System keeps failing under Prime 95 in Blend test at stock settings

[RussianSensation]

Update (Sept 16, 2009): Other people are having same BSOD issues as me with their MSI boards:
BSOD msi p5-gd65 core i5 team extreme ddr3 1333mhz ram
BSOD msi p55 gd53 + corei5 750 team elite ddr3 1333mhz 2gb x 2
123Bob on MSI boards claims to have Memory Issues with a new p55-cd53 board with an i7 860 CPU and Corsair RAM.

I am confident the MSI team is working towards addressing these problems as more users report their experiences. Still thought I would note it.

As a side note, there also appear to be issues with Gigabyte Boards (UD2 and UD4P) in showing only 4GBs of ram vs. 8GBs of ram when all 4 DIMMs are populated. It may be that due to the immaturity of BIOSes and the P55 chipset, it will take time for these problems to be sorted out.

It would be interesting to add more data points for Asus P7P55D.





-------------------------

Board: MSI P55-CD53 Bios 1.1 Version
Memory: OCZ DDR3 PC3-12800 Gold Low Voltage Dual Channel CL8-8-8-24 1.65V #OCZ3G1600LV4GK
CPU: Core i7 860 SLBJJ
Cooler: Stock Intel
PSU: Corsair 520 Modular
Videocard: MSI 4890 1GB factory OC

I have set everything to Optimized defaults in the latest bios.

- MemTest86+ v. 2.11 passes with 1 memory stick at DDR3 1600 8-8-8-24 1.651V (edited: Does not pass over 12 hour period with 2 sticks - fails Test #5 and #8)
- MemTest86+ v. 2.11 passes with 1 memory at DDR3 1066 7-7-7-18 1.651V (edited: Does not pass over 12 hour period with 2 sticks - fails Test #5 and #8)
- Tried setting CPU voltage to +0.200 from stock of 1.128V for a total of 1.328V (1.27V in CPU-Z 1.52)

Tried running the memory at 1066 and 1600. The system fails Prime95 Blended test within 1-2 minutes. Sometimes the mouse completely freezes on the screen when running Prime 95 Blend within seconds of starting the test requiring reboot. Otherwise the system BSOD at random even when I am browsing the Internet.

- CPU temperatures are 40*C idle and 74*C load, which is still under the threshold.
- 3.3V rail is 3.30V, 12V rail is 12.11V

The system also fails after enough passes of Resident Evil 5 Benchmark (fixed).
Same PSU and Videocard ran in the Q6600 @ 3.4ghz system (sig below) without any problems.


I am running out of ideas....mobo?

 

[Jumpem]

Try something other than Blend. Aren't there RAM only and CPU only tests? Try those and see if you can narrow it towards one or the other.

 

[BTRY B 529th FA BN]

What ratio are you running the ram at?

 

[SSWilson]

Try setting your command rate to 2N or comparable. Do these have an uncore voltage setting? Maybe look at that if equipped.

 

[RussianSensation]

Originally posted by: BTRY B 529th FA BN
What ratio are you running the ram at?

Ok after running the ram for hours, it keeps giving me errors at both 1066 and 1600 speeds. But BSOD are more frequent at 1600. However, both sticks pass in DIMM 1 when I tested them individually.

I have the rams in DIMMs 1 and 3 as per the motherboard manual. The problem is the mobo won't boot with 1 ram stick in DIMM 3 only since it asks to insert into DIMM 1 first if you are only using 1 stick. So I can't isolate if this is a problem with DIMM 3 in the slot, or if the board gets overstressed running 2 Rams at 1600...

Also there is pretty major undervolting on the CPU at load. I set 1.325 in BIOS and at load CPU-Z reports 1.27. There is not even an option on this board for Uncore vs. Core CPU voltage and no PLL adjustments... :brokenheart:

I am very unhappy with this setup. Thinking of just exchanging it for GSkill Ram and a Gigabyte or an Asus board. And I wanted to give MSI a try this build.

 

[terentenet]

Up the QPI voltage a notch, see if your ram will run (more) stable then. Maybe those Optimized defaults set some very low voltages.

 

[RussianSensation]

Update: Tested each ram stick in DIMM #1 (which is 2nd slot from the cpu cooler). Both pass. Left 2 sticks in DIMMs #1 and #3 overnight. Now it fails under prolonged use.

Ok guys I went for all the marbles here:

- Ram set at rated OCZ Gold low voltage 1600 speed 8-8-8-24 @ 1.651V 2T (from 1T)
- CPU voltage from 1.124V to 1.322 (+0.198V)
- PCH 1.8V --> increased to 2.0V
- PCH 1.05V --> Increased to 1.20V
- CPU spread spectrum Disabled
- there is no option at all for QPI or Uncore Voltage on this board (maybe CD65 or CD80 have it). There is also no option for Load Line Calibration.

All PSU voltages are steady but Prime95 fails core by core within minutes until the whole system just completely crashes. CPU temperatures reach 86*C on the dinky heatsink from idle of 37-39*C. I tried resetting the heatsink fan. No changes whatsoever.

At this point I am going to return the board and the ram and restart my build. I am not going to fault MSI's new boards across the board here because I might have just ended up with a dud and this is bottom of the barrel CD53 model.

However, it is confusing that they labeled PCH 1.8V instead of CPU PLL voltage and No Uncore voltage option seems to be available.




-------------------------------------------------------
Edit:

- I couldn't get this board and this particular ram to work with the newest BIOS and voltages way up. Since Ram passed when I tested each memory stick individually in DIMM#1, I am inclined to think this is a MOBO issue as it fails when 2 banks are populated in Memtest86+ over a 12 hour period test without any CPU load.
- Going to return them both to the store.
- Ordered Gigabyte GA-P55-UD3R and G.Skill Ripjaws 4GB DDR3 1600 CL7.
- Will report back on this new setup.

Thanks for all the help.

 

[RussianSensation]

bump

 

[Mr Fox]

Originally posted by: RussianSensation
Update: Tested each ram stick in DIMM #1 (which is 2nd slot from the cpu cooler). Both pass. Left 2 sticks in DIMMs #1 and #3 overnight. Now it fails under prolonged use.

Ok guys I went for all the marbles here:

- Ram set at rated OCZ Gold low voltage 1600 speed 8-8-8-24 @ 1.651V 2T (from 1T)
- CPU voltage from 1.124V to 1.322 (+0.198V)
- PCH 1.8V --> increased to 2.0V
- PCH 1.05V --> Increased to 1.20V
- CPU spread spectrum Disabled
- there is no option at all for QPI or Uncore Voltage on this board (maybe CD65 or CD80 have it). There is also no option for Load Line Calibration.

All PSU voltages are steady but Prime95 fails core by core within minutes until the whole system just completely crashes. CPU temperatures reach 86*C on the dinky heatsink from idle of 37-39*C. I tried resetting the heatsink fan. No changes whatsoever.

At this point I am going to return the board and the ram and restart my build. I am not going to fault MSI's new boards across the board here because I might have just ended up with a dud and this is bottom of the barrel CD53 model.

However, it is confusing that they labeled PCH 1.8V instead of CPU PLL voltage and No Uncore voltage option seems to be available.




-------------------------------------------------------
Edit:

- I couldn't get this board and this particular ram to work with the newest BIOS and voltages way up. Since Ram passed when I tested each memory stick individually in DIMM#1, I am inclined to think this is a MOBO issue as it fails when 2 banks are populated in Memtest86+ over a 12 hour period test without any CPU load.
- Going to return them both to the store.
- Ordered Gigabyte GA-P55-UD3R and G.Skill Ripjaws 4GB DDR3 1600 CL7.
- Will report back on this new setup.

Thanks for all the help.

The issues that you are having are being created in TWO key ways....

1,) - OCZ Memory ?????? WTF were you thinking.... JEDEC Spec is @DDR3 1333@1.5 Volts

Overclocking Memory has no real advantage with i7/i5

Scaling Memory is basically Jerkin yer gherkin...

http://www.tomshardware.com/re...y-scaling-i7,2325.html

You buy memory that runs @1.65 you take away any advantage as far as Tweaking voltages etc.

That OCZ is Overvolted CAS 9 bottom of the barrel Ram that they are overvolting the piss out of...

2,) 86 C ?? No damn wonder the UnCore is very sensitive to voltage, and that is why you are getting the temperatures

Sorry to be Blunt, but Ci5 is great with a stock set-up, but it is a major compromise in many ways.

People are finding out the hard way. The PCI-E Controller on Die is a very bad thing.

Originally posted by: Tom's Hardware


The memory industry, which typically focuses on offering faster (and more expensive) memory products, must find these results rather annoying?there is simply very little benefit to fast DDR3 memory on a fast processor such as the Intel Core i7-975. More than ever, premium RAM can be compared to a high-end sports car: it may provide better performance, but the benefit in everyday life is often very limited.

DDR3 memory kits capable of running beyond DDR3-2000 speeds at increased voltage are the best you can get. However, we limited the testing to DDR3-1600 speed, and stayed at a voltage level of 1.65 V for this article, which represents a reasonable maximum for most users (hitting 2133 in our i7-975 review required a screaming 1.75 V QPI). Our testing included all other selectable RAM speeds, all the way down to DDR3-800, trying both tight and relaxed latencies for each of the RAM clock speed settings.

High End Memory Is Not Worth The Money

The results are obvious: going from one memory speed to the next, e.g. from DDR3-1066 to 1333, does not provide major benefits. Even the replacement of slow DDR3-800 RAM by DDR3-1600 memory will mostly yield disappointing results. While the performance advantage is measurable, it is never noticeable.

Exceptions, however, do exist. Compressing files with WinRAR is significantly quicker on fast, low latency DDR3-1600 RAM. Some applications, such as games, can at least take minor advantage of the upgraded memory horsepower.

What If?

In this light, we decided to add a few more benchmark results at overclocked processor speeds. We decided to accelerate the CPU by one clock speed increment, which reflects exactly what happens if you decide to purchase a faster processor instead of high performance memory.

Click here for all faster CPU benchmarks results.

As you will see, a higher CPU clock speed typically provides better performance in most of the benchmarks?but not in all of them.

Memory Recommendation

Knowing all of these results, it is obvious that highest speed DDR3 memory only makes sense for serious enthusiasts, or for those with unlimited budgets. Everyone else should focus on mainstream clock speeds of DDR3-1066 or DDR3-1333, and go for a trustworthy brand and the quickest timings their budget allows.

 

[RussianSensation]

Hey Mr Fox, ya I read from Anandtech and at Tom's that faster memory does not impact performance for the majority of applications. However, the price difference between DDR3 1333 and DDR3 1600 was only $5.

On the MSI board, the lowest speed QPI:RAM ratio I found is 2:8 or effectively 133 QPI : 533 (DDR3-1066) ram. Therefore, if I am aiming at QPI of 190 x 21 Multiplier, wouldn't I need DDR3-1520? That's why I wanted the extra headroom that DDR3-1600 offers, so I wouldn't have to overclock the RAM. I realize at stock speeds DDR3-1600 is a waste.

What's wrong with OCZ memory? I've used them in the past and found them to be excellent performers. Although I had the Platinum Series of DDR2-800. Curious as to how you found out the ram is only rated at 1333 and is a CAS9 ram? In any case, the ram fails at both 1066 and 1333 even at CL8 and 1.65V which is above the 1.5V spec you noted above.

The ram is rated at 1.65V CL8-8-8-24 from the factory:
http://www.ocztechnology.com/p...w_voltage_dual_channel

I didn't run this ram overclocked beyond DDR3-1600 at any point in time.

In any case, I ended up getting G.SKILL Ripjaws F3-12800CL7D-4GBRH, which should run DDR3-1333 CL7 at 1.5V or DDR3-1600 CL7 at 1.65V.

In regard to the temperatures, the stock Intel heatsink is a joke. I'll be getting an aftermarket one at some point.

 

[Mr Fox]

JEDEC standard for DDR3 is 1066, and 1333, and the reference Test Specification Voltage is 1.5v....

Anything else is Non-standard..

If you were to test that DRAM @ 1.5v the CAS would be at least 9 possibly even 10...

Good DDR3 DRAM is CAS 7 @ 1.5v

OCZ was taking very crappy DRAM,, and overvolting it to get tighter CAS...(8)

1.65 volts is .15 volts overvolted from the JEDEC Standard.

that .15 volts does not seem like much, but it raises the UnCore Voltage, at the same time.

That is where the heat is being generated...

There have been a few threads here that have noticed the temp rise related to raising the VDimm voltage

Just food for thought !!!

 

[phil0sofa]

Excuse my possible ignorance - I'm learning 1156 overclocking too, but isn't some of the uncore voltage set by the core voltage? That aside, afaik the PCH 1.8v isn't the CPU PLL voltage, but rather the interconnect segment of the uncore, with the PCH 1.05v being the memory controller? At least that's the impression I got from this article. Don't quote me on that however.



For myself I'm quite happy with my G.Skill set, which runs happily at 1600 CL7 (down from CL8) @ 1.60v. The 860 is a damn fine chip, and with a bit of coaxing it can reach some rather impressive speeds, albeit with good cooling; I'm using a GD65 - even with a lot to learn about 1156 it hits 4.2 relatively easily. BOL with your overclock mate.

 

[RussianSensation]

Hey phil0sofa, right now I am trying to understand all the intricacies of the Core i7. I may have misunderstood that PCH 1.8V is a CPU PLL voltage. However, I found this from Bit-Tech.net:
http://www.bit-tech.net/hardwa.../msi-p55-gd65-review/5

"Pushing for maximum CPU frequency we hit a stability wall at 190MHz base clock, with a 22x multiplier and 1.45V core voltage, 1.42V VTT, 1.9V PLL and 1.12V PCH voltage."

Since my MSI board had a PCH 1.8 and PCH of 1.05 settings, I couldn't help but correlate their use of "PLL" towards 1.9V setting to the PCH 1.8V setting. Having said that, my board is CD53 so I am not sure if you have PCH 1.8, PCH 1.05 and PLL settings (this one I don't have). This would explain my flawed logic.

Glad to hear your chip overclocked very well. What are the temperatures you are experiencing in Prime95 or other program at load on that 9900? Right now I am still stuck with stock intel cooler until I decide on what to get. Likely leaning towards TRUE once 1156 support is added or the Noctua NH-U12P SE2.

 

[Jumpem]

Are you guys leaving turbo on so your 860 is ~4GHz on 1 core, and ~3.23GHz without turbo?

Or are you turning turbo off completely?

 

[RussianSensation]

It's going to be hard to justify leaving Turbo mode ON for the Core i7 860 because:

Intel Core i7 860 2.80GHz

3.23GHz (154 x 21.0)

3C/4C Active: 3.54GHz
2C Active: 3.85GHz
1C Active: 4.00GHz

vs.

3.99GHz (210 x 19.0) without turbo mode.

 

[Shmee]

wow, I am glad I got a UD5 and a D0 instead of this stuff :/

 

[betasub]

Originally posted by: RussianSensation
It's going to be hard to justify leaving Turbo mode ON for the Core i7 860 ... {snip}

For those that want maximum performance, regardless, yes - Turbo OFF.

A lot of users will do fine with the balanced power/performance with Turbo ON.

And then there may even be a few low-power under-volters keep everything cool with Turbo OFF.