The AMD 3GHz Club

[CharmedLover84]

Alrighty, I'll give the 14/16 a try and go ahead and bump the preamble up along with the chipset power. I think I'll bump the HTT back up to 3. And yeah since the PC4800 is in there I can screw with the memory divider a bit, long as I don't get it up past 600 it will still boot LOL

 

Thanks Thraxz, any other suggestions would be greatly appriciated

[CharmedLover84]

Welp just tried the following with the PC4800 and no dice, gonna play around with the divider though cuz for some reason it's only showing as DDR500 in Everest/CPUz, where as it should be showing as DDR560 if you multiply the divider times two then add in the amount I'm upping the FSB twice LOL.

 

BIOS Settings:

 

FSB Bus Frequency............................. - 334

LDT/FSB Frequency Ratio....................... - 3x

CPU/FSB Frequency Ratio....................... - 9x

PCI eXpress Frequency......................... - 101Mhz

 

CPU VID StartUp Value......................... - Auto

 

CPU VID Control............................... - 1.525V

CPU VID Special Control....................... - + 104.8%

LDT Voltage Control........................... - 1.30v

Chip Set Voltage Control...................... - 1.62v

DRAM Voltage Control.......................... - 2.87V

 

DRAM Configuration Settings:

 

DRAM Frequency Set............................ - 150

Command Per Clock (CPC)....................... - Enable

CAS Latency Control (Tcl)..................... - 2.5

RAS# to CAS# delay (Trcd)..................... - 4

Min RAS# active time (Tras)................... - 10

Row precharge time (Trp)...................... - 4

Row Cycle time (Trc).......................... - 14 Bus Clocks

Row refresh cyc time (Trfc)................... - 16 Bus Clocks

Row to Row delay (Trrd)....................... - 03 Bus Clocks

Write recovery time (Twr)..................... - 03 Bus Clocks

Write to Read delay (Twtr).................... - 02 Bus Clocks

Read to Write delay (Trwt).................... - 03 Bus Clocks

Refresh Period (Tref)......................... - 4708 Cycles

DRAM Bank Interleave.......................... - Enabled

 

DQS Skew Control.............................. - Auto

DQS Skew Value................................ - 50

DRAM Drive Strength...........................Auto

DRAM Data Drive Strength...................... Auto

Max Async Latency............................. - 8ns

DRAM Response Time............................ - normal

Read Preamble Time............................ - 5.5ns

IdleCycle Limit............................... - 16 Cycles

Dynamic Counter............................... - Disable

R/W Queue Bypass.............................. - 16 x

Bypass Max.................................... - 04 x

32 Byte Granularity........................... - Disable(4 Bursts)

 

Failed on both cores almost immediately. AM3 spit out 121K though :-D LOL

[oc newbie]

One way to try and tell if it is the cpu or ram is to look at which fft length prime fails in and then choose custom and set it to that fft. Set it to use alot of ram and start it and see how quick it fails. Go back in and set it to use very little ram with the same fft length and see if the cpu is passing that fft length. Takes awhile to do but seems to work pretty good. Thanks to cp for that advice helped me find out my ram is not clocking as high as it used to.

[CharmedLover84]

Okay, this may sound coocoo crazy BUT is it possible giving my CPU too much voltage would cause this problem?

 

I just backed the voltage off to 1.58 (1.55x102.4%) instead of 1.59 (1.525x104.8%) and it booted right up still. And when I give it 1.608 (1.5x107.2%) it won't boot up past the last BIOS screen.

 

Got the ram up to 273x2 now and it's bootin' right up.

[CharmedLover84]

Okay I did as you suggested just now OC, and it failed both, but it failed instantly on the CPU... now what does that tell me? :- I really think I may be givin' the CPU too much voltage as crazy as that sounds.

 

Okay, it fails immediately on Core1 with it set to very little, after a few seconds on core1 with it set to 400MBs of usage, and it failed around the same time with it set to very little and 400 on Core0. Seems like unlike most people my Core1 is gonna be the bigger problem, not the Core0 LOL

[OldGuy]

Keep in mind that the dividers don't always work out exactly right...least of your worries though Make sure the memory is running good even at a slower rate and then concentrate on the cpu...first and foremost/last but not least. What some folks forget when they tell you "just drop a divider on the memory" and find your max cpu is that some settings, if left on auto, will tighten up and create their own little problems. I always suggest booting at 200mhz and taking a snap shot using A64 tweaker and setting everything manually based on 200mhz before dropping the divider. I'd leave max async and read preamble on auto at this point and see where you can get. These two items will dynamically adjust upwards as you increase the bus speed/htt (or whatevever you want to call it).

 

Good luck on your 3ggg quest..............

 

Edit: Yes, too much voltage brings it's own problems. I have an Opty 150 that peaks at 2.9ghz and 1.58v. Any increase beyound 1.58v starts to take it backward as far as stability... BSOD festival the higher I take it. Some chips just have a max limit on how much voltage they can take and others will just keep climbing as long as you can keep them properly cooled...of course at some point you'll reach one limit or another (mhz or voltage).

[farksy]

Why don't you run your OCz TCCDs at their stock speeds ? It's 9:10 divider, 2.5-4-4-10 at that setting.

 

[The setting in which HTT = 334]

[Kingfisher]

Have you tried setting a 2T command rate? Worth a shot perhaps.

Get that 3 ghz OC already for the sake of Pete.

[CharmedLover84]

I'm gonna try to Prime as I have it right now, and I just tried the TCCD at a full 600, and guess what? AM3 wouldn't even pass LOL

 

Trying it backed off to 273x2 on the ram (According to Everest/CPUz) right now. If that fails I think I'll try setting the preamble and Max async to auto like Old Guy suggested, see where that gets me, then I'll try 2T with that if that also fails.

 

Never know, one of them may indeed be the difference maker. If it fails like that, I'll try lowering the ram until I'm way back down at 167x2, and if that fails, I'll switch back to my Patriot and see if all the adjustments I've made will actually worth with them LOL

[farksy]

I'll suggest you stick with one pair first. Try inverting the ram sticks, and also the ram slots, so you should have tried all 4 combinations. If you OCZ is not working fine, either you gotta RMA it of your computer's memory controller is a little bonkers which is pretty impossible.

 

Also, i think you should get your Cpu stable first [not sure whether you done that] instead of trying to get the whole setup to be stable at once.

[CharmedLover84]

I'm a bit insane when it comes to that. I try and get everything stable at once.

 

And that's what I'm trying to do by adjusting the ram divider, get the CPU stable, cuz it's ... kinda and kinda not.

 

I'm now at 1.57Vs and it seems to be happier than it was at 1.59 for sure.

 

We'll see.

 

I would like to focus on the processor but unfortunately I can't do that until I know for sure that the ram isn't the problem.

[farksy]

I'm a bit insane when it comes to that. I try and get everything stable at once.

 

And that's what I'm trying to do by adjusting the ram divider, get the CPU stable, cuz it's ... kinda and kinda not.

 

I'm now at 1.57Vs and it seems to be happier than it was at 1.59 for sure.

 

We'll see.

 

I would like to focus on the processor but unfortunately I can't do that until I know for sure that the ram isn't the problem.

I liked to get everything stable initially, but it got me no where. So it really depends.

 

I'll rather make sure your processor is 3Ghz do-able on both 1.57V and 1.59V. Then i'll use 1.57V for purely clocking the Cpu while 1.59V will be used when i'm clocking the rams, so the on-die mem controller gets more voltage.

 

Just to make sure, do some Prime95 small ffts. If it fails, it's your cpu already. If not, it's most probably your rams.

 

*A more systematic approach will help you go further.