Not all is well in SKY OC land.

[VirtualLarry]

Had reason to reboot my G4400 @ 4.455 / PCI-E M.2 SSD rig, after about a month.

Monitor said "no signal".

I was like, hmm.

Had to hit the reset button, and waited, and waited, and eventually it beeped, got a BIOS message that it had failed booting several times, and to hit F2 to go into BIOS setup.

So I did, double-checked voltages and settings and stuff, all looked up. Saved and exited.

Booted Win7, showed up in low-res, like the video drivers weren't loading, and my USB mouse / keyboard wouldn't work.

Was very confused. Thought maybe RAM instability had caused a registry glitch, and it lost my hardware config.

Then I realized, I had a SATA6G SSD in the box too, and it tried to boot off of that, rather than my PCI-E SSD.

So I had to go back into BIOS setup, and fix the boot order. (Had to make "Boot Manager (PCI-E M.2 Samsung)" the first entry.)

So now it seems to be working, but I'm a little nervous. Hope everything is OK long-term.

 

[ehume]

I've had that happen loads of times through the years and with various versions of Windows. Sometimes if you have a USB inserted, you'll get an "Insert Boot Disk" message. You did exactly the correct thing. You should have no problems

 

[VirtualLarry]

Yeah, I guess in hindsight, the "SKY OC" feature isn't perfect, and sometimes it fails to boot properly.

One other glitch that I noticed was, my CPU temp in the BIOS reached 81C. And the CPU fan didn't spin right away. It was sort of jerking, a bit, then it finally started spinning.

There isn't really any dust on it, and it was installed new a few months ago. (It was purchased new on ebay. It came with the factory paste pre-applied.)

 

[Denithor]

I had an even more fun one recently. A system I set up for buddy at his shop quit booting, it would get to the Windows splash screen then reboot as soon as the splash animation started. Tinkered with startup options, nothing seemed to be wrong, scratching head. Then figured it out. Somehow in BIOS the SSD got switched from AHCI to RAID mode. Set back to AHCI, Windows booted right up, everything fine. No idea how that happened. Not an overclocked system. Kinda stumped. Told him what to do if it happens again.

 

[escrow4]

I stuffed a new GPU into my box (HA!) and no display, no nothing, no keyboard either. Was working fine on the initial boot then it decided not to. Cleared CMOS and no issues since. Mystery really. Although I don't overclock anymore. They are never truly stable.

 

[nerp]

I stuffed a new GPU into my box (HA!) and no display, no nothing, no keyboard either. Was working fine on the initial boot then it decided not to. Cleared CMOS and no issues since. Mystery really. Although I don't overclock anymore. They are never truly stable.

I feel the same way. Every time I've OC'd in the past I felt like I was walking while trying hard to not fart. Eventually, you fart. No matter what.

Every time I'd encounter a bug, or a program crashed, or I saw a tiny flicker somewhere, I'd think "is it the oc? Or just a bug?" and troubleshooting felt pointless because what if it was the OC? And usually it was the OC.

I've had stable OCs that ran fine for a long time though but even then, eventually, something goes wrong.

Today, the speed improvement you get is not worth it to me. It's not like the old Celeron days. Back then, OCing could mean the difference of getting $1,000 cpu performance for $200. Today, the difference is what, $200 performance for $100 + extra electricity and cooling and all that time wasted futzing with settings. Not a value proposition.

 

[ehume]

I feel the same way. Every time I've OC'd in the past I felt like I was walking while trying hard to not fart. Eventually, you fart. No matter what.

Every time I'd encounter a bug, or a program crashed, or I saw a tiny flicker somewhere, I'd think "is it the oc? Or just a bug?" and troubleshooting felt pointless because what if it was the OC? And usually it was the OC.

I've had stable OCs that ran fine for a long time though but even then, eventually, something goes wrong.

Today, the speed improvement you get is not worth it to me. It's not like the old Celeron days. Back then, OCing could mean the difference of getting $1,000 cpu performance for $200. Today, the difference is what, $200 performance for $100 + extra electricity and cooling and all that time wasted futzing with settings. Not a value proposition.

I agree with you in principle. However, the i7 4770k has a stock speed of 3.5GHz. While mine is stable at 4.5GHz, for thermal reasons I run it at 4.3GHz -- a decent OC while perfectly stable; no glitches at all. My i7 4790k has a stock speed of 4.0GHz. It runs at 4.5GHz. While one could say that is an improvement, one could also agree with you that the OC speed is a minimal improvement considering the risk of instability.

It looks like Intel is taking the speed capacity of it chips into consideration while setting stock speeds, so that will limit overclockability on average. It will be interesting to see the average overclockability of the Skylake chips. Looks good so far. If it is like Devil's Canyon, Kaby Lake will improve on that. But still, I agree with you that the days of 100+% OC are gone.

 

[hojnikb]

Thats what happens when you use unsupported features. I'm surprised intel didnt kill OC with forced microcode updates via windows update.

Besides, buying system around the idea of unsupported over clocking is maddening for me. I mean its great as a bonus, but nothing else. If you need the power, cough up some more $$$ and get the next best thing.
Besides, OCing always had a price, even in old days when OCing was unlocked by intel. Some people just fail to realize that.

 

[PhIlLy ChEeSe]

I wouldn't worry @4.4550 unless you were using crazy voltage. Have a Gigabyte now every time it boots, it seems to loss boot assignment.

 

[know of fence]

Today, the speed improvement you get is not worth it to me. It's not like the old Celeron days. Back then, OCing could mean the difference of getting $1,000 cpu performance for $200. Today, the difference is what, $200 performance for $100 + extra electricity and cooling and all that time wasted futzing with settings. Not a value proposition.

I haven't been much of an OC'er, myself. But wasn't the responsiveness of a system determined by the HDD or your internet connection back in the day anyway. In those dark ages before heatpipe cooling and PWM fans.
Now that GPU performance has grown much faster than CPUs, and we have SSDs and plenty of RAM, CPU clock matters more than it ever did in the past.

Fine tuning Voltage, offset undervolt and power saving features should make your PC more power efficient, long lived even with an OC enabled. So it's worth it for this alone. And the SL non K OC is a good stand in solution for hopefully less crappy 14nm chips to follow.

I think Turbo Overclocking should have been a bigger topic with quad, hex and octo-cores, instead of the brute force all cores, all max all the time approach. It's definitely a completely underrated solution for the power/performance/noise compromise. I'd like to see 4 to 8 core CPUs run at efficient 3.2 GHz and turbo up to 4.2+ on a single core. Instead of the comically small turbo of the 6700K.

 

[wpcoe]

I had an even more fun one recently. A system I set up for buddy at his shop quit booting, it would get to the Windows splash screen then reboot as soon as the splash animation started. Tinkered with startup options, nothing seemed to be wrong, scratching head. Then figured it out. Somehow in BIOS the SSD got switched from AHCI to RAID mode. Set back to AHCI, Windows booted right up, everything fine. No idea how that happened. Not an overclocked system. Kinda stumped. Told him what to do if it happens again.

Not to veer too far off topic (but I will... ()), I thought AHCI was a subset of RAID. If so, why would a changed setting to RAID interfere with booting?

 

[VirtualLarry]

Not to veer too far off topic (but I will... ()), I thought AHCI was a subset of RAID. If so, why would a changed setting to RAID interfere with booting?

It wouldn't, necessarily, IF you had previously installed the RAID drivers for the chipset in the OS. I'm guessing that they didn't do that, because they didn't have the need at the time the OS was installed.

 

[wpcoe]

Okay, that makes sense. I was just checking my understanding of the relationship of AHCI & RAID. Thanks!

 

[nerp]

I agree with you in principle. However, the i7 4770k has a stock speed of 3.5GHz. While mine is stable at 4.5GHz, for thermal reasons I run it at 4.3GHz -- a decent OC while perfectly stable; no glitches at all. My i7 4790k has a stock speed of 4.0GHz. It runs at 4.5GHz. While one could say that is an improvement, one could also agree with you that the OC speed is a minimal improvement considering the risk of instability.

It looks like Intel is taking the speed capacity of it chips into consideration while setting stock speeds, so that will limit overclockability on average. It will be interesting to see the average overclockability of the Skylake chips. Looks good so far. If it is like Devil's Canyon, Kaby Lake will improve on that. But still, I agree with you that the days of 100+% OC are gone.

Yeah, that is a decent OC, actually. Nearly an extra GHZ would probably be noticeable in some scenarios.

Aside from stability, my other concern is power consumption. I don't like disabling power management features and like the amazing progress that has been made in recent years in efficiency.

 

[escrow4]

Yeah, that is a decent OC, actually. Nearly an extra GHZ would probably be noticeable in some scenarios.

Aside from stability, my other concern is power consumption. I don't like disabling power management features and like the amazing progress that has been made in recent years in efficiency.

If you have an MCE mobo you could set all cores to 3.9GHz anytime with no voltage or fiddly increases.

 

[VirtualLarry]

If you have an MCE mobo you could set all cores to 3.9GHz anytime with no voltage or fiddly increases.

Well, yes and no. Oftentimes, the mobo will set MCE if you set XMP for RAM, and auto-boost the vcore of the CPU by some margin because of that.

I think that this forum gets (or used to get) a couple of new posters a month, wondering why their shiny new 4790K was at 80C+ under gaming load, and most of the time it's because of auto-enabled MCE. (The other few times, were poorly installed heatsinks.)

 

[Madpacket]

So I get the impression these Z170 bus speed over clocks may have more risk associated with them than simple unlocked multipliers? Is this inherit to all boards or just maybe AsRock's implementation?