5 Replies Latest reply on Dec 11, 2016 10:58 PM by savagebeastzero

    Whats the maximum SAFE recommended MV for an nitro r9 390x for daily use.Recommended MV for R9 390x

    paddy4444

      Whats the maximum SAFE MV for an nitro r9 390x for daily use.

        • Re: Whats the maximum SAFE recommended MV for an nitro r9 390x for daily use.Recommended MV for R9 390x
          savagebeastzero

          Hey paddy4444,

           

          To ensure clarification, if by "MV", you're referring to millivolts (mV), then that question isn't quite as simple as one definitive answer. Firstly, simply requesting a figure in millivolts (mV) is extremely vague and isn't directly requesting anything specific. When speaking of electrical measurements with the intention of applying said measurements without requesting a specific aspect of the card you desire to gain knowledge of applying it to (which there are many), you're at risk of damaging the card, your system and/or risk injury to yourself. So, I'm unsure if whether this is simply a request for general information, which it appears to be based on your query or a detailed figure which you are requesting that I submit to you, to assist you in overclocking your graphics card(s) with the intention of operating said adjustments as your daily use multipliers.

           

          So, since you're weren't specific, I'll try to touch on both possible queries, but it's difficult for me to determine any specifics to the latter, I'll explain why later, but I will provide you with the necessary information for you to determine that yourself. Yet, regardless the information I'm writing here is simply touching on both subjects to hopefully get you started and is in no way what I would consider the full wealth of knowledge I could provide if given the time. With that being said, I've been asked by many within the AMD community to begin working on an in depth guide to overclocking for months and the vagueness of your request finally prompted me to begin today.

           

          So, I'll consider this my first entry and will continue to expand it as time goes on, just in a more permanent location here in the forums.

           

          So, to all interested in overclocking, but wanted somewhere to start, read on....you'll be in a slightly better place utilizing this guide as knowledge required to truly understand how overclocking works. If this guide is successful in explaining the fundamentals, it should give you a better understanding of what your trying to achieve and how you're going to achieve it. No more dangerous guess work, you'll be learning, while applying and pending you're utilizing the guide through every step, everything will take place in a much safer environment, reducing risk of damage to your components.

           

          Once this guide is placed in its permanent location, I will continue to expand upon the knowledge it contains. From beginner, intermediate and advanced levels of knowledge that most if not all will be able to learn something from. I hope everyone reading this first edition, enjoys the content. I apologize I haven't included a table of contents and will ensure when I have time to plan its permanent format, it will be curated in proper form. This began as a basic reply and five hours later, here we are.

           

          Wish me luck and thank you for you time.

           

          Out of Box Guaranteed Maximums

           

          If speaking in terms of factory out of box guaranteed maximum operation, then that information may be provided by the manufacturer themselves via their website, through direct correspondence (query) and/or via third party web sites that specialize in such specification details. Generally speaking this information is readily available across the internet, as each model must be able to maintain a set standard of operation. Manufacturers must determine the exact components and circuitry to be used on the board, finalize their designs and submit those also for approval prior to release to meet differing government(s) standards, as each are slightly different. Once said standards are met and approved, the manufacturer begins production and cannot alter the approved designs, unless they once more submit proposed changes, meet government standards and are approved once more.

           

          Thus, you can ascertain that the data available by normal means and to the general public is certifiable.

           

          Please, set fan control to 100% when unexperienced and testing the knowledge you learn below

           

          Please, I ask when attempting to utilize any of the information herein to overclock your system by any of the means described, please always and at all times ensure you leave your fan speed set to 100%, until you are able to determine the maximum operating temperature when under load with each new overclock. Once you deem the work you perform stable, then you can begin to reduce fan speeds by 10% increments, test again under load, if temperatures are deemed safe once more, reduce by another 10% increment etc. Once you determine the maximum fan speed necessary to maintain safe operating temperatures, you can then determine the maximum fan curve needed for daily use.

           

          Manual Overclocks: Understanding how and why overclocks work

           

          Yet, now if you are speaking in regards to assistance in achieving the highest daily use overclocks via manually increasing the millivolts (mV) to your core clocks and memory clocks through the use of a third party or first party overclocking tool, then this gets even more complicated, as no card is the same whatsoever, regardless if they're a reference model or a third party manufacturer design.

           

          When speaking in terms of increasing the millivolts (mV) to the Core Clock with the expected result that you'll receive greater performance, this isn't always necessarily possible or should I say, provide the desired result.

           

          Here's why...

           

          Each card is different as stated above, some require less voltage, are more stable, but offer little in the way of positive (mV) overclocking potential by simply raising it generously via (+mV). This would be categorized as a GPU with High Application Specific Integrated Circuits Quality, also know as a High ASIC Quality. This is a card that the general consumer would die for, even if they didn't know it. It consumes less power, requires less voltage to maintain stability and reaches a higher overclock via standard means of cooling such as that found on a R9 390X Nitro, which possesses an efficient, but general cooling solution. So, so far it sounds like this is the perfect overclocking silicon, you've hit the lottery, right?

           

          No, not exactly, but even that depends on perspective.

           

          Then there's the alternate ASIC Quality rating, Low ASIC Quality. Low ASIC Quality rated silicon consumes far more power, requires much more voltage to maintain stability, receives lower overclocks on standard air cooling solutions, usually requires a substantial memory clock overclock maintain its stability as well but.....here's the caveat, if this seemingly volatile Low ASIC Quality GPU is being maintained properly on a high end enthusiast grade cooling solution such as water/liquid cooling, which is coupled by an equally high end copper block, it can far surpass any overclock fathomable by a High ASIC Quality GPU, thus this makes them highly desirable for those with systems able to handle them.

           

          The true "silicon lottery"  CPU/GPU that so many speak of is truly that rare that the Low ASIC/High ASIC doesn't even truly apply to it, because it consumes far less power, requires far less voltage to maintain stability, can reach and/or exceed even the highest of overclocks comparable to Low ASIC Quality rated silicon and can do this on both cooling solutions because, the temperatures generated by it under load when overclocked are manageable regardless. It wastes far less and produces far more. A true rarity, these gems aren't seen even on a per wafer basis or even close. Thus 99% of overclocking consumers that believe they possess one are simply operating their system with improper settings, because they subconsciously desire to be in possession of one so badly, that they're applying settings that create an unstable environment for their system and aside from basic validation, are never truly stress tested, nor verified.

           

          So, don't be that guy in forums across the net, that swears up and down that you've received one that overclocks 40% higher than even world record holders can achieve, because it just makes you appear ignorant and inexperienced. Overclocking is about science and pushing the limitations of your hardware, even if those limitations are lower than you had hoped on your first try.

           

          It's o.k, you'll get better as you gain experience.

           

          World record overclockers allocate binned silicon for competitions, utilize LN2 for cooling and at times have entire teams to reach those numbers. So remember, you're not competing against them. You're simply challenging yourself to learn and the science of overclocking can become very addictive, as well as rewarding in and of itself. This last paragraph was solely written, because so many new overclockers fall into the trap of feeling insufficient and overwhelmed, because they're surrounded by overclockers that have been doing this like myself for many years. Don't boast, nobody cares. Instead, ask questions and converse with us.

           

          We're all still learning every day to this day and if someone experienced says they've learned everything there is to learn, they're not asking the right questions anymore.

           

          Overclocking via manual voltage increases

           

          Now that you understand what ASIC Quality is, you realize the importance of the data it tells us. So, with that being said. Download GPU-Z and once downloaded to the top right you'll see an icon represented by three horizontal bars. This is the settings menu, select it and once there, you'll see three tabs. We're looking for the "ASIC Quality" tab. Select that tab and you will be presented with your GPU's ASIC Quality. By using everything you learned above, determine your GPU's ASIC Quality.

           

          Once you have determined your GPU's ASIC Quality, you know what direction you're likely headed in, whether it be High ASIC Quality or Low ASIC Quality, you now have data to point you in the right direction.

           

          Now, I'll provide a basic outline of what percentages mean, so you have the tools to determine what your ASIC Quality may be.

           

          High ASIC Quality: As a rough estimate, High ASIC Quality begins at or above +78-80%. Use the data in the previous section to determine what that means and how you should plan your overclock prior to beginning application, testing and stress.

           

          Low ASIC Quality: As a rough estimate, Low ASIC Quality begins at or below -73%. Use the data in the previous section to determine what that means and how you should plan your overclock prior to beginning application, testing and stress.

           

          So what about an ASIC Quality between 73% - 77%? That's somewhere in between and results can vary. Just ensure that you utilize everything within this guide to make your adjustments accordingly. Remember, everything should be applied in small increments at all times in increments of 10.

           

          Video memory overclocking in relation to stable Core Clock overclocking

           

          Now, let's remember memory overclocking utilizes the auxiliary voltage (mV) measurement as well and can on many occasions be the deciding factor as to whether or not your Core Clock adjustments are successful. When improperly adjusting video memory, artifacts will present themselves on your display providing a visual warning that more (+mV) voltage adjustments need to be made for stability to be met. Yet, one must be careful when overclocking video memory, as exceeding the memory clocks without sufficient  (+mV) voltage increases can create system lock ups and if severely out of threshold of operation, the end user may have to operate in safe mode to remove said last applied settings.

           

          So, as you can see, memory overclocking can be sensitive, but if done in increments of 10-15Mhz until you see even the slightest of artifacts and once discovered via your display, increase your memory voltage by 10 (+mV) until the behavior is stabilized. Continue this process until satisfied. Yet remember to ensure to monitor your temperatures, because it's quite easy to go beyond the intended means of operation and fry your card through memory overclocking. So, no need to over due your memory overclocks until you're experienced enough to justify the risks. Simply, stick to overclocking video memory to gain stability for your Core Clocks, because the gains are moderate at best from video memory overclocking alone. So, ensure that it's being done to stabilize, not to gain.

           

          Always remember, the ability of your video memory has just as much to do with your settings as it does the manufacturer. As an example, the company Elpida is known to not provide video memory that overclocks well, but Hynix does an incredibly well at that. Why mention this? They on many occasions have provided AMD with video memory for the same series cards, thus just because you have a certain card, it only means the provider met a minimum standard of quality needed to supply AMD to be awarded a contract, as it did perform within a standard required for the design.  So, if your having a hard time overclocking your video ram, it might simply be that there's a limitation to the hardware itself and nothing more. Just stabilize your Core Clocks and move on if this occurs.

           

          Power Limit/Power Limitation overclocks

           

          Power limit overclocks are much less confusing and less worrisome in comparison to manually adjusting voltages. Generally speaking, the threshold set is safe when set to a maximum of +50% and incrementally increasing your Core Clocks until stability is lost. This safety net only holds true when not manually increasing Core Clock Voltage. So independently without additional manual voltage, it's 99% safe to maximize it to maximum value of +50%. I say 99% safe, because nothing is 100% and no guarantees are made when overclocking. So remember, raise your power limit to a maximum of +50% and adjust your Core Clocks by increments of +10Mhz until you monitor instability. Once instability is monitored, decrease your Core Clock by -10Mhz, stress test the overclock and once determined stable, you've achieved the maximum available stable overclock without having to manually adjust Core Clock Voltage.

           

           

           

          Thank you for taking the time to read this first guide entry if you chose to do so. Once again I apologize for its lack of content, any typo's and unintended mistakes but this wasn't planned as it began as a reply five hours ago and I've been awake for over 30 hours to boot so, forgive me for anything I might have missed while writing it. When I place the guide in its proper permanent location, I will be sure to edit this section, add more detailed information and dedicate more time to ensuring everything is in a working and easily understandable format.

           

          Thanks again and have a nice day all