1055T OC help/understanding

M

mmmmmdonuts

Gawd
Joined
Oct 7, 2009
Messages
821
I got a 1055T a few months ago and did some basic overclocking on it with my Gigabyte 870U3, and now have gotten my OC to [email protected] (300*12.5), 2400 NB, 2100 HT, and 1600MHz ram speed.

I know these settings seem to fall in the ideal range for basically all the components I have. I want to try for a higher overclock but I have been having some problems because I don't think I understand how overclocking a AMD processor occurs correctly.

I am used to OC Intel processors so I think of terms of FSB instead of CPU-NB being combined.

Tell me if I understand this correctly:

Your speed of your RAM is tied to your NB Frequency and NOT your FSB even though in the BIOS it allows you to use the multi by either (4, 5.33, 6, 8 for 800, 1066, 1333 and 1866 respectively)

So in my case I have a NB frequency of 2400/3 = 800 DDR - from my (3:8) ratio, I get my RAM at 1600, which is it's default.

I also was on this site, trying to further understand CPU-NB with the frequency relationship. The way I understand it is you want your NB frequency +/-200MHz = (CPU frequency*2)/3.15, which falls very close to what my settings are right now with my NB. So if I was to go for 4 GHz I should be around 2600-2700MHz for the NB right?

And because I did that I should lower my RAM setting to 4x where it will run 1:2 ratio which would bring it to 1300-1350 right?

I also think I understand I want my HTT to be around 2000 MHz plus or minus 150 MHz or so as a constant.

Now I have these voltages in my BIOS that I would change:

NB Voltage control - 1.1-1.8V
NB PCIE - 1.8-2.2V
CPU NB VID Control
CPU Voltage Control
Normal CPU VCore

Can someone explain these a little better to me in terms of what I want to control exactly?
I know the CPU Vcore is the max voltage for the CPU, but when NB starts to get integrated I start getting very confused.

Additionally should I be going for the highest FSB possibly with the lowest multi (aka 3900 (12*325, vs. 13*300). (I believe this matters in things like DC community where I would like to maximize production)

Thanks for all the assistance.
 
the same methods apply to non Black Edition CPU's for AMD as for most intel, however they have "FSB" holes at lower levels.......go with highest multi and back down as needed, same methodology.
 
I think I may just write a really easy/short guide to ocing all the AM3 chips since this seems to come up quite a bit
 
QuiteSufficient said:
I think I may just write a really easy/short guide to ocing all the AM3 chips since this seems to come up quite a bit
Click to expand...

probably a good idea.....you may want to pick CCityInstallers brain on this ;) he is pretty dam good with AM3 Overclocking
 
mmmmmdonuts said:
I know the CPU Vcore is the max voltage for the CPU, but when NB starts to get integrated I start getting very confused.
Click to expand...
The CPU-NB is part of the CPU and the general rule is to use no more voltage than [CPU voltage - 0.1v]. Thubans (x6) are more sensitive to CPU-NB voltage than Deneb (x4) so you will probably reach your CPU-NB ceiling between 1.25v and 1.35v. As you may already know ~1.5v is safe 24/7 for the CPU. There have been reports (rare) of degradation above 1.5v.

mmmmmdonuts said:
Additionally should I be going for the highest FSB possibly with the lowest multi (aka 3900 (12*325, vs. 13*300). (I believe this matters in things like DC community where I would like to maximize production)
Click to expand...

That isn't the case with AMD. With all things else the same there is no reasonable performance benefit to using a low(er) multiplier and a high(er) HT reference clock*. In a hypothetical with all other speeds the same your example of (12*325, vs. 13*300) would be identical in performance. There will also be a ceiling to which you cannot raise the HT reference clock any further regardless of voltage or the dropping of multipliers.

The final HT clock does not add any performance beyond 2000 Mhz. Your HT overclock of 100mhz is nothing to worry about now but if you keep going higher with the HT reference clock you may have to drop it further.

*Unless doing so puts the CPU-NB into a sweet spot you can't reach otherwise with the CPU-NB multipliers.
 
QuiteSufficient said:
I think I may just write a really easy/short guide to ocing all the AM3 chips since this seems to come up quite a bit
Click to expand...


lol i already did but i posted it a few weeks ago.

well now that i re-read what i posted, its not really a guide but it should get the op'er started on understanding the whole overclocking process for a locked AMD processor.

http://hardforum.com/showpost.php?p=1036460424&postcount=3

if something confuses you or ya have any other questions ask. hell i might even re-write it if i have to.

the only thing missing out of that semi half assed guide i made is the CPU-NB clock. if your board gives you the option to lower your NB multiplier then lower it. if not then keep an eye on it. the 1055T maxes out at 3Ghz on the NB and will most likely require you to raise cpu-NB voltage to reach that number.
 
Last edited:
Thanks for the help guys. Really the part that was confusing me is the CPU-NB and what role that played in the whole equation. Now since my CPU really seems to like 300 FSB, I should really just fiddle with the CPU multi's, and voltage (since I am only at 12.5 right now and can go to 14)?

I guess in my first post in the site I linked he said sometimes for higher overclocks I may have to use a higher CPU-NB frequency as well to obtain those speeds. (the sweet spot I guess, DedEmbryonicCe??)

the only thing missing out of that semi half assed guide i made is the CPU-NB clock. if your board gives you the option to lower your NB multiplier then lower it. if not then keep an eye on it. the 1055T maxes out at 3Ghz on the NB and will most likely require you to raise cpu-NB voltage to reach that number.
Click to expand...

Now for folding (and BOINC as well) I want that as close to 3GHz as possible correct?

To be honest with you last night I was doing a little to much reading on this subject, and most of it was for the Denabs or earlier Phenoms, so I think I really just confused myself more than necessary. The problem with most of the guides is most of them are very, very vague on the NB frequency and just say lower it till it is stable, but once you start pushing the chip close or near to its limits I think that no longer seems to hold true.

Thanks for the help though. I appreciate it.
 
with that board you should be able to hit 3 ghz on the NB.....try for 4.0 on the CPU speed and 3.0 on the NB with corresponding voltage.......my THuban needs 1.5v to get 4.0, and the NB needs 1.4v to get 2.8Ghz(however it is a 7** series chipset so this is a big part of the issue)
 
I have been trying to find the sweet spot on my 1055T also. I think I am running 280 x 13.5 atm because I my setup didn't seem to like 300 fsb. This puts me at 3,780 Mhz on the cpu,and 2800 on the HTT and NB frequency. Don't forget to tighten those memory timmings also. My G-skill are running at DDR3 746 Mhz @ 7-8-8-12. What really makes me like these numbers is the fact that my cpu can do this at 1.35 vcore :)
 
sirmonkey1985 said:
lol i already did but i posted it a few weeks ago.

well now that i re-read what i posted, its not really a guide but it should get the op'er started on understanding the whole overclocking process for a locked AMD processor.

http://hardforum.com/showpost.php?p=1036460424&postcount=3

if something confuses you or ya have any other questions ask. hell i might even re-write it if i have to.

the only thing missing out of that semi half assed guide i made is the CPU-NB clock. if your board gives you the option to lower your NB multiplier then lower it. if not then keep an eye on it. the 1055T maxes out at 3Ghz on the NB and will most likely require you to raise cpu-NB voltage to reach that number.
Click to expand...

I was more thinking a more full-fledged deal with suggested starter settings for the x2s, 3s, 4s, and 6s - since I've done all of them. Not sure how popular these processors even are though, seems like the intel forum is absolutely on fire in comparison.
 
QuiteSufficient said:
I was more thinking a more full-fledged deal with suggested starter settings for the x2s, 3s, 4s, and 6s - since I've done all of them. Not sure how popular these processors even are though, seems like the intel forum is absolutely on fire in comparison.
Click to expand...

I think a lot of people have AMD processors but don't try and get the maximum overclock they can relative to the Intel group or maybe don't overclock them at all. AMD processors to me tend to be a lot more finicky than their Intel counterparts in overclocking as well in my experience. The amount of BE chips available for AMD make for crank the multi overclock probably the most common.

For example I can fold for weeks straight (24/7 for 19 days) but then I can restart my computer and the BIOS would say its unstable. Go to stock settings, boot into windows, restart the computer and OC again and be able to go as long as I want again.

I honestly think a guide would be terrific especially considering how different each of the processors and chipsets (700 series vs 800 series) are. Each one your going to have different max values and I think it makes a huge difference on what to expect.

A guide that could summarize the chipsets and processors and their combinations would be a lot of work but it would go a long way in helping people overclock their chips to the maximum potential.:D
 
What i would like to know is how you folks are setting the mutipier lower or higher, with a 1055T.

these are multiplier locked cpus.

Regardless overclocking the am3 cpu is alot different than other processors. I am currently at 3.7ghz on my 1055t, with a FSB of 270. x 14.

I am on a older am2+ board, with ddr2 memory with the memory at 5.33 mutipier, giving me ddr2 1439mhz

Biggest problem with the am3's is the on die memory controler only rated for ddr3 1333. So anything above that is a overclock. Disable turbo mode if your motherboard allows you to. And run at full oc speed with cool and quiet.

Performance comes from the HT speed, and memory speed. My chip goes to 4ghz with ease with a small bump in voltage (.050) but i have to lower the memory mutipier to 4 which dropps my memory speed 1160mhz. MY system is faster at 3.7ghz on the cpu and 1439 mhz on the ram. Good thing at 3.7ghz cpu voltage is stock.
 
ebduncan said:
What i would like to know is how you folks are setting the mutipier lower or higher, with a 1055T.

these are multiplier locked cpus.
Click to expand...

Multiplier locked means it can't go above a certain multiplier which is 14 in the case of the 1055. It most certainly can go lower than that.
 
I tried to go with lower multi (13) and higher bus (300) but my setup didn't seem to like that. I'm sure it depends heavily on the persons setup. Some people seem to have no problem hitting 300 fsb while I was unable to achieve this.
 
interesting, i guess i find another flaw in my motherboard, i can set the multiplier to anything and it is always at 14 no matter what. Worked fine with my 9950 agenda core. this older am2+ board seems to be rather limited on the hexcores, but man is a great board.

oh well just going to have to wait till the am3+ mobos come out.

not like it would matter anyways highest i can get the fsb to is 340. i would have to lower the mutipier down to 12x for 4ghz on the cpu and set the ram mutiplier to 4. would give me 1360mhz on the ram.

I still think my 1439 ram speed with cpu at 3.7 would out perform the 4ghz cpu and 1360 on the ram..DDR3 ram just sucks for AMD users right now. I will take my ddr2 at this speed vs any ddr3 1333, or 1600 overclocked, its rather lucky to get 1800+ ddr3 speeds with current am3 cpus.
 
most of teh AM3's will run the Thubans at 4.0, with the Ram at 1600, thus beating your set up ;)

however that is a nice OC on the RAM you have, dont see that high on DDR2 too often anymore
 
actually ddr2 at 1439 will out perform all ddr3 at 1600, I have much lower timings.

5-5-5-15 timings.

vs ddr 3 timings of 7-7-7-24 (good ddr3) ddr3 1600 will loose to my ddr2 @1439, You need ddr3 1700+ speeds to out perform it.

most good ddr2 1066 spec ram has no problems going to these speeds.
 
Can you personally tell an honest difference between the timings? other than in benchmarks ?
 
yes, its the difference between x4 aa and no aa in metro 2033 @ 1080p
but otherwise no i cannot tell the difference. If i had a stop watch handy for when i am doing excel sheets i am sure there would be a good difference there as well.
 
for me i have a very similar setup, bit higher CPU speed, and i can do x4aa in Metro at 1080p with very loose timings (last i checked 9-9-9 or so) ..... and ya a stop watch to get the few seconds......i have never been able to see a noticeable difference, but if you can the more power to you.
 
I'm having the same issues of getting that special blend of the greatest overclocking performance. So let me get this straight, having a higher NB (eg. 3000mhz) increases performance? Cause most reviews always reference leaving the NB and HT around 2-2.1Ghz for stability. I game quite often and I'm trying to configure my rig optimally. See specs for rig below.
 
TooEasy! said:
I'm having the same issues of getting that special blend of the greatest overclocking performance. So let me get this straight, having a higher NB (eg. 3000mhz) increases performance? Cause most reviews always reference leaving the NB and HT around 2-2.1Ghz for stability. I game quite often and I'm trying to configure my rig optimally. See specs for rig below.
Click to expand...

What is your FSB and Multi along with your voltages, NB and HT because I have the same RAM/CPU and I am curious what your settings are at.
 
I'm having the same issues of getting that special blend of the greatest overclocking performance. So let me get this straight, having a higher NB (eg. 3000mhz) increases performance? Cause most reviews always reference leaving the NB and HT around 2-2.1Ghz for stability. I game quite often and I'm trying to configure my rig optimally. See specs for rig below.
Click to expand...

based on you sig, your on a 32" hdtv. with 4870's in crossfire, you really shouldn't have any problems running all your games at max settings at 1080p. You have more gpu muscle than i do with a single 6870.

peak performance will only be noticed in benchmarks. Try and push the HT link to 3ghz, just like any over-clock it can compromise stability.

or me i have a very similar setup, bit higher CPU speed, and i can do x4aa in Metro at 1080p with very loose timings (last i checked 9-9-9 or so) ..... and ya a stop watch to get the few seconds......i have never been able to see a noticeable difference, but if you can the more power to you
Click to expand...

are you at very high settings? I am a bit picky when it comes to FPS, i don't like it to dip below 30fps. Regardless i suspect some sarcastic comments from you, which is totally uncalled for.
 
ebduncan said:
based on you sig, your on a 32" hdtv. with 4870's in crossfire, you really shouldn't have any problems running all your games at max settings at 1080p. You have more gpu muscle than i do with a single 6870.

peak performance will only be noticed in benchmarks. Try and push the HT link to 3ghz, just like any over-clock it can compromise stability.



are you at very high settings? I am a bit picky when it comes to FPS, i don't like it to dip below 30fps. Regardless i suspect some sarcastic comments from you, which is totally uncalled for.
Click to expand...

You sir, should honestly be quiet..You are spouting your opinion as fact, when in fact, you couldn't be more wrong...Where to begin.....

actually ddr2 at 1439 will out perform all ddr3 at 1600, I have much lower timings.

5-5-5-15 timings.

vs ddr 3 timings of 7-7-7-24 (good ddr3) ddr3 1600 will loose to my ddr2 @1439, You need ddr3 1700+ speeds to out perform it.

most good ddr2 1066 spec ram has no problems going to these speeds.
Click to expand...
You are totally wrong..You *MIGHT* see a slight difference in some old benchmark, but I highly doubt it...For real world performance, especially F@H, raw speed is way better then running 400mhz slower with tighter timings...

2..
yes, its the difference between x4 aa and no aa in metro 2033 @ 1080p
but otherwise no i cannot tell the difference. If i had a stop watch handy for when i am doing excel sheets i am sure there would be a good difference there as well.
Click to expand...
If you expect anyone to believe this, then I have some ocean front property in AZ I'd love to give you a great deal on...

3..
not like it would matter anyways highest i can get the fsb to is 340. i would have to lower the mutipier down to 12x for 4ghz on the cpu and set the ram mutiplier to 4. would give me 1360mhz on the ram.

I still think my 1439 ram speed with cpu at 3.7 would out perform the 4ghz cpu and 1360 on the ram..DDR3 ram just sucks for AMD users right now. I will take my ddr2 at this speed vs any ddr3 1333, or 1600 overclocked, its rather lucky to get 1800+ ddr3 speeds with current am3 cpus.
Click to expand...

There is no game, program, or real world scenario where the same cpu clocked @ 3.7Ghz with 100mhz faster ram is going to out perform a 4.0Ghz clock with ram 100mhz less...It is simply NEVER GOING TO HAPPEN..

I'm having the same issues of getting that special blend of the greatest overclocking performance. So let me get this straight, having a higher NB (eg. 3000mhz) increases performance? Cause most reviews always reference leaving the NB and HT around 2-2.1Ghz for stability. I game quite often and I'm trying to configure my rig optimally. See specs for rig below.
Click to expand...

The higher the NB frequency, the better with Thurbans...Anandtech did a great article showing the difference with NB clocks from 2Ghz to 3Ghz, and the free performance boost is insane! You can get some awesome performance gains just from o/c'ing the NB while leaving the cpu @ stock speeds...

Performance comes from the HT speed, and memory speed.
Click to expand...

You are totally wrong on the first part..Leave the HT clock at 2Ghz vs 2.6Ghz makes ZERO difference...How do I know? I benched the shit out of my X6 with dozens of combination's and there is no difference...

On the second part, you proved my point..Raw memory speed with an on die controller will always outperform sub par speed with tighter timings...

I'm not trying to single you out, but you did an excellent job for me..You seem to be attached to your setup, and have decided that you know how much better it is over a faster DDR3 setup..Please remind me how many DDR3 setups with C3 X4's or Thurban's you have personally o/c'd?:rolleyes::rolleyes::rolleyes:

 
ccityinstaller said:
You sir, should honestly be quiet..You are spouting your opinion as fact, when in fact, you couldn't be more wrong...Where to begin.....


You are totally wrong..You *MIGHT* see a slight difference in some old benchmark, but I highly doubt it...For real world performance, especially F@H, raw speed is way better then running 400mhz slower with tighter timings...

2..
If you expect anyone to believe this, then I have some ocean front property in AZ I'd love to give you a great deal on...

3..

There is no game, program, or real world scenario where the same cpu clocked @ 3.7Ghz with 100mhz faster ram is going to out perform a 4.0Ghz clock with ram 100mhz less...It is simply NEVER GOING TO HAPPEN..



The higher the NB frequency, the better with Thurbans...Anandtech did a great article showing the difference with NB clocks from 2Ghz to 3Ghz, and the free performance boost is insane! You can get some awesome performance gains just from o/c'ing the NB while leaving the cpu @ stock speeds...



You are totally wrong on the first part..Leave the HT clock at 2Ghz vs 2.6Ghz makes ZERO difference...How do I know? I benched the shit out of my X6 with dozens of combination's and there is no difference...

On the second part, you proved my point..Raw memory speed with an on die controller will always outperform sub par speed with tighter timings...

I'm not trying to single you out, but you did an excellent job for me..You seem to be attached to your setup, and have decided that you know how much better it is over a faster DDR3 setup..Please remind me how many DDR3 setups with C3 X4's or Thurban's you have personally o/c'd?:rolleyes::rolleyes::rolleyes:

Click to expand...

Schools in session.


You just got e-raped.
 
To the OP..Now that I have cleared up the fact from fiction, I would like to get this thread back on topic and help you out..Please posted some detailed info for me, Voltage settings etc, and I will help you get things straightened out..

These cpu's and the 800 series actually o/c very easy once you get an understanding of where to begin...

 
personally sounds like someone smoked to much crack today.

how many setups have i benched? well over 10 different setups.

but i'm not going to argue over some stupid forums, so will just give you my e-penis. Yes 100mhz ram is way better than 200-300mhz on th cpu. Games are not cpu bottlenecked, just gpu and ram.

you go ahead and read your reviews. Why don't you look up inside-hardware.net its a old computer website i was admin at. I know my way around.

if you ran so many benchmarks post them up.
 
first those games are fully mutithreaded.

and your own charts show nothing more than 1-2 fps difference between 200-300mhz of clock speed. You will get much better results with higher speed ram, end of story.


nice try though.

I am not going to argue with you folks, if you want to express you noobness go right on ahead. Like I said before i will be happy to hand you my e-penis. This is exactly why i don't post more often on these forums. Bunch of idiots running around acting like they know the world.
 
Sc2 uses 1.5 cores btw, and both the sc2 and civ5 charts demonstrate at least a 10% gain for 400mhz (the difference between 3,600 and 4,000)

Since you are calling on others for proof, where is yours? Are there any benchmarks on the web to prove the superiority of ram speed? Have you done any of your own?
 
By sc2 I meant starcraft not supreme commander, my apologies

Out of curiosity I have just performed a test attempting to duplicate the gaps we are discussing,

First I ran pi on my own system

4.1ghz / 1440mhz DDR 3 at very slow 10-10-10-24 timings

This yielded a 40 second time to calculate 2m units

firstpi.jpg


Next I turned my processor down to 3,760 and my DDR3 down to 1,152 with 7-7-7-20 timings. I have never really benched my ram so I do not know what the fastest timings it can hit at a given speed

secondpi.jpg


This yielded a 43 second time to calculate 2m units.

I will attempt more benchmarks tomorrow in more gaming oriented applications, but it seems you are attempting to compare DDR2 at a fast speed (~1,400) with 5-5-5 timings to DDR3 at 1,600....1,800....or even 2,000 mhz with 8-8-8 or 9-9-9 timings. I do not see this penciling out, but my tests may show otherwise.
 
In 3dmark06,

4,100mhz processor / 1,440mhz memory @ 10-10-10-24: 19,140 marks

3,780mhz processor / 1,152mhz memory @ 7-7-7-20: 18,387 marks

3,780mhz processor / 1,440mhz memory @ 8-8-8-20: 18,436 marks

Not the most modern tool I understand
 
ccityinstaller said:
To the OP..Now that I have cleared up the fact from fiction, I would like to get this thread back on topic and help you out..Please posted some detailed info for me, Voltage settings etc, and I will help you get things straightened out..

These cpu's and the 800 series actually o/c very easy once you get an understanding of where to begin...

Click to expand...

Thanks for the help. I tried fiddling last night and got no where stable.

First off: MB
RAM (4x4gb)

Current Settings: (Been stable 24/7 running BOINC/F@H no better stability IMHO)
CPU FSB: 300
CPU Multi: 12.5
CPU Frequency: 3750MHz
NB Multi: x8 (2400MHz)
HT Link (1.4GHz) (basically a x7 multi) (2100MHz)
PCIE Clock: 100MHz
Memory Clock: 5.333x (1600MHz)
RAM: 9-9-9-9-24-2T-33
DRAM Voltage: 1.5V
NB Voltage Control: 1.1V (Default Normal setting)
NB PCIE Voltage Control: 1.8V (Default Normal setting)
CPU NB VID Control: 1.225V
CPU Voltage Control: 1.475V

My temps are never cross 45C on the chip using LinX, and about 40C on the individual cores according to HWMonitor. System is cooled by a H50 in Push/Pull.

Additionally I bought passive heatsink fans to put over the power mosfets on the NB so those never cross about 80C.

I tried these fooling around with these settings last night as well (boots without a problem just shuts down when stress testing)

Current Settings: (Been stable 24/7 running BOINC/F@H no better stability IMHO)
CPU FSB: 308
CPU Multi: 13
CPU Frequency: 4004MHz
NB Multi: x8 (2464MHz)
HT Link (1.4GHz) (basically a x7 multi) (2156MHz)
PCIE Clock: 100MHz
Memory Clock: 5.333x (1644MHz)
RAM: 9-9-9-9-24-2T-33
DRAM Voltage: 1.6V
NB Voltage Control: 1.1V (Default Normal setting)
NB PCIE Voltage Control: 1.8V (Default Normal setting)
CPU NB VID Control: 1.275V
CPU Voltage Control: 1.5V


I guess I have two questions: Which of my two NB voltages is the one I care about to get a high NB? I think its CPU NB VID Control. Am I right?
My upper bound on my FSB seems to hit around 325-330 so I have plenty of headroom if needed.

Thanks for the help.
 
I guess I have two questions: Which of my two NB voltages is the one I care about to get a high NB? I think its CPU NB VID Control. Am I right?
My upper bound on my FSB seems to hit around 325-330 so I have plenty of headroom if needed.
Click to expand...

Cpu nb vid control is your main intergated memory controller voltage(on the cpu). The NB voltage is the actual voltage of your northbridge.

hope that helps.


4,100mhz processor / 1,440mhz memory @ 10-10-10-24: 19,140 marks

3,780mhz processor / 1,152mhz memory @ 7-7-7-20: 18,387 marks

3,780mhz processor / 1,440mhz memory @ 8-8-8-20: 18,436 marks
Click to expand...

3dmark06 has seperate values in benchmarking the cpu. I understand i am loosing some performance by reducing the actual cpu speed, that is not what i am concerned with, i am concerned with fps in games. Yes I am comparing the speed of high clocked low timing ddr2 vs higher clocked ddr 3 with more loose timings. I am not entirely crazy. But for example lets use your 3dmark06 scores. The last two specifically, there is a difference of 49 marks between the 2. For a difference of 288mhz on the ram, reducing your timings down just 1 across the board pretty much matched its performance.

I don't think ddr2 @ 1440mhz with timings of 5-5-5-15, can match 1800+ ddr3 though. it would Trade blows at 1800, anything faster for ddr3 it would win.
 
DrDoU said:
AOD has a tuning reference at their web site.It will give you some idea what may be achieved.The sampling pool was small at the time,so overclocks may vary.It also gives definitions of various terms.
Click to expand...


Late to conversation so here is a guide
 
mmmmmdonuts said:
Thanks for the help. I tried fiddling last night and got no where stable.

First off: MB
RAM (4x4gb)

Current Settings: (Been stable 24/7 running BOINC/F@H no better stability IMHO)
CPU FSB: 300
CPU Multi: 12.5
CPU Frequency: 3750MHz
NB Multi: x8 (2400MHz)
HT Link (1.4GHz) (basically a x7 multi) (2100MHz)
PCIE Clock: 100MHz
Memory Clock: 5.333x (1600MHz)
RAM: 9-9-9-9-24-2T-33
DRAM Voltage: 1.5V
NB Voltage Control: 1.1V (Default Normal setting)
NB PCIE Voltage Control: 1.8V (Default Normal setting)
CPU NB VID Control: 1.225V
CPU Voltage Control: 1.475V

My temps are never cross 45C on the chip using LinX, and about 40C on the individual cores according to HWMonitor. System is cooled by a H50 in Push/Pull.

Additionally I bought passive heatsink fans to put over the power mosfets on the NB so those never cross about 80C.

I tried these fooling around with these settings last night as well (boots without a problem just shuts down when stress testing)

Current Settings: (Been stable 24/7 running BOINC/F@H no better stability IMHO)
CPU FSB: 308
CPU Multi: 13
CPU Frequency: 4004MHz
NB Multi: x8 (2464MHz)
HT Link (1.4GHz) (basically a x7 multi) (2156MHz)
PCIE Clock: 100MHz
Memory Clock: 5.333x (1644MHz)
RAM: 9-9-9-9-24-2T-33
DRAM Voltage: 1.6V
NB Voltage Control: 1.1V (Default Normal setting)
NB PCIE Voltage Control: 1.8V (Default Normal setting)
CPU NB VID Control: 1.275V
CPU Voltage Control: 1.5V


I guess I have two questions: Which of my two NB voltages is the one I care about to get a high NB? I think its CPU NB VID Control. Am I right?
My upper bound on my FSB seems to hit around 325-330 so I have plenty of headroom if needed.

Thanks for the help.
Click to expand...

Boost the CPU NB VID to 1.3V and try again..If not, reboot and boost it again..You are running a fairly high FSB (even though that isn't the proper term anymore), and you will need a good bit of NB VID to get it stable IMO...

Is the machine powering completely off when you stress test or just rebooting with a BSOD?

 
ccityinstaller said:
Boost the CPU NB VID to 1.3V and try again..If not, reboot and boost it again..You are running a fairly high FSB (even though that isn't the proper term anymore), and you will need a good bit of NB VID to get it stable IMO...

Is the machine powering completely off when you stress test or just rebooting with a BSOD?
Click to expand...

Thanks for the help. I will try it a little bit tonight. Typically I have been getting about a 60/40 mix of blue screen (stop code 124) to complete shut off.
 
mmmmmdonuts said:
Thanks for the help. I will try it a little bit tonight. Typically I have been getting about a 60/40 mix of blue screen (stop code 124) to complete shut off.
Click to expand...

The BSOD is related to memory issues on this platform, which too little NB VID can indeed cause..

In my findings, powering completely off or freezing is related to cpu Vcore..

Are you sure you have a quality PSU in the box?
 
You must log in or register to reply here.
Back
Top