GA-890GPA-UD3H,1090T & Ballistix Sport

[Tasa]

Hi to all of you Gigabyteians 

Let me first start with configuration that I have:

MOBO > GA-890GPA-UD3H (rev. 2.1) Bios Fgf
CPU & Cooler > AMD 1090T, CM 212+ push/pull (PWM)
MEMORY > Crucial Ballistix Sport 8GB DDR3-1600 CL9 (2X4) X2 (2 separate kits) 16Gb TOTAL
(9,9,9,24@1.5v)
POWER SUPPLY > OCZ Fatal1ty 550W
VIDEO CARD > EVGA GTX 460 FPB
HDDs/Optical drives > WDC WD5001AALS-00E3A0 500Gb, OCZ-AGILITY3 60Gb, LG BD Burner
O/S > Win7 Ultimate SP1
Case > CM 690
Fans > 2x CM 120mm top exhaust, 1x CM 120mm front intake, 1x AC(Arctic Cooling)92mm 9PWM side intake, 1x AC92mm 9PWM back exhaust

My computer needs are not big....I do sound recording and editing, video editing and games....and up to now...all was good with no glitches.
With 8Gb OCZ I had it stable and cool up to 3.8G

My nightmare started about week ago when I decided to upgrade RAM from 8Gb OCZ Gold 1333Mhz to 16Gb Sport 1600Mhz (2 kits 2x4Gb).
As I would normally do....Installed new RAM...loaded default settings. Few things I changed were: manual to 1600Mhz, Minimum Ras Time to 24 since SPD was showing 30, all regular things disabled (C1, Virt., C&Q,). Everything else was auto. CPU stock 3.2.
So I got USB flash drive with MEMTEST86+ , rebooted and started testing each of those sticks in first RAM slot. Did only 1 pass on each of those, just to make sure that at least initially ram was ok. I was going to do same more serious testing later on. After about 2 hours, and all of them passed first pass, I decided to load them all and do some PRIME95. Started Blend test, temps were all ok (44C max on CPU)....after 31 mins, worker 4 failed me.
I really didn't wanna see this....I hoped that I would have luck like I did for years now. Never had to RMA or return anything that I ever bought for computers that I was doing for myself.
Back to MEMTEST....started test on 2 sticks and after about 8mins (test 5) it started showing errors...I stopped the test.
At this point I was so pissed off that I decided to stop all testing and do some shooting (BF3) to chill out....about 15mins into nice action...BF3 crashes. It happened last time long time ago....I knew it was ram...so I restart computer and now BLACK screen welcomes me. No Post!
Pulled all RAM out and started getting one by one back in.... 2 sticks died....Since I bought 2 kits - first was in 1&3 slots, second was 2&4 slots...second kit was dead. I switched first kit from slots 1&3 with 2&4 just to check if slots were ok and it was ok. BIOS shows 2 sticks.
Went back to Microcenter and replaced kit with new one....Nightmare goes on!
I was never into OCing...never had a need to do so. So far with OCZ RAM that I had, I managed to do 3.6 just by getting multiplier to 18 and all was good....But now I started reading about it in more details cause I wanted to see how much I can push it. At this point I started looking into some more detailed settings in BIOS....My Normal CPU Vcore is at 1.425v - CPUID monitor reads different voltages from 1.22 to 1.42v depends on load, RAM is set in BIOS for 1.50v but CPUID monitor reads 1.49v....I know that all those monitoring/ sensor softwares lie a bit. So my next step was to see if boosting RAM voltage in bios would make change. After few trials, I managed to to do 3.8Ghz , DDR V to 1.53v (CPUID reads 1.5v), CPU NB VID to 1.225 and no problems with BF3 crashing on me. MEMTEST fails still (had it over night and it failed on pass 2 test 5 -  this is longer than b4), Intel Burn passes on both standard and high (5 and 10 passes). One thing to add....both BIOS Fgf and FF have same issues...just temps are bit different.
I don't think that MOBO is problem, i don't think CPU is problem....
I hope that some of you will be able to guide me through process of setting this up.
Good thing is that I took photos of every thing I ever changed throughout all stages and all tests.
I have tried from stock 3.2 to 3.9 and it was more or less stable....4.0 tried and failed.
Before I go to return this RAM, I would like to give it a last few shots....
One thing that bothers me is DDR settings in BIOS....if those are not set the way they should be...any test will fail.
My timing is set the way it should be...9 9 9 24 @1.5v and that is not working....so it might be that sub-timings are not set.
Or Crucial sucks badly! 
Crucial forums are so ridiculous....people ask questions and they don't answer them even after few days.
One of the guys had trouble with setting timings...answer was "just set it as it is stated on your stick".



Thanx,

Tasa









 

[absic]

Hi there,

as you are running a 1090T processor you might like to read this thread: http://forum.giga-byte.co.uk/index.php/topic,2515.0.html if you haven't already done so. Running RAM above 1333MHz with these CPU's is problematic to say the least and really not advised by AMD unless you are using just two sticks of RAM, one stick per channel. Even then, there is a chance that the memory controller on the CPU will fall over.

Do you have the same problem if you run the new RAM at 1333MHz?

[Tasa]

Hi,

Thanx for the reply....i will try that...will let you know.
That is probably the reason why my OCZ 8gb (2x4) 1333 was running so good!

[Tasa]

Results:

Last night, after your suggestion to switch to 1333, I did Prime and it failed after about 35mins.

So I've decided to try Memtest and let it run overnight.

This test was done on 3.8G, ddr 1333 with 8 8 8 20 @ 1.53v.
Health screen in BIOS shows 1.504v for RAM.
Memtest passed overnight after 7h 35min with no errors.

Few things I noticed - Temps were down.
Can you tell me what is TMPIN2? This sensor shows lower temp also (max with this setup was 42 - total max was 57 with some different settings )
Max on CPU was 42....max i ever reached on it was 51 (MOBO Sensors).
CPU sensor max was about 45c.

I will post the update soon.

Thanx,
T

[Tasa]

Update:


Prime after 1h 6mins no errors!
This is the longest I ever got so far. Usually fails after about 25-35mins.
This might be the sweet spot.
I will run another test over night.

[absic]

Pleased you seem to be getting somewhere and I know how annoying it can be. I know that I felt really angry with my 1090T processor when I discovered its limitations with RAM.

TMPIN2 might be the Northbridge temp. Sorry, but I can't say this with any certainty as TMPIN2 is one of those readings that is elusive to get full details for. The best way to check is to watch the reading when you are doing some heavy graphical work, if it is the Northbridge the temperature will increase.

[Ripshod]

TMPIN0 = system temp; TMPIN1 = CPU temp; TMPIN2 = NB temp - From the Gigabyte GA-890FXA-UD5/UD7 Owners Club
You can be certain now.

[Tasa]

Hi,

I was not in town for few days.

Thanx for the info. What is the NB max temp?
NB temp was the only close to 60 (57) when i was testing.
I didn't have time to do more testing but it is stable.
It is not crashing while playing BF3 and that is good sign.
I managed to get to 4Ghz. I was not able to do it on 1600.
I will do more testing and let you know.

Thanx

[Ripshod]

I've seen reports of NB temps hitting 70C with no ill effects. If you was to touch it's heatsink I'm sure it'd feel warm, but not hot.

additional:

Both the Northbridge and VRM heatsinks run fairly hot on the 890GPA which is understandable considering it uses the latest and greatest AMD chipsets. It may be in Gigabyte's best interest to use taller, larger heatsinks in the future.

  from an article on silentpcreview. The article also clearly states the max temp for the NB is 86C.

If you are concerned about that heat you could maybe add a spot fan to blow on the heatsink