTL REFs to bump up your stable OC. Take a look here: understanding what GTL REFs do and here: Guide to overclocking with GTL REFs.
Hi again HT, still here too?! Yeah, guess 1900 is a fiasco. It was dated January already, but released recently, even the (eventually) extensive testing didn't help. 1902 was a record for Intel, released right after ready I guess, usually they only release after about 2 weeks or so, according to the dates. I guess they had to rush out this one to solve problems brought by 1900?
Sean, those were my thoughts exactly, AT is being developed for the SmackOver, still beta I believe. They have both versions for that board v4.3 and v4.4 beta with AT if I'm not mistaken. Guess AT will be coming to us as well, I hope. At least the guy from Intel promised me as well that development would be resumed in March the latest, as soon as they get it right with the DX58SO.
hi guys, i have a lite question, I bought a Q9550 an when i go to My PC --> right click --> Properties it says that i have a Xeon 2.0Ghz runing @ 2.6Ghz, but when i run everest it shows the correct info about the Q.... does any one have a clue?
I also updated BIOS. I noticed weired behavior.
Update was OK, system rebooted, then I entered BIOS to check version. Well when I was in BIOS after long time I changed settings "Primary Video Adaptor" from "Auto" to "Ext. PCIE Graphics (PEG)" and also to clean event log.
Save, shutdown (like you change memory timings), turn on, boot and error that there was unsucessful POST atempt. I say "N", because I know that settings work. On next boot everything was OK.
Do you guys know what are these settings for?
"Primary Video Adaptor" (I gues forces to use PCI-E or PCI graphic card)
"Compiliance test patern" (Don't know enabled/disabled no change)
I would say it's Windows issue, probably wrongly detects this CPU. Here I would trust more Everest or CPUID. Whats your Windows version?Quote:
Originally Posted by Avantlord
I am going to say that you should try updating your x38's bios...still running 1804?Quote:
Originally Posted by Avantlord
Hi Avantlord
Maybe the BIOS is not reporting correctly, you should maybe try updating the BIOS to a newer version. Still on 1804 like mentioned in your sig? Maybe Q9550 was released after BIOS 1804, can't be sure. Everest must read proc info in a different way than Windows.
Sean, sorry, you beat me to it. I had the reply but forgot to post... just realised.
Guma, Primary Video Adapter should be just that as you said I guess, if you have more than one vid card maybe you can select which one is the primary, to drive your primary display. Maybe this is a sign that SLI is coming to all Intel boards. Not sure though with all the war with nVidia regarding the QPI license and all. I guess Intel was waiting for SLI to be licensed for DX58SO to start legal procedures. Smart guys!!
Regarding Compliance Test Pattern, your guess is as good as mine. I believe it's got to do with video compliance testing before boot, again maybe SLI? This would be too good to be true!!!!
Funny how they don't mention any of this in the BIOS readme.
Manel, those video options have been in the BIOS for as long as I can remember...sorry to break your hope.
Sean
Primary Video Adapter should be there already, never really messed with it, no need to. I thought there were new options reading Guma's post. Guess I misunderstood.
Ragarding the Compliance Test Pattern, that one I don't remember for sure.
Anyway, nobody cares for older hardware so SLI for us should be out of the question anyway.
Hope was back up after reading Theo Valich's opinion on this, if that counts for anything. He believes very shortly SLI will open up for any capable platform.
He expects a driver out to enable SLI for all shortly.
Maybe...
I have been reading through all these threads and I am trying to get my Corsair TWN3X4G-1600C9DHX G memory to run at 1600.
If somebody has been successfull could you please post screen shots, timings etc...
Thanks and also has anybody noticed any improvements with the newest 1902 bios?
BlackKnight00, we are going to need a bit more information about your systemt in order to help you.
1. dx38 or dx48?
2. What processor?
3. Are you running the processor at stock settings or overclocked?
4. In BIOS, what are your processor settings? Multiplier and Core Clock Freq
5. By 1600Mhz memory, are you meaning the resulting memory frequency as seen in CPU-Z or Everest, or are you meaning the memory setting in BIOS where you can choose 1600?
Well, we have officially been relegated to a "legacy board" on intel's site
System specs added to sig + see screen shots
BK, 1600Mhz resulting memory speed should be no problem for you, but there are a few things you may need to do to get there.
the most important is to make sure that your memory DIMMs are in the BLACK slots on your motherboard NOT the blue ones. I know that the instructions tell us that they must be in the blue, but trust me here and move them to the black slots if you dont have them there already.
Once you have moved them, go ahead and turn on and let windows load with your current settings. Restart, enter BIOS, make sure that Watchdog is DISABLED, then go to memory settings and change mem freq and mem ref freq to 1333 and 333, save and exit. your board should do the powerup/powerdown/powerup thing. If you fail to POST, turn off/unplug power supply and let sit for 15-20 secs, turn PSU back on, and power up (if it continues to freeze without POSTing, try holding downt the power button after it freezes to shut down, then push again to turn back on...since Intel cant seem to get their :banana::banana::banana::banana: together with how this board responds to changing settings and/or cutting the power, you may have to do this a few times to see if it will POST with these memory settings).
If simply changing the mem settings and trying the restart and power drain methods do not boot, dont worry there are other things to try that will get you there.
let us know what happens.
Sean
PS - what kind of cooling are you using?
So, we are now officially legacy, hem? Hope this doesn't mean no more BIOS updates and IDCC final version as promised.
I also find it odd that Intel can't fix the issue with the power disconnection. This is an old issue and after so many BIOS updates still it persists. At least once they have officially fixed it in the readme, though. Are these guys for real?
Quote:
Originally Posted by Sean-E-Boy
YOU WILL BE PWNED!
what a words...
So, Intel. We are?
Sean,
Thanks! My memory is in the black slots and I will try disabling watchdog and do the power on power off and let you guys know what happens.
I was not able to get the board to post when I changed the memory and disabled watchdog. The on off powering did not work either.
BTW for cooling I am using a Zalman cnp9700nt on cpu with fan control enabled. I am also using a Thermaltake MCH/Northbridge cooler and a 60mm fan mounted on the memory heatsinks.
So far so good with the 1902 BIOS. I've been using it for a little while. Seems to be pretty stable for me.
I'm going to get some faster DDR3 soon as I currently only have 1333MHz Patriot memory. Anyone have recommendations for some really good, but not too expensive DDR3? I'd like to add as much as possible to, just to have breathing room for a while. I'm confident I should be able to push the clocks even further on my Q6600 with the right memory.
BlackKnight00, HighTest
My dream is to fill ALL FOUR slots with memory, but no way, yeah?
So what's conclusion? Two 1600 mhz memory in JUST BLACK slots or four right 1333 memory banks in FOUR slots, right?
I'm talking about 4x2Gb banks, in target to get 8Gb of memory.
I'm not surprising, since DDR3 memory always was point of problems.
BlackKnight, there are many angles from which to approach this:
If all you want is to be at 400 x 8 with memory running at 1600 follow this:
1. go to your memory settings and first choose "auto" to have the board set all the timings to SPD defaults, then change auto to manual (two options here, you want the one that doesnt change any of the timings), set the mem freq to 1333 and the ref to 333, do not change any timings at this point and make sure the voltage is set to what is noted on the ram themselves...if you cannot choose the exact voltage, choose 1 step higher...next set Command Rate to 2T (if you have this on auto, sometimes it causes problems with higher memory clocks). save and exit.
2. if you booted in #1, try changing the command rate manually to 1T (the tighter of the settings). if it didnt boot, go back into BIOS and go from 1333x333 in the memory settings to 1333x266 (slower, but stay with me). If that boots (and it should) restart and change back to 1333x333 (sometimes this board is funny with memory settings...who would have thought that?!?!). if that doesnt work, go to bios and try setting memory to 1600x400 (results in the same memory frequency, but loosens the internal timings of the MCH a bit).
2. if 1600x400 works, good, if not, you will need to go back to 1333x333 and try #3
3. go into bios, change the multi on your CPU to the minimum (6?) and keep the host clock at 400 (will keep the memory at 1600). save and exit.
--something to remember here is that the MCH has to deal both with the I/O of the RAM and some of that of the CPU, by lowering the multi you are checking if the memory and the MCH can work together properly at the given bus speed, memory speed, and timings.
If this works, you know that running at the particular combination is possible and that you may need more vMCH and likely more vFSB. I know that more vFSB sounds silly since you are not changing the FSB, but there is an interaction between the MCH and the vFSB and this interaction can be controlled by either adding more vMCH, vFSB, or altering how the MCH uses the vFSB, also known as the MCH GTL REF, in this interaction. this setting is found in the processor settings page near the bottom..."adjust reference voltages" or something like that and there are several settings.
so at this point you have 2 options: you can increase the vMCH and the vFSB (i would bump up this one first) one at a time until you can boot to 400x8 at 1600Mhz ram, or you can take the next step in tuning and read this post which explains GTL REFs, how they work, and how they make certain voltage settings inter-related. Once you get a grasp on the concept, read this post and start playing.
there are a couple advantages to adjusting GTL REFs in addition to other voltages compared with just adjusting voltages alone:
first and probably the best reason, by adjusting GTLs you should be able to run stable at a given hostXmulti with less vCORE than by leaving GTLs at auto, which means lower temps and voltages and longer life...but if you dont really care about that,
second, you can move up your maximum stable FSB by ~5-8% (may not sound like much, but for me went from 460x8.5 to 485x8.5, or .212 Ghz of processor frequency),
third, the absolute weakest part of the x38/48 board is the MCH (which is why intel went to Corei7, effectively redesigning how the memory and the CPU use the MCH, but thats another story) and by adjusting MCH GTL REF, magic happens and all of the sudden stability increases and so does your memory frequency capacity. With 2 dimms i was able to get my mem frequency to 1940Mhz at 8x multi, but i wanted the 8.5x multi for the processing power (which also increases memory throughput) and since the MCH cannot handle 4 cores at 485 and 2x2g at 1333x333 i had to go to 400x1333 and take a memory freq hit to 1617.
HighTest and Dukzcry, few things to consider here: if you are overclocking and are currently using 2 dimms, 4 dimms will cut your max OC down quite a bit or will force you to use a lower resulting memory frequency to remain at the same hostclockXmulti.
second, more RAM will not increase performance in most things except for video encoding or severe photoshoping of a really large (20+ megapixel) image (unless you have ancient hard drives, in which case investing $ to them would increase performace more).
third, if during your most intensive activity, be it games or photoshop or whatever, you are not using more than say 85-90% or your currently installed memory, adding more will be of no benefit and is like throwing money away.
think of a computer like a desk at an office...the person sitting behind it is the CPU, the surface of the desk itself is the RAM, and the drawers and filecabinets are the hard disks. now you can access your hard disks and remove files and place them on your desk so long as there is room to do so...with too little RAM (too small of a desk), you constantly have to keep putting files back in the drawers in order to look at a new one. but if your programs are not too intensive and with too much RAM, no matter how hard you run them, half/40% of the desk remains empty. now in an office this is fine because you can say "look how big my desk is...you know im important" but with computers it is just a waste of money.
Sean-E-Boy, very good comparison, but i thought that for virtualization the more memory desired. Although, i haven't any needs in more than one guest system.
At least few words about silliness and costing away buyer's money. In our russian review sites board with 12Gb of RAM for i7 considers as negative side of board, they want more and more :(
for sure if you are dealing with servers more ram is a bonus, since part of the goal there is to make sure that everything fits into ram to avoid access delays.
I lowered the cpu multi to 6 and the board still will not post with memory at 1600. I went as high as 1.5v on the mch and 1.35v on fsb. I am beginning to think that this memory just cant run at that speed.
i have had several sticks of corsair, and they are were able to go at least a little over what they say they are "verified to operate at". are you at 1600x266 or 1333x333? you can also try increasing the timings from 9s to 10s and Command Rate to 2T.
Darkknight,
sorry, i havent had much time in the last week or so to help you with this. here is one thing i suggest:
put the memory at 1333x333 10-10-10...2T, processor at 333x6, and leave the voltages where they are.
if it boots, check with cpu-z or something to make sure that your memory is running at 1333, if it is, restart, go into BIOS and start increasing the FSB by 10mhz steps.
when you get to the point that it wont boot, back it off 5 mhz and see if it will...if it boots, bump the FSB up 2...if it doesnt, go down 2, etc, until you figure out where the max is that it will boot to and take note of that as well as what happens when it wont. For example, does it try to load windows and freeze, blue screen, restart, not even POST.
now that you know what the max is under the current conditions, and if it is anywhere close to 400mhz FSB (within 10 or 15), bump it up 5.
--if a) you know that it is just going to freeze on restart and not even attemp to load into windows, you need to start playing with voltages, probably the vFSB (refer to the GTL REF guide for details as to why you would do this) save and see what happens.
--If b) you think that it may try to load into windows, go ahead and let it load, but keep track of how many times the little bar below the windows logo or name makes full passes as well as how far into the next cycle it goes.
In the case of a), play with the voltages and see if anything makes a difference and let us know.
In the case of b), if it doesnt load all the way or freezes after login or when you stress it with prime, go back to BIOS and on the CPU page, got to reference voltages, and increase the MCH GTL REF by 2 and see if it lengthens the time it takes to freeze either upon loading or once in windows. if it shortens the time (unlikely), then go down 2 from the original setting and keep testing, the longer before freezing the better, but if it still freezes then you have a voltage issue and you should try increasing vFSB by 2 or 3 steps without changing anything else and see if it helps.
these boards are more than capable of running ram @ 1600 mhz, (at 8x multi i can get mine into the high 1800's, but doesnt produce the same overall speed that i get by going to 1617 at a higher multi) it just takes some playing and understanding to get there...but one thing to consider is that ram frequency is not the holy grail...you can have a much faster system running wth the memory at 400x1333 and increasing the FSB and working on tightening the memory timings at a lower memory frequency than having tight fast memory at lower FSB.
its kinda like having a guy with a shovel and a guy with a wheelbarrel...with faster ram you are moving the wheelbarrel off and dumping it as fast as possible regardless of how full it is and possibly spilling some along the way...but if you slow down the wheelbarrel (the memory) and increase the FSB (the guy with the shovel) you can still get the same amount of dirt to the destination in the same time, but much more efficiently (and without the spills)! and sure, sometimes getting the dirt there faster regardless of how much you have is better (some gaming) but 99% of the time it really doesnt matter, and if you are dealing with encoding movies or some other large file manipulation, it is better to get as much there at a time so it can be processed and work can start on the next part.
Sean
ps. what is the rating of your power supply?
I am sooo glad that I found this thread. Lots of reading to do.
I am up in the air as I recieved the DX48BT2 in combination with a QX9770.
I have been trying to research the overclocking abilities of the board and I am on the fence an whether to keep the board or sell it in efforts to fund the purchase of an Asus Ramp. Extreme.
I am not an extreme overclocker, but I do want to make the most out of the components I purchase. I currently have a watercooling setup what has served me well and enabled me to OC a q6700 to 3.8ghz at 1.425V on a Asus P5K Deluxe.
I guess my initial question is do I keep the MB? My goals are to get the qx9770 to 4.0ghz and my memory to run 1600+.
Your thoughts are appreciated.
Castlehawk,
Welcome!
4Ghz and 1600Mhz with that chip and board should be cake. the most diffictult thing here is that the Intel x38/x48 boards, while very soilid once you get them configured correclty, can be a bit tempermental in the early days and weeks of OCing.
You might be the first person in this thread with a processor that runs 1600mhz front side bus at default settings, so that should be interesting to see how the x48 handles it and the memory capabilities along the way.
What RAM do you have? Be aware that filling all 4 slots on this board (as with any other, i would think) will severely limit your overclock and also that the x48 chip itself is the weakest part of this board...you are not going to break it, but it is where all the limitation for OCing this board is.
when installing your RAM (assuming 2 dimms, not 4), be sure to put it in the BLACK slots, not the blue ones.
Sean
Hey Hawk, welcome.
That QX9770 was the proc I was eyeing, but 130W is overkill I believe, I'm more on the green camp now. Too bad these babies, and a lot more C2D are being discontinued, PDNs have been issued by Intel I believe.
Still, it will be interesting to see how they do with these boards. Please do keep us posted on your experiments.
Intel keeps updating the DX58SO BIOS, as expected of course, while our boards will probably get left behind more and more. At least the power issue Intel, give us that fix, you know it's a "bug".
Hope soon enough there will be an update to IDCC for the SmackOver, maybe then Matt keeps his promise and updates ours too.
No new feedback on 1902? Is the AC power loss issue fixed or not? I believe someone said no. Intel has (they say) fixed it in 1893 already.
Anyway, just noticed that Intel has updated the readme of 1902 to pdf, before was doc. Maybe this was a while back and I didn't pay attention. They did change a bit the text though, nothing important, same meaning.
So has anyone else tried the new 1902 bios? Does it fix anything? I was also wondering...has anyone played with the Intel Desktop Control Center? The website says it is incompatible with any bios after 1814...which was a LOOOOONNNNNGGGGGG time ago. I was just wondering if it was worth it...I have ATI's Catalyst Control Center for my video card and it's freakin sweet...on the fly OCing, temp monitoring, and fan control! If only my MoBo could do the same...::sigh::
Anyway, back to my original question...is 1902 worth it?
soccergeek, I dont see any advantages or disadvantages to 1902...most of the updates deal with improved ability of the BIOS to read hardware IDs. So if you are having issues with a certain device always or commonly being detected as new hardware and having to be configured upon windows boot, this may help. other than that, same ol yttihs BIOS!
Hiya everybody! Soccergeek! I'm still on 1893. System is working, that's enough for me. And after reading about 1900, that it is maked for prevent user from load "custom bios", i think - no way. First i must know what is that - custom bios (still have no info about it)... And about oc'ing software, AMD was released for their users very good looking utility, named OverDrive. For us, Intel's dearest users - nothing similar. It's sad.
Guys, guys... regarding IDCC not all is lost yet, I hope. Matt at Intel promised they would go back to our version of IDCC and work on it again, just after they work out the DX58SO version, v4.3 and v4.4. However, it's been a long time since even those versions of IDCC ahve had an update. They were shooting for the end of Q1 I believe, so maybe in April we'll see some work done. Matt, if you're reading this don't let us down.
Regarding BIOS, I really had high hopes when John from OCZ got involved with the BIOS development, with some mem optimizations in the oven, but I guess that stayed in the drawer since nothing really new in that departmente surfaced. Too bad.
I can't understand how a company like Intel, a giant in the semicon and platform design and implementation, does not have the resources to make things like this happen, at least more quickly. Don't they have good enough engineers? How come even the taiwanese have better development teams? Intel, having desined the silicon and studied the platform, should have the expertise to do much better. I'm not saying they have bad boards or hardware in general, excelent in my opinion in terms of stability (not performance so much) without those much hyped techs from the taiwanese manufacturers, but the software to go with it... well, you know how it is.
IDCC is a cool tool, just wish it was working like it should.
I won't even say a word regarding those BIOS issues that are still to be dealt with!!!
Now bear with me here...I like to play devil's advocate, as I think it is important to look at the other side and try to understand their experience...
Lets consider this: compared to Asus and the likes, how long has Intel allowed its consumers to really have any control of the board? not very long, in my recollection. Maybe this is part of the teething process for them...they realize that they needed to get in on the portion of the market that likes to "play with it until it breaks," but didnt realize exactly what that meant.
Just think about it in terms of aftermarket performance stuff for your car...10 years ago it was almost impossible to get anything for your car, from the dealer, that would increase its performance (aside from japanese and a select german makers). And even including the japanese and germans, if you could get such an item it was essentially a "no-liability, no-support, buy and install at your own risk" kind of affair. Even today, where they will sell and install performance parts, if you tweak that performance up a little, it lands outside of the scope where the dealer is going to help you with any issues without them charging you an arm and a leg, because the manufacturer is more concerned about building and maintaining reliable cars intended to perform at the level they designed...not those teetering on the brink of explosion!
now correlate that back to our situation. I think Intel has done a tremendous job at providing support (for free, no less) that allows us, with multiple configurations of parts, to improve our performance and our stability and have done so multiple times. now with cars, it took them nearly 90 years to get to the point where they will assist with this to ANY degree...and the fundamentals of the car and the factors relating to performance have almost not changed in that timeframe. With computers, almost every day comes with a new technology that changes the way the entire system, or at least a major part of it works. and tie that in with taking the components to the 99 percentile of their possible performance and you would have an undeniable headache with trying to make sure it all works together ok.
Am i saying that we are wrong in demanding more from Intel? absolutely not, because it will help to improve (as long as they are taking notes) their future products and help them to understand not only how overclockers are a portion of the market that will pay a little more just so we can "tweak it," but also how we are a portion of the market that continually demands more and more FOR FREE. the only incentive they have to service us is the hope that we will continue to buy their products, nothing more, nothing less....and i think they have done a good job at playing catch-up with the other manufacturers out there.
sean
ps. what i really wish would get fixed is how XS logs you out after 5 minutes and you have to log in again to post what you have been typing....i mean, seriously?!?!?! my BANK lets me stay logged in longer before it kicks me off!
Sean
You are of course correct, that is normally my opinion too. Sometimes I just get carried away and spill out my frustrations.
I work for such a big company too and I know how hard it is to convince people that, although your product might not be up there with your competitors in terms of features and performance, the reliability of the product or solution is what should really matter.
Yes, it's not been long since Intel has opened this type of control to users of their boards, if I'm not mistaken the BadAxe must have been the first with some tweakability. Maybe Intel shouldn't have called their line Extreme right off and let it mature first, it might be misleading, but then again for Intel it's extreme really.
For my needs, I'm not that extreme, it serves me good, very good indeed. I'm more into stability right now in fact.
Still, in the software department they are slow, but again, Intel is more of an hardware company, can't really blame them.
If going after the enthusiat crowd is their aim now, they will eventually have to provide the necessary resources to be more agressive, otherwise they will loose the race, at least to those die-hard guys that need the most tweaking available. Those guys will definatelly go the taiwanese way, with the super-charged BIOS, the heatpipes and all.
For me, and although I keep :banana::banana::banana::banana::banana:ing about the power issue (which I still would like to see fixed, this is not some additional feature, it's a well known and documented bug), this is still a great board.
and yes, logout so quickly? Man, this is a nuissance, once you write up such a long post... just happened to me now!!!
Its necesary to conect a molex in the 4 pin auxiliary power conector located in the motherboard (Intel X48BT2) ???
i have two 4870 pci express cards
fredoariaudo, the 4 pin power connector on the mobo (the one by the PCI slots) is ONLY needed when devices connected via PCI do not have their own power connectors and are drawing the board power down to the point that you are seeing issues. I do not recall anyone in this thread having said that they have had to use it.
you would probably have to have most of the PCI slots filled with devices like non-powered graphics cards (unlike your 4870s which have 1 or 2 pci-e power connectors each) or highly power consumptive RAID cards to need to use it.
are you experiencing any issues with both of your 4870s installed?
sean
Sean,
I agree with what you said. Intel hasn't been in the game very long..but to play "devil's advocate ;) " right back, no car manufacturers (at least until recently) would boast about all the after-market add-ons you could do. Intel's site was very specific on the "here's what you can do with this board!" advertisement. They all but said the IDCC could do dishes for you (embelishing a wee bit of course). It wasn't until after you get ready to download and use the IDCC that you get a little warning about it only working on a specific version. Now that was MY fault for updating the bios before reading the manual...but that comes from forced habit of work practices (update bios and firmware first); however, we us proven system setups so I take full responsibility for not reading the fine print before beggining my little adventure into the wonderful world of x48. That is really my only complaint about the board (well, that and the 4 Dimm slots NOT being able to run at 1600Mhz...but that isn't really a big issue because my processor is an E8500 which only runs at 1333) :off:
I love my new system. I have a friend who has an i7 and he complains all the time about the problems with it and he has the Intel x58 board. I don't have any stability issues at the moment (even with Vista). And everything is working like it's supposed to. For that, I tip my hat to Intel. It is very difficult to make a board that is:
1.) Overclockable AND stable
2.) Works out of the box with the 1,000,000 different configurations people are going to use on it
3.) Pretty, oh so pretty (yes, I broke out a little "West Side Story")
Bottom line...I agree with you...and with Manel. I'm happy with the board. I would be happy with a Porshe too...but that wouldn't stop me from wanting a Ferrari! :rofl:
Hopefully one day, we will live in a perfect world and all will be well with Intel's updates...but until that day...we can still dream...:yepp:
One other thing I forgot to mention...Sean...your analogies are off the charts man...where do you come up with that stuff? It's like pure magic! I love them. Keep 'em coming.
Soccergeek, thanks for the props...mostly they have developed through experience while trying to explain difficult or foreign concepts to my friends and family. My educational background is in the biological sciences, and many people that ask me related questions often doesnt understand how complex the explanations behind a simple thing can be, nor do they often (if ever) used that part of their brain and so then the struggle to relate it to something they can grasp ensues.
My wife is particularly guilty of this and has aided tremedously in the fine-tuning of my analogy skills!
sean
SGeek, Sean really is the analogy man. He has shown that in several occasions.
For me this board was like love at first sight. Simple yet beautiful as hell. No heatpipes and bright colors like some other manufacturers out there.
For me it was the no legacy finally that convinced me. Still, the Marvell controller could be gone, although for some the extra ports might come handy of course. If FiewWire was on PCI Express instead of regular PCI better yet, but no more lanes.
Another thing Intel might have changed would be the capacitors used, I'd prefer the Japanese solid caps, only because the electrolytic ones can spill, happened before. The longer lifespan is really a marketing trick, your board will probably not last that long in your hands anyway. Regarding power consumption, really can't tell if it's really that better.
I have a problem also with the placement of the front panel connectors, my 900 cables are not long enough, they're too far off.
Another issue is the power loss issue but that's something they should have fixed long ago.
Maybe the audio was not the best option, although I find it good enough, but the driver updates from IDT (or maybe Intel doesn't provide them) are almost not existent.
Other than that... great board.
Also, Intel could have provided the 40mm fan for the NB.
The 4 mem slot at 1600 issue is, I believe, an XMP issue, not this particular board I guess. Not sure if other manufacturers support 1600 in all 4 slots.
Sound drivers are horrible, just lost the ability to select 5.1 setup in the drivers. Also, DD never worked. Crappy drivers, look horrible too.
Speaking of sound, still couldn't figure out what sound coded the DX58SO uses, I've seen the same IDT but also Realtek.
Intel has some problems with providing the correct information.
Intel changed their website, downloads page at least...
Nothing new though!
IDCC even for DX58SO sees no updates latelly, I guess Matt's prediction for final version for late Q1 is coming closer to a no go. That means that our final version will not be coming soon, if ever. I guess we're stuck with a beta version that doesn't even work with the latest BIOS.
From what I understood from Matt's email, they would be unifying all versions for X38/X48/X58, not sure though. Auto-tuning for us also sounds good!
Quote:
Originally Posted by Manel
Manel, I agree with the fan setup. I know of at least one board that has 4 1600 MHz fsb mem dimms. But it cost $400!
I have never been able to get my surround sound to work. I always just figured it was my speaker system (it's from 2001) going bad. After reading your post, I check it out and they still work on my laptop (supports 5.1, SPDIF, and optical). Then I went back to my PC, ran the sound setup, the TEST part worked fine, but playing music (itunes and winamp) and movies (PowerDVD) didn't play the Surround Sound. So yeah, my sound driver isn't working properly either.
Also, is anyone running their processor with the stock heat-sink and fan? If I'm not going to be over-clocking for a while, is it be worth getting a bigger heat dissappation setup?
SoccerGeek, the general concensus is that "cooler running processors last longer," but if you aren't adding additional voltage even if you are mildly overclocking, you aren't going to be increasing the heat generated by much and the stock fan should be fine. If you plan to continue running at stock or on a mild (no voltage increase) OC, the stock fan should keep the processor running for much longer than you will want to keep it. Also, the stock fans tend to be significantly quieter than the aftermarket ones, and some of the aftermarket ones do not have the 4-pin connector and cannot be easily controlled via the mobo or software.
Sean
The reason I ask about cooling is I haven't been able to get an honest temp reading while gaming. I know playing a game, like Supreme Commander or Crysis, I am using at least 90% of core 1 and anywhere from 50% to 80% on core 2...but when I try to get a capture of the temp for the processor, I get figures that are all over the charts.Quote:
Originally Posted by Sean-E-Boy
I usually play for an hour or two at a time. It's very rare that I would play for longer (usually because my wife will kick me off the computer kinda like this = :slap: ). Once, the temp read was at 39ēC...the next time it was 58ēC (that's a pretty drastic difference). I DID use artic silver when mounting the stock heat sink and fan.
In the BIOS it seems idle right at 30ēC (slightly warmer than room temperature).
And you're right, I only plan on keep this processor until 1.) Extreme Processors come down in price, or 2.) More programs take advantage of quad-core processors.
Not sure what happened to my post?? I am still having issues OCing my board. It is like the bios settings don't take affect when I enter windows. I use CPUZ and it tells me my processor is running at 2.83 under load....even though I have the bios set to 3.5........I have the C1E disabled and watchdog off.......I've done everything I can think of to try and make this work.....I'm beginning to think my board is crap. I ran into similar issues with a e8400 processor, I couldn't get it over 3.2 Ghz.
Right now I'm trying to use a q9550. I really am at a loss for words.
I thought maybe the memory was to blame so I ordered some crucial Ballistix memory 10600 and let the bios detect its settings and its fine. I never have reboots now which is a good sign. The only issue is that it just won't OC. There has to be some critical step I'm missing. This is the first board I've ever tried to OC.
Oh, on a side note, I did get it to OC to 3.5 using the IDCC.......one time.......ever since then when I reboot the IDCC fails to run. Some error message about some other program tuning my machine.........which there isnt.......
I'm really getting frustrated here.
My current build is:
DX48BT2
Q9550 2.83
2x2 crucial ballistix 10600
HD 3870 x 2
running vista 64.........
help.
BTW, i read first 20 posts of this thread and pulled out some of useful tips, than tired and went to page, where GTJ bought his DX48BT2 one, then read a couple of pages from the end.
Main problem of the thread, that in the past it was considering more on problems of particular folks, more of which were released when new version of BIOS had come.
It's simply to say, but it's nothing good for newcomers than reading this thread from first till end, but it's would have been better to has an Wiki pages or just well documented guide with tips made actually for last BIOS versions.
Sean-E-Boy, thanks, i'll prefer reduced size and better OC potential, but i'm interesting though what speed
were achieved with 4x2Gb quads?
Thinking of what HW parts 'll be next.
The memory is still the question, the tw3x4g1800c8df that Sean-E-Boy has is overpriced and GTJ is currently uses only 2GB of RAM.
Any conclusion on checked decision of 2x2Gb pair?
WOW! we lost a TON of stuff. unfortunate, but not a total loss.
First things first:
From Matt @ Intel
Hi,
You are receiving this email because you sent a request to idccfeedback@intel.com requesting information on the Intel Desktop Control Center (IDCC) availability for the Intel Desktop Board DX38BT or DX48BT2. If you no longer wish to receive email updates on this product, please reply with UNSUBSCRIBE in the subject line.
As you know, the IDCC beta version v4.1.0.168 has been available for over a year, but it has some limitations. Due to architecture changes in the underlying Extreme Tuning Utility (XTU) Service (upon which, IDCC relies), this version of IDCC is only compatible with BIOS version 1814 or older. We have been very slowly working our way toward a new version of XTU, IDCC, and BIOS that are compatible with one another. We are now down to one critical show-stopper issue in XTU. While it's only one bug, it's difficult to resolve and I don't know how long it will take. We are working on it and will get a new version of IDCC posted as soon as we get this issue resolved.
We will get IDCC released for your motherboard as soon as possible. Thanks for your continuing patience.
--MATT
Matt Alford | Software Product Manager| Intel Corporation
would I be better off just selling this board and getting an easier board to OC ?? I bought this thinking it made alot of sense with a intel processor and that the OCing would therefore be easier, because I'm a noob at this......
Second,
Moto, most likely you are experiencing problems related to your memory...now i am not saying your memory itself, but there are some things to try:
1) make sure that you memory dimms are installed in the BLACK slots. I know the manual says blue, but just imagine you didnt read it.
2) if they aleady are in the black slots, or if that didnt help, you should loosen the timings on your memory. I experience the same issues--have system set to 4+Ghz and while fiddling with memory timings all of the sudden the system boots to 2.85 (stock) and all of the memory settings are at defaults too. so, go into the BIOS and to the memory settings page and increase the first 3 numbers by 1. save and reboot and see if you are still having issues.
3) if #2 didnt help, adjust the reference frequency and the mem frequency to something that results in a slower memory speed as seen in the first screen of the BIOS. I am going to guess that you are trying to keep the memory at 1:2, ie resulting memory frequency as seen in CPU-Z equal to the FSB seen in CPU-Z. if so, you are trying to run your memory at 1647 MHz. while this is possible with this board, the additional confounding factors of an 8.5 multi and a quad core processor simply overload the MCH. Usually, however, trying to run the memory too fast will result in crashes at the desired speed, whereas the system booting to the stock settings is a problem with memory timings being too tight and the MCH being unable to keep up.
4) now pc-10600 memory is "designed" to run at or below 1333 Mhz, i am not sure what the overclockability of the memory you have is, but i am going to doubt that they can run much faster than 1550 Mhz let alone 1600+. but do not let this bother you...while fast memory speeds are great for things like transncoding a 6gb movie where huge chunks of data are tossed around, most other things you do will not be affected.
now, my end suggestion is this:
find the settings recommended for your memory as well as the voltage required and write down. you can usually find this on the memory itself.
while you are there, make sure that they are in the BLACK slots only.
reboot, go into BIOS to memory, choose manual, set the mem ref to 400 and the memory speed to 1333, set the timings to those on the memory. they will usually give you something like 7-7-7-24 1T, the first 4 are the first 4 listed from top down, the last (1T) can be set in the "command rate" field. now set your memory voltage to what is listed on the memory...be sure to set it 1 option higher than the closest or actual match (dont worry about why).
save, reboot (system may freeze upon reboot, this unfortunately is normal with this board. just shut down manually with power button and boot up again...may have to do this 2-3 times) and you should be working fine.
if not fine, let me know.
sean
dont give up, everyone is a noob once
can anybody help me? i am running a DX38BT-Q9300-OCZ Extreme Edition (2X2GB) and i cant get my settings right. if anybody is running this same setup could they post their settings for me. here are my current settings
vcore default
Vfsb 1.3
mult 7
Vmch 1.425
1333/default 7-7-7-20-19-1.74v
when i do a restart it begins to boot and then shuts down then starts up again completing the boot.
caneymud, does it do that every time or only after changing something in the BIOS?
everytime
dukzcry, the speed that you can achieve with all four dimms populated depends on a couple things.
probably the most important has to do with what processor your are using (i see that you are talking about quads) as in the specific model, and what its highest multi is--or at least the highest you plan on using.
with the quads, the MCH has to handle much more I/O than with dual core processors, and because of this added overhead it simply cannot keep up beyond a certain point.
also you have to consider that there are slight differences in the quality of MCH from board to board. for example, the x48 is simply a higher binned version of the x38. essentially all this means is that is is the exact same as the x38, but it just so happens that it is capable of "better performance" than the average x38, according to sources interviewing Intel folk.
as for how fast you can run 4 dimms with quads, my answer is going to have to be that it varies. For example, i used to have a Q6700 (1066 fsb) with 4x1gb super fast capable ram, but i could never get it 100% stable above 1580 or so with the 10x multi. now that was before i learned how to adjust GTL refs and voltages appropriately so maybe i could have got it stable at 1600 or 1610....but there is only so much tweaking you can do for stability when the processor is running flat out and everything is close to its max.
now i have a q9550 (1333 fsb 8.5x multi) and with those original 4x1 dimms i ran into similar issues at like 1602 MHz for the RAM, which was frustrating because going from 333 core clock to 400 is hardly an overclock, in my opinion. so i ditched the 4dimms and went for 2x2gb and it made a huge difference.
with 2x2g and an understanding of GTL Refs and how vFSB, and vCore, affect stability, especially with MCH and high memory speeds, i was able to reach and run a few tests (not fully stable, but pretty good) at 1840 MHz with a max stable @ 8.5 multi somewhere around 1760-1780.
now the other issue at hand is this:
i know that you are thinking about 8gb of RAM, but why? unless you are transcoding a movie with something like Premiere Pro, while at the same time working on some super complex 3d CAD thing all while having 20 tabs of IE8 open with Java on them and playing Crysis, you DO NOT need that much RAM.
Its like having Z-rated pirelli tires on your Yugo...sure they can handle speeds over 180 Mph wihtout flying apart and are so sticky that the car will hold on a 60 degree angle, BUT ITS A YUGO! Bicycle tires would provide more performance than the car could handle...
a good exercise is to open the things that you commonly do at one time or even the most that you could feasibly do at once and simply look at the windows task manager. If you have less than 10-12% free memory, then yes, more ram would be helpful, otherwise adding RAM is a waste of money. Its like having a Yacht on a 10 Acre lake....or a 7 bedroom house and youre single with no kids, no pets, no friends who stay over, and you sleep on the couch.
If youre looking for a faster system, and you already have a quad, and you are already overclocking, and you have a good graphics card (if youre a gamer) the next best place to put your money is in hard drives. they are the true bottleneck in today's computers. throw some money at some good overall performing SSDs or buy 6 raptors...2 in raid 0 for your OS and 4 in raid 0 for your games or movies or such.
sean
first off, make sure that if using 2 memory dimms that they are in the BLACK slots only, not the blue ones.Quote:
Originally Posted by caneymud
second, dont set any memory settings to default or auto...this will only screw you.
set your ram to 333 1333 command rate 2T
if this works, try command rate 1T.
if none of this helps, tell us more about your setup...is the processor at stock or overclocked, if overclocked, what is the host clock and the multi.
what is the wattage of your powersupply? how many video cards and hard drives, etc?
here are my memory settings
333-1333-7-7-7-20-19-6-8-5-6 1.74v 2T
after changing settings it tried to start 4 times and finally got to windows logo and crashed. started again went down one more time and back up and windows started. i will post back in a few to let you know how it goes on multiple restarts. i am running 2X500gb 7200rpm hd in raid 0. one nvidia card with 1gb ram on it, 750w zalman PSU, thanks for the help i wil post back with my fsb and processor settings.
I'm gonna try to overclock the i7920 with DX58SO once I get home. Hoping for decent results!
http://img25.imageshack.us/img25/553...rd02xvzphb.jpg
Sweet results for a DX58SO. I've updated to the latest bios and overclocked it to 3.6ghz with stock voltage! Temperature never went past 55c with IBP-Z001 cooler (dualfans mod) + Noctua thermal paste.
Only problem I had was slow bootup and instability when trying to go over1600 RAM speed.
Thank you for such exhaustive post.Quote:
Originally Posted by Sean-E-Boy
The see-pee-you in question is a q9550 w e0 stepping, with locked multiplier, so memory with good OC potential will be best offer.
It was the memory host frequency that was stopping me. And I tried to put the memory in the black slots and it won't boot at all......odd
Now my dilemma is heat......
I got it to OC to 3.78 GHZ giving me a 19,000 on 3dmark (stoked)
The issue I have is I ran real temp and monitored it under load and I just about hit TJ MAX......it hit 98 and I stopped the test....went back to bios and dropped the voltage. Now I still get hot but not like that. idle its around 50.....thats way too hot...I'm thinking my cooling fan is crap...it isn't very loud and the RPMs won't even break 2000.......I gotta do something about that.....
Any suggestions??
Moto,
DAMN THATS HOT! What were you using to monitor temps? You mentioned IDCC earlier, are you using that for fan control?
So I take it that youre on the stock CPU fan? If so, i suggest that you:
1) Write down all the values that are on the CPU page in BIOS.
2) Set vCore, multi, and host clock to stock values, make sure that enhanced power slope is Enabled and that the other power option to increase the voltage by xxx is disabled
3) save, reboot, re-enter BIOS
4) go back to the cpu page in BIOS and look at the bottom where it says what the current vCore is, now manually set the vCore to that value.
5) save, reboot, load into windows.
6) come back here and post the values from step 1, the default vCore value, and what temp monitoring program you are using.
7) download http://www.ultimate-filez.com/files/IntelBurnTest.zip
8) open the zip file and extract the LinData folder and the IntelBurn Test.exe to your desktop.
this program is the most intensive stress testing software that I have found and it produces TONS of heat as well as good confirmation of stability.
9) while having a temp monitoring program running and nothing else, double-click IntelBurn Test.exe
10) type N, <enter>
11) type 1, <enter>
12) type at least 5 (higher is better, like 10 or 15, but for the sake of time start with 5)
13) wait and monitor temps.
14) if test completed without crashing computer or freezing (each run of the test takes up to 2 mins, so be patient) make sure that all the numbers in the far right column are the same (see attached). If they are not, then you have stability issues (probably the ram) but you shouldnt run into this for a little while.
15) if everything went ok, restart computer, go back to BIOS and without adjusting Vcore, increase host clock by 20 or so.
16) save, exit, repeat tests, repeat increase if all ok.
17) once you run into stability issues while runing test, increase vCore by 1 step. keep doing this until you reach a stable point where you are satisfied with temps. I would say dont let the CPU get over 80C, others will tell you lower, so I will leave the final decision up to you.
If after all this you are not happy with your OC, and i imagine you wont be, save up some money and get this:
http://www.amazon.com/Swiftech-H20-2...0503171&sr=8-4
it takes some time to install, and you need to be sure that your case is big enough, but itll keep your stuff cool and it has enough capacity to add an MCH water block. For example, my Q9500 @ 4.124 Ghz or 8.5 x485 and 1.55 vCore doesnt get above 82 with the IntelBurnTest.
You can read more about the package here:
http://www.swiftnets.com/
Well, sorry that there is no image for you to see....for whatever reason, we have lost the ability to upload pics from file, AND linking to URL doesnt work either...sigh
you can see it here, though:
http://fourpawhikes.com/images/IntelBurnTest.JPG
Been away for a while but it seems we lost some posts around here.
Sean, received the exact same info from Matt @ Intel. Maybe we're close on the final version od IDCC after all, if they can solve the XTU issue that is. I understand they are still working on a new BIOS version as well, maybe this is the one we all have been waiting for!!!
Caney, I believe you have a low vMem issue, yoou mention OCZ Extreme mem, I believe it's the Intel Extreme series. I'm on the Plats and they require 1.8V, although I'm actually running them at 1.78V. Maybe you can try a higher voltage, just 1 or 2 notches, the OCZs usually require around 1.8V, not the newer ones for triple channel though at only 1.65V as "required" by the i7 (not exactly but hey...). Anyway, I was cheking out OCZ site and it needs 1.75V @ 1333 or 1.8V @ 1600. Maybe this is the issue. Try it.
I'm still also using the stock fan and my temps are miserable, might have to re-apply paste and/or use a new cooler. I like the Extreme series ones but they're too loud I hear and you can't buy them separate anyway.
From my experience, if you are overclocking either the fans will be noisy or they simply wont cool well enough.
Prior to my H2O setup, i was using a Zalman CNPS9700 NT type fan (not sure if the model is exact) and while quiet at idle, it would make some pretty good noise at full tilt. With the H2O, i can set the fans to one of 3 levels by changing the connector and restricting the volts, but i am running it full out now since under load my cpu can get around 80C.
the noise isnt particularly annoying...sounds like mid 1990s gateway...and while it is a bit louder than the zalman at idle, it is much quieter and more effective at load...plus i have a cooler on the MCH, which is nice. Before, i had a little 40mm or whatever fan on it and that alone was louder than my zalman and the 2900xt fan together.
maybe one day someone will figure out how to make some liquid in which the entire board can be submerged without shorting and a few "fans" that direct the flow across the cpu and other parts.
as far as IDCC goes, i really dont care anymore. would be nice to tweak from windows, but it can just as easily do it from the BIOS and i no longer have any need for fan control.
Sean
There is such a liquid for submerging the entire board already, for some time now. It's a non conductive liquid of course, a type of oil or parafin or something, can't remember what it was exactly. I believe some german (I think) system buider sells such a setup, the case was quite nice I believe. You extracted the board from the top, with a handle, but problem is it usually was a mess if you had to do so to change a vid card or so. Cool though.
Hey, So i got on here a long time ago to try and over clock my dx38bt. however It was my only computer and I got scared so I stopped.
Well now I got watercooling installed. I just would like to achieve a mild stable overclock. Honestly if someone would be willing to get on aim or something that would be really awesome.
My build is in my sig. Obviously I would have to pull out the 2 1gb and get it stable then think about putting those back in.
I'm thinking I want to start with getting the board to 1600 fsb and lowering the cpu multi. which would put me at 2.85 ghz. If someone could give me a general outline of what settings to start with and what to tweak to get it stable that would be extremely helpful. I think I understand most of it, just not the math involved with timings etc. Thanks so much!!
Pontius,
Ill be more than happy to help you out. I know that this board is tricky to OC, so you must have patience.
A few things to note:
1) there is the possibility that you will run into issues with only 500W power supply, but you may be ok seeing that the 3870 is a relatively low draw card and that your HDs are 7200 rpm, no?. Not sure what brand your water cooling is, but those pumps can draw some juice too. Just something to keep in mind.
2)
a) if you plan on using all 4 dimms at some point, you might as well start the OCing process with them in...getting stable with 2 and then adding 2 more will not be fun.
b) overclocking with all 4 memory slots filled will be limiting, but seeing that you said "mild stable overclock" you should be ok, but dont expect to keep the memory 1:2 with the FSB (ie. 1600Mhz fsb and 1600Mhz RAM speed as seen in say CPU-z) much beyond 1600
3) do all 4 of your RAM dimms have the same timings and voltages listed on the memory itself? if not, put the details in your sig. Also, are they all PC3-12800 (1600Mhz)?
4) with your water cooling, did you install a water block for the MCH or just the CPU?
5) in BIOS on the CPU page, with vCore set to "default" (and saved and rebooted, if it isnt already at default) what value is displayed for current CPU voltage? its towards the bottom of the page.
6) what BIOS version does your board have? have you regularly updated other components (<---this one is key, make sure that the chipset drivers are up to date...AND when updating chipset drivers, the super intelligent folk @ Intel didnt build an installer that actully updates everything that it is supposed to. so first run it like you think you should, then you must extract the exe (download 7zip, maybe winrar to do this...right click installer, select extract...) to a folder somewhere...desktop works...then right click My Computer, click manage, device manager, at the top click view then devices by connection, then expand ACPI....then Microsoft ACPI...then PCI Bus and then everything within PCI bus right click, choose update driver, then choose browse or i will find it myself or whatever it says, then select that folder you extracted (make sure search subfolders is checked, if available) then ok. do this for each item in the PCI bus category...many wont update, but several will that should have with the installer but didnt.
thats all i have for now, post back the details and work on updating drivers, etc.
the fun (and anguish) will start soon.
Sean
I forget about MCH fan... It was discussing, but too much past.
Is something like redundant 40x40mm Delta will fit?
Hello there!
Here is my another try to OC DX38BT.
http://img403.imageshack.us/img403/6937/dx38bt.th.png
Still unable to OC memory to 400/1600,
it is working at 400/1333 now and with only those timings
(if timings are set to it's default 9-9-9-24, mobo wont boot at all).
Dangerously hot, isn't it? I must install water cooling, i know.
I want to run memory at 1:1. If i'll rise voltage on the memory to,
for example, 1.8v (default 1.5)? May it help me to run memory at 1600?
"If your system application requires more MCH cooling than that provided by the MCH
passive heat sink, you can add a fan to the heat sink by using the included MCH heat
sink fan mounting bracket. This fan mounting bracket is designed to accomodate a
40 mm x 10 mm or 40 mm x 20 mm, 12 V dc fan. You can use the board’s 3-pin MCH
fan header to supply power to the MCH fan."
its a 40x40mm square 10 or 20mm tall.
the sound that they make is really annoying, though.
petr0id,
you probably dont need water cooling. the TDP for your E8400 is only 64 Watts, whereas the quads are much higher...like 95+ watts (about 50% more). while water cooling would keep you very nice and cold and you could put an MCH water block on too, a good aftermarket CPU fan like a zalman or something should do just fine.
as for the memory, it is unlikely that more volts will increase your speed much (some memory respond well, some dont), usually increasing mem voltage helps when trying to lower timing values.
what is the model number of your RAM?
also, what are your voltages in BIOS (mch, fsb, etc.)?
Sean! Thanks for your reply!
Here is some pictures:
http://img26.imageshack.us/img26/2804/001jjn.th.pnghttp://img18.imageshack.us/img18/4387/002egz.th.pnghttp://img510.imageshack.us/img510/9910/003q.th.pnghttp://img266.imageshack.us/img266/2506/004g.th.pnghttp://img408.imageshack.us/img408/7138/005wjl.th.pnghttp://img142.imageshack.us/img142/4448/spd.th.png
Watchdog and EISST disabled, voltages are manually set to it's stock values. About temperatures: i'm not sure who's lying - RealTemp or BIOS, but temperature values are different: in BIOS (how you can see on the second pic.), it is only near 50C, but RealTemp showing something near 70C.
[Edit]:
BTW, i can make good pictures of all BIOS pages and post it here, only if we need it to be posted here, in that thread.
Petr0id,
excellent shots! first, like you said earlier, stock cooler may hold you back here, but it doesnt mean that you cannot start figuring some things out.
Question: what vCore were you using when you ran the Prime95 test a few posts ago?
Next, there is a progression of testing that helps you to determine some limits such as max FSB, Max (more or less) CPU frequency, etc.
the first thing to do is to start with all voltages on the CPU page at default...setting them manually is fine, make sure enhanced power slope is ON and Voltage offset is OFF,
now set your multiplyer to the lowest possible and put your host clock to 333 (stock),
next, go to the memory page and set your memory to 400/800 and raise timings...i know you said that you had issues with running higher than 7-7-7-21, but try to set it to 9-9-9-28 or higher and be sure to set the Command Rate to 2T. Memory in Black slots, right?
also, just to keep out the possibility of your video card causing issues, put the PCI burn in back to 100Mhz for now.
save and exit and boot into windows. remember...this board has issues with changing memory settings and being able to boot fully the first try, so you may have to power cycle a few (a few could be up to 5) times to get into windows.
once you get into windows, download this: http://www.ultimate-filez.com/files/IntelBurnTest.zip
open the zip and extract the LinData and intelburntest.exe files to your desktop or to a folder somewhere. we will use this program a little later, but good to get it now b4 i forget.
if you were running at 400x9 earlier on stock voltages, go ahead and restart the computer, enter BIOS and jump to 400x{lowest multi} without changing anything else. save and exit.
if you were not running 400x9 with stock voltages, restart computer, enter BIOS, go to ~360x{lowest multi} without changing anything else. save and exit.
so the idea here is that with the processor multiplyer set to its lowest, the memory speed set to its lowest, and the memory timings more loose than normal, we can increase the host clock of the CPU which increases the FSB and know pretty well that when we reach a point of instability that it is caused primarily by the FSB and related voltages rather than by some other thing. this process of OCing takes more time, but is less frustrating than the "just crank-it-up" style. Plus, you will have a better understanding of what your configuration is capable of in multiple categories.
the way to proceed here is to gradually increase the FSB (by 10-30 each time) until windows freezes on boot or a quick Prime95 (version 25.8 or 25.9) small FFTs or in place large FFTs results in an error or crash/bluescreen, ect.
using a piece of paper, or preferably a second computer, keep a spreadsheet of the settings that you are using and note whether testing was successful or failed/froze/bsod. the more information you have here the easier it will be later to troubleshoot issues.
[EDITED] once you get to that point, then increase the vFSB by 1 step, change nothing else, save, exit, reboot, enter windows, try testing again. at a certain point (and this could be quite high...Host Clock 450,470 maybe higher with dual core), increasing the FSB will not fix instability and you will need to raise vCore or vMCH (see last 2 sentences in this post). but once you figure out if vCore or vMCH was needed, increasing host clock should again cause instability that can be fixed by vFSB.
one very important practice to get into is making sure to only change one thing at a time when trying to troubleshoot instability....if you raise vcore, vfsb, vmch all at once, sure it may fix the issue, but what WAS the real issue? understanding what is causing the problem give you a better grasp overall of what is going on.
[ADDED] also, when you get to the point where windows freezes upon loading and you are still at default vCore and no increase in vFSB fixes the issue WRITE ALL VOLTAGES AND SETTINGS DOWN. this point allows for some more advanced tuning that we can go back to later.
Thanks, Sean!
Your small manual on oc'ing is very useful for all, i think.
I will give it a try 8)
Answer: vCore wasn't changed. It is at 1.3250V.
Yes, memory is in the black slots and i think, i'll gonna buy 1600Mhz memory.
Memory and a stock cooler are the weakest points in my system for now.
First of all - temperature. CPU must be cold in the summer time.
glad to help. you may want to hold off on buying memory for a bit. I know that with the quad cores and stock multis that the MCH cannot handle mem frequencies much over 16 or 1700, but with the dual cores you may be able to stay 1:2 up to 1800 or so.??? something to consider.Quote:
Originally Posted by petr0id
Hmm. Do You think my memory is overclockable?
Agreed.Quote:
Originally Posted by Sean-E-Boy
I will continue my experiments and will post here results (if there will be no fire onboard) :D
http://img140.imageshack.us/img140/783/84047038.th.png
Thanks to Sean-E-Boy, tonight i was able to run my rig at 8X450,
at stock voltages (vFSB - 1.2000, MCH/ICH - 1.250) without further problems (eg reboots, freezes, BSOD's etc).
Memory was set to 400/800, 9-9-9-24, 1.58v (+1 notch to default mem voltage).
Windows won't start with higher FSB (missing system .dll's, .sys's at bootup), so i leave it at 450 and save those settings as custom defaults.
Next step is to run those settings with the memory at higher freq's (i tried to run it at 400/1600, but no luck).
[Edit]:
I'am worrying about my HDD's. All those power ON/OFF's...
Updated audio drivers ver. 20001.0 is available for download.
But not a word about new Intel Chipset Software Installation Utility 9.1.0.1014
pert0id,
the failure to boot b/c of missing files is caused by a voltage issue...dont remember which, but probably vFSB. it tends to happen after system crashes during boot, then next boot it says bad files. drop the FSB down and its ok, then increase vFSB, increase FSB and try again.
what is the lowest multiplyer for your E8400? how high did you get the host clock at the lowest multi? also, dont be affraid to increase the vFSB some, maybe even up to 1.3 it shouldnt affect temps much.
your hard drives should be fine. if you are concerned, you can always unplug whichever ones dont have the OS on them, but my 6 have been power cycled probably several hundred times just from trying to OC, and (crossing fingers) so far so good.
what is the actual model number of your RAM?
Alright Sean, thanks so much for your reply! :D
Heres what I got written down:
bios version (i know its unnecessary but heres the full thing) BTX3810J.86A.1554.2008.0501.1628
processor voltage is 1.325
also, FSB volt is 1.225
MCH/ICH is 1.250
my memory timings are the same and the speeds are the same. I made sure they would be the same before I got them.
ref freq. 333 MHz
mem freq. 1333 MHz
tcl: 7
tRCD: 7
tRP:7
tRASmin: 24
(will someone let me know if these next ones are important i haven't messed with them and don't know if they matter..)
tRFC:60
tRRD: 4
tWR: 10
tWTR: 5
tRTP: 5
oh and my voltage is 1.82
command rate: auto
I went through and did the driver updates like you suggested. Thats so ridiculous of having to do that manually. I always assumed the inf update utility did everything. psht. oh well.
Yes I have a waterblock installed on northbridge as well as CPU.
If i have to get a new powersupply oh well. I've already poured enough money into this thing. I think it should be fine though.
It is strange, because i did not touch any voltage settings, only mem voltage is at 1.58Quote:
Originally Posted by Sean-E-Boy
Lowest is 6. I was able to run it at 450MhzQuote:
Originally Posted by Sean-E-Boy
PC3-10700H (667MHz), Samsung, M378B5673DZ1-CH9Quote:
Originally Posted by Sean-E-Boy
petr0id, just dl'ed the new audio drivers. At last, it seems Intel heard us here.
Funny thing, there were several updates in between that Intel never released. Also, the latest available drivers before this last one are not mentioned in the readme, 5790.3_5713.7_IDTGUIv83. Those IDT guys are also very confusing in the descriptions. The legacy_aus tag doesn't comfort me either, why legacy, and aus is for australia only?
Latest versions are not WHQL?
Is the GUI updated from the old design (ugly as hell)?
Have you installed it yet? Is it working properly?
Hope it solves my issues, can't select 5.1 anymore, some scratching also. No Dolby controls in the GUI either.
Where did you see the INF 9.1.0.1014?
Also mew Intel Integrator Assistant.
Pontius,
Good, sounds like most everything is in order....I would update the BIOS to 1902, though. Since you are using water cooling, there is no real need to have an older BIOS to be compatible with IDCC for fan control, and sure its nice to have for monitoring temps and voltages, but everest does pretty well for that.
after updating BIOS, i would suggest that you refer to the above post to petr0id: http://www.xtremesystems.org/forums/...postcount=1884 to determine your cpus max fsb.
in short, this is what you do:
0. make sure Watchdog is DISABLED
1. take memory to 400/800 and raise the fisrt three timings to 9 and the next one to 28. also, make sure command rate is manually set to 2T.
2. go to the cpu page, i am going to guess that 1.325 vCore is the default, but manually set it to this level if it isnt already.
3. And since you are on water, increase the vFSB to wherever you are comfortable...large gains in FSB speed are seen with relatively small increases in vFSB up to 1.4 vFSB, beyond that more voltage is helpful, but less performance is gained per step....above 1.45 is really for squeezing those last few Mhz out of the FSB.
as a cautionary note, overvoltage can be just as damaging as overheating....but i wouldnt let that scare you. 1.4 should be pretty safe...I have been running my 9550 vFSB (and vMCH) at max for 5 months now, still running...and i am going to guess that it is going to last longer at this voltage than i will want to keep it.
ANYWAY, set the vFSB to where you are comfortable and since you are running 4 dimms, kick the vMCH up to 1.35 or so.
4. set the multi and host clock settings to manual, take the multi to its lowest, put the host clock to stock 333.
5. save, exit, reboot into windows, verify with cpu-z or everest that you are in fact locked at the lowest multi (had problems with my q6700 and this). do so by having cpu-z or everest open and then open a program like word or photoshop or something that takes a bit of time to load and watch the mulit or the processor frequency and make sure it doesnt jump up and down.
6. if this checks out, restart, go to BIOS, increase host clock by 30 mhz and change nothing else. save, exit reboot into windows.
7. download this: http://www.ultimate-filez.com/files/IntelBurnTest.zip extract the lindata and intelburntest.exe to your desktop, double click intelburntest.exe, type n, then 1, then 5. it will take a little while to run, but when it is done, make sure that all the numbers in the far right column are the same. ***during the tests is also a good time to monitor cpu temps as this is the most tempurature intensive test i have found***. if they are the same, restart, go back to BIOS, increase host clock by 30 again.
8. repeat.
9. once you get to 400mhz, you may want to start increasing host clock by steps of 10-15.
10. post back results and problems
petr0id,
i know you didnt touch voltage, but you did increase the host clock. increasing the host clock requires more vFSB at low multis to be stable, just as increasing the multi from 6-8 at host clock 450 will require more vCore to be stable. its all related and the missing file error is a sign of undervoltage.
Hiya Manel
I was recently bought E-MU 1212M, so audio is disabled in BIOS
and i can't test new Intel's famous audio drivers.
About 9.1.0.1014, here, for example.
well i adjusted the memory timings, and set the voltages like you outlined. turned off eist and watchdog. it did appear that the multiplier was indeed locked at the lowers. i went to open outlook and the computer shut off. i increased the fsb voltage a little bit to see if that helps.. watching it boot right now.
also i stepped up the voltage on the memory to 1.9 i think, thats what ocz says the 2gb sticks should be at anyway. apparently there is a slight difference in the timing of those, the 4th timing on the 1gb sticks sould be 20 and it should be 24 on the 2gb. i've never had any issues with it.
hrm im just kinda trying messing with some of the voltages that you had suggested to alter. no real progress. Its interesting that it will make it all the way to windows, but will crash when I try to load a large application. that makes me think it's memory but i have no idea.
EDIT: Alright. well i pulled out the two 1 gig sticks. booted. things went smoothly, i was able to open outlook and even fallout 3. i went to run just a brief pass of the intel burnin utility and my computer just turned off (although I think thats what my computer does when it blue screens). so i adjusted the speed of the memory back to 1333/333 and it seems to be running just find now, the burnin test is running right now.
and this is all at 333 host clock with lowest multi?
yeah i was very confused about it, because technically it was an underclock. perhaps those memory don't like at running at a slower speed though?
but after I put the memory back on 400/1333 it worked just fine. had no problem running stress tests after that. i left the timings at 9-9-9-28 2t and was able to easily bring it up to 400mhz on the lowest multiplier(6).
here's the voltages I was using, some of them may be a bit high cause i was kinda just guessing
cpu volt 1.3375
fsb v 1.375
MCH 1.35
i even put the other 2 sticks in there, brought the multi up to 7 giving me a 20% oc, at 2.8 ghz. stress tests pass. currently im trying to tighten up the timings on the memory and bring that up to the 1600 fsb speed but im running into the issue where windows says file not found, so im gonna re-read what you guys were talking about a few posts back.
EDIT: sorry I'm probably being confusing. yes originally i couldnt open outlook at lowest muliplier at 333. I've gotten past that though :D thank you SO much for your continued help.