On the Heels of the G.SKILL "Trouble" -- A new question

[BonzaiDuck]

Is anyone aware of some limitation or bug in PRIME95 -- particularly the BLend-Test -- under this scenario: VISTA-64, 2x2GB RAM?

Here's the current configuration:

680i chipset, August 08 BIOS revision
2x2GB RAM (Any: I've tried the G.SKILL and Corsair kits at completely stock settings)
VISTA-64 Service Pack 1

PRIME95 "small-FFTs" runs rock-solid at least 10 hours error-free.

replacing a 2x2GB kit with a 1x2GB kit allows PRIME95 Blend-Test to run flawlessly.

So I cannot see how there would be "other hardware" issues if the last statement is "true." with a 2x1GB kit, the configuration tests out perfectly.

I will be SOOOO greatful for any insight and assistance here.

Incidentally, I have two BFG 9600 GT graphics cards in SLI, with the latest drivers installed and SLI configured. Push comes to shove, I can probably removed one card at a time and re-test, but I cannot see how this would matter, given the different result with the 2GB versus 4GB kit. Perhaps someone has some useful info about this..


Meanwhile, I will try other memory stress-test options and see how that goes . . . .

 

[BonzaiDuck]

OH, SHEEE . . ****TTTT!!

I THINK I'm using PRIME95 32-bit version on a 64-bit OS, and there's a 64-bit version I should've used!!

GEEZ!! the trail of tears, the sweat, worry and trouble, the RMA to G.SKILL of a PERFECTLY GOOD SET of stix . . . . .

I think I better double check this -- because I don't remember following the 32-bit v 64-bit distinction when I downloaded the new PRIME95 version . . . .

 

[BonzaiDuck]

Nope -- that's not it -- the "About" screen in PRIME95 shows "Windows-64."

Does ANYONE have insight into my little crisis here?

Running Everest Ultimate 4.50's memory-stress-test, I discover something that vaguely troubled me before the BSODs under PRIME95. I couldn't shut off the screen-saver under VISTA 64. Now, Windows reports a problem that the "Logon Screensaver is unable to respond" while Everest stressing continues in the background.

It seems that there are more subtle difficulties that are causing my PRIME95 sessions to fail, and difficulties getting my desktop back under "more stable" stress-testing.

These are posts for "CPUs and Over-Clocking," or "Operating Systems." But they involve the stress-testing of RAM. I'll put up threads under those forums.

If anyone has any insights here, though, feel free to throw in your two-cents worth.

Meanwhile, I've got to send e-mail to G.SKILL and suggest some options for rectifying the mistaken RMA.

 

[Viper GTS]

What version of Prime95 are you using? There is a newer version than what is listed on their web site, you should be using 25.6 64-bit.

http://files.extremeoverclocking.com/file.php?f=205

Some of the earlier versions had issues on 64-bit systems, that may be what you are running into.

Also, if you run MemSet what TRFC are your DIMMS running? I had some older OCZ memory that benefitted greatly from a SPD flash to TRFC of 54. This is more of an issue if you're running 4x2GB like I am, but still could be an issue with 2x2GB.

Gregg

 

[nevbie]

You spawn new threads like a spider.

 

[BonzaiDuck]

Originally posted by: nevbie
You spawn new threads like a spider.

Nevbie -- you're right -- it displays my panic and frustration over something I thought I'd always had an intuitive knack for -- troubleshooting new OS installations.

I've discovered the source of the trouble. It is a bug deriving from interaction between the VISTA-64 screensaver -- the PRIME95 Blend-Test component -- and possibly the graphics driver and SLI-configuration.

Fact is, I had always turned off the screensaver when running stress tests, as well as the AV program and even the software firewall.

Viper -- I'm using the latest 64-bit update to PRIME95 -- I forget the version number, but it's the latest.

NOWWWWW!! Everything is A-OK, copacetic, hunky-dory, fine, marvelous and WUN-der-fulll.

Sorry about the thread-spawning, folks.

 

[LOUISSSSS]

bonzai, what did u do to solve your problems? was it a combination of new ram and OS installation?

glad ur good to go. now do some overclocking and gaming

 

[BonzaiDuck]

Again -- I mis-diagnosed it, thinking it was a misbehaving VISTA64 screensaver and PRIME95 with the 4GB RAM kit. I am humbled. NEVER . . . NEVER . . . . take short-cuts over-clocking.

This 680i board was certified for processors stock at 1333 FSB. But when it hit the street, most of what was available were 1066-FSB Conroes -- duals and quads. It's a testament to the Striker that it was released with an objective of being Penryn-ready, a year before Penryn was released. As it stands, with 2008 BIOS upgrades, it is Wolfdale, but not Yorkfield compatible. But that shortcoming was common to all 680i boards

Anyway, I was over-confident and impatient, taking short-cuts with the over-clocking, and making an assumption that had -- three years ago -- been proven false to me between 1GB and 2GB DDR kits. I assumed that a 4GB DDR2 kit would mostly require the OC settings proven with a 2GB kit.

And I've seldom had BSODs while over-clocking, and even with these Striker boards, never had more than a stopped worker-thread in PRIME95.

So . . . . here I am . . . . "Whoops! it's bad memory!" And I RMA the G.SKILLs showing lack of patience, intuition or insight. Then, after they're in the mail, with a Corsair 4GB kit, I discover -- Ohhhh!! It's NOT bad memory!!"

Then, I think it's "interaction between screensaver and PRIME95," or the OS, or some other nonsense. But I only tested the RAMs and everything this morning at the E8400 stock setting, the RAM settings for the 4GB kit being essentially the same as the over-clock regimen for the E8400. When I finally get it back up to 3.6 Ghz, screensaver turned off, lo and behold -- BSOD in about 30 minutes.

Then it dawned on me. I'd never really had experience running this mobo up to 1600 FSB, because the old Conroe processors and this motherboard seemed to top out around 375 to 380. People had complained about "FSB holes" -- all sorts of things -- and if they ever got beyond 400/1600 -- they'd dropped the multiplier -- yes, it would run fine with a lower multiplier at the higher FSB. I bought the dAMN thing because solid benchmark reviews showed E6600 with multiplier 7 and FSB moving between 400 Mhz and 500 Mhz -- but-- BUT -- NOT with Multiplier = 9!!!

So far, I've got the OC back up to 12.5 Mhz short of the 400 CPU-FSB where the error occurred, or 387.5 Mhz. But I thought about it, and tweaked the CPU_VTT and 1.2V_HT voltages so that the monitored values in BIOS monitor show 1.35V instead of 1.24 or whatever.

If I haven't shown a BSOD running at this speed for going on an hour-and-a-half, FSB = 1550 -- what do you think are the chances I'll get things back where I wanted them in the first place -- rock stable? The termination voltage tweaks aren't excessive -- I know that -- at LEAST -- from my year-long experience with this board. More extreme settings are between 1.45 and 1.55V for those items, and 1.35V is ... pretty limp for what the board can do.

LESSON LEARNED AND LESSON RE-LEARNED: NEVER, NEVER take short-cuts OC'ing a board you're familiar with when you have a new-generation processor, double the memory and a different OC objective; and don't assume that "rock-stable" with a 2GB RAM kit means that you can just stick in the 4GB without needing to tweak . . . . something . . . .

EDIT: And, like the CARTOON channel character in Dexter's Laboratory, " . . . . I fee-ul so STEWWWW-Pid!! Sooooo . . . STEWWWW-Pid!!" Ya -- I'm feelink that way . . . .