Swapped i5 for i7, could it be faulty?

Phil87 · 4 Dec 2018 at 08:51

Edit: I found that the following steps helped to fix this problem (there was nothing wrong with the CPU):
- Check stability with Memtest86, Prime95 etc
- Reset BIOS settings to default
- Reinstall Windows

Original post:
Hi all, I replaced my i5 3570K for an i7 3770 and I've had a couple of stability issues which make me wonder if the new CPU could be faulty. The main reason for the upgrade was that the i5 was struggling with Shadow of the Tomb Raider, but it's noticeably smoother with the i7. The full spec is in my signature, and nothing is overclocked unless you count enabling XMP in the BIOS.

I've seen the game crash to desktop a few times, mainly when when it loads a new area or autosaves, but I had a couple of similar errors before the CPU swap when returning to the main menu, so I'm not sure if it's the game which is to blame. I've managed to play it for a few hours over the last couple of days with no more issues though.

Most worryingly my PC completely locked up around a day after the upgrade - the screen froze and nothing would respond, so I had to press the reset button which actually turned the system off, turned it back on and it seemed like nothing was wrong after that. This was while watching a video on TouTube and with Steam downloading a game in the background, so nothing too taxing on the system. There was nothing in Event Viewer or BluescreenView to give any info about the cause of the freeze.

Also I ran an older version of Prime95 and it was giving rounding errors on basically every thread, at around the same time which makes me think that could be a false positive based on the info on their website. I ran it again with the latest version and there were no errors after around an hour (these were just running the default blend test). I also ran the Windows memory diagnostic, and Memtest86 for around an hour (one pass) with no errors.

Does any of this point to the i7 being faulty?

Phil87 · 10 Dec 2018 at 08:33

I think I've solved this one myself - the problem was my misunderstanding of what happens when you enable XMP in the BIOS. I noticed yesterday that this also changes the turbo ratio settings for the CPU, which allows all four cores to go up to 3.9, whereas with default settings they can individually hit 3.9 but with all together they only go up to 3.7. So my mistake of blindly enabling XMP effectively overclocked the CPU slightly, and although it seemed to increase the core voltage slightly it clearly wasn't enough.

All I had to do was set the turbo ratio back to auto, and now all seems fine although with a ~5% reduction in CPU performance, but that doesn't seem to be a noticeable difference. I think the lesson to take from this is to understand exactly what a BIOS setting does before changing it :rolleyes:

pandem0nium · 10 Dec 2018 at 09:02

You should have the option to bump the all core boost up to 4.1 with that cpu on a z77 board. I could with mine anyway.

Phil87 · 10 Dec 2018 at 11:13

Yeah I might have a look at bumping it up a bit, but for now I'll leave it at stock to make sure the problem is actually fixed. Yesterday before I made the change it was failing Prime95 within minutes, although a few days ago it ran for an hour with no errors. If I can go for a week or 2 without any game crashes I'll be happy that it's fixed.

Phil87 · 1 Jan 2019 at 15:59

Happy new year everyone

. I've had some more issues with this today, thought it was stable but had a complete lockup this morning while watching YouTube. The system completely froze and wouldn't restart, it wouldn't even respond to the reset button so I had to hold in the power button to turn it off.

After doing some digging I found out that enabling XMP actually overclocks the memory controller on the CPU, so that might explain why the system still wasn't stable. Maybe the IMC on my old 3570K was better than the one in this 3770, does that make sense?

MassiveJim · 3 Jan 2019 at 15:48

what cooler are you using ?
what temps are you hitting under load ?
what voltage are you running CPU at ?
what ram are you running ?

is the chip unfettled ? (ie not lapped or delidded etc)

Those chips do run fairly hot but should be good for a decent stable overclock , so running 5/10% o/clocks should not be troubling the CPU at all.

Phil87 · 14 Jan 2019 at 22:24

Sorry for not replying, I tried entering the memory settings manually instead of XMP which seemed to make it stable, for over a week anyway. I just had another blue screen today while watching Youtube and got errors in Prime95, so set all BIOS settings back to defaults (1333 MHz RAM). The cooler is a Hyper 212 Evo and temps barely hit 60 under load, haven't touched the voltage. The RAM is Corsair Vengeance DDR3 rated at 2133 MHz. The 3770 is non-K but is 2nd hand, so I don't know its history but assume it wasn't tampered with, probably came from an office PC or something.

slim01 · 16 Jan 2019 at 18:25

pandem0nium said:
You should have the option to bump the all core boost up to 4.1 with that cpu on a z77 board. I could with mine anyway.

mine used to hit 100 degrees after 10 seconds of prime with a 4.1 OC and a h80i... Hence my delid. From 100deg max down to 57 - biggest drop in temps I will probably ever see.

LadFromWales · 19 Jan 2019 at 00:33

wow 100c! I just popped round my brothers as his 3770K system was being unstable and he went and messed around in the bios and made stuff worse.
Short story is I reconfigured it how it was before but with a tiny bit more CPU voltage (hardware is getting on a bit now, I imagine the power delivery isn't as tight as it once was...), XMP2133 with 1T at 1.65v and the CPU all core turbo to 4.5GHz with a +0.125 offset, bringing the load voltage to just over 1.25v...max temp I saw while running prime was 78c! Cooler is the 212...

pingu666 · 19 Jan 2019 at 04:21

you should run memtest86, i had a ramstick go faulty recently which surprised me

Phil87 · 19 Jan 2019 at 19:41

I ran Memtest for 1 pass when I first discovered this problem and had no errors. I know that's not enough to prove the memory is stable but it certainly seemed to be more likely to cause errors in Prime95 when XMP was enabled, so at least I can get somewhat consistent results from running that. No issues so far with default BIOS settings, so I'm going to keep it like that for now.

I'm hesitant to increase the voltage, partly because the system was unstable at basically no load i.e. playing a YouTube video. Would increasing the voltage offset have any effect when the CPU isn't under load?

Phil87 · 3 Feb 2019 at 22:30

I'm still having occasional crashes, a couple last last weekend and a blue screen just now. The blue screen code was 0x00000124, caused by driver hal.dll, crash address ntoskrnl.exe+1b3440. One thing I've noticed every time it crashes is that in Event Viewer there is always a "EnhancedStorage-EhStorTcgDrv - A TCG Command has returned an error" one second before a "Kernel-Power" error. Could that mean this is a hard drive issue?

PiKe · 3 Feb 2019 at 22:37

Any minidump files in c:\Windows to look at?

slim01 · 3 Feb 2019 at 22:53

Increasing the voltage will increase stability. Regardless, you should only really run your memory at rated volts anyway.
What memory is it? What voltage and speed is it rated at?

Not all memory will overclock, even to its rated speed on some systems. And that doesn't even mean it's faulty - is luck of the draw.
Did it used to run at xmp? It shouldn't matter if its old or not. Ram doesnt really age like other parts with moving parts (hdd, optical drive, psu etc).

I would run a proper run of memtest first. And test at stock settings, THEN oc using manual or xmp. Xmp is an overclock don't forget

Phil87 · 3 Feb 2019 at 23:08

Here is the minidump: https://drive.google.com/file/d/1YT3CstXBsGzYPmqH1OxiSpLf3QeDEUH5/view?usp=sharing

Here is the memory info from CPU-Z: https://drive.google.com/file/d/19aCsdoXIj_Y-3EF01KNMyffKd7A3NgNe/view?usp=sharing

All the BIOS settings are at default, so I'm not looking at XMP or anything until I can figure out what the problem is.

Phil87 · 5 Feb 2019 at 08:11

I had another freeze yesterday evening, it wouldn't restart by itself and there was no blue screen, just the usual storage error following by kernal power error in event viewer. I also noticed the HDD activity LED was on constantly while it was frozen.

I ran the latest version of Memtest86 overnight, it took around 4 hours to do the default 4 passes with no errors, is that enough to assume the memory is ok? I had to find that in the log file after restarting because there was no signal on the monitor when I checked on it this morning. I've seen that happen before with Memtest so can only assume the GPU went to sleep after so long with nothing changing on the screen, unless that could point to a problem with the graphics card. The keyboard responded to num lock so the PC didn't freeze, it just didn't display anything at that point.

Also ran Prime95 overnight around a week ago, which went for about 9 hours with no errors.

MartinPrince · 5 Feb 2019 at 14:37

Are you using the 2501 bios for your motherboard?
You should note the 2133Mhz is not officially supported by the CPU 1600Mhz is the max but the motherboard does some jiggery pockery to 'support' up to 2400Mhz.

Try the Intel chipset drivers from the manufactures website if that doesn't' help then also try the latest ones on Intel's site.

Phil87 · 5 Feb 2019 at 17:18

Yes I have the latest BIOS, and the memory is just running at the default 1333Mhz for now. I haven't tried updating chipset drivers yet, can they be downloaded directly from Intel or just from the mobo manufacturer?

MartinPrince · 5 Feb 2019 at 23:50

Phil87 said:
Yes I have the latest BIOS, and the memory is just running at the default 1333Mhz for now. I haven't tried updating chipset drivers yet, can they be downloaded directly from Intel or just from the mobo manufacturer?

They will probably be different version drivers with the Asus drivers being older. I would try those first and run with them for a while and if still no joy try the newer ones from the Intel website.

Vince · 6 Feb 2019 at 00:02

Hal error so hardware abstract layer. Did you reinstall Windows after the cpu change? I wonder if Windows is having issues with a change in the number of cpu cores?

Could also be a hard disk issue as well.