BSOD f3-2400c11q-32gxm

[Sylvester]

Would be grateful for any pointers for a fix.

I have had a set of F3-2400C11Q-32GXM (this is the ID# on the packaging card) 4x8GB for a while (since June 2014 from GSkill on Amazon.co.uk ) in the following mobo (both QVLs match) and it has been throwing BSODs about twice a month since I began to use it full time in April this year. This is a problem because I like to help out by beta testing games and I cant make useful bug reports if my rig is misbehaving.

Mobo: Gigabyte GA-Z97MX-Gaming 5 (1.0) BIOS:F4
VGA: GTX 770 (GV-N770OC-2GD)
OS: Win 7 Ult x64 SP1
CPU I7 4790K #SR219
RAM GSkill 32GB F3-2400C11Q-32GXM
PSU: Seasonic 860W Platinum SS-860XP2
SSD: M550 1TB & BX100 1TB (latest firmware)

Previous testing found no RAM errors and I started a thread (with all the BSOD info) to try to find the problem as below.

http://forums.guru3d.com/showthread.php?t=403048

Recently more extensive retesting with Memtest86+ and Memtest86 v6.2.0 have shown rare errors from the RAM at XMP timings and clocks with mobo vRAM set 1.66v and BIOS monitor reading 1.69v. (When the BSODs started the volts were on auto / XMP and the mobo showed 1.65v set, I switched to manual to try to stabilise the system and the BIOS was +/- 0.02v so it was either 1.64v or 1.66v).

Re cooling and clocking etc: the CPU is not overclocked at all and is running at stock with turbo disabled in order to ensure there is no overheating even in stress testing with OCCT and AIDA64. Neither the CPU nor the RAM has ever been overclocked as the intention was to make a lightweight gaming box hence mATX form factor and air cooling. The CPU heatsink is a Noctua NH-L12 low profile and it has a big Noctua fan which makes a current of air over the RAM. Case is open sided for testing and it is a LianLi PC-V354 with 2x120mm stock fans in at the front and Akasa Viper S-Flow 120mm+140mm out, so well ventilated.

So the question is how do I fix the errors? Given that Memtest86 v6.2.0 tests CPU cache as well as RAM it isnt good logic to assume it is the RAM just yet. But it is reasonable to see if the errors can be stopped by remedies directed at the RAM.

Interestingly Memtest86 v6.2.0 was reading it as PC-18200 not PC-19200 as stated on the packaging. Does that mean it is an overclocked chip or is Memtest86 making an ID error? Is Memtest86 v6.2.0 reliable and/or is this a known thing about this model, if so is there a well trodden path to success?

Either way I just want it to work without errors and would appreciate advice. I am happy to tweak it if that works. So I am asking is it advisable to boost the volts some to try to stabilise it? If so how far can it safely go bearing in mind the mobo monitor is reporting higher volts than set. Or should I play it safe and try to get proof to invoke the lifetime warranty?

Report Date	2015-12-15 12:22:57
Generated by	MemTest86 V6.2.0 Free (64-bit)
Result	FAIL
System Information

EFI Specifications	2.31
CPU Type	Intel Core i7-4790K @ 4.00GHz
CPU Clock	4000 MHz
# Logical Processors	8
L1 Cache	4 x 64K (209669 MB/s)
L2 Cache	4 x 256K (60830 MB/s)
L3 Cache	8192K (45375 MB/s)
Memory	32815M (24718 MB/s)
DIMM Slot #0	8GB DDR3 XMP PC3-18200
/ F3-2400C11-8GXM
10-12-12-29 / 1142 MHz / 1.650V
DIMM Slot #1	8GB DDR3 XMP PC3-18200
/ F3-2400C11-8GXM
10-12-12-29 / 1142 MHz / 1.650V
DIMM Slot #2	8GB DDR3 XMP PC3-18200
/ F3-2400C11-8GXM
10-12-12-29 / 1142 MHz / 1.650V
DIMM Slot #3	8GB DDR3 XMP PC3-18200
/ F3-2400C11-8GXM
10-12-12-29 / 1142 MHz / 1.650V
Result summary

Test Start Time	2015-12-14 23:26:44
Elapsed Time	12:56:00
Memory Range Tested	0x0 - 82FE00000 (33534MB)
CPU Selection Mode	Single: CPU # 0
# Tests Passed	34/35 (97%)
Lowest Error Address	0x49B51AC4C (18869MB)
Highest Error Address	0x49B51AC4C (18869MB)
Bits in Error Mask	0000000000010000
Bits in Error	1
Max Contiguous Errors	1
Test	# Tests Passed	Errors
Test 0 [Address test, walking ones, 1 CPU]	3/3 (100%)	0
Test 1 [Address test, own address, 1 CPU]	3/3 (100%)	0
Test 2 [Address test, own address]	3/3 (100%)	0
Test 3 [Moving inversions, ones & zeroes]	3/3 (100%)	0
Test 4 [Moving inversions, 8-bit pattern]	3/3 (100%)	0
Test 5 [Moving inversions, random pattern]	2/3 (66%)	1
Test 6 [Block move, 64-byte blocks]	3/3 (100%)	0
Test 7 [Moving inversions, 32-bit pattern]	3/3 (100%)	0
Test 8 [Random number sequence]	3/3 (100%)	0
Test 9 [Modulo 20, ones & zeros]	3/3 (100%)	0
Test 10 [Bit fade test, 2 patterns, 1 CPU]	3/3 (100%)	0
Test 13 [Hammer test]	2/2 (100%)	0

Last 10 Errors
[Data Error] Test: 5, CPU: 0, Address: 49B51AC4C, Expected: 68608451, Actual: 68618451
Certification

[Sylvester]

Update, I have tested the RAM at XMP DDR2400 and also underclocked to DDR2200. I used test#5 exclusively as this was the one it failed.

@ 2400 it gives errors even at 1.704v reported and failed 10x Memtest86 v 6.0.2 test#5
@ 2200 it has no errors at 1.66v reported and has passed 4 nights of testing including >160x Memtest86 v6.0.2 test#5

So this indicates the RAM is the source of the fault and I am running it underclocked at DDR2200 to see if this stops the BSODs.

Would failing Memtest86 v6.2.0 at XMP be good enough justification for an RMA?

If I narrow it down to a single stick which is failing at XMP, is it possible to RMA a single stick, or does the entire set need to go?

[Sylvester]

Update, I have been in touch with the developers of Memtest86 v6.2.0 and they are taking a look at the log to figure out why the RAM is reporting PC-18200 etc. Will let you know what they find.

http://www.passmark.com/forum/showth...facturers-spec

[Sylvester]

Passmark have got back to me and said that the RAM numbers are in error because Memtest86 cannot decode XMP 1.3 which this RAM is using and is currently able to use up to XMP 1.1 but is being updated.

However they also say this does not effect the test which is still valid. So the errors at full speed are valid.

[GSKILL TECH]

Use memtest86+ or another memory diagnostics software to see what results you get.

[Sylvester]

After testing and a couple of months of real world use without BSODs this issue has been resolved satisfactorily by boosting I7 4790K CPU volts as well as RAM volts, plus a little extra voltage with CPUring and SysAgent as below.

So the RAM is now working at 2400 XMP timings.

vCore @1.050v
DRAM @1.68v set
CPU ring @1.075v
SysAgent @+0.025v

A unique factor in this set up may be that CPU turbo is disabled so that at full load @4GHz there is no step up of frequency or voltage and it may have been this which meant that the CPU proved to be underpowered at maximum load, its not possible to say for sure that maximum memory was a factor but anecdotally I have heard that a little volts tweak sometimes is necessary to get a full set of 32Gb working with Haswell type set ups, bearing in mind the memory controller is on the chip it makes sense that the CPU might need some more volts too.