-
Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!
OC gone bad?
- Thread starter Itachiii
- Start date
- Thread Starter
- #42
Well,I noticed the voltage doing 1 sec drops at random moments. At stress tests( 5 minutes the longest one), it was keeping steady 1,414 - 1,428V. So, just to sum things up, OC'ing the CPU with the multiplier then tweaking individually CPU_NB,HT Link and RAM timings will be easier and probably a stable system?
- Thread Starter
- #44
Yea, I was hoping that maybe when i was gonna stabilize the system I could play around with max speed. But 4.5GHz @ 1.414V doesn't leave much space for more squeeze, does it?
- Joined
- Apr 20, 2004
- Location
- JAX, Mississauna
Well if the board has a decent CPU_LLC circuit attached to Cpu voltage input, then you don't see those sort of swings. And catching the voltage doing that swing is hard to actually do without using a DMM and watching the voltage for a while. The software on computers is only polling/looking at voltages at some predetermined rate and not all the time.
The other big thing attributed to Giga was cpu throttling at which time the multiplier usually dropped and also the Vcore in order to make for a weak VRM circuit. It is so hard to actually give solid advice to one about a setup that we have not seen very successful in the forum. Not that in the rest of the world something is not okay...just we have not seen it here.
On my Asus CHV I can disable APM and enable HPC and everything is fine. It is said on Giga AM3+ that you must leave APM enabled and enable HPC or the board double boots everytime you start it up. So just crazy arse stuff that the majority of boards do not do is found too often in the Giga AM3+ boards.
It is my understanding that the newer 970 UD3P Rev 1.0 since it is the first of that series, is a fairly decent mobo that acts much more like other boards, but again we have seen just a few.
I still suggest you take F12s pics and downsize them and attach to the forum so we can see all the overclocking/over-volting options. The deal is you have to scroll down thru the menus and capture along the way and keep captures together so they make sense. WYSIWUG with the captures. Every F12 is exactly what is on the screen and no more. So scrolling and capturing is a must. We can download a manual and see pics of bios bUt the later bios may change options and we still have no clue exactly what a user may really have for bios menu options. Of course that is work to do. I know. Been there and done it. But it maybe the only way you actually get a fully reliable overclock with a board that seems bound to act oddly at first glance.
RGone...ster.
The other big thing attributed to Giga was cpu throttling at which time the multiplier usually dropped and also the Vcore in order to make for a weak VRM circuit. It is so hard to actually give solid advice to one about a setup that we have not seen very successful in the forum. Not that in the rest of the world something is not okay...just we have not seen it here.
On my Asus CHV I can disable APM and enable HPC and everything is fine. It is said on Giga AM3+ that you must leave APM enabled and enable HPC or the board double boots everytime you start it up. So just crazy arse stuff that the majority of boards do not do is found too often in the Giga AM3+ boards.
It is my understanding that the newer 970 UD3P Rev 1.0 since it is the first of that series, is a fairly decent mobo that acts much more like other boards, but again we have seen just a few.
I still suggest you take F12s pics and downsize them and attach to the forum so we can see all the overclocking/over-volting options. The deal is you have to scroll down thru the menus and capture along the way and keep captures together so they make sense. WYSIWUG with the captures. Every F12 is exactly what is on the screen and no more. So scrolling and capturing is a must. We can download a manual and see pics of bios bUt the later bios may change options and we still have no clue exactly what a user may really have for bios menu options. Of course that is work to do. I know. Been there and done it. But it maybe the only way you actually get a fully reliable overclock with a board that seems bound to act oddly at first glance.
RGone...ster.
- Thread Starter
- #46
I had to restart the system from windows couple times and it would just turn off the pc once gone from the shut down screen, and then it would come on again. Ok so i'll be getting screens and i'm gonna try bumping each freq individually
- Joined
- Apr 20, 2004
- Location
- JAX, Mississauna
Well you really need to get off this kick of bumping the other frequencies at all until you get to some sort of stable speed by raising the multiplier. That is what they unlocked the multiplier to be able to do. I would guess based on testing and benches that 90% of the extra performance comes from getting the cpu speed up and that leaves only a measly 10% gain somewhere among three other busses. Not so much really.
RGone...ster.
RGone...ster.
- Thread Starter
- #48
Ok, so I OC'd the CPU to 4.5@ 1,41V (wouldn't stabilize at 4,6 same Vcore and would go 1,52V for 4.7!),temps would not go over 41C,set the timings 11-9-9-27-39 and then I tried to tweak the CPU_NB but it wouldn't reach 2200MHz (stock was 2000MHz) even at 1.3V so I left it on auto. Ran P95 for about 2 hours, no errors and then got my friend playing BF4 for 2 hours or so and the system crashed/BSOD. It might have been something random..? I also tweaked the GPUs (Xfire 2x HD6850 820/1100@ 1.150V)
might have caused the issue?
might have caused the issue?
- Thread Starter
- #50
Everything I wrote above, I did step by step. First OC'ed the CPU, checked for stability, then I tweaked the timings on the RAM, checked for stability,then I OC'd the CPU_NB and the system was getting errors at P95 so I reverted everything back to auto. Ran P95 for 2 hours blend test and the everything was ok, no errors, no warnings. The only thing I did not check was the GPU OC, I got my friend to play Batman Arkham City and BF4 just to check for GPU temps and stability, and everything seemed ok! Then I left from his house and about 1-2 hours later he got a BSOD. I don't think I was rushing this time. Besides, I knew more or less my max speeds @ x Volts.
- Joined
- Apr 20, 2004
- Location
- JAX, Mississauna
Maybe you went step by step...
...part of the problem is the rig is not in your hands nor close enough for us to view the situation either. Windows Event Viewer may give clues to what is causing BSOD and if BSOD is told to save memory dump, the memory dump is about a 7 of 1-10 tool for determining what is going on at time of BSOD.
KHX1600C9D3/4G
Jedec #4 shows 9, 9, 9, 27, 39 at 1.5Volts.
Jedec #5 shows 10, 9, 9, 27, 39 at 1.5Volts.
Jedec #6 shows 11, 9, 9, 27, 39 at 1.5Volts.
Now how in the hale can that be? I mean really. tRAS and tRC are the same for those 3 Jedecs. You could look at probably a 100 other types of ram and never ever see such occur. Because most ram has to change at least 4 out of the 5 of those timings or generally so if you change the heck out of the CAS.
I would believe Jedec #5 much more if it were written as 10, 11, 10or9, 27, 39 and 2T for CAS. We have seen too too many users with Kingston HyperX with issues with AMD in this forum section about AMD cpus. So much so that of course we would not recommend such for the average user.
I think I would certainly set the ram voltage to 1.55V instead of 1.5V. Maybe even 1.6V which either of which has not hurt any users ram lately that we have seen.
Then I would go into the ram timings and set 10, 11, 10, 29, 41, 2T CAS and retest with P95 Blend mode for the minimum we suggest of 2 hours. Should be no failures of any sort during that 2 hour testing time frame. Temps should be CPU of less than 72c and Core/Package of less than 62c as viewed in HWMonitor FREE which is running during the P95 Blend mode testing so it can log min/max volts and temps.
These changes may do well. They may not and if you wish us to try and help we need to see the attached capture of HWMonitor open with all voltages shown and down thru cpu package temp. Then CPUz in three captures of CpuZ > CPU Tab > Memory Tab > SPD Tab. Without those for our viewing miles and miles from your computer...well you might as well commune with ghosts.
RGone...ster.
...part of the problem is the rig is not in your hands nor close enough for us to view the situation either. Windows Event Viewer may give clues to what is causing BSOD and if BSOD is told to save memory dump, the memory dump is about a 7 of 1-10 tool for determining what is going on at time of BSOD.
KHX1600C9D3/4G
Jedec #4 shows 9, 9, 9, 27, 39 at 1.5Volts.
Jedec #5 shows 10, 9, 9, 27, 39 at 1.5Volts.
Jedec #6 shows 11, 9, 9, 27, 39 at 1.5Volts.
Now how in the hale can that be? I mean really. tRAS and tRC are the same for those 3 Jedecs. You could look at probably a 100 other types of ram and never ever see such occur. Because most ram has to change at least 4 out of the 5 of those timings or generally so if you change the heck out of the CAS.
I would believe Jedec #5 much more if it were written as 10, 11, 10or9, 27, 39 and 2T for CAS. We have seen too too many users with Kingston HyperX with issues with AMD in this forum section about AMD cpus. So much so that of course we would not recommend such for the average user.
I think I would certainly set the ram voltage to 1.55V instead of 1.5V. Maybe even 1.6V which either of which has not hurt any users ram lately that we have seen.
Then I would go into the ram timings and set 10, 11, 10, 29, 41, 2T CAS and retest with P95 Blend mode for the minimum we suggest of 2 hours. Should be no failures of any sort during that 2 hour testing time frame. Temps should be CPU of less than 72c and Core/Package of less than 62c as viewed in HWMonitor FREE which is running during the P95 Blend mode testing so it can log min/max volts and temps.
These changes may do well. They may not and if you wish us to try and help we need to see the attached capture of HWMonitor open with all voltages shown and down thru cpu package temp. Then CPUz in three captures of CpuZ > CPU Tab > Memory Tab > SPD Tab. Without those for our viewing miles and miles from your computer...well you might as well commune with ghosts.
RGone...ster.
- Thread Starter
- #52
Hey RGone ( and the rest of the guys ofc) thanks a lot for your help! I contacted with my friend, and after that random BSOD, the system is running just fine.. Checked everything (voltages, freqs etc) and eveything is as I set them. The way he told me ai though he was getting crashes all the time, but it was one random. He's playing now for 5 hours straight, Cpu temps not going over 42c and sockey temps over 47c.
- Joined
- Apr 20, 2004
- Location
- JAX, Mississauna
Itachiii, you are in the least enviable position known to computer users. Unpaid IT Tech for somone.
RGone...ster.
RGone...ster.
- Thread Starter
- #56
Welp I was able to stabilize it finally. Seems like the "sweet spot" is 4.4GHz @ 1.418V. Unfortunately 4.5GHz would need 1.5V or higher to be fully stable and since the user is gaming 7-8 hours a day and got the chip "second-hand" it wouldn't really last long. Left the memory timings at 11-9-9-27-39. Tried to tighten them a bit (my first attempt was 8-9-8-27-39 but got rounding error in P95 after 3 minutes or so)
- Joined
- Apr 20, 2004
- Location
- JAX, Mississauna
Well stable is stable even if it is a speed short of what one would hope for. Might be shett luck of the draw on the cpu or still just how that Rev 3.0 is working. Most of the one's we saw with the Rev 3.0 just had to settle for a lesser clock. Not really a bad thing since your bud does not seem real computer literate. Upped overclocks take a little looking at now and again.
Good luck to the both of you and some happy gaming.
RGone...ster.
Good luck to the both of you and some happy gaming.
RGone...ster.
- Thread Starter
- #58
Yea seems like it. Well, he's satisfied anyway.. He was gaming on a FX4130 for 10 months so this is actually a really big boost for him. I was actually hoping to get that chip higher tbh. I was able to get my brother's 6100 at 4.5Ghz @ stock voltage and since he ditched it,I got his 6350 at 4.9GHz@ 1.389V without breaking a sweat. Both chips sitting on an Asus m5a99fx pro rev.2
The specific MSI mobo is not really loving I suppose. Neither is the RAM! Seems like it won't let me tighten the timings at 1600MHz at least. Or I might need to take each timing down a notch, really slowly and patiently to see what's causing the trouble.
Similar threads
