Not another EVO thread! This one is GOOD!

[jdirty]

LoL


I have a M4A79XTD EVO with the F3-128000CL9D-4GBRL

No problems! WoW!

Manually set timings in BIOS to 9-9-9-24 2T

ROCK SOLID STABLE @ stock

Notice, the @ stock part.

Pushing CPU multi ONLY, I'm able to get it to 3.6GHz, STABLE!

Good memory, thanks GSKILL!

I liked it so much, I bought another matching kit to fill the 2 empty slots on my mobo!

Now, whatcha think voltages should be with all 4 DIMMS filled?

And if you recommend a "NB" voltage, please be sure to specify which NB, the CPUNB, or the mobo NB.

[jdirty]

Anybody (not just a GSkill Tech), know what the NB,CPU/NB,HT voltages need to be when running all 4 sticks? I would really like an answer, I'm waiting to install the new memory until after I get confirmation on the information I need

[Tradesman]

Try it at stock (your current settings), may have to raise both voltages by .05 to .1 . Also check you tRFC, currently should be around 127 will need to go up to 195

[jdirty]

tRFC0 and tRFC1 both dont have an option to set at 195.

The only available settings are 90 160 300.

So I set it at 160ns

Prime failed on 1 core within 5 minutes (blend)

Prime stable for 8 hours on small FFT size.

This shows the CPU is stable, possible problem with memory.

Everything is stock except CPU multi.

BTW: I have no problems with this EVO rig and GSKILL memory, for example, I dont get blue screens, I dont have cold boot issue's, and everything is rock solid at stock.

[Tradesman]

May jump the tRFC to the 300 then for 4

[jdirty]

setting it to 300 causes instant reboot as soon as prime hits 100% core utilization.

Leaving it on AUTO sets it at 110. That is what it is currently running at.

On 110, Prime only fails instead of an instant reboot.

This is not a memory/cpu compatibility problem, keep in mind that the memory is 100% stable under Prime95 64bit, and OCCT Linpack 64bit for 12+ hours at stock 1600 speeds. EVERYTHING is still the same, except the CPU multi. I'm not overclocking the base clock, HT link, or the NB. Everything is stock, except the cpu multi.

I'm thinking it's a voltage issue, somewhere. Not sure where though. I've seen and read so many various voltage recommendations, I dont know what to believe. I actually seen a GSKILL tech recommend setting the NB voltage to 1.4 for this model MB, and that's almost silly. 1.4 is the absolute maximum for the mobo NB, and the CPU has it's own memory controller and nb independent of the motherboards nb, so bumping the CPUNB to 1.4 is also an odd recommendation, because bumping that value above 1.3 only causes extra heat and instabilities.

So which NB should I set at 1.4? Niether, both NB's simply dont need that much juice, and the CPU's NB actually suffer's from overvolting.

I know the motherboard's NB will need some type of voltage bump when filling all the dimms, but max it out.. no way. Maybe 1.25 - 1.3, anything above 1.3 is heavy for that component, it simply does not need that much voltage.

One more inquisitive question...

GSKILL recommends setting the timings to 9-9-9-24-2T for 800MHz (ddr3 1600).
But SPD reports that 1333 should use 9-9-9-25-2T (685MHz)

Thats a looser timing for a significantly faster MHz jump. The 25 value is tightened to 24 with a 115 MHz speed increase, doesnt seem like it makes sense, but that's how it is.

[jdirty]

No answer for 24 hours, so I'm bumping this thread.

Any response to the timings question? To simplify:

GSkill recommends 9-9-9-24 for 800MHz operation
SPD recommends 9-9-9-25 for 685MHz operation

That is one clock faster for a 115MHz speed increase, see what I'm saying?

The other question was in regard to a voltage recommendation for the NB (not the cpu nb) in order to properly handle 4 DIMMs, instead of 2. It seems I have found a good setting for my setup, All modules passed 4 pass's of memtest86 v4.00, but I would still like to here GSkill's response on why they have recommended setting the NB voltage to 1.4, as this can likely damage the NB because 1.4 is the maximum voltage allowed to be supplied to NB. If they were not referring to the motherboard NB, then the only other NB is on the CPU die, and feeding it 1.4v is ALOT of voltage for that component, which has been tested and verified to only cause instabilities and extra heat if set above 1.3. And the CPU/NB is independent from the motherboards NB, so voltage increases to CPUNB would have nothing to do with actual physical memory stability. The motherboard NB is what needs the voltage bump, but 1.4 seems like an outrageous number. Mine is ROCK solid right now at 1.3 NB Voltage, running 4x2GB stock. The system is overclocked (only CPU multi overclock) to 3.6 GHz right now (up from 2.8) with a 1.45 vcore on air with temps that I havent seen get above 40c. My northbridge stays nice and cool, but 1.4 would cause it to get hot, along with the MOSFETS that power it, just causing more heat for no reason.

[GSKILL TECH]

Manually set the memory to it's specifications. DDR3-1600 (800MHz), 9-9-9-24-2T 1.50V

The NB Voltage is necessary, it can be the highest setting for some motherboards since they are not technically qualified for 4 sticks of DDR3-1600, but if you want to operate them at that you will need to. What you can do is find the lowest stable value, but for testing purposes, we suggest 1.40V first. Or, simply downclock them to DDR3-1333 with 8-8-8-24 timings if you are concerned about the NB.

Thank you
GSKILL SUPPORT