Overclock.net banner

321 - 340 of 9977 Posts

·
Registered
Joined
·
53 Posts
I have the Prime X370-Pro, so it'll be a bit different in some of the power system area, but I'm gonna give this a go and see what I can come up with. Using Hynix 2400 Corsair Vengeance stuff, which I have successfully tweaked to 2933 with a ton of trial and error and guess work on my part. Gonna try 3066 with the timings and settings from the calc and see if that'll work for me. I'll post my results in a couple of days.
 

·
Registered
Joined
·
721 Posts
<div class="quote-container" data-huddler-embed="/t/1640919/ryzen-dram-calculator-overclocking-dram/270_30#post_26457559" data-huddler-embed-placeholder="false"><span>Quote:</span>
<div class="quote-block">Originally Posted by <strong>1usmus</strong> <a href="/t/1640919/ryzen-dram-calculator-overclocking-dram/270_30#post_26457559"><img alt="View Post" class="inlineimg" src="/img/forum/go_quote.gif"></a><br><br>
I decided to write a small article explaining why some systems work fine at 3200+ and others do not at all<br><br><span style="font-size:36px;"><b>Influence of СLDO_VDDP on MEMCLK "holes"</b></span><br><br>
CLDO_VDDP is a voltage regulator for the module (physical interface) of encoding and decoding of the transmitted and received data stream. The purpose of coding is to simplify the process of restoring the data stream of the receiver. It determines the signals, signal ratios and time parameters necessary for transferring control information, reading and writing data to DRAM devices. In plain language, CLDO_VDDP is the voltage that regulates the memory access at a certain frequency. "Hole" in turn - the frequency gap on which the memory controller can operate with our RAM.<br><br>
Consider this simple picture:<br><br><a class="H-lightbox-open" href="http://www.overclock.net/content/type/61/id/3154333/"><img alt="" class="lightbox-enabled" data-id="3154333" data-type="61" src="http://www.overclock.net/content/type/61/id/3154333/width/500/height/1000/flags/LL" style="; width: 500px; height: 281px"></a><br><br><br>
It shows 3 identical systems (motherboard + RAM + processor). All 3 systems were overclocked and received the following results:<br><br>
1) The system was perfectly dispersed to a frequency of 3333 MHz<br>
2) The system was accelerated to 2933<br>
3) The system did not start at all<br><br>
If all systems are the same, why such results? Let's understand. The bottom line is that each memory controller (IMC) has its own technical characteristics ("voltage" and time) and at the same voltage / frequency it will behave differently, namely it will have different access to memory. Red marked our MEMCLK holes, these are the very hole-mediators through which our memory controller communicates with RAM, and if there is no hole in the frequency range chosen by us - the system does not start or start, but the memory runs with errors. At you I think there was a question as these holes to move and expand - all is very simple, voltage CLDO_VDDP allows to spend the given manipulations. The only difficulty is that these holes can not be mathematically calculated. A vivid example of CLDO_VDDP 866 which is magical for many. The hole of this voltage is in the region of 3300-3500 MHz, but again not for all systems. As shown by our internal tests, not all of it works, I repeat all the IMC are different and require a different voltage CLDO_VDDP to achieve the same frequency.<br><br>
In view of the fact that the shape of the voltage CLDO_VDDP is wave, the minimum voltage change can drastically change the stability of the system. The voltage step is 1 mv. Borders from 700 to 975.<br><br>
I also want to publish a list of CLDO_VDDP, which can help stabilize your memory<br><br><a class="spoiler-link H-spoiler-toggle" href="#"><strong>CLDO_VDDP list (volts)</strong> <span class="spoiler-help">(Click to show)</span></a><div class="spoiler-hidden">0.562<br>
0.568<br>
0.573<br>
0.579<br>
0.585<br>
0.590<br>
0.596<br>
0.601<br>
0.607<br>
0.613<br>
0.618<br>
0.624<br>
0.630<br>
0.635<br>
0.641<br>
0.646<br>
0.652<br>
0.658<br>
0.663<br>
0.669<br>
0.675<br>
0.680<br>
0.686<br>
0.691<br>
0.697<br>
0.703<br>
0.708<br>
0.714<br>
0.720<br>
0.725<br>
0.731<br>
0.736<br>
0.742<br>
0.748<br>
0.753<br>
0.759<br>
0.765<br>
0.770<br>
0.776<br>
0.781<br>
0.787<br>
0.793<br>
0.798<br>
0.804<br>
0.810<br>
0.815<br>
0.821<br>
0.826<br>
0.832<br>
0.838<br>
0.843<br>
0.849<br>
0.855<br>
0.860<br>
0.866<br>
0.871<br>
0.877<br>
0.883<br>
0.888<br>
0.894<br>
0.900<br>
0.905<br>
0.911<br>
0.916<br>
0.922<br>
0.928<br>
0.933<br>
0.939<br>
0.945<br>
0.950<br>
0.956<br>
0.961<br>
0.961<br>
0.967<br>
0.973<br></div>
</div>
</div>
<br>
Thank you for this, but I have to note two things:<br><ul class="bbcode_list"><li>While I am impressed at the range of languages you are conversant with, this was the most difficult to follow comment of yours that I have read. In addition to the "hole-mediator" commentary, the use of the terms "dispersed" and "accelerated" in the list are confusing.</li>
<li>In the terminology used in the C6H thread (first place that I had ever read of a memory hole), my understanding was that the memory hole was a range of frequencies for which the memory would <b><i>not</i></b> work. That seems to be the opposite of your usage, assuming that I understood what you meant above. I used the term memory peak when reporting regions where best (or sometimes any) performance was found on my system when trying combinations of CLDO_VDDP and Proc_ODT.</li>
</ul><br>
I raise these points because I am uncertain whether I have understood you, or misunderstood you.
 

·
Registered
Joined
·
73 Posts
<div class="quote-container" data-huddler-embed="/t/1640919/ryzen-dram-calculator-overclocking-dram/290#post_26458791" data-huddler-embed-placeholder="false"><span>Quote:</span>
<div class="quote-block">Originally Posted by <strong>1usmus</strong> <a href="/t/1640919/ryzen-dram-calculator-overclocking-dram/290#post_26458791"><img alt="View Post" class="inlineimg" src="/img/forum/go_quote.gif"></a><br><br><b><span style="font-size:36px;">Ryzen DRAM Calculator 0.9.9 v6<br></span></b><br><br>
<a href="https://drive.google.com/file/d/1zVtLtmSzEaLE4iC7CVdVk1KOax-EABfZ/view?usp=sharing" target="_blank">https://drive.google.com/file/d/1zVtLtmSzEaLE4iC7CVdVk1KOax-EABfZ/view?usp=sharing</a><br><br>
* Added support for 8 RAM modules<br>
* Redesigned saving algorithm (now after updating the program all previously entered data will be available)<br>
* fixed tRTP calculation for all calculators<br>
* fixes in Termination block<br>
* Block Drive strength added (auxiliary parameters for memory stabilization)<br>
* Added a block of memory interleaving (a small increase in processor performance)<br>
* partially reworked Extreme preset (memory bandwidth is reduced, but latency is improved)<br>
* small visual improvements</div>
</div>
<br>
Thank you again @1usmus !<br><br><b>The new recommendations in v6 has allowed me to stabilize 3466 CL15 CR1 for the first time with 4 sticks!</b> <img alt="thumb.gif" class="bbcode_smiley" src="http://files.overclock.net/images/smilies/thumb.gif"><br><br>
Just for the record:<br><br><a class="H-lightbox-open" href="http://www.overclock.net/content/type/61/id/3155881/"><img alt="" class="lightbox-enabled" data-id="3155881" data-type="61" src="http://www.overclock.net/content/type/61/id/3155881/width/500/height/1000/flags/LL" style="; width: 500px; height: 224px"></a><br><br>
It has survived 2hs of AIDA64 stress test and I've been playing around and testing games with no problems at all <img alt="biggrin.gif" class="bbcode_smiley" src="http://files.overclock.net/images/smilies/biggrin.gif"><br><br>
While trying to get there I tried every termination block & CAD BUS block recommendation of v6<br><br>
ProcODT 53 > RTT_NOM RZQ/7 > RTT_WR OFF > RTT_PARK RZQ/5 > CAD_BUS AUTO = Failed AIDA64 Stress test in 5 min.<br><br>
ProcODT 53 > RTT_NOM RZQ/7 > RTT_WR OFF > RTT_PARK RZQ/6 > CAD_BUS AUTO = Failed AIDA64 Stress test in 7 min.<br><br>
ProcODT 53 > RTT_NOM RZQ/7 > RTT_WR OFF > RTT_PARK RZQ/7 > CAD_BUS AUTO = Failed AIDA64 Stress test in 10 min.<br><br>
Sticked with the one that survived the longest and started playing around with CAD_BUS.<br><br>
So...<b>ProcODT 53 > RTT_NOM RZQ/7 > RTT_WR OFF > RTT_PARK RZQ/7</b><br><br>
CAD_BUS 24-24-24-24 = Failed AIDA64 Stress test in 10 min. (I guess this would be the same as AUTO)<br><br>
CAD_BUS 30-30-30-30 = Failed AIDA64 Stress test in 18 min.<br><br>
CAD_BUS 30-30-40-60 = Failed AIDA64 Stress test in 14 min<br><br>
CAD_BUS 40-40-40-40 = Survived 2 hs of AIDA64 Stress test, then games without crashing or BSODing<br><br>
Fun fact, never even tried 40-40-40-40 before. It was just a lucky guess I made and tried when I saw that 30-30-40-60 failed earlier than 30-30-30-30.<br><br>
Maybe my RAM @3466 just likes 40 ohms in the CAD BUS? Who knows? <img alt="tongue.gif" class="bbcode_smiley" src="http://files.overclock.net/images/smilies/tongue.gif">
 

·
Registered
Joined
·
28 Posts
Thank you so much for this! With the new BIOS and your calculator, I am able to boot up with faster settings than ever before!<br><br>
I have one question that I'm sure is stupid... Where is BGS in the BIOS? I was able to find everything else.<br><br>
So far, I've been able to boot with my RAM (4x F4-3866C18-8GTZR) at 3600MHz but with a crash in Prime95 after a few minutes. I seem stable at 3333MHz with safe timings and I'm currently testing 3466MHz with safe timings. My previous best on BIOS 1701 was 3200MHZ RAM with ****ty timings so I already have a great improvement.
 

·
Premium Member
Joined
·
138 Posts
<div class="quote-container" data-huddler-embed="/t/1640919/ryzen-dram-calculator-overclocking-dram/320#post_26461712" data-huddler-embed-placeholder="false"><span>Quote:</span>
<div class="quote-block">Originally Posted by <strong>kaseki</strong> <a href="/t/1640919/ryzen-dram-calculator-overclocking-dram/320#post_26461712"><img alt="View Post" class="inlineimg" src="/img/forum/go_quote.gif"></a><br><br><ul class="bbcode_list"><li>While I am impressed at the range of languages you are conversant with, this was the most difficult to follow comment of yours that I have read. In addition to the "hole-mediator" commentary, the use of the terms "dispersed" and "accelerated" in the list are confusing.</li>
<li>In the terminology used in the C6H thread (first place that I had ever read of a memory hole), my understanding was that the memory hole was a range of frequencies for which the memory would <b><i>not</i></b> work. That seems to be the opposite of your usage, assuming that I understood what you meant above. I used the term memory peak when reporting regions where best (or sometimes any) performance was found on my system when trying combinations of CLDO_VDDP and Proc_ODT.</li>
</ul><br>
I raise these points because I am uncertain whether I have understood you, or misunderstood you.</div>
</div>
<br>
You're not alone with your confusion. I had the very same difficulties making sense of 1usmus comment and choice of words. I can't even tell what the diagram is supposed to show.<br>
elmor and The Stilt introduced memory holes as rather narrow frequency ranges where the memory does not work. 1usmus seems to invert that concept.
 

·
AMD overclocker/developer
Joined
·
1,651 Posts
Discussion Starter #326
<div class="quote-container" data-huddler-embed="/t/1640919/ryzen-dram-calculator-overclocking-dram/320_20#post_26461712" data-huddler-embed-placeholder="false"><span>Quote:</span>
<div class="quote-block">Originally Posted by <strong>kaseki</strong> <a href="/t/1640919/ryzen-dram-calculator-overclocking-dram/320_20#post_26461712"><img alt="View Post" class="inlineimg" src="/img/forum/go_quote.gif"></a><br><br>
Thank you for this, but I have to note two things:<br><ul class="bbcode_list"><li>While I am impressed at the range of languages you are conversant with, this was the most difficult to follow comment of yours that I have read. In addition to the "hole-mediator" commentary, the use of the terms "dispersed" and "accelerated" in the list are confusing.</li>
<li>In the terminology used in the C6H thread (first place that I had ever read of a memory hole), my understanding was that the memory hole was a range of frequencies for which the memory would <b><i>not</i></b> work. That seems to be the opposite of your usage, assuming that I understood what you meant above. I used the term memory peak when reporting regions where best (or sometimes any) performance was found on my system when trying combinations of CLDO_VDDP and Proc_ODT.</li>
</ul><br>
I raise these points because I am uncertain whether I have understood you, or misunderstood you.</div>
</div>
<br>
Thank you for being with me and helping!<br>
It really happens to me (sometimes) is difficult to correctly present technically complex information.<br>
I understand you, in my article the terms "hole" may be incorrectly used, but the main idea is that there are several dead regions. This conclusion I made on the basis of the cases when the memory of Samsung B-die was not responding at a frequency of over 3200 on default CLDO_VDDP (I had several 1700 processors, one of them could work with frequency memory 2933 and 3600, between these frequencies was a dead region). Therefore, my theory is slightly different from the information that was in the thread C6H.<br><br>
this article is my guess, to check it need an oscilloscope ... <img alt="frown.gif" class="bbcode_smiley" src="http://files.overclock.net/images/smilies/frown.gif"><br><div class="quote-container" data-huddler-embed="/t/1640919/ryzen-dram-calculator-overclocking-dram/320_20#post_26461864" data-huddler-embed-placeholder="false"><span>Quote:</span>
<div class="quote-block">Originally Posted by <strong>neur0cide</strong> <a href="/t/1640919/ryzen-dram-calculator-overclocking-dram/320_20#post_26461864"><img alt="View Post" class="inlineimg" src="/img/forum/go_quote.gif"></a><br><br>
You're not alone with your confusion. I had the very same difficulties making sense of 1usmus comment and choice of words. I can't even tell what the diagram is supposed to show.<br>
elmor and The Stilt introduced memory holes as rather narrow frequency ranges where the memory does not work. 1usmus seems to invert that concept.</div>
</div>
<br>
you correctly understood me, this conclusion was based on practical tests
 

·
AMD overclocker/developer
Joined
·
1,651 Posts
Discussion Starter #327
<div class="quote-container" data-huddler-embed="/t/1640919/ryzen-dram-calculator-overclocking-dram/300_20#post_26461620" data-huddler-embed-placeholder="false"><span>Quote:</span>
<div class="quote-block">Originally Posted by <strong>ZeNch</strong> <a href="/t/1640919/ryzen-dram-calculator-overclocking-dram/300_20#post_26461620"><img alt="View Post" class="inlineimg" src="/img/forum/go_quote.gif"></a><br><br>
i have at the momment in the last days 3 freeze screen (not bsod, not black screen) i think i need more Ram voltage.<br>
What do you think?<br><br><br>
(i have the last official bios of prime x370 pro.)<br><br>
My rams are clocked with 16-16-16-18-36-56-1T @3200mhz 1.4v (now 1.41 to test) Mdie SR.<br><br>
my TRFC is 560 in windows but bios show 312 (i have in auto this settings, all other settings manual)<br><br>
ps: one of these freezes is with mkv "render" (i add other audio). other with virus scan. different disk, and different cable sata.</div>
</div>
<br>
I think it is necessary to expose 1,415, there is no threat to the system with a specified voltage<br><br>
Do not pay attention to what BIOS suggests <img alt="wink.gif" class="bbcode_smiley" src="http://files.overclock.net/images/smilies/wink.gif">
 

·
AMD overclocker/developer
Joined
·
1,651 Posts
Discussion Starter #328
<div class="quote-container" data-huddler-embed="/t/1640919/ryzen-dram-calculator-overclocking-dram/320_20#post_26461769" data-huddler-embed-placeholder="false"><span>Quote:</span>
<div class="quote-block">Originally Posted by <strong>Ex0cet</strong> <a href="/t/1640919/ryzen-dram-calculator-overclocking-dram/320_20#post_26461769"><img alt="View Post" class="inlineimg" src="/img/forum/go_quote.gif"></a><br><br>
Thank you again @1usmus !<br><br><b>The new recommendations in v6 has allowed me to stabilize 3466 CL15 CR1 for the first time with 4 sticks!</b> <img alt="thumb.gif" class="bbcode_smiley" src="http://files.overclock.net/images/smilies/thumb.gif"><br><br>
Just for the record:<br><br><a class="H-lightbox-open" href="http://www.overclock.net/content/type/61/id/3155881/"><img alt="" class="lightbox-enabled" data-id="3155881" data-type="61" src="http://www.overclock.net/content/type/61/id/3155881/width/500/height/1000/flags/LL" style="; width: 500px; height: 224px"></a><br><br>
It has survived 2hs of AIDA64 stress test and I've been playing around and testing games with no problems at all <img alt="biggrin.gif" class="bbcode_smiley" src="http://files.overclock.net/images/smilies/biggrin.gif"><br><br>
While trying to get there I tried every termination block & CAD BUS block recommendation of v6<br><br>
ProcODT 53 > RTT_NOM RZQ/7 > RTT_WR OFF > RTT_PARK RZQ/5 > CAD_BUS AUTO = Failed AIDA64 Stress test in 5 min.<br><br>
ProcODT 53 > RTT_NOM RZQ/7 > RTT_WR OFF > RTT_PARK RZQ/6 > CAD_BUS AUTO = Failed AIDA64 Stress test in 7 min.<br><br>
ProcODT 53 > RTT_NOM RZQ/7 > RTT_WR OFF > RTT_PARK RZQ/7 > CAD_BUS AUTO = Failed AIDA64 Stress test in 10 min.<br><br>
Sticked with the one that survived the longest and started playing around with CAD_BUS.<br><br>
So...<b>ProcODT 53 > RTT_NOM RZQ/7 > RTT_WR OFF > RTT_PARK RZQ/7</b><br><br>
CAD_BUS 24-24-24-24 = Failed AIDA64 Stress test in 10 min. (I guess this would be the same as AUTO)<br><br>
CAD_BUS 30-30-30-30 = Failed AIDA64 Stress test in 18 min.<br><br>
CAD_BUS 30-30-40-60 = Failed AIDA64 Stress test in 14 min<br><br>
CAD_BUS 40-40-40-40 = Survived 2 hs of AIDA64 Stress test, then games without crashing or BSODing<br><br>
Fun fact, never even tried 40-40-40-40 before. It was just a lucky guess I made and tried when I saw that 30-30-40-60 failed earlier than 30-30-30-30.<br><br>
Maybe my RAM @3466 just likes 40 ohms in the CAD BUS? Who knows? <img alt="tongue.gif" class="bbcode_smiley" src="http://files.overclock.net/images/smilies/tongue.gif"></div>
</div>
<br>
very interesting find, I'll try to check your values <img alt="smile.gif" class="bbcode_smiley" src="http://files.overclock.net/images/smilies/smile.gif"> thanks
 

·
Registered
Joined
·
1,488 Posts
Hey just wanted to chime in, i'm using G.SKill RIpjaw V 3600 CL16 Kit x2 (4x8Gb B-Die).<br><br>
When plugging in my kit data i get Mem quality 91%. I presume because out of box it's a low latency 3600 kit<br><br>
The point of this post is purely around me noticing a substantially improved ability to Train my 4 Dimms with a modest speed of 3466 without any effort or tweaking.. system has been operational for 4-6 hours so far no issues, slept, hibernated and woke a couple of times without issue... yet... I will stab test later on. I think my 1700 might be a good one, so I will give 3600 a stab at some point too... Safe preset for sure..<br><br>
I was able to work my way into 3466 1000% HCI some time ago but after a cold start or two i lost the ability to get it going again? It is of note it would take 4-8 attempts to train at 3466 previously.<br><br>
I tried out 3466 Safe i believe and i found much more solid training. i suspect my issue may have been voltage before as i was 3466 CL16 1000% @ 1.41v. Now i am running 1.44v<br><br>
I powered down system and booted a few times from hard off.<br><br>
1st - second training attempt > BOOT<br>
2nd - first training attempt > BOOT<br>
3rd - second training attempt > BOOT<br><br>
i didn't both trying again as previously i'd hit 4-8 attempts with about 50% chance of it default to 1866.. <img alt="tongue.gif" class="bbcode_smiley" src="http://files.overclock.net/images/smilies/tongue.gif"><br><br>
I previously was just using timings directly from my XMP but wasn't hitting CLDO/VDDP as much..<br><br>
either way, here is dump from the Calc i entered in the EFI this time.<br><br>
If i understand the values correct some are variable and we just need to fiddle..<br><br>
DRAM 1.44v<br>
DRAM BOOT 1.44v<br>
VDDP 0.900<br>
CLDO_VDDP 866<br>
tRFC 450<br>
SOC Volt 1.075v<br>
Termination block Ohm > Rec. settings Top to Bot >> 53 > RZQ/7 > OFF > RZQ/5<br>
CAD BUS 30-30-40-60 (this seemed to work for me before..)<br><br>
Every other settings was set as per below screenshts. CPU is @ 3.9Ghz 1.35v (it can run 3.95 at around 1.31-1.33v but i couldn't be bother with any chance of instability this time..)<br><br><a class="H-lightbox-open" href="http://www.overclock.net/content/type/61/id/3156038/"><img alt="" class="lightbox-enabled" data-id="3156038" data-type="61" src="http://www.overclock.net/content/type/61/id/3156038/width/350/height/700/flags/LL" style="; width: 350px; height: 239px"></a><br><a class="H-lightbox-open" href="http://www.overclock.net/content/type/61/id/3156037/"><img alt="" class="lightbox-enabled" data-id="3156037" data-type="61" src="http://www.overclock.net/content/type/61/id/3156037/width/350/height/700/flags/LL" style="; width: 350px; height: 239px"></a><br><a class="H-lightbox-open" href="http://www.overclock.net/content/type/61/id/3156039/"><img alt="" class="lightbox-enabled" data-id="3156039" data-type="61" src="http://www.overclock.net/content/type/61/id/3156039/width/350/height/700/flags/LL" style="; width: 350px; height: 239px"></a><br><a class="H-lightbox-open" href="http://www.overclock.net/content/type/61/id/3156044/"><img alt="" class="lightbox-enabled" data-id="3156044" data-type="61" src="http://www.overclock.net/content/type/61/id/3156044/width/350/height/700/flags/LL" style="; width: 350px; height: 420px"></a>
 

·
Registered
Joined
·
580 Posts
<div class="quote-container" data-huddler-embed="/t/1640919/ryzen-dram-calculator-overclocking-dram/300#post_26462229" data-huddler-embed-placeholder="false"><span>Quote:</span>
<div class="quote-block">Originally Posted by <strong>1usmus</strong> <a href="/t/1640919/ryzen-dram-calculator-overclocking-dram/300#post_26462229"><img alt="View Post" class="inlineimg" src="/img/forum/go_quote.gif"></a><br><br>
I think it is necessary to expose 1,415, there is no threat to the system with a specified voltage<br><br>
Do not pay attention to what BIOS suggests <img alt="wink.gif" class="bbcode_smiley" src="http://files.overclock.net/images/smilies/wink.gif"></div>
</div>
<br>
Thanks,i raise my vSoc to 1.2v and render some times the same file, if i get other error i try with 1.415v in Ram.<br><br>
The soc is used to pciex16+sata+ram, but the principal reason to raise it is ram.<br>
If i use my ram with lower latency, i need more soc voltage? or it only depends to memory frequency?
 

·
Registered
Joined
·
149 Posts
I can train my RAM to 3466CL14 easily, but it usually crashes out on AIDA64 memory stability testing within half an hour. 3333CL14 runs flawless as far as I can tell, any variation of 3600 doesn't even POST so far. Still, 3333mhz with the timings your DRAM Calculator generated runs ****loads better than it did with the D.O.C.P. automatic timings!
 

·
AMD overclocker/developer
Joined
·
1,651 Posts
Discussion Starter #332
<div class="quote-container" data-huddler-embed="/t/1640919/ryzen-dram-calculator-overclocking-dram/320_20#post_26462369" data-huddler-embed-placeholder="false"><span>Quote:</span>
<div class="quote-block">Originally Posted by <strong>ZeNch</strong> <a href="/t/1640919/ryzen-dram-calculator-overclocking-dram/320_20#post_26462369"><img alt="View Post" class="inlineimg" src="/img/forum/go_quote.gif"></a><br><br>
Thanks,i raise my vSoc to 1.2v and render some times the same file, if i get other error i try with 1.415v in Ram.<br><br>
The soc is used to pciex16+sata+ram, but the principal reason to raise it is ram.<br>
If i use my ram with lower latency, i need more soc voltage? or it only depends to memory frequency?</div>
</div>
<br>
I have a stable system at 0.98750, I do not think that values above 1.05 can help <img alt="smile.gif" class="bbcode_smiley" src="http://files.overclock.net/images/smilies/smile.gif"><br><div class="quote-container" data-huddler-embed="/t/1640919/ryzen-dram-calculator-overclocking-dram/320_20#post_26462371" data-huddler-embed-placeholder="false"><span>Quote:</span>
<div class="quote-block">Originally Posted by <strong>Disassociative</strong> <a href="/t/1640919/ryzen-dram-calculator-overclocking-dram/320_20#post_26462371"><img alt="View Post" class="inlineimg" src="/img/forum/go_quote.gif"></a><br><br>
I can train my RAM to 3466CL14 easily, but it usually crashes out on AIDA64 memory stability testing within half an hour. 3333CL14 runs flawless as far as I can tell, any variation of 3600 doesn't even POST so far. Still, 3333mhz with the timings your DRAM Calculator generated runs ****loads better than it did with the D.O.C.P. automatic timings!</div>
</div>
<br>
try it 15 15 15 15 30 54<br><br><div class="quote-container" data-huddler-embed="/t/1640919/ryzen-dram-calculator-overclocking-dram/320_20#post_26462325" data-huddler-embed-placeholder="false"><span>Quote:</span>
<div class="quote-block">Originally Posted by <strong>blair</strong> <a href="/t/1640919/ryzen-dram-calculator-overclocking-dram/320_20#post_26462325"><img alt="View Post" class="inlineimg" src="/img/forum/go_quote.gif"></a><br><br>
Hey just wanted to chime in, i'm using G.SKill RIpjaw V 3600 CL16 Kit x2 (4x8Gb B-Die).<br><br>
When plugging in my kit data i get Mem quality 91%. I presume because out of box it's a low latency 3600 kit<br><br>
The point of this post is purely around me noticing a substantially improved ability to Train my 4 Dimms with a modest speed of 3466 without any effort or tweaking.. system has been operational for 4-6 hours so far no issues, slept, hibernated and woke a couple of times without issue... yet... I will stab test later on. I think my 1700 might be a good one, so I will give 3600 a stab at some point too... Safe preset for sure..<br><br>
I was able to work my way into 3466 1000% HCI some time ago but after a cold start or two i lost the ability to get it going again? It is of note it would take 4-8 attempts to train at 3466 previously.<br><br>
I tried out 3466 Safe i believe and i found much more solid training. i suspect my issue may have been voltage before as i was 3466 CL16 1000% @ 1.41v. Now i am running 1.44v<br><br>
I powered down system and booted a few times from hard off.<br><br>
1st - second training attempt > BOOT<br>
2nd - first training attempt > BOOT<br>
3rd - second training attempt > BOOT<br><br>
i didn't both trying again as previously i'd hit 4-8 attempts with about 50% chance of it default to 1866.. <img alt="tongue.gif" class="bbcode_smiley" src="http://files.overclock.net/images/smilies/tongue.gif"><br><br>
I previously was just using timings directly from my XMP but wasn't hitting CLDO/VDDP as much..<br><br>
either way, here is dump from the Calc i entered in the EFI this time.<br><br>
If i understand the values correct some are variable and we just need to fiddle..<br><br>
DRAM 1.44v<br>
DRAM BOOT 1.44v<br>
VDDP 0.900<br>
CLDO_VDDP 866<br>
tRFC 450<br>
SOC Volt 1.075v<br>
Termination block Ohm > Rec. settings Top to Bot >> 53 > RZQ/7 > OFF > RZQ/5<br>
CAD BUS 30-30-40-60 (this seemed to work for me before..)<br><br>
Every other settings was set as per below screenshts. CPU is @ 3.9Ghz 1.35v (it can run 3.95 at around 1.31-1.33v but i couldn't be bother with any chance of instability this time..)<br><br><a class="H-lightbox-open" href="http://www.overclock.net/content/type/61/id/3156038/"><img alt="" class="lightbox-enabled" data-id="3156038" data-type="61" src="http://www.overclock.net/content/type/61/id/3156038/width/350/height/700/flags/LL" style="; width: 350px; height: 239px"></a><br><a class="H-lightbox-open" href="http://www.overclock.net/content/type/61/id/3156037/"><img alt="" class="lightbox-enabled" data-id="3156037" data-type="61" src="http://www.overclock.net/content/type/61/id/3156037/width/350/height/700/flags/LL" style="; width: 350px; height: 239px"></a><br><a class="H-lightbox-open" href="http://www.overclock.net/content/type/61/id/3156039/"><img alt="" class="lightbox-enabled" data-id="3156039" data-type="61" src="http://www.overclock.net/content/type/61/id/3156039/width/350/height/700/flags/LL" style="; width: 350px; height: 239px"></a><br><a class="H-lightbox-open" href="http://www.overclock.net/content/type/61/id/3156044/"><img alt="" class="lightbox-enabled" data-id="3156044" data-type="61" src="http://www.overclock.net/content/type/61/id/3156044/width/350/height/700/flags/LL" style="; width: 350px; height: 420px"></a></div>
</div>
<br>
you have a really good memory controller, a great rarity of 3466 for 4 modules<br>
I'm waiting for results from 3600 <img alt="smile.gif" class="bbcode_smiley" src="http://files.overclock.net/images/smilies/smile.gif">
 

·
Registered
Joined
·
149 Posts
<div class="quote-container" data-huddler-embed="/t/1640919/ryzen-dram-calculator-overclocking-dram/330#post_26462406" data-huddler-embed-placeholder="false"><span>Quote:</span>
<div class="quote-block">Originally Posted by <strong>1usmus</strong> <a href="/t/1640919/ryzen-dram-calculator-overclocking-dram/330#post_26462406"><img alt="View Post" class="inlineimg" src="/img/forum/go_quote.gif"></a><br><br>
try it 15 15 15 15 30 54</div>
</div>
<br>
I'll give it a try soon, thanks <img alt="smile.gif" class="bbcode_smiley" src="http://files.overclock.net/images/smilies/smile.gif"><br>
For what it's worth this is what I have <a href="https://www.newegg.com/Product/Product.aspx?Item=N82E16820232530" target="_blank">https://www.newegg.com/Product/Product.aspx?Item=N82E16820232530</a><br><br>
Update: it's hit the 30 minute mark without failing - will leave it overnight and see how it goes. Fingers crossed slightly loosening those timings as you suggested will do the trick.
 

·
Registered
Joined
·
721 Posts
<div class="quote-container" data-huddler-embed="/t/1640919/ryzen-dram-calculator-overclocking-dram/300_30#post_26462222" data-huddler-embed-placeholder="false"><span>Quote:</span>
<div class="quote-block">Originally Posted by <strong>1usmus</strong> <a href="/t/1640919/ryzen-dram-calculator-overclocking-dram/300_30#post_26462222"><img alt="View Post" class="inlineimg" src="/img/forum/go_quote.gif"></a><br><br>
.....<br><br>
this article is my guess, to check it need an oscilloscope ... <img alt="frown.gif" class="bbcode_smiley" src="http://files.overclock.net/images/smilies/frown.gif"></div>
</div>
<br>
Well beyond my willingness to pay to eke out 10% more bandwidth, but here is an example of what you need loaded on the oscilloscope that can run it: <a href="https://www.tek.com/datasheet/dpojet-datasheet-1" target="_blank">https://www.tek.com/datasheet/dpojet-datasheet-1</a>
 

·
Registered
Joined
·
247 Posts
Honestly I'm at the point where methinks using memtest86+ is no longer a viable solution, what is everyone else using? I see people talk about memtest but is there other viable options for test memory that isn't that or its pro version? I certainly don't have the patience to sit there and open 16 instances of the freeware version and I rather not pay 14 bucks for the bootable version as nice as it would be.
 

·
Registered
Joined
·
1,488 Posts
<div class="quote-container" data-huddler-embed="/t/1640919/ryzen-dram-calculator-overclocking-dram/330#post_26462406" data-huddler-embed-placeholder="false"><span>Quote:</span>
<div class="quote-block">Originally Posted by <strong>1usmus</strong> <a href="/t/1640919/ryzen-dram-calculator-overclocking-dram/330#post_26462406"><img alt="View Post" class="inlineimg" src="/img/forum/go_quote.gif"></a><br><br>
you have a really good memory controller, a great rarity of 3466 for 4 modules<br>
I'm waiting for results from 3600 <img alt="smile.gif" class="bbcode_smiley" src="http://files.overclock.net/images/smilies/smile.gif"></div>
</div>
<br>
I think i have a good chip all around, it seems to need less voltage than most report when i really try for lower voltages. If you are interested it's a 1724 Malaysia 1700 (I opted for this in the hopes it'd need less voltage, as the 1700 is binned to be lower than 65watts rather than X variants binned for clock speed. Consuming less power to me would indicate a potentially higher circuit quality.<br><br>
I also have better memory than most who are running 4 modules (most buy 3200 or lower from what i've seen) as mentioned these are CL16 3600 kit, they weren't cheap, but well worth it given Ryzen's thirst for memory clock.<br><br>
I'll try 3600 at some point soon to see if it's bootable.... As mentioend this time.. 3466 was only a 1-2 train with your calculator. Thanks a ton for working it all out, it's saved so many people so much time.
 

·
Registered
Joined
·
53 Posts
I'm using Hynix M-die memory. I followed the instructions and I couldn't get any variation of timings to work that came out of the calculator. Hynix is the HQ XMP correct? I basically used the 2400 timings of CL14-16-16-31 at 2933 and that worked better than the timings from the calc. Most of those timings wouldn't even POST, and I spent hours listening to the 3 beeps as I tried damn near every itteration of timings and alternatives from the calc. Except I can't get it 100% stable. Won't pass IBT on high, but will on standard. Also ran a P95 last night while sleeping and woke up to a black screen...<br><br>
So yeah, it's frustrating because I had 2933 stable before and those settings aren't working anymore. I was trying to get 3066 to work, but I know the Hynix (the RAM was free, otherwise I would of bought some Samsung B-die kits) stuff is just not really good beyond 2800 or so. Speaking of 2800, can't seem to get that 100% stable either. I updated to 1201 BIOS, from 0810, so I don't know if it's the BIOS, or the recent windows updates screwing with things.<br><br>
I'm about to say screw it and go back to the DOCP of 2400, which I found to have messed up timings for a few settings that I have to manually tweak. I'll just get some proper B-die stuff when prices have come back down to earth.<br><br>
Overclocking RAM is way more a pain in the butt than I thought it would be. It's a fun challenge, but when you can do certain things on your PC and everything seems fine, but you know there is that slight instability just waiting to make you crash, it's just frustrating and hard to figure out why.
 

·
Registered
Joined
·
1,488 Posts
<div class="quote-container" data-huddler-embed="/t/1640919/ryzen-dram-calculator-overclocking-dram/330#post_26463442" data-huddler-embed-placeholder="false"><span>Quote:</span>
<div class="quote-block">Originally Posted by <strong>Acidstorm</strong> <a href="/t/1640919/ryzen-dram-calculator-overclocking-dram/330#post_26463442"><img alt="View Post" class="inlineimg" src="/img/forum/go_quote.gif"></a><br><br>
So yeah, it's frustrating because I had 2933 stable before and those settings aren't working anymore. I was trying to get 3066 to work, but I know the Hynix (the RAM was free, otherwise I would of bought some Samsung B-die kits) stuff is just not really good beyond 2800 or so. Speaking of 2800, can't seem to get that 100% stable either. I updated to 1201 BIOS, from 0810, so I don't know if it's the BIOS, or the recent windows updates screwing with things.</div>
</div>
<br>
Please fill out your system specs, referring to BIOS revisions when we don't know anything about your system is not going to get you much help or direction to make the right settings changes.<br><br>
I recommend just investing in B-Die as you mentioned. Anything else I don't really think is worth the time (my opinion no offence to anyone without B-Die <img alt="tongue.gif" class="bbcode_smiley" src="http://files.overclock.net/images/smilies/tongue.gif">). But.. Free RAM is free <img alt="biggrin.gif" class="bbcode_smiley" src="http://files.overclock.net/images/smilies/biggrin.gif"> so I'd use it too!!! haha<br><div class="quote-container" data-huddler-embed="/t/1640919/ryzen-dram-calculator-overclocking-dram/330#post_26463203" data-huddler-embed-placeholder="false"><span>Quote:</span>
<div class="quote-block">Originally Posted by <strong>LightningManGTS</strong> <a href="/t/1640919/ryzen-dram-calculator-overclocking-dram/330#post_26463203"><img alt="View Post" class="inlineimg" src="/img/forum/go_quote.gif"></a><br><br>
Honestly I'm at the point where methinks using memtest86+ is no longer a viable solution, what is everyone else using? I see people talk about memtest but is there other viable options for test memory that isn't that or its pro version? I certainly don't have the patience to sit there and open 16 instances of the freeware version and I rather not pay 14 bucks for the bootable version as nice as it would be.</div>
</div>
<br>
I've been using Run MemTest Version 2.5 (Dang Wang version) I cannot remember where i got it.. i am pretty sure thinking about it's just memtest86 :S which won't really help you...<br><br>
This auto starts 16 copies for each thread with a set memory value..<br><br>
HCI is the got to though these days..
 

·
Registered
Joined
·
149 Posts
Seems like setting the timings to 15-15-15-15-30-54 was the key to stability - 14 hours and counting with no errors. Thanks <a data-huddler-embed="href" href="/u/540936/1usmus">@1usmus</a> <img alt="smile.gif" class="bbcode_smiley" src="http://files.overclock.net/images/smilies/smile.gif">
 

·
AMD overclocker/developer
Joined
·
1,651 Posts
Discussion Starter #340
<div class="quote-container" data-huddler-embed="/t/1640919/ryzen-dram-calculator-overclocking-dram/320_20#post_26462850" data-huddler-embed-placeholder="false"><span>Quote:</span>
<div class="quote-block">Originally Posted by <strong>kaseki</strong> <a href="/t/1640919/ryzen-dram-calculator-overclocking-dram/320_20#post_26462850"><img alt="View Post" class="inlineimg" src="/img/forum/go_quote.gif"></a><br><br>
Well beyond my willingness to pay to eke out 10% more bandwidth, but here is an example of what you need loaded on the oscilloscope that can run it: <a href="https://www.tek.com/datasheet/dpojet-datasheet-1" target="_blank">https://www.tek.com/datasheet/dpojet-datasheet-1</a></div>
</div>
<br>
Thank you, with all the desire and respect for this study, it takes a lot of time ... I unfortunately do not have it. Maybe it will be for Christmas holidays<br><div class="quote-container" data-huddler-embed="/t/1640919/ryzen-dram-calculator-overclocking-dram/320_20#post_26463203" data-huddler-embed-placeholder="false"><span>Quote:</span>
<div class="quote-block">Originally Posted by <strong>LightningManGTS</strong> <a href="/t/1640919/ryzen-dram-calculator-overclocking-dram/320_20#post_26463203"><img alt="View Post" class="inlineimg" src="/img/forum/go_quote.gif"></a><br><br>
Honestly I'm at the point where methinks using memtest86+ is no longer a viable solution, what is everyone else using? I see people talk about memtest but is there other viable options for test memory that isn't that or its pro version? I certainly don't have the patience to sit there and open 16 instances of the freeware version and I rather not pay 14 bucks for the bootable version as nice as it would be.</div>
</div>
<br>
i use hci memtest pro (5$) + special launcher<br><br><br><div class="quote-container" data-huddler-embed="/t/1640919/ryzen-dram-calculator-overclocking-dram/320_20#post_26463442" data-huddler-embed-placeholder="false"><span>Quote:</span>
<div class="quote-block">Originally Posted by <strong>Acidstorm</strong> <a href="/t/1640919/ryzen-dram-calculator-overclocking-dram/320_20#post_26463442"><img alt="View Post" class="inlineimg" src="/img/forum/go_quote.gif"></a><br><br>
I'm using Hynix M-die memory. I followed the instructions and I couldn't get any variation of timings to work that came out of the calculator. Hynix is the HQ XMP correct? I basically used the 2400 timings of CL14-16-16-31 at 2933 and that worked better than the timings from the calc. Most of those timings wouldn't even POST, and I spent hours listening to the 3 beeps as I tried damn near every itteration of timings and alternatives from the calc. Except I can't get it 100% stable. Won't pass IBT on high, but will on standard. Also ran a P95 last night while sleeping and woke up to a black screen...<br><br>
So yeah, it's frustrating because I had 2933 stable before and those settings aren't working anymore. I was trying to get 3066 to work, but I know the Hynix (the RAM was free, otherwise I would of bought some Samsung B-die kits) stuff is just not really good beyond 2800 or so. Speaking of 2800, can't seem to get that 100% stable either. I updated to 1201 BIOS, from 0810, so I don't know if it's the BIOS, or the recent windows updates screwing with things.<br><br>
I'm about to say screw it and go back to the DOCP of 2400, which I found to have messed up timings for a few settings that I have to manually tweak. I'll just get some proper B-die stuff when prices have come back down to earth.<br><br>
Overclocking RAM is way more a pain in the butt than I thought it would be. It's a fun challenge, but when you can do certain things on your PC and everything seems fine, but you know there is that slight instability just waiting to make you crash, it's just frustrating and hard to figure out why.</div>
</div>
<br>
I advise you to install a new modified BIOS, it should help<br>
<a href="http://www.overclock.net/t/1640394/unlocked-amd-cbs-for-ryzen-motherboard/0_20">http://www.overclock.net/t/1640394/unlocked-amd-cbs-for-ryzen-motherboard/0_20</a>
 
321 - 340 of 9977 Posts
Top