Boot failure from CPU deadlock at 4.5GHz 5820K
Boot failure from CPU deadlock at 4.5GHz 5820K
My thermals perform well but I experience a system crash at 4.5GHz within one minute during stress tests. Right now the stable frequency is 4.375GHz. CPU Vcore is 1.270v with an LLC level of 5. The thermals show idle temperatures between 24-31°C under stress, but reach a maximum of 74°C. I play games such as Just Cause 3 and GTA5, yet rarely exceed 59°C after extended gaming sessions.
I feel I have sufficient headroom to reach 4.5GHz, but whenever I try I get a BSoD with a CPU Watchdog_Deadlock error from Win10 x64.
Steps I attempted to reach 4.5GHz:
1. Gradually increased the Vcore by 0.5v starting from 1.270v and stopped at 1.29v, refusing to go to 1.3v for minor improvements.
2. After each BSoD, I disabled the LLC controller, reducing it from level 5 down to 3 then to 1 (which is off), but the issue persisted.
All BIOS information appears normal. I’m unsure why this happens, as I’m conducting stress tests while writing this.
Tools used include AIDA64, RealBench, and Memtest86+ for hours without problems, but once I increased settings to 4.5GHz, everything broke.
Any suggestions I might have overlooked? Full specs included in the note.
I finally understood a few things.
I achieved a CPU Cache speed of 3500MHz and passed stress tests.
I discovered that my RAM at 3000MHz with a 5820K cannot adjust the BCLK beyond 125, which is just a limitation affecting all RAM and MOBOS models. Everyone I've met with a 5820K + 3000MHz RAM had to reduce it to either 2666 or 2444MHz to allow manual BCLK tuning for better CPU performance.
That's the situation.
I'm stable with:
CPU at 4375GHz @1.270v, vcore multiplier 35x, 125BCLK
CPU Cache at 3500MHz, voltages remain on auto
LLC set to lv3, but I might be fine at lv5 or lv1
RAM at 3000MHz with speeds 14-15-15-34 at 2T @1.35v
I just wish my memory read/write/copy consistency matched other overclockers
AIDA64 shows me a READ speed of 44GB/s WRITE...
Have you experimented with 1.3 volts to check its functionality? A jump from 1.29 to 1.3 isn't too drastic, though I get your concern.
Alternatively, you might consider a chip that doesn't overclock well. The silicon selection is quite unpredictable.
what freq and multiplier are your trying to get 4.5ghz?
also remember voltage increase is not linear. so to go 100mhz faster may use an extra .001 volt but to go 200hmz faster may be .003 and 300mhz may require .004
at this point you are far away from the base freq by almost 1ghz so .02 volts your trying might not be enough to gain 125mhz. you also could be at the end where no more voltage will gain more speed depending on the cpu lottery you got.
just also make to check your cpu input voltages/load lines some times a small bump up there can help. theres no clear cut answer because every oc is diffrent
Have you experimented with 1.3 volts just to check? A jump from 1.29 to 1.3 isn't too drastic, though I get your concern. You might also consider using a chip that doesn't overclock well—the silicon lottery is tough.
Yeah, I'm willing to try that. I've heard online that 1.3v is safe for long-term use with a 5820K. So far, my thermals are performing well considering the overclocking and the mid-range water cooler I have. I'll test 1.3v and if it causes any issues, I won't try more. Deadlocks seem to be a big problem for the CPU.
I'll get back to you in a few minutes.
Maxwellmelon is asking about the frequency and multiplier needed to reach 4.5ghz. He notes that voltage increases don’t rise linearly, so a small extra voltage might be necessary for higher speeds. He mentions that going much faster could require adjustments like an additional .001 volt or more. He also points out that being close to the base frequency means the current voltage boost might not be sufficient, and that the final speed limit depends on the CPU model. He suggests checking the CPU’s input voltages and load lines occasionally, as a slight change can help. He clarifies that there’s no single correct answer since each OC is different. He shares that for 125MHz a voltage around 1.3V is a big jump and could cause noticeable frame drops, while 4.25GHz was smoother, and he’s testing if 4.5GHz is feasible.
maxwellmelon :
what freq and multiplier are your trying to get 4.5ghz?
also remember voltage increase is not linear. so to go 100mhz faster may use an extra .001 volt but to go 200hmz faster may be .003 and 300mhz may require .004
at this point you are far away from the base freq by almost 1ghz so .02 volts your trying might not be enough to gain 125mhz. you also could be at the end where no more voltage will gain more speed depending on the cpu lottery you got.
just also make to check your cpu input voltages/load lines some times a small bump up there can help. theres no clear cut answer because every oc is diffrent
It just won't hit 4.5
I upped to 1.3v and it lasted 3m in stress vs 1m but BSoD
Then I dropped to 1.297v vcore but upped the CPU cache from 1.15 to 1.25v and lasted 3m til BSoD
I don't understand, my thermals are not the issue so what is?
BCLK mode set to high
LLC off (lv1)
Vcore tried 1.297 and 1.3
SA offset 0.409 (default)
CPU Cache voltage upped from 1.15 to 1.25
RAM slightly upped from default 1.35v to 1.38v
Only thing I can think of is dropping RAM from 3000 to 2400MHz which is not helpful for what I need
Anything else I can do? It really looks like undervoltage is the BSoD but isn't going past 1.3v crazy?
What do I even gain from a 125MHz increase? I saw noticeable improvement in screen recording from 4GHz to 4.375GHz + dropping RAM timings from 15-17-17-35 to 14-16-16-34 but is 125MHz worth it by itself? I'm wondering if the RAM is my weakest link since writing 4K raw data is bandwidtrh intensive but dropping to CL13 just can't happen, I've tried and failed Memtest utterly with 0 errors after hours at CL14 to 1000s of errors within seconds at CL13
Have you attempted 1.3 volts just to check? A jump from 1.29 to 1.3 isn't too drastic, though I get your concern. You might also consider a chip that doesn't overclock well—silicon is really inconsistent. I've shared again in case you need more tips.
What is your BCLK setting? Are you using values above 100? Occasionally, varying BCLK and CPU multipliers can influence overclock stability.