GA-EP45T-DS3R DDR3-1600 So far so good

[profJim]

I found this at the Corsair forum:
Recommended Corsair DDR3 memory for P45 chipset motherboards

We’ve received some feedback from users with motherboards based on the Intel P45 chipset who are unable to run 1600MHz stable, even with modules that are rated at that speed.

Testing done in our labs has concluded that users looking for the best overclocking results at 4GB on a P45-based DDR3 motherboard will have the best results using the following parts:

TW3X4G1800C8DF
TW3X4G2000C9DF
TW3X4096-1600C7DHX

The following part may top out at a speed short of 1600MHz on a P45-based motherboard:

TW3X4G1600C9DHX

The modules that we found unable to achieve 1600MHz on a P45-based motherboard were able to achieve 1800+ on an Asus P5E3 Premium which uses the X48 chipset.

This issue is isolated to only motherboards using the P45 chipset. Please note, this type of issue is something that’s often able to be remedied with a BIOS update from the motherboard manufacturer.

We’ll continue to update this thread as more information becomes available.

========
Based on this info, I'd suggest getting your system fully stable at lower memory speeds using conservative (i.e. SPD) memory settings. Once your system is stable, you can then try higher memory speeds or using tighter memory timings.

The link to the memory review was with a P5E3 (X48) mobo and their XMS memory had a slightly different model # and SPD.

You might want to try the following setup:
8x440MHz = 3.52GHz
3.20Bx440MHz = 1408MHz memory speed
Using the "B" multiplier, you might be able to use tRD=7 (Static tRead Value) after doing initial testing with your current tRD=8.

**edit** Monitoring software often has "issues" with the 12 volt reading. You can try HWMonitor (v114) or SpeedFan (v438) to see if they do better. I wouldn't be concerned with the different Vcore temperatures at idle, let us know what the differences are under full loads.

[ghost707]

ProfJim,
Thanks so much for taking the time to investigate the RAM for me.

Made your exact changes - and nothing else.

IBT runs 20 passes with no problem.

Running OCCT failed within the first minute and rebooted.

Changed only FSB down to 430 and nothing else.
OCCT errored after 13 minutes. (not sure by looking at the graphs what caused the error yet.

My feeling is that the RAM is in tolerance now but the CPU may be throwing a wrench into the works - my guess is VCore is maybe too low (at 1.25625).

OCCT reported the following under 100% CPU load:

CPU 1: 55 2: 55 3: 60 4: 65

Let me know what you think. Meanwhile, I am going to see what HWINFO says.

Again, thanks so much for your help. This is a great learning experience.

[ghost707]

Update:
Increased VCore to next increment (1.26250V) and OCCT failed after only two minutes.
Temperatures remained the same as 1.256250V).

HWINFO shows exact same data (Core Temps) as OCCT.

The plot thickens.

[gampamu]

Ghost707, sorry to hear you are experiencing issues. For troubleshooting purposes, you could take the CPU out of the picture by reducing the CPU multi to 7 and then run the OCCT. If the OCCT runs error free, the problem was CPU related: if the problem persists, you have either chipset or memory stability issues.

[ghost707]

Yeah, my thoughts exactly.
If the CPU passes some OCCT testing (@3.00GHZ), I am going to take the Ram down to 1066 (FSB 333 with same timings) and then up the CPU frequency - leaving the memory unchanged and see what happens.

Thanks gampamu.

[profJim]

For specific troubleshooting and testing with OCCT:

1) Select CPU:OCCT on the main screen.
2) Click on the "Test Type" field and select "Custom".
3) Click on the "Test Mode" field and select:
--> Small Data Set for "small FFT" cpu testing
--> Medium Data Set for "Blend" testing of cpu and ram
--> Large Data Set for "large FFT" ram testing
4) Set the Custom Test Duration time.

There are similar Custom settings using the CPU:Linpack test option where:
--> Low (25% free mem) uses the most memory
--> Medium (50% free mem) uses less memory
--> High (90% free mem) uses the least memory

Watch your cpu vcore temperatures when testing with the Linpack option.
20 - 30 minutes of Linpack testing should be sufficient for each Linpack option.

Post your bios settings with your test results, additional screen-shots help to eliminate possible errors and/or confusion.

[ghost707]

We had some storms come through Austin last night and this morning - so I have had everything unplugged from the wall - so I am back up and running now.

I did some baseline testing with OCCT using the LinpacK test with memory @1066 7 7 7 20 FSB @333 and CPU @3.33 GHZ - OCCT reports hottest core temp @61.
Tested Flight Simulator X for 2 hours (same configs) with affinity mask set to 14 (to use 3 cores) and HWINFO reported highest core temp @50.

The next step is to test the CPU @ 3.66GHZ (333x11) and test leaving RAM @1066. Reading through the info ProfJim found - it looks like testing the RAM past 1333 may be a waste of time. I know Corsair tries to make their chips compatible - so lower speeds/higher timings may be the compromise to achieve this.
It looks like DDR3 1600 (2GB sticks) may be a gray area with respect to P45 chips.
I noticed on Crucial's website that they don't even make matched pairs (2x2GB) of 2GB DDR3-1600 - only 1GB sticks. Interesting.

Will report findings later this evening.

Thanks again ProfJim for the primer on OCCT.

[ghost707]

OK - here is the latest,

I think maybe the problems I have encountered have been due to an undervoltage of VCore. Memory has been set to run at 1066 - to take it out of the equation.

I can run OCCT small data set @3.00GHZ (333X9) stock voltages with no problem, 3.33GHZ (333X10) with a small increase in Vcore to 1.2652V.

When I get the CPU up to 3.66GHZ (333X11) the system will reboot about 1 minute into the test at anything less than 1.3125V.
OCCT fails after about 5 minutes into the test - but does not reboot system (VCore@1.3125V).

However, there is discrepancy between the Vcore setting and actual voltage shown.
In Bios VCore is set to 1.3125 - in PC health screen Bios shows actual VCore @1.268.
In Windows: OCCT shows unloaded CPU @ 1.26, CPUZ and Easytune both show 1.264V.
Not sure if this is just normal variations or the engineers skewing the voltage a little so less people fry their CPU's.

So, my thinking is that VCore is still a little low in real voltage - which is causing the failure - I didn't want to take it any farther until I got ProfJim's input.

This is where I sit now:

MB Intelligent Tweaker(M.I.T.)
Robust Graphics Booster .............: [ Auto ]
CPU Clock Ratio .....................: 11 X  
Fine CPU Clock Ratio.................: 0  
CPU Frequency .......................: 3.66GHZ (333x11)

Clock Chip Control
Standard Clock Control
CPU Host Clock Control...............: [ Enabled ]
CPU Host Frequency (Mhz) ............: 333  
PCI Express Frequency (Mhz) .........: 100 
C.I.A.2 .............................: [ Disabled ]

Advanced Clock Control [Press Enter]
CPU Clock Drive......................: [ 800 mV ]
PCI Express Clock Drive..............: [ 900mV ]
CPU Clock Skew (ps)..................: [ 0ps ]
MCH Clock Skew (ps)..................: [ 0ps ]

DRAM Performance Control
Performance Enhance..................: [ STANDARD ]
(G)MCH Frequency Latch...............: [ AUTO ]
System Memory Multiplier ............: 3.20B  
Memory Frequency (Mhz) ..............: 1066
DRAM Timing Selectable ..............: [ Manual ]

Standard Timing Control
CAS Latency Time.....................: 7 
tRCD ................................: 7 
tRP'.................................: 7 
tRAS.................................: 20 

Advanced Timing Control
tRRD.................................: 5 
tWTR.................................: 5 
tWR..................................: 10 
tRFC.................................: 70 
tRTP.................................: 5 
Command Rate (CMD) ..................: 2


Channel A
Static tRead Value...................: 8
tRD Phase0 Adjustment................: 0 [ AUTO ]
tRD Phase1 Adjustment................: 0 [ AUTO ]
tRD Phase2 Adjustment ...............: 0 [ AUTO ]
tRD Phase3 Adjustment................: 0 [ AUTO ]
Trd2rd(Different Rank)...............: 9 [ AUTO ]
Twr2wr(Different Rank)...............: 7 [ AUTO ]
Twr2rd(Different Rank)...............: 8 [ AUTO ]
Trd2wr(Same/Diff Rank)...............: 8 [ AUTO ]

Channel B
Static tRead Value...................: 8
tRD Phase0 Adjustment................: 0 [ AUTO ]
tRD Phase1 Adjustment................: 0 [ AUTO ]
tRD Phase2 Adjustment ...............: 0 [ AUTO ]
tRD Phase3 Adjustment................: 0 [ AUTO ]
Trd2rd(Different Rank)...............: 9 [ AUTO ]
Twr2wr(Different Rank)...............: 7 [ AUTO ]
Twr2rd(Different Rank)...............: 8 [ AUTO ]
Trd2wr(Same/Diff Rank)...............: 8 [ AUTO ]


Motherboard Voltage Control
Voltage Type...............
CPU
CPU Vcore............................: 1.31250V
CPU Termination….....................: 1.240V
CPU PLL..............................: 1.570V
CPU Referen..........................: 0.760V [ AUTO ]


MCH/ICH
MCH Core.............................: 1.260V 
MCH Reference........................: 0.760V [ AUTO ]
MCH/DRAM Ref.........................: 0.750V [ AUTO ]
ICH I/O......................….......: 1.570V 


DRAM
DRAM Voltage ........................: 1.800V
DRAM Termination ....................: 0.750V [ AUTO ]
Ch-A DATA VRef.......................: 0.750V [ AUTO ]
Ch-B DATA VRef.......................: 0.750V [ AUTO ]
Ch-A Address VRef....................: 0.750V [ AUTO ]
Ch-B Address VRef....................: 0.750V [ AUTO ]

Advanced Settings
Limit CPUID Max. to 3...............: [ DISABLED ]
No-Execute Memory Protect...........: [ ENABLED ]
Load-Line Calibration...............: [ DISABLED ]
CPU Enhanced Halt (C1E).............: [ DISABLED ] 
CPU Thermal Monitor 2(TM2) .........: [ ENABLED ]
CPU EIST Function...................: [ DISABLED ] 
Virtualization Technology...........: [ ENABLED ] 

USB Legacy Support..................: [ DISABLED ]

some pretty pictures:

[ghost707]

Some more pictures:

[profJim]

A simplistic expanation:
vdrop = the difference between your cpu vcore setting and what you see when your system is at idle
vdroop = the difference between the observed cpu vcore at idle and when fully loaded

Here is an excellent article that deals with Vdrop and Vdroop:
Overclocking Intel's New 45nm QX9650: The Rules Have Changed (page 5)
You'll probably want to read the complete article.

Your "normal" cpu V.I.D. is usually displayed in your bios:
CPU Vcore .................. 1.?????...........: [AUTO]

Given the P45 chipset issues (from post #11) with your memory, I recommend the following bios changes:

1) set primary memory timings to 8-8-8-24 for now (it's easier to overclock using conservative memory timings)
2) set DRAM voltage to 1.60 or 1.70 volts (more isn't always better)
3) you can try increasing CPU Vcore incrementally, it's not likely that settings higher than 1.33 - 1.34 volts will be needed
4) set CPU Termination to 0.04 - 0.06 less than your CPU Vcore setting (this might help with stability)

I would start with steps 1 & 2 and try steps 3 & 4 if needed.
Load Line Calibration can be tried later if necessary.

The following are the results on my system with a 48% overclock:
CPU Vcore ............ 1.35000 ........ [1.36875]
PC Health: Vcore = 1.348
cpu-z @ idle ___ = 1.312
OCCT @ load __ = 1.298 - 1.312 (small data set)

The ET6 screenshot --> Tuner --> Advance --> Voltage is useful because it displays the actual applied bios voltage settings and lets us verify that these settings match the posted bios settings in your template.

Not so pretty but useful picture: