Overclock.net - An Overclocking Community - Reply to Topic
Thread: [Official] Ryzen 2700 and 2700X Owners Club 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)
03-31-2020 09:36 AM
yrelbirb Thank you, i will be continuing tests on my cpu

yes, ibt results are varied bcoz i continued to use my pc (a bad habit i guess)

i will sleep and leave the pc to p95 with your suggested settings ,

--

SORRy, couldnt do what i said, slept forgetting test

tried prime95 with your custom settings but vrm temps worried me a bit.

https://prnt.sc/rr0b9m

i think for me 20 min is enough i will try intel burn test again though and see if its balanced and stable (when not using computer)
03-31-2020 09:27 AM
The Sandman
Quote: Originally Posted by yrelbirb View Post
its a gigabyte mobo bug, when an offset is set hwinfo reads the SV12 TFN voltage wrong (without offset)

my cpu Tdie maxes out at 80 degrees in alot of heavy stress tests and in gaming 55-60 degrees

same for vrm, it maxes out at 85 90 degrees but in gaming its usually around 45-60 degrees

i just stress test to make sure its stable, i dont plan on using any cpu intensive applications in near future (so gaming wise, vrm temps fine for me). any thoughts on that? am i wrong to think like this_?

i cant downloda prime95 28.10

ftp://mersenne.org/gimps/p95v2810.win64.zip

site is unreacheble

---

i upped the voltage offset +0.90 just to be sure though

i also added manually +0.09 offset to hwinfo sv2 tfn reading. it seems to be accurate along with vcore;

https://prnt.sc/rq1kzi

i started ibt maximum stress test i will do 10 runs (i guess?) and return with results

10 runs passed;

https://prnt.sc/rq2hjn

i know vrms are toasty and wiill add a modification for them in future but for now i think its stable_? for gaming at least. highest i see in ac odiyssey and battlefield 5 was 60 degrees (vrm s)
Was not aware of the Giga bug thanks for sharing. I honestly can't keep all the bugs/manf are having straight. My C6H is real bad on the Super IO Chip granularity which is about the opposite, most all voltage values are off under the WMI header and need to be offset in HWInfo using a DMM lol.

For gaming etc you should be okay at those temps. Iirc most VRM's max closer to 110c. Lower is always better, especially if you're riding the edge.

Prime95 v28.1 attached below if you're interested

Good you passed IBT! Makes a good start. Sad part is there's always more to the story.
IBT is old as dinosaur poop (Linpack) and now-a-days all I use it for is a quick tune before the longer runs that P95.
While you did actually "Pass" the test this is what I see:

Notice how your values in the "Time" and "Speed" column vary? Might be due to back ground processes (always minimize these during a stress test) while they are nothing more than a reference they do aid in optimizing an OC.
You'll find that the more balanced (finer tuned) your OC is these values will flatten out thus saving time as test time is much shorter in IBT.
All this helps make P95 less work time wise. Hope this makes some sense.

Compare my IBT snip to yours and you'll see more of what I mean. This OC currently passes 6hrs RamTest, 90 minuts Y-Cruncher, 4 hrs HCI, countless hrs of P95 both 128/128 and w/90% mem.
Note: rather than Maximum I use "Custom" using 13350MB (90%) for 16GB of memory.
03-31-2020 09:05 AM
rdr09
Quote: Originally Posted by deankenny View Post
You know what, instead of always worrying about the temp, I'm just going to leave it and not even look at temps, I know for sure when gaming hard I barely see it above the 60s, and I'm sure there's no danger there, if I can stop worrying and caring so much I can just go enjoy my computer.
I agree.
03-31-2020 08:49 AM
deankenny You know what, instead of always worrying about the temp, I'm just going to leave it and not even look at temps, I know for sure when gaming hard I barely see it above the 60s, and I'm sure there's no danger there, if I can stop worrying and caring so much I can just go enjoy my computer.
03-31-2020 02:40 AM
yrelbirb
Quote: Originally Posted by The Sandman View Post
The most accurate CPU Voltage reading will be from SVI2 TFN which is reporting the voltage the VRM controller is seeing (not the Vcore value under ITE IT8686E header).
source: https://www.overclock.net/forum/25889679-post1.html

Add active cooling for the VRMs. An old case fan with zip ties mounted very close will help a lot. If your case has the option, adding another fan behind the mobo blowing on the back side on the CPU socket and VRM area will also lower temps. You have to think that going under h2o you've removed a lot of the internal case air flow which should be replaced. My VRMs rarely see 50c under load as an example. It all adds up, more heat requires more power, more power makes more heat.

Prime95 Blend is alright but is not the final answer by any means.

Here is how "The Stilt" recommends to run P95 on Ryzen.

"For testing Zen's stability I highly recommend using Prime95 28.10 version, with custom FFT config (Min & Max = 128, with run FFTs in place selected). The other versions or settings I've tested do not stress the CPU properly (the resulting power draw is around 70% of the real maximum). I haven't checked the most recent beta versions thou. Also Linpack is a rather poor stability test, since it won't sustain high stress levels very long / constantly, unless you use extremely large problem sizes (requiring more than 16GB of RAM). In addition regardless the problem size Linpack idles around 20% of the time required for the total computation, while allocating memory."Source: https://www.overclock.net/forum/26265802-post25177.html

I like running P95 with a "Custom" run using 90% memory after a bit of Blend.

Before I get into P95 I use IBT AVX version found here https://www.overclock.net/attachments/13202 to help find the easy instabilities a little faster.
Source:https://www.overclock.net/forum/10-a...ners-club.html

Y-Cruncher is another good test utility I recommend.
Memory should be tested with HCI MemTest (free, but Pro version is more user friendly) and /or Karhu RamTest ($10.00 but worth it imho) I use both.

As for the Vdroop you mentioned, if this was noticed while running P95 it's very possible it is just the difference in the load P95 creates. It is not a constant load and will vary. To a point this fluctuation is normal and intended.

Will yours run stable 4.1 @ 1,265v...? Maybe but I wouldn't hold my breath at this point. Depends which value you're calling 1.265v to actually be. Your current snip is showing Vcore of 1.154v for SV12 TFN. To me this is low and probably hurting performance.

Also be sure to read CPU Tdie for temp.

This is how my setup looks at 4.2GHz (PE3 OC) running P95 with 90% memory utilized to give you something to gauge against.

its a gigabyte mobo bug, when an offset is set hwinfo reads the SV12 TFN voltage wrong (without offset)

my cpu Tdie maxes out at 80 degrees in alot of heavy stress tests and in gaming 55-60 degrees

same for vrm, it maxes out at 85 90 degrees but in gaming its usually around 45-60 degrees

i just stress test to make sure its stable, i dont plan on using any cpu intensive applications in near future (so gaming wise, vrm temps fine for me). any thoughts on that? am i wrong to think like this_?

i cant downloda prime95 28.10

ftp://mersenne.org/gimps/p95v2810.win64.zip

site is unreacheble

---

i upped the voltage offset +0.90 just to be sure though

i also added manually +0.09 offset to hwinfo sv2 tfn reading. it seems to be accurate along with vcore;

https://prnt.sc/rq1kzi

i started ibt maximum stress test i will do 10 runs (i guess?) and return with results

10 runs passed;

https://prnt.sc/rq2hjn

i know vrms are toasty and wiill add a modification for them in future but for now i think its stable_? for gaming at least. highest i see in ac odiyssey and battlefield 5 was 60 degrees (vrm s)
03-30-2020 07:17 PM
umeng2002 Silicone lottery on these chips is like playing Cash 3 instead of Powerball.
03-30-2020 06:34 PM
The Sandman
Quote: Originally Posted by deankenny View Post
I knew it, why I always got to lose the silicon lottery, FFS!!!!


You have to remember especially while running a PE OC it's all about the cooling solution being there at boot.
This is when the system determines which multiplier to apply based off thermals.

My setup has a mono block plus a Noctua NF-A12x25 PWM (2000 rpm) mounted directly behind the CPU/VRM inside of the right cover and as system boots
it's real close to max rpm before going silent at 900 rpm. Add this to 3 rads and a D5 and things do act a little more friendly.

If it where me I wouldn't be so quick to judge the chip but perhaps take notice of how much things may change by lowering temps.
I really don't feel my chip is anything special trust me. Here's another snip, (previous snip was 100 Bclk) same test but with my current OC using a 101.8 Bclk.
Notice the difference in the multiplier/Vcore and "Intake Air Temp" between snips?
03-30-2020 12:08 PM
deankenny I knew it, why I always got to lose the silicon lottery, FFS!!!!

03-30-2020 11:42 AM
rares495
Quote: Originally Posted by deankenny View Post
Memory is on 3400mhz now and tighter timings, and seems to be stable so far, 48 hours system uptime with gaming use.



As for case, yes rad is on intake. Bearing in mind I've done all the same testing with side panel off etc with no drastic change.





Those are older pics, i have new exhaust fans and new cables for gpu, but pretty much remains the same.

The GPU cables in the last image are horrible. Eww.
03-30-2020 11:27 AM
The Sandman
Quote: Originally Posted by yrelbirb View Post
Hi people, i've yet again made some changes on my 2700x overclock

i was 1.32v 4.1 ghz and tried some undervolting

currently i'm down to 1.265v at 4.1 ghz

https://prnt.sc/rpacxu

currently im at a 7 hours of blend test and its still going strong (i know vrm temps are toasty, in gaming its tend to hover around 50 55 degrees though)

can it be expected to hit 4.1 ghz at this voltage?

also i saw that it my vcore drops to 1.248 and sometimes jumped to 1.32. is this an acceptable vdroop? sadly gigabyte board doesn't have any llc option

here's my cpu temps as well;

https://prnt.sc/rpaei6

also; is blend test can be consired a good stability along with my RAM?

my timings goes like this; https://prnt.sc/rpagzr

(not sure about tRAS. some say it should be tCL+tRCD+tRP. some say tCL+tRCD+2. im stable at both values but latency is same. some say it can hurt performance in some cases when tRAS is set to low. no idea on that

The most accurate CPU Voltage reading will be from SVI2 TFN which is reporting the voltage the VRM controller is seeing (not the Vcore value under ITE IT8686E header).
source: https://www.overclock.net/forum/25889679-post1.html

Add active cooling for the VRMs. An old case fan with zip ties mounted very close will help a lot. If your case has the option, adding another fan behind the mobo blowing on the back side on the CPU socket and VRM area will also lower temps. You have to think that going under h2o you've removed a lot of the internal case air flow which should be replaced. My VRMs rarely see 50c under load as an example. It all adds up, more heat requires more power, more power makes more heat.

Prime95 Blend is alright but is not the final answer by any means.

Here is how "The Stilt" recommends to run P95 on Ryzen.

"For testing Zen's stability I highly recommend using Prime95 28.10 version, with custom FFT config (Min & Max = 128, with run FFTs in place selected). The other versions or settings I've tested do not stress the CPU properly (the resulting power draw is around 70% of the real maximum). I haven't checked the most recent beta versions thou. Also Linpack is a rather poor stability test, since it won't sustain high stress levels very long / constantly, unless you use extremely large problem sizes (requiring more than 16GB of RAM). In addition regardless the problem size Linpack idles around 20% of the time required for the total computation, while allocating memory."Source: https://www.overclock.net/forum/26265802-post25177.html

I like running P95 with a "Custom" run using 90% memory after a bit of Blend.

Before I get into P95 I use IBT AVX version found here https://www.overclock.net/attachments/13202 to help find the easy instabilities a little faster.
Source:https://www.overclock.net/forum/10-a...ners-club.html

Y-Cruncher is another good test utility I recommend.
Memory should be tested with HCI MemTest (free, but Pro version is more user friendly) and /or Karhu RamTest ($10.00 but worth it imho) I use both.

As for the Vdroop you mentioned, if this was noticed while running P95 it's very possible it is just the difference in the load P95 creates. It is not a constant load and will vary. To a point this fluctuation is normal and intended.

Will yours run stable 4.1 @ 1,265v...? Maybe but I wouldn't hold my breath at this point. Depends which value you're calling 1.265v to actually be. Your current snip is showing Vcore of 1.154v for SV12 TFN. To me this is low and probably hurting performance.

Also be sure to read CPU Tdie for temp.

This is how my setup looks at 4.2GHz (PE3 OC) running P95 with 90% memory utilized to give you something to gauge against.
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