EX58-UD5 and (possible) RAM problems

Toggle sidebar Toggle sidebar
flervk

flervk

Distinguished
Sep 25, 2009
13
0
18,510
hi all.
first off, excuse my lousy english; I'll try my best to explain, but can't guarantee

anything :)

well, I'm not really sure what exactly is wrong with my new setup, so I'm sorry if this

thread is in the wrong place.
I'm running Win7 RC on i7 920, GB EX58-UD5, 12 GB DDR3 1333 Corsair XMS-3 (new stuff),

and Corsair HX520, nVidia 9600GT, 3 x WD HDDs (from before) - the problem is, OS keeps

resetting randomly all the time, and it happens during some basic operations (longer

installations, surfing, files copying etc.; I don't know how it would handle some heavy

rendering, I didn't manage to install my 3D applications yet), or being idle for some

time (btw, there's no BSOD, even though I checked that Don't restart immediately upon

error option).

now, connections of hardware to PSU and MBO are double-checked. all of RAM and HDDs are

visible in both BIOS and Windows. nothing is overclocked. all drivers are up-to-date.

temperatures are normal, I believe (the case is Antec p182, properly cooled with 3 case

fans and Scythe Kabuto on the CPU).
I've tried many tests for hours (Prime, OCCT, memtest, various HDD utilities, Cinebench

etc.), and absolutely everything passed with 0 errors.
I'd like to point out 2 things: in safe mode everything works (installations complete,

gigabytes of files are copied around just fine, and system is stable in idle), which

makes me think that the problem is software (driver?) related.
secondly, RAM is operating at 1066, which, I've read, is normal for i7 setup; however,

people also mention that the RAM voltage values should be slightly modified to improve

stability (which I don't dare to do by myself - I'm not into OCing and really don't know

much about it). but if that's the case, how come the memory passes all the tests?

I'm really confused with the situation and would appreciate any help, thanks in advance.
😱
 
Setting the BIOS to the proper DDR voltage is not overclocking - it's setting to required manufacturer specs. The specs on that RAM want 1.6V - set the BIOS to provide that. It'll also likely make the motherboard "see" the RAM as 1333. The 1066 is not normal, just want many settle for.

First thing is to go into BIOS and set it to Optimal Defaults. You can modify from there if you want, but doing that makes sure all basic settings are proper. Then adjust the RAM voltage.

Post back and let us know how it's going.

(Your English is wonderful! But no need to put the extra lines between lines - makes it harder to read)
 
The 1066 is not normal, just want many settle for.
Click to expand...
1066 is the only speed supported by Intel for the i7 CPUs. That is not to say many people haven't got it running faster - simply that you're 'noman's land' doing it... Also, there is no appreciable, subjectively visible reason to do so - you won't be able to tell the difference between i7 RAM at 1033 versus 1600 - homey don' play 'dat! :
http://www.tomshardware.com/reviews/memory-scaling-i7,2325-5.html
 
They didn't seem to include the cost of that test!

I don't know a thing about SSDs - but I realize that if I can stick a 32GB flash drive in my pocket, it won't be long until I'm putting one into the computer permanently.
 


okay, thanks for the answer.
three sticks are taken just out for now, the system is perfectly stable with 6GB;
I'm leaving messing with the voltage for later today:)




thank you, kind sir :)
those extra lines weren't supposed to be there; constant restarts forced me to write the text in wordpad and then c/p it to forum, so..
but (almost) everything is well now.
 
those extra lines weren't supposed to be there; constant restarts forced me to write the text in wordpad and then c/p it to forum, so..
Click to expand...
Ack - I though it was just peculiar to my system - I have the same problem with 'canned' posts edited in notepad!

Much like it's predecessor, the MCH, the X58 IOH needs a bit of extra voltage to accomodate extra DIMMs - the setting is on the "Advanced Voltage Control" sub-page of the "MB Intelligent Tweaker(M.I.T.)" page of the BIOS:

"IOH Core" to "1.200V", or possibly "1.250V", if that doesn't do it...
Another thing that will aid in stability with six DIMMs is to 'bump up' the tRFC that the "Load Optimized" set in 'auto' by eight or twelve for stability, then 'walk it down' two at a time, until stability is affected...

Here's an excellent GB X58 BIOS guide:
http://www.mediafire.com/?txzi1tmf22g
 
To fix the problem of extra CR's in Notepad, maybe Wordpad which I don't use, here's whatcha do. With Word Wrap turned on, just adjust the size of the window. The text will re-wrap and the issue is gone. Sometimes, toggle from Wrapped to un-Wrapped.

You can also save defaults of all sorts in Notepad by saving the file "notepad.ini" in the folder you start Notepad in. I use shortcuts to run Notepad from C:\Data\Notepad and keep both a copy of Notepad.exe there and a copy of a blank test file called notepad.ini with various pre-sets, like default font, style and size.

I have an handy old utility called The Cleaner, now apparently called e-Cleaner, that will fix many problems with text files, not just the <<< forwarded email symbols it's advertised to fix. Handy for messed up text files, recovered data, etc... I'm using v1.02 but I see a newer version here:
http://www.pcworld.com/downloads/file/fid,6492-order,1-page,1-c,alldownloads/description.html
 
heh, I know about wrapping, but I was in a hurry to post the post before restart so I didn't pay much attention to details :)
 
It's all good flervk! The post was mainly to bilbat - he knows a lot more about most hardware stuff than I do and so I jumped at the chance to maybe fix a problem he's been having.
 
And it worked! Thanks mongox - I will now create four notepad'x' directories, under ProgramFiles, with four copies fine-tuned for the various tasks I use it for! I use UltraMon to manage my four-screen desktop, so I can even get each one to open on the monitor I usually use for the chosen specific tasks... WOOT!
 
Glad to help in return for a change. I figure all us old-style tech guys like using Notepad, it's one of a very few things I keep in quick launch. I also know a few trick with Explorer and have an old shut-down MS-DOS shortcut.

LOL, I still remember a few DOS Edlin commands... and my KB tricks for DOS.
 


I beg to differ. I saw a solid 10% increase in overall benchmarks going from DDR3 1066 to DDR3 1600. Thats substantial if you ask me. Especially considering that most can accomplish this with just 1.3v QPI and 1.6v RAM which are perfectly safe voltages that will assure your i7 runs "forever" as Intel claims.

While you can get RAM up to DDR3 2000 that is advertised as compatible with the i7 I tend to think DDR3 1600 is a good place to be at because as I said you can get it without having to up the QPI voltage over Intel's recommended max of 1.35v. Of course this is doubtless a very conservative "maximum" and this should not dissuade people from running DDR3 1800 or 2000 just be aware that you will most likely need your QPI set up to 1.375 - 1.475.

Devin
 
I beg to differ. I saw a solid 10% increase in overall benchmarks going from DDR3 1066 to DDR3 1600. Thats substantial if you ask me.
Click to expand...
But, you stated elsewhere
from DDR3 1066 7-7-7 to DDR3 1600 7-7-7
Click to expand...
Which, as I pointed out, is comparing apples to orangutangs. That 1066 7-7-7(RAM 'A'), assuming it will run at 1600, will need to run at 1600 10-10-10, or 11-11-11; conversely, the 1600 7-7-7(RAM 'B') will run at 1066 5-5-5 (and maybe, at 5-4-5, or 5-4-4); either way, even if comparing RAM A at 1600 with RAM B at 1066 (and 5-5-5), RAM B will be a lot faster - but there won't be an appreciable difference between RAM B at 1600 7-7-7, and RAM B at 1066 5-5-5... (And there will be an attendant increase in power consumption, current through the memory controller, and heat dissipation all around at 1600)
 
bilbat,

Agreed. My only issue was the *implied* perception of the article you linked to that running > 1066 ram on an i7 machine has no real world benefits. The logic here is nonsense to me because running the same RAM at different speeds will give little or no performance increase regardless of the CPU/mobo/etc. Everyone knows that. ....don't they?

I just wanted to clarify for future readers that higher speed RAM will in fact increase performance, sometimes substantially, so long as timings are taken into consideration.

But be warned -- you will need after market cooling. The stock cooler for the i7 is meant to be *just enough* to keep the CPU under ~ 80 degrees celcius with stock settings. If you want to take advantage of higher speed RAM you will be overclocking the CPU's QPI/Uncore which means you will need a better cooler.

I got the Cogage TRUE Spirit. You can find it for just $35 and it includes the fan unlike its counterpart, the Thermalright TRUE 120, which goes for $55 without a fan. Reviews of the TRUE Spirit show it performing very close to as well as the TRUE 120, for much less $$. With the Cogage I lowered my temps by a whopping 25 degrees celcius. I now max out at 62 degrees running Prime95 and 65 degrees running LinX. I am now safe to up my QPI voltage to 1.3 to get DDR3 1600 speeds at tight timings which in my opinion is a safe voltage and should allow my proc to run for years and years.

Having said that, I am having the same issue that others have reported where my i7 will not run completely stable at any settings. I have replacement RAM arriving today but I will most likely not solve the problem until I RMA the CPU itself.

In over 10 years of building, tweaking, overclocking and stressing computers I have never once seen a bad processor. This one was bad straight out of the box. And others have had the same problem.

Very disappointing....

I would never recommend anyone buy an i7 OEM. Get the retail box-- you will likely need the 3 year warranty.

Cheers,

Devin
 
The logic here is nonsense to me because running the same RAM at different speeds will give little or no performance increase regardless of the CPU/mobo/etc. Everyone knows that. ....don't they?
Click to expand...
Therein lies the rub! We had a posting here from someone who toasted an i7 and a pricey X58 MOBO because he didn't know to use the standoffs! Screwed the board directly to the mounting tray! 😱 I have learned to (pretty much, hopefully) assume nothing about posters' background knowledge and abilities - and try to make everything as clearly understandable (for anyone) as I can...
I just wanted to clarify for future readers that higher speed RAM will in fact increase performance, sometimes substantially, so long as timings are taken into consideration.
Click to expand...
You kind of see my point, and kind of deny it, all in the same statement; what I'm trying to point out is that the clock speed of RAM, especially for the i5/i7, is irrelevant - if you run that 1600 7-7-7, as I said, at 1066 5-4-5, you will see no appreciable difference - you may see a minor gain in 'synthetic' memory bandwidth testing (and it will be minor), but if you time any OS operations, or (preferably) a suite of them - nada! Latency is latency, and it's worth paying to reduce, if you're doing things that (like video encoding) will benefit from throughput enhancement; but latency is not a number of counts programmed into the SPD/EPP - it is a set of timing constraints in nanoseconds and picoseconds that are the operating limits of the physical DIMM itself, and the size of those 'windows' don't change at all with regard to clock speed. The only thing that improves at higher speeds is the 'integer rounding error' inherent in having to set up latency as an integer clock count. The reason I say your 1600 7-7-7 may run 1066 5-5-5, or 5-4-5, or 5-4-4, is that when I convert seven cycles at a 1600 clock to nanoseconds, and back again to cycles at 1066, I get 4.66 cycles, which says the most demanding timing constraint, column address strobe, will likely need to be 5, but you may get away with either of the two less demanding ones, row address strobe to column address strobe, and row precharge, at 4. The rounding advantage is this: the more clock cycles per unit time (faster clock), the smaller the 'distance' between the integers you must select. It's pretty intuitive - the 'distance' between four and five is twenty-five percent of four; the 'distance' between seven and eight is only fourteen percent of seven - so, the more clocks you have to 'play with', the closer you can get to 'hitting that nail on the head'!

Your point about cooling holds true for all processors; the 'thermal envelope' that the included rotary postage-stamp is designed for is stock speed and voltage with all the 'green', down-clocking/down-volting features enabled; if you want to turn the 'green crap' off for stability, and 'push it' a bit - you want aftermarket, and any aftermarket solution will be head-and-shoulders above the free junk they sent you! I also point out that people obsess way too much about which cooler can extract that last erg of heat (which, of course, makes the HSF makers happy and rich), but, unless you're really gonna 'crank it', it won't matter! And, if you are 'really gonna crank it', you probably would be better served to put it 'under water' to begin with! (And, water is really a lot simpler and less demanding than the 'enthusiasts' [and radiator marketers] would have you believe, as well!)
 
bilbat,

Again, agreed. I do see your point and I do not deny it. What I meant to express is the fact that using high-speed low-latency memory will increase performance whether you run it at higher speeds or at stock speed (1066) because of the fact that you will be able to use tighter timings with higher rated memory.

My recommendation to people who dont want to run faster than the default of ddr3 1066 is to buy higher speed low latency RAM anyway and run it at 1066 with tighter timings, as you've mentioned I could do with my ddr3 1600 7-7-7. For me the fool-proof answer to this problem is: "Buy faster RAM if you want to increase speed." Its as simple as that. Faster RAM *does* increase i7 performance but it must be understood that "faster" RAM is determined by speed AND latency.

Agreed? 😉

Devin
 
Agreed! There is a place where high speed, versus low latency, will be an advantage - any operations that require large, sustained, reads from and writes to RAM - like, as I mentioned, video transcoding... I always consider my 'pass/fail' system stress test to be: watch/pause one HDTV stream off a networked ATSC tuner, while recording a second stream off a PCI NTSC tuner, while transcoding and 'de-commercialing' a third stream to an NAS media server... But, for the vast majority of people, for the vast majority of use, this is not the case. What's going on behind the scenes: the task scheduler is scurrying around, busier than a centipede learning to tap-dance, counting 'ticks': ...tick... yo - over there, you gotta finish up, your tick is over, push your environment, that's a good fella; oops - cache snoops says we've got an incoherency - grab me a meg for him from over there; you - get me the address of the block being used by {F92BFB9B-59E9-4B65-8AA3-D004C26BA193}, will 'ya; yeah - UAC says he has permission - I dunno - we'll just have to trust him; damnit - everybody listen up, we've got a pending interrupt request, everyone drop what you're doing, and you - over there - query interrupt handler for a vector - this is important!!! ...tick.... And the most fascinating (scary) thing about it all, is that, at some synaptic, neural level, we're doin' the same thing! (...though, the older I get, the less dependable my interrupt return mechanism is - I repeatedly find myself at the bottom of the basement steps, wondering "now what did I come down here for?!")
 
Soooo, does this mean that both of you agree that I did the right thing in buying good quality DDR2 RAM rated 1.9V, 1066 @ 5-5-5-15 - then took advantage of its high quality and cranked it down to 800 @ 4-4-4-10?

Seems faster that way to me - odd thing is that it always POSTS faster at the 800 @ 4 setting, runs right thru the HD search slick as silk. And I have Win Explorer open on startup, it's normal procedure is to open showing current folder, then sit and sulk a bit while other things open - like Weatherbug, then come back and show all the drives. When set at 800, it shows the drives right away. Odd eh?
 
mongox:

Yes, that is strange.

Yes, I do think that you're example is exactly what I was talking about. Of course, if you were on an i7/DDR3 machine you could take even more advantage -- say, buying a DDR3 2000 8-8-8 kit and setting it at something as awesome as DDR3 1066 4-4-4 (you're same timings but a step faster in speed/mhz and triple channel!)

But yes, you get my point exactly.

Also, if you were on an i7 system (or plan to build one in the future) you wouldn't want 1.9v RAM. In fact, it probably wouldn't even work. Thus the 'low-voltage' DDR3 memory going around. To make the i7 happy you need to be around 1.5 - 1.65v.

Just something to keep in mind.

Cheers,
Devin
 
Now, if I find that thread where this guy argued with me about this, I would sic ya on him!

Fact is, none of these things should make a BIG difference. Compared to making sure your dual/triple channel is working or having enough RAM for OS overhead or beefing up the CPU speed - RAM speed and to a lesser degree timings are small stuff.

But boy do folks like to talk about it!

Yes, I'm well aware of the voltage issues. I think they're gonna start calling me the RAM-Voltage guy because I always bring it up and it often solves the problems. The DDR2 standard is 1.8V and i7 and even i5 is 1.5V... with max of 2.1V and 1.65V respectively. But have to watch that 2.1V even, since my Gigabyte won't deliver more than 2.1V and some 2.1V RAM likes a bump to 2.2V. Took me 3 tries to get good low-voltage/timings RAM for this board. And there's also a big push for low-voltage DDR2 going on - hence some places dumping their OCZ Plat 2.2V stuff because they don't think it'll sell once the 1.8V is stocked. They say it's because stock on the 1.8V isn't available yet!

I'd like all the DDR3 RAM at places like NewEgg to be labeled either i5/i7 or AMD. The difference being the voltage requirements.
 
So... I'm still a bit new on this i7 stuff...

Let me get this right,
The only reason to buy fast i.e. 2000MHz RAM is so that when you clock it down, you can set better (lower) timings?

Also,
With faster RAM, you can OC more? (with more stability)?

Say I wanna do the popular i7 920 4GHz overclock... would I need 2000MHz RAM?
or would 1066MHz RAM be able to do the same thing?


Sorry if this question is stupid.... 😛
 
allen,

Yes, 1066 RAM will be able to do the same thing as far as the 4ghz overclock because you will be overclocking the cpu core and not the qpi/uncore. You are welcome to get the ddr3 2000 low-latency RAM but from what I've seen most who are overclocking the i7 have left their RAM relatively low and clocked up the core clock to 3.4-4ghz. This may be due to the fact that the i7's memory controller is very inconsistent when overclocked.

It up to you. If you have the extra $$$ I say but the faster, lower-latency RAM and see how fast you can get the RAM running along side your overclock. Otherwise just get the 1066 and only worry about overclocking the core. The latter is the safer and more full proof option but its your call.
 


I see. I haven't had or used a i7 before, so I'm very n00bish with it's overclocking etc. It seems quite complex!
I'm looking at prices of the Kingston HyperX and the 2000Mhz CL8 RAM is quite expensive ($300+) while the 1600Mhz CL8 RAM is about $200.

I know you can't compare latencies across different speeds, but the CL8's are the lowest timings on any HyperX ram.
The other brands with timings of CL7 etc at 1600 and 2000mhz are just wayy too expensive, at $400+

Maybe I'll get the 1600s then? since I can save about $100 and maybe get myself another Hard Drive? 😛
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom