Overclock.net - An Overclocking Community - Reply to Topic

Thread: Skylake-X/Kaby Lake-X Combined Discussion Reply to Thread
Title:
Message:

Register Now

In order to be able to post messages on the Overclock.net - An Overclocking Community forums, you must first register.
Please enter your desired user name, your email address and other required details in the form below.
User Name:
If you do not want to register, fill this field only and the name will be used as user name for your post.
Password
Please enter a password for your user account. Note that passwords are case-sensitive.
Password:
Confirm Password:
Email Address
Please enter a valid email address for yourself.
Email Address:

Log-in


  Additional Options
Miscellaneous Options

  Topic Review (Newest First)
06-14-2019 09:34 PM
ThrashZone Hi,
Almost pulled 8900 dang it

Edit single core score just finished 502 mp ratio 17.70x
06-14-2019 11:48 AM
ThrashZone Hi,
Well 1.2v is what is needed for 4.5

5.0 I estimated 1.37v would be needed for me on 7900x temp wise wasn't going to happen
1.34v for 4.9 on a cool day was as good as I ever got.
06-14-2019 11:30 AM
kingofblog
Quote: Originally Posted by ThrashZone View Post
Hi,
4.5 shouldn't need any avx offset at all
If you crash vcore isn't enough but 1.2v should be plenty
Still crashing you have other issues like memory or cache causing the crash or as said cooling not working well in some way/ thermal paste spread....

I don't and probably never will use prime for testing.
Blender demo files classroom/ bmw/.. are more realistic work loads.

https://www.blender.org/download/demo-files/
If I set offset to 0, it uses the VID for 4.0 GHz at 4.5 GHz, which is 1.0 V. On the other hand, if I use offset 5 with 5.0 GHz (4.5 GHz effective), the VID somehow becomes 1.2 V.
06-14-2019 10:24 AM
ThrashZone
Quote: Originally Posted by Mysticial View Post
It's not the board. The VIDs are specified by the CPU. My 14 core has cores on both extremes (leaky/low-vcore, and not-leaky/high-vcore)

The two TB3 cores are stable 4.7 GHz (non-AVX) with only 1.24 vcore. While there's 4 others that can't do it at even 1.35v.
Hi,
Heck post hwinfo at what ever clocks and manual voltage you've used and lets see the core temp spread.
If the core temps are way off too like more than 10c could be a liquid metal application problem lord knows I had that problem too from silicon lottery it only lasted 9-10 months before I had to redo it.
06-14-2019 10:16 AM
Mysticial
Quote: Originally Posted by ThrashZone View Post
Hi,
Wow that's a bad spread
Might be worth rma have you seen other giga boards that bad ?
It's not the board. The VIDs are specified by the CPU. My 14 core has cores on both extremes (leaky/low-vcore, and not-leaky/high-vcore)

The two TB3 cores are stable 4.7 GHz (non-AVX) with only 1.24 vcore. While there's 4 others that can't do it at even 1.35v.
06-14-2019 10:12 AM
ThrashZone
Quote: Originally Posted by Mysticial View Post
I have a Gigabyte board so I don't have per-core voltage. Yes, there's a huge vcore spread across the different cores (0.130v spread @ 4.5 GHz), but there isn't anything I can do about it.

The only option is to lock the vcore for all cores to a specific voltage regardless of load or clock which is not what I want.
Hi,
Wow that's a bad spread
Might be worth rma have you seen other giga boards that bad ?
06-14-2019 10:05 AM
Mysticial
Quote: Originally Posted by ThrashZone View Post
Hi,
Even with offset I had to tune in each core to get the vid's closer together otherwise same deal as adaptive occurred :/
A lot of tuning to get 10 cores happy
9940x are a lot better and about the same 4.5 no avx offset needed
But I only use blender demo files/ classroom too and then lastly real bench...
I have a Gigabyte board so I don't have per-core voltage. Yes, there's a huge vcore spread across the different cores (0.130v spread @ 4.5 GHz), but there isn't anything I can do about it.

The only option is to lock the vcore for all cores to a specific voltage regardless of load or clock which is not what I want.
06-14-2019 09:59 AM
ThrashZone
Quote: Originally Posted by Mysticial View Post
Offset voltage, IIRC +0.030.
Hi,
Even with offset I had to tune in each core to get the vid's closer together otherwise same deal as adaptive occurred :/
A lot of tuning to get 10 cores happy
9940x are a lot better and about the same 4.5 no avx offset needed
But I only use blender demo files/ classroom too and then lastly real bench...
06-14-2019 09:52 AM
Mysticial
Quote: Originally Posted by ThrashZone View Post
Hi,
On manual core voltage ?
On adaptive voltage I'd probably agree if the vid's get to far apart it does or did for me cause instability
But that was on higher clocks 4.6....

The vid spread between cores was pretty bad +-0.040 that is usually the problem.
+-0.020 and it gets a lot more stable.
Offset voltage, IIRC +0.030.
06-14-2019 09:51 AM
ThrashZone
Quote: Originally Posted by Mysticial View Post
Then you have a really good chip. On both of mine, there are multiple cores that can't do AVX512 at 4.5 GHz. My 7940x has at least one that won't do it at even 3.8 GHz even with higher than stock vcore.
Hi,
On manual core voltage ?
On adaptive voltage I'd probably agree if the vid's get to far apart it does or did for me cause instability
But that was on higher clocks 4.6....

The vid spread between cores was pretty bad +-0.040 that is usually the problem.
+-0.020 and it gets a lot more stable.
This thread has more than 10 replies. Click here to review the whole thread.

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off