- last edited on by
ROGBot
I don’t have any question or problem I’m asking for help about, just a story to tell. The on-topic point is that the Maximus VI Formula lets me do more in the way of stable overclocking at either end of the speed spectrum than my first motherboard.
I think that Haswell CPUs fall into four bins for overclocking:
Tigers we all hope for can run at 4800MHz 24/7 and reach 5 GHz when pushed.
Horses, whether you think of thoroughbred or Clydesdale, are a little slower, but they work.
Dogs are most numerous, run at 4400MHz and might get to 4600 with a push and good cooling.
Turkeys … ‘nuf said.
I’ve had one of each. My first is a dog. The second, a turkey, went back to the store. I sold a horse on ebay. My fourth try got me a very low-voltage tiger -- 4600MHz unconditionally stable at 1.23Vcore. It ran my mixed-make DDR3-2400 with no trouble. The integrated graphics core ran at 1700MHz with 1.19v. The problem was that it wouldn’t go faster, giving 101-code bluescreens when voltage was raised high enough to try 4800MHz. I wanted to try higher bclk straps to go easier on the PLLs. The A****k motherboard would not set the integrated graphics clock multiplier low enough to run with a high bclk. That CPU deserved a better motherboard.
I got the M6F about two weeks ago. Within a day, I had the tiger purring in Realbench 4600MHz at 1.200Vcore and unconditionally stable at 1.22v. With the help of overclocking guides and advice I read on these forums, I found I had been running the system agent voltage too high. Reducing that let the memory run with tighter timing. The realbench submission appears to be the highest score for a 4770K and integrated graphics. Something was working right.
Next step was 4800MHz. Don’t know what I did—too big a step at once, typo in a voltage value, entry in the wrong box – whatever, it ran amok part way through applying some changes. The tiger died. Q code 00.
I had used the dog as a learning device for delidding. It’s still around, so I dropped it in. The M6F went most of the way through POST, but stuck at A2 IDE detect and had the VGA Q led lit. SATA ports and the integrated display adapter are in the Z87 Southbridge. The disaster had taken out the motherboard. Fortunately it happened on day 14 - a day before the end of the return period for Microcenter. I got another M6F right away.
The dog works as it always did at 4400MHz, 1.33 Vcore. It took a day to get the dog to 4600MHz which the old motherboard could never do. Graphics clock and DRAM timing are better, too. The dog needs 1.425 Vcore for Realbench and 1.450 v for unconditional stability. That’s a big step in voltage and gives the water cooling loop a real workout.
Now I’m back in the Haswell lottery because this motherboard deserves a better CPU. Wish me luck.
Jeff
I think that Haswell CPUs fall into four bins for overclocking:
Tigers we all hope for can run at 4800MHz 24/7 and reach 5 GHz when pushed.
Horses, whether you think of thoroughbred or Clydesdale, are a little slower, but they work.
Dogs are most numerous, run at 4400MHz and might get to 4600 with a push and good cooling.
Turkeys … ‘nuf said.
I’ve had one of each. My first is a dog. The second, a turkey, went back to the store. I sold a horse on ebay. My fourth try got me a very low-voltage tiger -- 4600MHz unconditionally stable at 1.23Vcore. It ran my mixed-make DDR3-2400 with no trouble. The integrated graphics core ran at 1700MHz with 1.19v. The problem was that it wouldn’t go faster, giving 101-code bluescreens when voltage was raised high enough to try 4800MHz. I wanted to try higher bclk straps to go easier on the PLLs. The A****k motherboard would not set the integrated graphics clock multiplier low enough to run with a high bclk. That CPU deserved a better motherboard.
I got the M6F about two weeks ago. Within a day, I had the tiger purring in Realbench 4600MHz at 1.200Vcore and unconditionally stable at 1.22v. With the help of overclocking guides and advice I read on these forums, I found I had been running the system agent voltage too high. Reducing that let the memory run with tighter timing. The realbench submission appears to be the highest score for a 4770K and integrated graphics. Something was working right.
Next step was 4800MHz. Don’t know what I did—too big a step at once, typo in a voltage value, entry in the wrong box – whatever, it ran amok part way through applying some changes. The tiger died. Q code 00.
I had used the dog as a learning device for delidding. It’s still around, so I dropped it in. The M6F went most of the way through POST, but stuck at A2 IDE detect and had the VGA Q led lit. SATA ports and the integrated display adapter are in the Z87 Southbridge. The disaster had taken out the motherboard. Fortunately it happened on day 14 - a day before the end of the return period for Microcenter. I got another M6F right away.
The dog works as it always did at 4400MHz, 1.33 Vcore. It took a day to get the dog to 4600MHz which the old motherboard could never do. Graphics clock and DRAM timing are better, too. The dog needs 1.425 Vcore for Realbench and 1.450 v for unconditional stability. That’s a big step in voltage and gives the water cooling loop a real workout.
Now I’m back in the Haswell lottery because this motherboard deserves a better CPU. Wish me luck.
Jeff
So I tried the Haswell lottery again. I might lose – get a CPU worse than my trusty dog – or I might win – get a horse or a tiger. Little did I dream I would lose so badly.
The M6F did the normal stutter boot, but gave Q-code 00 on the second try. It didn’t even start POST the next time I tried to start. The PSU shut down.
The new CPU was not only DOA, it was poison. It killed my M6F and my original non-Asus MB when I tried to troubleshoot. A spare PSU and my “known good” dog behaved the same way with both MBs.
I returned and exchanged the CPU. The local Microcenter was out of M6F and M6E stock. I settled for a Hero that someone else had returned – cheaper, but … The newer CPU, the Hero and both PSUs work properly. The dog doesn’t. The poison CPU killed my dog.
The RMA hasn’t come through yet, but I already miss the Formula. Fully Manual Mode, Latency Boundary and swizzling are BIOS features I missed immediately. Audio is known to be less good. Displayport video connection ... the list of differences is long. The added price does buy something in this case.
The newer CPU isn’t as good as the dog – 4.2GHz at 1.24Vcore and hits 100C throttling before reaching stable voltage for 4.3. I won’t be advancing my Realbench score or doing any more Rookie Rumbles for a while. This CPU isn’t delidded and isn’t worth the effort or risk. I intend to dispose of it when I get a better one. Ebay is my friend.
I miss my M6F. The dog – not so much.
Jeff
The M6F did the normal stutter boot, but gave Q-code 00 on the second try. It didn’t even start POST the next time I tried to start. The PSU shut down.
The new CPU was not only DOA, it was poison. It killed my M6F and my original non-Asus MB when I tried to troubleshoot. A spare PSU and my “known good” dog behaved the same way with both MBs.
I returned and exchanged the CPU. The local Microcenter was out of M6F and M6E stock. I settled for a Hero that someone else had returned – cheaper, but … The newer CPU, the Hero and both PSUs work properly. The dog doesn’t. The poison CPU killed my dog.
The RMA hasn’t come through yet, but I already miss the Formula. Fully Manual Mode, Latency Boundary and swizzling are BIOS features I missed immediately. Audio is known to be less good. Displayport video connection ... the list of differences is long. The added price does buy something in this case.
The newer CPU isn’t as good as the dog – 4.2GHz at 1.24Vcore and hits 100C throttling before reaching stable voltage for 4.3. I won’t be advancing my Realbench score or doing any more Rookie Rumbles for a while. This CPU isn’t delidded and isn’t worth the effort or risk. I intend to dispose of it when I get a better one. Ebay is my friend.
I miss my M6F. The dog – not so much.
Jeff
