Overclock.net › Forums › Intel › Intel CPUs › Broadwell-E thread
New Posts  All Forums:Forum Nav:

Broadwell-E thread - Page 113

post #1121 of 7252
Quote:
Originally Posted by vibraslap View Post

Stop making me feel inadequate lol..

Seriously though, any tips? I couldn't even get it stable on XMP. Manual settings weren't stable either. Did I miss something maybe?
SA? Auto showed me I was playing games toying around with SA much lower than were going to be needed. With 128G it is ~1.2v on auto. From past experience, I expect slightly lower with 64 and lower again with 32/16.

I've left it on auto for now. One curious thing I saw was that Aida would show me stable at 0.9 SA manually, but stressapp in Linux would quickly find errors. I started ramping up 10mV at a time but ran out of time to tune so left it at auto for now.

Also, I found if SA was the ultimate deficiency, then more VDIMM often made it worse. So start with SA and then add VDIMM has been my experience with both HWE and BWE.

Ditto on the cache 3.7/3.8 wall and I need a big jump in volts to get to 3.8 stable and corresponding heat.

Filmguy, I am finding the BWE chip to run cooler for any given OC than HWE even though it has more cores, so I'd be surprised that you are seeing more heat than an even larger Si feature size, but I don't have a 4xxx to compare.

At least the ASUS bios is throwing plenty of volts at everything in auto which makes it work, but obviously produces more heat. Check your SA and cache voltages under load? If they look reasonable then I'd check for cooling loop issues.
post #1122 of 7252
FWIW, regarding SA, where I left off was a single bit error in an hour of stress app at auto -0.05. So 1.15-1.16v was almost enough for 128G.
Edited by cekim - 7/7/16 at 9:17am
post #1123 of 7252
1.45V on RAM, ouch! Won't it hurt the memory controller on the long run?
post #1124 of 7252
How reliable is the temp/voltage data from AISuite? It says that my CPUIN jump up to 1,98V under load with LLC on auto while it's set to 1,95 in BIOS.
post #1125 of 7252
Quote:
Originally Posted by Martin778 View Post

How reliable is the temp/voltage data from AISuite? It says that my CPUIN jump up to 1,98V under load with LLC on auto while it's set to 1,95 in BIOS.

As reliable as A/D usually are, which is to say only roughly unless you are talking about a very expensive circuit which these measurements aren't.

That sort of variance is typical. Even relatively tame compared to what I've seen on my systems.

Speaking only for myself, I've rarely found CPUIN to be the magic sauce that gets me to a stable OC with HW/BW. So it's almost always auto or among the last to be tuned.
post #1126 of 7252
Quote:
Originally Posted by cekim View Post

Filmguy, I am finding the BWE chip to run cooler for any given OC than HWE even though it has more cores, so I'd be surprised that you are seeing more heat than an even larger Si feature size, but I don't have a 4xxx to compare.

At least the ASUS bios is throwing plenty of volts at everything in auto which makes it work, but obviously produces more heat. Check your SA and cache voltages under load? If they look reasonable then I'd check for cooling loop issues.

Appreciate the input! Yeah, I've wondered about a potential issue in the loop, but I'm not sure what it could be. My GPU is hitting at about the temps I'd expect, so I would think that would help rule out any larger issues with the loop itself.

I guess the best way to boil my question down, for anyone who has done a similar switch recently, is: should I expect to be seeing a ~10-15 degree bump in temps going from an Ivy-E hex core to a Broadwell-E hex core on the same loop? Seems like that shouldn't be the case.
post #1127 of 7252
Quote:
Originally Posted by cekim View Post

As reliable as A/D usually are, which is to say only roughly unless you are talking about a very expensive circuit which these measurements aren't.

That sort of variance is typical. Even relatively tame compared to what I've seen on my systems.

Speaking only for myself, I've rarely found CPUIN to be the magic sauce that gets me to a stable OC with HW/BW. So it's almost always auto or among the last to be tuned.

Thanks for the explanation. I am more concerned about the voltages being unnecessarily high while set on auto.
Since I started overclocking the 6700K on a VIII Ranger I got a little bit allergic to automatic voltage as it often was way too high.
post #1128 of 7252
Quote:
Originally Posted by Martin778 View Post

How reliable is the temp/voltage data from AISuite? It says that my CPUIN jump up to 1,98V under load with LLC on auto while it's set to 1,95 in BIOS.


Set a manual LLC
post #1129 of 7252
Quote:
Originally Posted by Martin778 View Post

Thanks for the explanation. I am more concerned about the voltages being unnecessarily high while set on auto.
Since I started overclocking the 6700K on a VIII Ranger I got a little bit allergic to automatic voltage as it often was way too high.
Can't blame you, there are specific voltages I've seen go nuts like VDIMM that I only use auto during tunning and then I lock them down. So far CPUIN hasn't been one of them. I've only found that one getting stretched with very high OC 4.8-5.0GHZ on HWE, so even if it were to fix it, I'd start relatively low

Filmguy, questions on the loop included cooling head mounting issues, but sounds like you've ruled that out. I've run my 5930k up to 4.8Ghz under aid64 even an H110 AIO was able to keep it under 85C.
post #1130 of 7252
Can someone remind me - lower LLC level (LLC1-3 for example) meant higher voltage boost?
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel CPUs
Overclock.net › Forums › Intel › Intel CPUs › Broadwell-E thread