Degraded 4770K

[Kougar]

Either I was unlucky, or these chips are more fragile than is commonly accepted. I thought 1.26v was safe given how many people are running 1.3v or higher (even >1.4v for the XS folks), but apparently not.

A few weeks ago I began getting the typical "clock interrupt not received on a secondary processor" BSoD, so I backed off my OC. Did this until I was at 4Ghz @ 1.18v, but earlier today it happened again. Pretty sure the CPU is well on its way to failing outright.

For technical details I'm running a 3x140mm radiator with an Apogee XT and MCP655, and a GTX 480 in the same loop. My 24/7 settings were 4.2Ghz @ 1.26v, with full system load temps varying between 60-75c due to room temps. Idle was in the 30's. Was one of the horrible launch day batches as the chip maxed out at 4.3Ghz with 1.3vcore, didn't like the temps so I backed off to the above settings.

I'm no stranger to overclocking... my first chip was a 2.8Ghz Northwood that I ran up to 3.4Ghz for years. That 10 year old system still worked perfectly right up until I sold it this year. Upgraded in 2006 to a launch day E6300, which I ran at a 100% OC (3.8Ghz) for 24/7 folding@home purposes for a few years. That chip never degraded an iota and is still in use as my father's PC with a mild 3Ghz air OC on it.

Out of the dozen chips I've OC'd this is the first I've ever personally had degrade, and it's only four months old!

 

[chimaxi83]

Good thing you bought the $25 Performance Tuning Plan

 

[BallaTheFeared]

What are you doing with your uncore, and what is the BCC code?

101 is generally uncore related, I'm pretty sure the error is 101, but if it isn't let me know.

That's really now how they fail typically.

 

[Rvenger]

It was probably unstable from the start. This happened to my other Haswell setup. It ended up being the board not able to provide stable voltage between idle and load situations.

 

[Techhog]

It was probably unstable from the start. This happened to my other Haswell setup. It ended up being the board not able to provide stable voltage between idle and load situations.

And there goes the theory about the board not mattering much for Haswell OCs.

 

[Kenmitch]

And there goes the theory about the board not mattering much for Haswell OCs.

It's more like AS much with Haswell.

Sounds like he did get a dud chip but I guess the only way to know would be to test it on another rig to verify.

The OP left a lot of things up to our imagination as far as other voltages, settings go....Kind of hard to determine what the issue is without more data.

 

[Kougar]

The motherboard is a Gigabyte Z87-UD5H. I liked several of the features, but it also is one of just a handful of motherboards using the top quality IR3550 VRM chips. My E6300 experience taught me the value of having a higher quality VRM/PWM assembly as that made the difference between 3.5Ghz and 3.8Gh for stable 24/7 use.

What are you doing with your uncore, and what is the BCC code?

101 is generally uncore related, I'm pretty sure the error is 101, but if it isn't let me know.

That's really now how they fail typically.

Huh, I didn't know that. Can't speak for all of them, but the most recent BSoD was indeed a 101 code.

I played around with uncore settings but when I gave up pushing frequency I ran the uncore at 1:1 parity. The uncore on my processor was already extremely weak, as all of my OC attempts often resulted in a "cache hierarchy error" or "internal parity error" notes in the Windows event logs. When I raised the uncore to 4Ghz I had to bump it to somewhere around 1.18v to get rid of those event log errors. I then had to nudge the Ringbus voltage up to 1.24v to get the chip stabilized at 4.2Ghz at 1:1 frequencies.

The OP left a lot of things up to our imagination as far as other voltages, settings go....Kind of hard to determine what the issue is without more data.

What other settings are ya looking for Kenmitch? Everything was vcore and Ringbus. Tried a VRIN of 1.9v and even 2v briefly but eventually figured out that didn't help anything so I left it at the stock 1.8v. VRIN loadline calibration was a mix between Auto and low settings for 24/7 use as I don't approve of the voltage overshoot risk disabling it entails at lower process nodes. I eventually raised it to medium after the OC destabilized but that didn't help. Memory is 32GB of half-height DDR3-1600 @ 1.35v.

Given 4Ghz is no loner stable I'm at stock frequency 1.18vcore, 1.1 ring, LLC Auto now with fingers crossed. Terrible time to RMA the chip as I don't have any other Haswell platform hardware. Been contemplating just getting a vanilla 4770 for VT-d + TSX support now, and selling the 4770k when it comes back from RMA...

 

[Kenmitch]

The motherboard is a Gigabyte Z87-UD5H. I liked several of the features, but it also is one of just a handful of motherboards using the top quality IR3550 VRM chips. My E6300 experience taught me the value of having a higher quality VRM/PWM assembly as that made the difference between 3.5Ghz and 3.8Gh for stable 24/7 use.



Huh, I didn't know that. Can't speak for all of them, but the most recent BSoD was indeed a 101 code.

I played around with uncore settings but when I gave up pushing frequency I ran the uncore at 1:1 parity. The uncore on my processor was already extremely weak, as all of my OC attempts often resulted in a "cache hierarchy error" or "internal parity error" notes in the Windows event logs. When I raised the uncore to 4Ghz I had to bump it to somewhere around 1.18v to get rid of those event log errors. I then had to nudge the Ringbus voltage up to 1.24v to get the chip stabilized at 4.2Ghz at 1:1 frequencies.



What other settings are ya looking for Kenmitch? Everything was vcore and Ringbus. Tried a VRIN of 1.9v and even 2v briefly but eventually figured out that didn't help anything so I left it at the stock 1.8v. VRIN loadline calibration was a mix between Auto and low settings for 24/7 use as I don't approve of the voltage overshoot risk disabling it entails at lower process nodes. I eventually raised it to medium after the OC destabilized but that didn't help. Memory is 32GB of half-height DDR3-1600 @ 1.35v.

Given 4Ghz is no loner stable I'm at stock frequency 1.18vcore, 1.1 ring, LLC Auto now with fingers crossed. Terrible time to RMA the chip as I don't have any other Haswell platform hardware. Been contemplating just getting a vanilla 4770 for VT-d + TSX support now, and selling the 4770k when it comes back from RMA...

I was wondering if you tried the other voltages is all. Does kind of sound like your chip is going south pretty quickly.

I'm running my 4670k at 4GHz 1.05v(fixed)

 

[BTRY B 529th FA BN]

Does "clock interrupt not received on a secondary processor" mean it's CPU related? I know it seems pretty obvious but this is new to me. The first issue that comes to me is the RAM being or going bad/degrading.

 

[BallaTheFeared]

Huh, I didn't know that. Can't speak for all of them, but the most recent BSoD was indeed a 101 code.

I played around with uncore settings but when I gave up pushing frequency I ran the uncore at 1:1 parity. The uncore on my processor was already extremely weak, as all of my OC attempts often resulted in a "cache hierarchy error" or "internal parity error" notes in the Windows event logs. When I raised the uncore to 4Ghz I had to bump it to somewhere around 1.18v to get rid of those event log errors. I then had to nudge the Ringbus voltage up to 1.24v to get the chip stabilized at 4.2Ghz at 1:1 frequencies.

Just do 3.6 or 3.8 or even lower if you have to for the uncore, it doesn't make a huge difference in performance but with mine it makes a rather large difference in vcore required.


For me I can do 4.8/4.8 around 1.325 on the vcore for my required stability (no crashing in gaming). With 4.8 core and 4.4 uncore I can do 4.8GHz on the core with just 1.24v... So for me anyways, it makes a huge difference.

Uncore can even make your core unstable with 124 (more vcore) errors, it's just not worth it.

 

[Zoeff]

Interestingly I have the same exact motherboard and also bought it with a 4770k on release day. I'm running it at 4.5Ghz core with 1.272v and 4.2Ghz uncore and with basically every intel/mobo feature disabled except hyperthreading. It never downclocks when idling and the voltage is stuck at 1.272v. It's also always on during the daytime, even if I leave the house. My sample actually gives me the occasional CPU related BSoD when I reset the bios to the defaults. I tried to RMA it because of that but they couldn't reproduce the blue screens... :/

Instead of writing down what settings I use I'll just give you some screenshots instead

I hope these settings give you some idea's on what to try, assuming you haven't already touched all these things already.

 

[Kougar]

Keep in mind I had the chip stable at 4.2Ghz for two months. After that destabilized I backed off to 4Ghz core/uncore and after a week or two, THAT began giving me the same BSoD even with 1.18 vcore / uncore settings.

Does "clock interrupt not received on a secondary processor" mean it's CPU related? I know it seems pretty obvious but this is new to me. The first issue that comes to me is the RAM being or going bad/degrading.

It's always meant insufficient Vcore. In the past one could easily pick up errors with Prime95 or Linpack tests before this specific BSoD would show up. But now with Haswell it just either works, or it doesn't. Never any middle ground anymore...

Zoeff, thanks for the screenies. Nice to see your HD interface works, mine still doesn't which ticks me off.

I will play with a few of those settings and see what the chip does, but honestly if 4Ghz isn't stable at 1.18v anymore then I don't think it would help. I've locked it to stock frequency @ 1.18v just to keep it stable. My current plan is to try and snag a 4770 during Black Friday, then I'll RMA this chip...

 

[BallaTheFeared]

Keep in mind I had the chip stable at 4.2Ghz for two months. After that destabilized I backed off to 4Ghz core/uncore and that was no longer stable even with 1.18 vcore / uncore settings.

Try resetting everything and starting over.

Also verify stock is stable, if it isn't... RMA time

 

[Zoeff]

Zoeff, thanks for the screenies. Nice to see your HD interface works, mine still doesn't which ticks me off.

I will play with a few of those settings and see what the chip does, but honestly if 4Ghz isn't stable at 1.18v anymore then I don't think it would help. I've locked it to stock frequency @ 1.18v just to keep it stable. My current plan is to try and snag a 4770 during Black Friday, then I'll RMA this chip...

The HD interface magically started working when I upgraded from a GTX580 to a GTX780. I wonder what exactly the difference is between these 2 cards that made the BIOS finally accept it...

Oh even at 4Ghz it's unstable now? Dang, that's degrading pretty fast then. Best of luck with your future 4770. :thumbsup:

 

[Kougar]

Balla, I'll give it a shot tomorrow. Was going to flash the backup UEFI as well just to be sure. Not sure I WANT to try stock voltages I don't want to be without a computer during all the Halloween stuff that's going to go on else I'd RMA it right now!

The HD interface magically started working when I upgraded from a GTX580 to a GTX780. I wonder what exactly the difference is between these 2 cards that made the BIOS finally accept it...

Oh even at 4Ghz it's unstable now? Dang, that's degrading pretty fast then. Best of luck with your future 4770. :thumbsup:

I use a 480, so I guess GB has problems with older NVIDIA hardware... I'm disappointed to say the least because there's no technical reason for it.

I made this thread after the 4Ghz OC began failing, that's what kinda sealed it for me that the chip itself was degrading. Thanks for the wishes... if I get a vanilla 4770 then as long as Hyper-V has no trouble with VT-d on my board I'll be happy.