My Experiences with OCing an i5 - 6400

[a3venom]

My first OCing experience with a CPU, extremely happy with the results so far

I got an Intel 6400 not so long ago, to experiment with the BCLK OCing, I was going to grab a used 3770k/2600k + Mobo setup but couldn't find a mATX mobo since all people use ATX for some reason.

You save around 80$ in Canada by buying 6400 instead of the K part.

Firstly, got a MSI Mortar Z170 and i5 6400 with a H60. Plugged it in and was easily able to get it at 4.4 Ghz around 1.35v, It would boot fine at 4.5 but wouldn't stay stable during stress test. It was stable at 4.5 @ 1.4v and I learnt about Vdroop and my CPU really only needed 1.35 to be stable but my mobo didn't have LLC.

Disappointed, I went to replace my Mobo and thought that I should try another time with the CPU since 4.5 ghz is good on a 2.7 ghz chip but maybe i could get better silicon lottery.

Returned the combo and got an ASUS Z170m Plus + 6400. It ran very well with 4.6ghz @ 1.3v and i just finished testing it at 1.34v / 4.7 Ghz (LLC lvl 6). That's almost 75% OC which I am pretty impressed with.


I am using OCCT 4.4.2's 2 tests to check for stability and my max temp was 82 with just OCCT and 83 with Furmark + OCCT linpack dumping my r9 290's heat into the case.

If i had a well ventilated case, I couldve got better temps with a similar priced AIR cooler but I am not sad with my H60 so far. I am using a Bitfenix Pandora which is sexy as fuck, so the tradeoff is worth it.


I am planning to Delid my CPU sometime later if I feel comfortable - as I read it drops temps quite a bit in majority of the cases.

 

[chameleoneel]

My first OCing experience with a CPU, extremely happy with the results so far

I got an Intel 6400 not so long ago, to experiment with the BCLK OCing, I was going to grab a used 3770k/2600k + Mobo setup but couldn't find a mATX mobo since all people use ATX for some reason.

You save around 80$ in Canada by buying 6400 instead of the K part.

Firstly, got a MSI Mortar Z170 and i5 6400 with a H60. Plugged it in and was easily able to get it at 4.4 Ghz around 1.35v, It would boot fine at 4.5 but wouldn't stay stable during stress test. It was stable at 4.5 @ 1.4v and I learnt about Vdroop and my CPU really only needed 1.35 to be stable but my mobo didn't have LLC.

Disappointed, I went to replace my Mobo and thought that I should try another time with the CPU since 4.5 ghz is good on a 2.7 ghz chip but maybe i could get better silicon lottery.

Returned the combo and got an ASUS Z170m Plus + 6400. It ran very well with 4.6ghz @ 1.3v and i just finished testing it at 1.34v / 4.7 Ghz (LLC lvl 6). That's almost 75% OC which I am pretty impressed with.


I am using OCCT 4.4.2's 2 tests to check for stability and my max temp was 82 with just OCCT and 83 with Furmark + OCCT linpack dumping my r9 290's heat into the case.

If i had a well ventilated case, I couldve got better temps with a similar priced AIR cooler but I am not sad with my H60 so far. I am using a Bitfenix Pandora which is sexy as fuck, so the tradeoff is worth it.


I am planning to Delid my CPU sometime later if I feel comfortable - as I read it drops temps quite a bit in majority of the cases.

View attachment 5273

Is your motherboard still running an old bios, allowing you to overclock the BCLK? Or are you able to do it some other way, with a newer bios?

 

[a3venom]

Is your motherboard still running an old bios, allowing you to overclock the BCLK? Or are you able to do it some other way, with a newer bios?

it came with a newer BIOS and i just flashed the old one, I didnt try it with the Bios it came with though

 

[rennyf77]

just picked up a used i5 6400 on the cheap to doink around with. i thought the bus overclocking bioses disabled reliable temp monitoring.

 

[a3venom]

just picked up a used i5 6400 on the cheap to doink around with. i thought the bus overclocking bioses disabled reliable temp monitoring.

Package temp is almost as reliable as core temp, you would not see any cases where package temp is really lower than the hottest core on Skylake in HWinfo or HWmonitor

 

[Araxie]

Package temp is almost as reliable as core temp, you would not see any cases where package temp is really lower than the hottest core on Skylake in HWinfo or HWmonitor

Package temp is based on Tcase instead of Tjunction so a software that really read Tcase probe temperature instead of an average of the four Tjunction temperature will show that package temps are always going to be lower than core temps, for most chips package temperature have a max between 62C and 72C, while Tjunction max temp of 100C they are and have entirely different thermal probes, and package temp (Tcase) should never and ever be higher than each individual core temps..

Ironically the Tcase max for the i5 6400 (72C) is greater than the for example i7 6700K (64C) this truly speak about the binning process for i7 chips, so yes in most case scenarios a chip reaching Tcase max will cause a more aggressive thermal throttling than Tjmax to the point it can just shutdown the machine as Tcase will always be very close to socket temp.

 

[a3venom]

Package temp is based on Tcase instead of Tjunction so a software that really read Tcase probe temperature instead of an average of the four Tjunction temperature will show that package temps are always going to be lower than core temps, for most chips package temperature have a max between 62C and 72C, while Tjunction max temp of 100C they are and have entirely different thermal probes, and package temp (Tcase) should never and ever be higher than each individual core temps..

Ironically the Tcase max for the i5 6400 (72C) is greater than the for example i7 6700K (64C) this truly speak about the binning process for i7 chips, so yes in most case scenarios a chip reaching Tcase max will cause a more aggressive thermal throttling than Tjmax to the point it can just shutdown the machine as Tcase will always be very close to socket temp.

If you lower your fanspeed and run prime or some other thermally intensive test and see if there is any delta between max core and package temp on HWmonitor? Because i removed the heatsink on my CPU to test this theory and found out Package was equal to or higher than highest core when running without heatsink at stock speed (to read core).

you got a 6700k easy for you to check and let me know please

 

[Araxie]

Hwmonitor read "package" as #Core 0 temperature, Open HwMonitor utilize Package as an average of all core temps, only app I found that read Tcase correctly as per own intel recommendation is Aida64... which show CPU, CPU package, Package and per core temp. is the only one able to read Tcase as far I found and yes my temps are about 15C up to 20C of difference between Tcase and Tjunction.