I was thinking about that. Amount of RAM, RAM latencies, RAM speed, FSB and clock speeds.Quote:
Originally Posted by AgentGOD
Printable View
I was thinking about that. Amount of RAM, RAM latencies, RAM speed, FSB and clock speeds.Quote:
Originally Posted by AgentGOD
Picture didn't show up for some reason:
http://i34.tinypic.com/2zgcl08.jpg
Hi AgentGOD,Quote:
Originally Posted by AgentGOD
thanks for replying. I am rather happy to see that there is something like Intelburntest. It dawned on me in the last 24h that the GFLOPs cannot only depend on the raw CPU power....
I am currently facing some issues here since I changed to an ASUS P5Q3 deluxe mainboard.
My current settings for CPU PLL Voltage, GPU Ref 0/2, 1/3, NB Voltage, and some others are so razor's edge that partially only one step up or down will make my system prime-unstable. Intelburntest however is a lot more tolerant and allows for a much wider range of settings without showing calculation errors.
What's the common opinion here? Can one dismiss Prime/Orthos/OCCT's errors? Priming has always been a neccessary evil to me and I would like to have it replaced by a better solution.
I saw Leghoof'd's opinion here, what is stable for you is stable enough. True. But I have seen effects at tiems on m operating systems that led me to believe that errors in the background that noone notices is slowly affecting th e reliability of one's OS overall, as in causing problems in the long shot that others would easily blame on Microsoft's incompetence. <g>
Would like an answer to this aswell. IBT passes me, but Prime95 randomly fails. I'm inclined to believe IBT because P95 doesn't stress the system as much as IBT (going by temps). But still P95 erroring out does hang in the back of your head.Quote:
Originally Posted by Amurtigress
There's also nothing saying that random Prime errors aren't just that...and not very dependable. I'm still inclined to stick with IBT for a good while, but i will still be running other stresstesting apps alongside IBT for stability verification for all of those nay-sayers out there. :)Quote:
Originally Posted by Singh400
Could be...However, when I increase FSB from 400 to 425 MHz and remain IBT stable but Prime95 goes from completely stable to errors inside of a minute, this does make me quite apprehensive.Quote:
Originally Posted by Cornelious0_0
Cornelious, Singh:Quote:
Originally Posted by Cornelious0_0
It would be a bit too simple to say one or the other application causes more or less stress when it comes to P95 and IBT.
IBT, for once, uses the whole memory and constantly shoves data through the northbridge. The NB is certainly the main point in Intel systems, and is being involved in everything crucial. Let's take for granted that IBT is putting extremely high stress on the system, and let's also take for granted that P95 is making one's NB hotter. Sure, it should still pass, but is P95 still appropriate for modern chipsets? Does prime stability mean adjusting to the needs of an outdated piece of software? How well does Prime 95 actually work when dealing with completely different ways of taskmanagement in Vista and Windows XP, which came many years after Win95/Win98? And what about 64 bit systems?
Frankly, I have no clue. I just mean to sum up points to be considered. IBT uses original Intel code, and Intel knows what they are doing about their own CPUs and chipsets.
Then again, just as Singh said, there is that P95 instability in the back of my head. I've been building systems for 17 years...mostly for a hobby, but still. It's such an old habit!
Raz,Quote:
Originally Posted by Razmatazz
such a harsh effect looks extremely unusual. 25 MHz are nothing for a 45 nm quadcore. I've had a Maximus Formula board and now the P5Q3, so I can roughly say about your P5E3 that this should not happen. And by now it looks to me that this is a general reaction of intel systems that run DDR3 RAM. My P45 is a dieshrink compared to your X38, yet, it seems so familar.
This really looks more like an issue of Prime 95, frankly.
Quote:
Originally Posted by Cornelious0_0
Thanks for the replies lads :up: :up: :up:Quote:
Originally Posted by Amurtigress
Guy's I do this everyday for a living,but this has me Stumped! I am running Supertalent T800UX2GC4 800MHz overlockers ram in my unit and even at stock speeds IBT passes, then fails! If i run it in Single Channel Mode it's fine,(Oced or Stock),but yet i put in 2 GB's of Transcend Jet ram in dual Channel mode and it passes everytime ,stock or Oced at 3.8 GHz! The ST Ram I've checked with memtest and every other test I have and it's fine! I ran Ortho's with 3D mark 06 looping 75 times on each benchmark continuosly and after 10 hrs of ortho's the supertalent is still running!! I don't get it,,why is such Good ram failing in Dual channel mode,Passing in single channel mode and any other Ram including my OCZ DDR2 800 Mhz passing? I tried what Stealth suggested and upped my Chipset Voltage No Diff,but this is only happening to the ST Ram! Don't get me wrong,,No Crashes,No BSOD's or any problems with the Supertalent Ram,,Just in IBT!! I thought I'd give you guy's a Smack at it as my Heads tired,lol!!
BTW,If I set my Ram usage from option 1 to just 20 MB below what IBT say's is max ram usuage,it Passes everytime! And also all Temps are very low as well,except CPU when in IBT go's to 58 degree's! Thanks to All in Advance for any Suggestions!
My Specs:
E8500@ 3.82 Ghz,Asus Striker ll Formula 0902,2 x WD 320 GB,EVGA 9800 GTX,2 GB ST RAM,SB X-FI EXTREME Music SC,PC P&C 750 Quad P/S,2 LG DVDRW’s,Zalman9700 CNPS Cooler,Antec 1200 Gaming Case
Prime95 is used in years to test stability. It's widely trusted. A stable pc run prime95 for any length of time you want without reporting error, If it failed your pc isn't stable period.
I agree that if you use your pc lightly you can't even see any BSOD or crashed but still your ALU's/FPU can't be trusted.
Can you ignore any prime error if IBT or whatever program you use runs just fine instead ? well yes, it's your pc, but if you want a trusted pc you can't.
Prime95 stress the pc in differente way to IBT, so it's reasonable that it can or can't show errors.
There's no One test to rule them all... the more different test you throw at your pc the better chances to uncover any instanbility.
There's no way to proof that a PC is stable, but more tests it passed higher is the probability that you pc is really stable.
For me make no sense ignore prime95 if it report error and trust only at IBT if the run ok, an Error is a error, if lowering the clock the test is passed that is a proof that the program is OK but your pc isn't.
I have 4gb (2x2gb) of ddr3 ram and when run under 32bit vista it crashes out soon as you start the test saying "This application has to close" but under x64 it works flawless. The system is stable (40+ IBT passes under x64 vista, 10+ hours of various prime95 tests and 20+ hours of memtest86)
any idea why its crashing in 32bit? is it cause it has more memory then the OS can see?
[edit] nvm, found the answer a few pages in, cant do max stress test under 32 bit, uses more memory then it has available.
Holy $#%#$
That was something to see my physical memory sky rocket like that in the task manager then see all 4 cores cap out at 100%.
Previously I ran Seti@home on 4 cores and I would get around 65*C. When I ran this, within seconds I was over 70* and stopped it so it didnt trigger my auto shutoff (Yes, I know. My new heatsink should be here tomorrow!)
Thanks Guys,,I found my Problem!I was suspicious that 1 of my 2 sticks may have a Bad sector or 2! Nothing that 2 New Sticks didn't fix! IBT ran 20-then 30 -then I ran 50,,Rock Solid now! Funny though Prime or Orthos didn't pck it up! Oh well perhaps not enough to show up in 9.5 hours! Thanks Again AgentGOD for the Program,,Very useful Tool in my books!!:up::D
Let me jut say one thing: The very nature of this thread does not allow me to explain all the circumstances. Just my 17 years of experience tell me that something has to be figured out with the P45/X48/X38 chipset and DDR3 memory, and it has nothing to do with my CPU clock. Been using the CPU on another board before I bought this one.Quote:
Originally Posted by mazzy
There's no need to state the seemingly obvious, but it doesn't simply WORK for the issues some are having here.
One little part of it is that the P5Q3 is plainly defaulting to DDR2 timings while using DDR3 RAM. Or how does one explain 5-5-5-15-5 appearing in the BIOS when turning the memory timings from AUTO to MANUAL.
Considering all this, your post has been pretty useless....:shakes:
Hello,
I seem to have solved the basic stability issues for my system now.
So far, any moving away from a NB voltage of 1.32V meant an increasing frequency of prime errors.
Last night my PC was priming for 7 h 20 min. stable with no errors at a NB volt. of 1.28V, using the 24.14 Prime for 64 bit Windows. the only thing I changed was....
PCI-E Frequency from 100 Mhz to 101.
I ahve heard before that this shall have solved stability issues since the ASUS P5B series, but it never did for me. Anyway, for those who are suffering prime issues, enable this for a test and let us know what it does for you. I know it doesn't sound very plausable, but it is entirely possible that this is changing a strap within your intel chipset that makes it all stable.
This applies most likely to all Intel chipsets and ASUS boards since the P965 chipsets. For example, P965, X975, P31/33/35, X38, X48, P45, etc.
System:
Intel Q9550 [3.4ghz (400x8.5) @ 1.264v]
DFI X48-T2R
8GB [4x2GB] G.Skill DDR2 [960 @ 2.1v]
Thermaltake Toughpower 850W
EVGA GTX 280 SSC
I was able to prime95 for 8hrs. Using IntelBurnTest, if I set the stress level to low (1/4 of memory) - it passed (50 passed). If I set the stress level to anything higher than 1/2 of memory, it failed (1 passed/4 failed). I set everything back to defaults. If I set the stress level to 1/2 of memory, it passed (5 passed). But if I set the stress level to Maximum stress level, it failed (1 passed/4 failed).
What does this mean?
That you aren't stable. ;)Quote:
Originally Posted by vannguyen0
Well... if I'm not passing the test at stock - does that mean I have bad memory, a dud for a cpu, etc?Quote:
Originally Posted by Brama
Or you can decide 8h of prime95 / Intellburner's 1/4 result (atleast it didnt end on BSOD!!)
No, there are other reasons too.Quote:
Originally Posted by vannguyen0
Possible memory incompatibility with your mainboard.
Memory not qualified to run with four sticks on your board.
Insufficient Northbridge voltage.
Mainboard using stupid defaults.
A defective peripheral device on the board, or card in the system, being broken and throwing your northbridge bus system off.
Your power supply being insufficient or dying (Dead elcaps!)
And possibly more issues.
Nope, i bet your system is overheating. For example, my q6600 @ 3.85ghz, when i use the IntelBurnTest. temperature were up to the 70s then system crash but when i ran OCCT for hours it never crashed.Quote:
Originally Posted by vannguyen0
yes... at least I've got that working for me :)Quote:
Originally Posted by carepolice
The BIOS release I'm on adds support for my memory (g.skill ddr2-1000). As for running it with four sticks -> :confused:Quote:
Originally Posted by Amurtigress
I'll up the northbridge voltage and see what happens. PSU should be more than sufficient - I have a Thermaltake ToughPower 850W.Quote:
Originally Posted by Amurtigress
Hmmm... the temps on my system does raise to about the 70s - but I do not experience any blue screens, random reboots, or system hangs. It finishes what it does and tells me that it failed 80% of the passes.Quote:
Originally Posted by Bail_w
Some testing here with Allendale C2D @3GHz using option 1 max stress and 2000x iterations;
375x8, MEM Linked mode 3:2 divider @1000MHz DDR, NB voltage @1.56v
@1.337v passes error free Prime95 24/7 non stop all tests (Small/Large FFT/Blend mode)
@1.337v error in IBT within 3 hours
@1.343v error in IBT within 4 hours
@1.350v error in IBT within 7 hours
MEM changed to Unlinked mode 1000MHz DDR
@1.350v error in IBT within 11 hours
@1.356v error in IBT within 16 hours
@1.362v FUBAR
@1.368v FUBAR
As you can see above, IBT stresses the system so much I had to keep increasing the vcore voltage.
Note that is also highlighted the boards NB weakness when using Linked Mode with divider @3:2, changing to Unlinked mode 1000MHz improved things although the vcore still need a little bump.
:up:
Edit:
Darnit something isn't holding up, I think it must be either the CPU's hard FSB wall of 375MHz failing, or the NB failing with that FSB/MEM divider, gonna try using a higher multiplier and reducing FSB.
333x9, MEM unlinked mode 1000MHz, NB voltage 1.56v
@1.356v Test in progress
update if anyone cares:
So I was able to pass 5 tests - the only change I had to make was for the memory to use a 1:1 divider. I was using a 5:6 divider (400fsb - 480 memory). My memory is rated for DDR2-1000, so I don't know what's going on there. But if I set my memory to DDR2-800, I can claim stability from the IntelBurnTest. I had the memory voltage at 2.1v, but reading up about the motherboard, it said to put the voltage at 2.15v to get 2.1v (real). So I'll play around with the memory voltage when I get home.
Question about memory dividers on DFI board for those who know: What's the difference between setting the memory divider to 333:666 vs 400:800?
Thanks!