Welcome Guest ( Log In | Register )

4 Pages < 1 2 3 4 >Bottom

Outline · [ Standard ] · Linear+

 AMD Ryzen, AM4 / AM5 Platform

views
     
lolzcalvin
post Dec 25 2019, 07:23 PM

shibe in predicament
******
Senior Member
1,583 posts

Joined: Mar 2014
From: 75°26'11.6"S, 136°16'16.0"E


QUOTE(goldfries @ Dec 24 2019, 04:01 PM)
What exactly is 3600MHz with terrible timing?

Nothing wrong with Hynix die either, don't be too obsessed with Samsung B-Die.
*
generally 18-22-22-42 variant which most sellers in Malaysia would sell for both trident z royal and trident z neo.

bought one just now anyway, as you all suggested hynix performs well in overclocking. I suppose I'll try tightening the timing to 16 and maybe post some results somewhere here.
lolzcalvin
post Dec 29 2019, 01:03 AM

shibe in predicament
******
Senior Member
1,583 posts

Joined: Mar 2014
From: 75°26'11.6"S, 136°16'16.0"E


hey all, maybe it's already answered 10 times here already but is 3900x idling at 45-50 celsius normal? running cinebench R20 will heat it up to 75-80 celsius.

This post has been edited by lolzcalvin: Dec 29 2019, 01:07 AM
lolzcalvin
post Dec 29 2019, 01:43 AM

shibe in predicament
******
Senior Member
1,583 posts

Joined: Mar 2014
From: 75°26'11.6"S, 136°16'16.0"E


QUOTE(SSJBen @ Dec 29 2019, 01:20 AM)
Air cooling? Many people have it in the 50s and they live in much cooler countries than msia. If you're getting 45c, that's not bad at all and is rather normal.
*
AIO cooled. Corsair H115i
lolzcalvin
post Dec 29 2019, 05:59 PM

shibe in predicament
******
Senior Member
1,583 posts

Joined: Mar 2014
From: 75°26'11.6"S, 136°16'16.0"E


kinda found something rather abnormal with the BIOS or maybe the chipset itself, but maybe it's normal though.

1. while idling my 3900x will hover at around 1.38-1.43v vcore (core voltage), and almost always >1.4V for VID which can reach as high as 1.47v.
however during load e.g. Cinebench R20, the vcore and VID will be downvolted compared to when idle, to safer voltages such as ~1.32v on vcore and ~1.26v on VID. this is not normal right?
» Click to show Spoiler - click again to hide... «



2. I've tried overclocking the chip and can run 4.3GHz all cores at 1.35v, but as the vcore goes 1.35-1.37v (latter being affected probably by load line calibration - set to high), the VID will always hovering around 1.09-1.1v. will such a low VID cause any unprecedented issue, or the vcore reading is already enough?

This post has been edited by lolzcalvin: Dec 29 2019, 06:09 PM
lolzcalvin
post Jan 2 2020, 07:09 PM

shibe in predicament
******
Senior Member
1,583 posts

Joined: Mar 2014
From: 75°26'11.6"S, 136°16'16.0"E


QUOTE(SSJBen @ Jan 2 2020, 04:59 PM)
Alright been using the 3900x for a week now, quite a happy camper here. Tuning memory is fun, haven't had this much focus on memory since the DDR2 days - quite a nostalgic feeling.

Stuck with AGESA 1003ABBA BIOS though due to some issues with PBO on 1004B. How are you guys doing with AGESA 1004B?
*
not sure, if i restart the PC then go straight to UEFI then save my settings, gigabyte will straight bork the settings to something i can't see (but can feel) and destabilise my system (RAM crash due to sudden undervoltage, borking the multipliers and voltage of CPU etc). had to shut down the PC and go into UEFI and then configure my settings instead, that's my current fix allowing the BIOS to read my settings correctly.

This post has been edited by lolzcalvin: Jan 2 2020, 07:09 PM
lolzcalvin
post Jan 3 2020, 04:14 PM

shibe in predicament
******
Senior Member
1,583 posts

Joined: Mar 2014
From: 75°26'11.6"S, 136°16'16.0"E


QUOTE(SSJBen @ Jan 3 2020, 04:12 PM)
Seems to me that manual OC is better than PBO at the moment for most apps - as long as your silicon quality is decent.

I got 4.4ghz all cores @1.325v vs PBO doing 4.55 for 2 cores and all others being 4.1 @ voltages spiking to 1.48v. The manual OC is much, much cooler under max load too (70c vs 79c, on a Noctua D15s). PBO gets like a 100 points more over manual OC in cinebench R20 and extra ~3fps for 99th avg percentile in AC Odyssey.
*
i can do 4.4GHz 1.325v vcore voltage too. however I notice the VID will be around 1.1v.
also, with this overclock above it'll run cinebench at 85c on my end.

did you disable PBO when you overclock your chip?

This post has been edited by lolzcalvin: Jan 3 2020, 04:19 PM
lolzcalvin
post Jan 3 2020, 04:37 PM

shibe in predicament
******
Senior Member
1,583 posts

Joined: Mar 2014
From: 75°26'11.6"S, 136°16'16.0"E


my result when OCed to 4.4GHz @1.325v vcore, XFR and PBO disabled.
idles at 46-55c, bit hotter than my usual stock multiplier with -0.1v offset vcore.
runs up to 80c+ when loaded with cinebench r20.
user posted image

This post has been edited by lolzcalvin: Jan 3 2020, 04:38 PM
lolzcalvin
post Jan 3 2020, 07:01 PM

shibe in predicament
******
Senior Member
1,583 posts

Joined: Mar 2014
From: 75°26'11.6"S, 136°16'16.0"E


QUOTE(SSJBen @ Jan 3 2020, 06:40 PM)
Are you on 1usmus power plan?
*
nope, don't know what is it until I googled it lol
lolzcalvin
post Jan 4 2020, 12:00 AM

shibe in predicament
******
Senior Member
1,583 posts

Joined: Mar 2014
From: 75°26'11.6"S, 136°16'16.0"E


OK, so I've tried out 1usmus power plan and as instructed I use the 1usmus universal plan since I'm running windows 10 v1909.

I've OCed to 4.4GHz @ 1.325v vcore and 1.1v SOC (stock is 1.2v).

idle is bout the same, 45 - 55c bounces. vcore will be at 1.332v, slightly higher than what is set in the bios (1.325v)

load temp is slightly better as it didn't go over 80c, but during load it's interesting, the vcore (shown in CPU-Z) actually dips to 1.284v. this dip likely caused one cinebench application error crash out of numerous tests. the VID went quite low too, 1.0375v. I don't really have a full idea of the correlation between VID and vcore and wonder if low VID would cause instability as well.

any sifus can suggest how to stabilise the vcore? it's not setting to the value i wanted to be in the BIOS, and actually dips about 40mV on vcore when under load.

user posted image

This post has been edited by lolzcalvin: Jan 4 2020, 12:01 AM
lolzcalvin
post Jan 4 2020, 12:13 PM

shibe in predicament
******
Senior Member
1,583 posts

Joined: Mar 2014
From: 75°26'11.6"S, 136°16'16.0"E


QUOTE(AMDAthlon @ Jan 4 2020, 11:10 AM)
Vcore drops/increase under load can be optimized using LLC setting in your bios. Check it.
*
like setting LLC to high?
lolzcalvin
post Jan 5 2020, 11:37 PM

shibe in predicament
******
Senior Member
1,583 posts

Joined: Mar 2014
From: 75°26'11.6"S, 136°16'16.0"E


QUOTE(SSJBen @ Jan 4 2020, 02:23 PM)
Turns out 4.4ghz @ 1.325v isn't 100% stable for me. Was encoding a video on handbrake and an error came up, eventhough it's minor.

Turned it down to 4.25ghz @ 1.30v - brought down load temps to 65c and losing only 2% overall perf. Not a bad trade off I guess.

EDIT*
1.325v isn't a magic number and it's NOT SAFE for Zen 2?
*
that's something new. time to reset CMOS.
lolzcalvin
post Jan 5 2020, 11:59 PM

shibe in predicament
******
Senior Member
1,583 posts

Joined: Mar 2014
From: 75°26'11.6"S, 136°16'16.0"E


my god, after 2 weeks of fiddling i finally found out the major reason my cpu always run hot and high VID voltages across cores during idle.

it's the iCUE software.
lolzcalvin
post Jan 9 2020, 12:53 AM

shibe in predicament
******
Senior Member
1,583 posts

Joined: Mar 2014
From: 75°26'11.6"S, 136°16'16.0"E


QUOTE(SSJBen @ Jan 6 2020, 04:47 PM)
Very common issue it seems. Razer synapse 3 also consumes high CPU power and simply boosts your CPU clocks to the highest for no reason.
*
oof, closing both icue and synapse immediately drops my voltage to 0.85v.

rclxub.gif
lolzcalvin
post Jan 14 2020, 11:19 AM

shibe in predicament
******
Senior Member
1,583 posts

Joined: Mar 2014
From: 75°26'11.6"S, 136°16'16.0"E


if you open razer synapse and corsair icue at the same time your cores will always stay active 24/7 and they don'y allow it to sleep.

2 software will keep the voltage over 1.4v for me on both VID and VCore.
closing both of them brings it down to 0.85v.
lolzcalvin
post Jan 14 2020, 08:39 PM

shibe in predicament
******
Senior Member
1,583 posts

Joined: Mar 2014
From: 75°26'11.6"S, 136°16'16.0"E


QUOTE(edmund_yung @ Jan 14 2020, 02:18 PM)
What monitoring software do you guys use to check the temperature without it going boost all the time? I’m already modding the fan curve like what you said too.
*
ryzen master
lolzcalvin
post Jan 18 2020, 11:47 PM

shibe in predicament
******
Senior Member
1,583 posts

Joined: Mar 2014
From: 75°26'11.6"S, 136°16'16.0"E


1.4v on idle is still not acceptable for me though. the main problem is that synapse and icue do not periodically boost frequencies -- it's constantly running at 1.4v without dipping at once.

my fix is now to actually run a very stable OC at 4.3GHz at 1.287v -- with low LLC it runs just at 1.272v, to actually stabilise the vcore voltage and to keep my silicon clean.

also, i wonder if anyone notices that CoreTemp and Ryzen Master report VID instead of vcore. my VID alwyas run at around 1.1v when overclocked, and this is the value showned by both the mentioned monitors. except for CPU-Z, it reports the actual vcore voltage.
lolzcalvin
post Jan 24 2020, 12:36 AM

shibe in predicament
******
Senior Member
1,583 posts

Joined: Mar 2014
From: 75°26'11.6"S, 136°16'16.0"E


i'm a user of aorus master as well.

it's an extremely solid motherboard, quality VRMs, true 12 phases, actual contacted heatsinks, cool, good DAC for audio, excellent overclocker, and I love the dual bios.

just that the BIOS itself, not only its UI, is shitty. it has bugs too, didn't save my settings correctly and downvolted the XMP voltage back to 1.2, made my 3600MHz crashed immediately.

good thing BIOS reset is as simple as pushing the reset bios button.

remember if you plug an M2 SSD at the bottom most M2 slot, the SATA ports proportional to that M2 slot will be unusable.

do read the booklet that came with the mobo because it's very useful icon_idea.gif

This post has been edited by lolzcalvin: Jan 24 2020, 12:44 AM
lolzcalvin
post Jan 29 2020, 07:19 PM

shibe in predicament
******
Senior Member
1,583 posts

Joined: Mar 2014
From: 75°26'11.6"S, 136°16'16.0"E


QUOTE(Hornet @ Jan 29 2020, 06:14 PM)
To those using x570 Aorus Master, do you guys have the issue where the graphic card is touching the chipset heatsink?

I can't see any sign that the heatsink is not fixed properly, it looks fine to me.
But it is definitely higher than the height of the PCIE slot towards the top portion, and since only the top PCIE slot is 16x, we dont really have any option to use lower slots.

So far the graphic card seems working fine for me though. I think it slots in just enough to make sufficient contact, but I can't get the PCIE clip to click
*
chipset heatsink? VRM heatsink or NVMe heatsink?
lolzcalvin
post Jan 29 2020, 10:12 PM

shibe in predicament
******
Senior Member
1,583 posts

Joined: Mar 2014
From: 75°26'11.6"S, 136°16'16.0"E


QUOTE(Hornet @ Jan 29 2020, 09:00 PM)
I mean the heatsink for the x570 chipset, the one that have a fan inside it.

» Click to show Spoiler - click again to hide... «

*
which gpu are you using? because mine doesn't block. here you can see the height between the latch and the chipset fan. I still have a little bit of gap between the fan and the GPU shroud/backplate.
» Click to show Spoiler - click again to hide... «


This post has been edited by lolzcalvin: Jan 29 2020, 10:15 PM
lolzcalvin
post Jan 31 2020, 09:35 PM

shibe in predicament
******
Senior Member
1,583 posts

Joined: Mar 2014
From: 75°26'11.6"S, 136°16'16.0"E


QUOTE(k!nex @ Jan 31 2020, 08:21 PM)
I might need your advice here. Ok the thing is, I have an R7 3700X . X570 Auros Master.
RAM I have 2 kits:
1. G.Skill Trident Z RGB F4-3600C18D-16GTZRX
2. G.Skill Trident Z Neo F4-3600C18D-16GTZN

I bought these 2 slightly different kit because of stock availability issues.

I can run 3800Mhz 18-22-22-42-64 1T with a ridiculous TRFC at 666 ( shown in hwinfo64) but only with 2 sticks of Trident Z Neo. FCLK 1900Mhz is stable under Asus Realbench for 4 hours. If I run 4 sticks of RAM, I cannot achieve stability.
Currently VDIMM at 1.38V, VSOC 1.1V , CPU use PBO only.

My main issue is I cannot tighten the timings using 1usmus's Ryzen DRAM calculator as guide (even safe preset, wanna go down at least CL16 ) or increase VDIMM to 1.45V like that. Cannot POST , mobo throws me a C3 code and I have to use the BIOS reset button to recover. From 4 sticks I reduce to 2 stick of RAM, same thing.

Using Thaiphoon burner, I can confirm all 4 sticks of RAM I have are Samsung B dies. Do they really that suck for lowly binned ones ?
*
3600MHz 18-22-22-42 may be a samsung b die reported from thaiphoon, but it's probably the poorest binned one, because i'm running one as well. and it's a damn trash tier overclocker.
some people claim the serial number for this RAM is a giveaway of what type the die is, such as if the serial number trails with "10B" then it's a samsung b-die. however, the 3600 CL18 version of G.Skill Trident Z Neo trails with "10C". really makes you think hmm.gif

it's another damn shame 95% of local resellers don't sell RAMs with better timings. I couldn't find the remaining 5%.

This post has been edited by lolzcalvin: Jan 31 2020, 09:42 PM

4 Pages < 1 2 3 4 >Top
 

Change to:
| Lo-Fi Version
0.2309sec    0.37    7 queries    GZIP Disabled
Time is now: 25th November 2025 - 04:15 PM