Page 46 of 191 FirstFirst ... 36 44 45 46 47 48 56 96 146 ... LastLast
Results 451 to 460 of 1907
  1. #451
    ROG Enthusiast Array
    Join Date
    Nov 2018
    Reputation
    10
    Posts
    55

    TBH, I don't really find it a good idea for Asus to have PE enabled as default for stock cooler use.

    For a Ryzen+ under PB2/PBO, either the Windows Balanced power setting should be used, or the Ryzen power plan with minimum cpu frequency set to the same 5% instead of default 90%. By doing this, it's possible to keep those cores with lower loads more steadily on a lower speed/voltage that gives room for PB2/PBO to push the most loaded threads to a higher frequency.

    If you check on oc .net the VII thread, the experience with 32GB is mixed. There's one user running a 2x16GB kit of B-die at 3533 C16, but C14 is a no-go; 4 single sided sticks don't seem to fare better. 2x8GB B-die (single sided) seems the only way to go for high speeds.

    New agesa revisions for ryzen+ have done a lot for older ryzen users in respect to memory compatibility (even and mostly on older boards), so by the time those new 32GB sticks are on the market along with Ryzen 2's, we might get benefits for our current systems.

    Edit/PS: I really hope asus brings a wmi sensor interface for this board and the prime, in the same way it's being tested on the VII. The current fan fail situation is worrying for anyone who overclocks and does long stress tests, or uses their system for long renders left unattended. All because there is also no shutdown temp that can be set on bios.
    Last edited by AlleyViper; 11-07-2018 at 05:11 PM.

  2. #452
    ROG Guru: Yellow Belt Array
    Join Date
    Jul 2018
    Reputation
    10
    Posts
    218

    Quote Originally Posted by AlleyViper View Post
    TBH, I don't really find it a good idea for Asus to have PE enabled as default for stock cooler use.

    For a Ryzen+ under PB2/PBO, either the Windows Balanced power setting should be used, or the Ryzen power plan with minimum cpu frequency set to the same 5% instead of default 90%. By doing this, it's possible to keep those cores with lower loads more steadily on a lower speed/voltage that gives room for PB2/PBO to push the most loaded threads to a higher frequency.

    If you check on oc .net the VII thread, the experience with 32GB is mixed. There's one user running a 2x16GB kit of B-die at 3533 C16, but C14 is a no-go; 4 single sided sticks don't seem to fare better. 2x8GB B-die (single sided) seems the only way to go for high speeds.

    New agesa revisions for ryzen+ have done a lot for older ryzen users in respect to memory compatibility (even and mostly on older boards), so by the time those new 32GB sticks are on the market along with Ryzen 2's, we might get benefits for our current systems.

    Edit/PS: I really hope asus brings a wmi sensor interface for this board and the prime, in the same way it's being tested on the VII. The current fan fail situation is worrying for anyone who overclocks and does long stress tests, or uses their system for long renders left unattended. All because there is also no shutdown temp that can be set on bios.
    Thanks, your post is chock full of info! I wasn't aware of no shutdown temp in bios, if the cpu overheats or if the cpu fan stops dont all modern motherboards automatically shut down? I had that feature even with my old AsRock build from 2007 that was done with the Core 2 Q6600 quadcore CPU. And if it doesn't automatically shut down if the cpu overheats or the cpu fan stops spinning is there a way to do that using third party software? I wonder if uninstalling monitoring software like AI Suite and HWiNFO is enough to ensure the fan stopping/cpu overheating bug won't happen? It did get rid of the black screen/motherboard freeze bug. I'm only using Core Temp and Afterburner from the task bar and HWMonitor on demand now. Another thing I'm worrying about is if using monitoring software actually shortens the life of the sensors on the board and the fan headers in the process.

    I'm really excited about the new memory, hopefully we can make full use of Windows 10 Home's ability to access up to 128GB of memory at some point. I see that G. Skill came out with a new 2x8GB combo for Ryzen that is clocked above 4000 so it looks like that that 2x8GB B-Die combo is still the best like you said. Would you say that 3533 at C16 is the limit for what can be done using 2x16GB memory? And I wonder if 2x16GB RAM has some stability issues.

    I switched to Windows Balanced and have seen it do wonders for stability and temps, now my CPU core temps stay in the 40s and 50s. The only thing I cant get it to seem to do is put the computer in sleep mode if I leave a simulation running for a few hours and fall asleep. Is there any way to force sleep mode if I dont register any keyboard or mouse activity for x number of hours? For now I just enabled a task in task scheduler to automatically shut the computer off late at night in case I fall asleep first.

    I only did some minor (but necessary) bios modifications, as on this motherboard for whatever reason extreme overclocking happens if the settings are left on auto which is why I put PE on default and modified the fan curves so the CPU fan wasn't constantly spinning at 6000 RPM and sounding like a jet propeller lol, now it spins at 3000 RPM or less and the temperatures are about the same. Even running simulations all night my CPU core temps stay in the 50s and my GPU temps stay in the 50s also. I haven't found a way to turn PE off if I'm only surfing the net- is there a way to do that? I hope having PE on doesn't shorten the life of the CPU or other components.

  3. #453
    ROG Enthusiast Array
    Join Date
    Nov 2018
    Reputation
    10
    Posts
    55

    Yes, it should shut down at some point or at least start to throttle a lot (>80ºC?), while still keeping a very high temp if it doesn't reach critical shutdown point. It's just that I'd prefer to set a lower limit before that even starts to happen, as older award bios would let you.
    I haven't much time on this board, but for what I understand is that it doesn't like both the asus EC to be polled a lot, or having the other ITE ic being polled concurrently by two apps causing it to crash along with fan control (and maybe for a long time too?). Those threads you've linked should have a lot of info about it. Don't worry about the IC's life shorten, that really shouldn't happen. IIRC in some thread someone also proved that the problem isn't even caused by an overheating IC.

    I really don't know the limits of this platform (very new to it, didn't search much), neither I have a top bin b-die kit or done any relevant testing other than finding my "defaults". Just went with a 3200C14 b-die 2x8 kit that is known to play well since Ryzen "1" days. I suggest you take a look at the CHVII thread or the ryzen memory calc thread on overclock.net (or around here), where you'll find many users testing a good variety of top hardware. That 3533 C16 on 16GBx2was just an example after a very quick search, so you might find more encouraging results.

    When you refer (further) disabling PE, I'm not sure I understand what you mean. Performance Enhancement levels 1-4 are a series of overrides to PB2/XFR2 (the stock function that sets your voltage/clock automatically according to load, max tdp available and temperature), that change how default performance boost works by raising some limits (via PBO and other tweaks see here). Setting it to "default", as you have it, leaves these tweaks "disabled" (while "Auto" seems to set lv1 or 2, behaving like 10x scalar PBO), so your CPU uses the its stock PB2/XF2 algorithm without extra leeway.

    To further cool your CPU (even without PE/PBO active), you can set a negative voltage offset that will lower the max voltages applied by stock PB2/XFR2. How low you can go will depend on your CPU, but something like -0,05 to -0,1V could be possible while maintaining stability. This might also result in faster speeds being applied depending on load, due to lower power consumption at same speed (also a must under PBO ).

    If you wish to tame your stock precision boost while browsing, you can even make a new Power Profile with a cap under 100% on maximum cpu state. But there's no need for it. If you're not messing with PE/PBO, etc, the higher voltages it sustains by very short time periods are expected by your CPU design.

    Regarding auto sleep, please check if a task like this will help (link), using hibernate alternatively (with /h instead of /s). For a proper sleep command go here and copy the "Standby" instruction (not the "Sleep" just bellow). At worst, you can set it to sleep or hibernate at a fixed time with that command if your program interferes with the idle timer.
    Last edited by AlleyViper; 11-08-2018 at 06:53 AM.

  4. #454
    ROG Guru: Yellow Belt Array
    Join Date
    Jul 2018
    Reputation
    10
    Posts
    218

    Quote Originally Posted by AlleyViper View Post
    Yes, it should shut down at some point or at least start to throttle a lot (>80ºC?), while still keeping a very high temp if it doesn't reach critical shutdown point. It's just that I'd prefer to set a lower limit before that even starts to happen, as older award bios would let you.
    I haven't much time on this board, but for what I understand is that it doesn't like both the asus EC to be polled a lot, or having the other ITE ic being polled concurrently by two apps causing it to crash along with fan control (and maybe for a long time too?). Those threads you've linked should have a lot of info about it. Don't worry about the IC's life shorten, that really shouldn't happen. IIRC in some thread someone also proved that the problem isn't even caused by an overheating IC.

    I really don't know the limits of this platform (very new to it, didn't search much), neither I have a top bin b-die kit or done any relevant testing other than finding my "defaults". Just went with a 3200C14 b-die 2x8 kit that is known to play well since Ryzen "1" days. I suggest you take a look at the CHVII thread or the ryzen memory calc thread on overclock.net (or around here), where you'll find many users testing a good variety of top hardware. That 3533 C16 on 16GBx2was just an example after a very quick search, so you might find more encouraging results.

    When you refer (further) disabling PE, I'm not sure I understand what you mean. Performance Enhancement levels 1-4 are a series of overrides to PB2/XFR2 (the stock function that sets your voltage/clock automatically according to load, max tdp available and temperature), that change how default performance boost works by raising some limits (via PBO and other tweaks see here). Setting it to "default", as you have it, leaves these tweaks "disabled" (while "Auto" seems to set lv1 or 2, behaving like 10x scalar PBO), so your CPU uses the its stock PB2/XF2 algorithm without extra leeway.

    To further cool your CPU (even without PE/PBO active), you can set a negative voltage offset that will lower the max voltages applied by stock PB2/XFR2. How low you can go will depend on your CPU, but something like -0,05 to -0,1V could be possible while maintaining stability. This might also result in faster speeds being applied depending on load, due to lower power consumption at same speed (also a must under PBO ).

    If you wish to tame your stock precision boost while browsing, you can even make a new Power Profile with a cap under 100% on maximum cpu state. But there's no need for it. If you're not messing with PE/PBO, etc, the higher voltages it sustains by very short time periods are expected by your CPU design.

    Regarding auto sleep, please check if a task like this will help (link), using hibernate alternatively (with /h instead of /s). For a proper sleep command go here and copy the "Standby" instruction (not the "Sleep" just bellow). At worst, you can set it to sleep or hibernate at a fixed time with that command if your program interferes with the idle timer.
    I prefer a lower limit too, in my experience a threshold that high can shorten the life of components even if it doesn't kill the CPU. Is that your experience too and why you'd like to set a lower limit too? I'd like to keep it at 70C. Fortunately I haven't experienced that particular problem and it sounds like my "black screen death"/motherboard freezing former problem with the fans still spinning and the lights still on is the preferable problem because it's just inconvenient while the fans stopping is downright dangerous. The power supply I'm using (Thermaltake Smart Power 850W) doesn't have thermal protection and is rated for only 40C although it has been review tested to work well up to 50C. To set a lower limit than 80C on the cores I might have to search around for some software that does that although that might cause problems for the sensor also, perhaps the monitoring software that I use for core temps (Core Temp) already has that option built in. It seems to be stable with this motherboard so I might as well just stick with that and Afterburner which I'm already using.

    I saw a thread on reddit where users posted the memory overclock settings that were known to work with Ryzen, I think thats what you mean by memory calc thread? I saw there were quite a few more options at 2x8GB than there are at 2x16GB.

    Thanks for the explanation on PE, so leaving it to default should keep CPU temps lower than the auto setting and is probably the best setting to use for regular browsing? It helped to keep my temps lower and the CPU fan doesn't spin all the way at max. Besides doing that I couldn't find any other way to disable PE or PBO. I'll try the slightly lower voltages and see what that does. I also keep my memory on auto, I wonder if that's a good idea rather than setting it at its rated speed. When I do run HWMonitor to see the speeds of the different cores, I see some as low as around 2.1 while others are as high as 4.4. I guess this is normal behavior and does not place any extra stress on the cores or shorten the life of the CPU?

    I was thinking of making a new power profile and capping max CPU state to 90% on it but like you said it's not needed because the voltage changes are part of the design and should not shorten the life of the CPU or any other components.

    Thanks for the links on hibernate and sleep. When I figured out that the motherboard crashing had nothing to do with the computer's sleep mode I decided to re-enable it however it hasn't come on since but what you linked to should help. I was wondering about sleep mode vs hibernate mode but I read somewhere that to not put extra writes on the SSD drive the two things that should be done is to move the page file to a mechanical HD and to use sleep mode rather than hibernate, since hibernate puts active memory on the SSD rather than sleep mode which puts it on RAM (although there should be a way to put hibernate active memory on the other HD, which is mechanical in my case.) Is there a difference in standby vs sleep mode? I thought the possibilities were sleep, hibernate and hybrid mode.
    Last edited by Alex0915; 11-08-2018 at 06:30 AM.

  5. #455
    ROG Enthusiast Array
    Join Date
    Nov 2018
    Reputation
    10
    Posts
    55

    Yes, but I'd force a shutdown at 80ºC, not much sooner. AVX loads can go easily above 70 under OC. AMD says max temp for my 2600X is 95ºC, and 85ºC for the 2700X, not that I'd like to see a CPU reach them. I shoot for under 75ºC in stress testing (so daily max loads are kept at 60's).

    The ryzen memory calculator thread is the support thread for that soft intended to optimize timings. Memory speed maters on Ryzen because it also dictates the speed for the Infinity Fabric, reducing latency in between cores (so 3000MHz is enough to have some gains from it). So if auto is setting you memory speed lower than rated, you could try to either use DOCP, or set main timings/speed/voltage manually (that's what DOCP does according to XMP). Beware that the higher the memory speed, the higher the SOC voltage will be auto set and you might not need it so high. On the linked thread, there should be people with a lot of knowledge to help you if you wish to go above rated, or tighten secondary timings.

    Yes, PE set on Auto is a form of overclocking (it sets lvl1 or 2, in the same way of changing PBO settings). Having it on "Default" disables it, and allows you to run cooler. Hwmonitor isn't the best for checking cpu speeds, you'll see more precise clocks under hwinfo. Those clocks are the expected for a 2700X. The small extra above 4.3Ghz is set by XFR2 according to (good) thermals and available tdp; non X CPUs don't do that above the regular PBO2 range.

    Standby/Sleep usually refers to the same. I just pointed the "Standby" instruction on that post, because it has a workaround that permits to execute a regular non-hybrid "sleep" via command line without permanently disabling hibernation beforehand.
    Last edited by AlleyViper; 11-08-2018 at 08:19 AM.

  6. #456
    ROG Guru: Yellow Belt Array
    Join Date
    Jul 2018
    Reputation
    10
    Posts
    218

    Quote Originally Posted by AlleyViper View Post
    Yes, but I'd force a shutdown at 80ºC, not much sooner. AVX loads can go easily above 70 under OC. AMD says max temp for my 2600X is 95ºC, and 85ºC for the 2700X, not that I'd like to see a CPU reach them. I shoot for under 75ºC in stress testing (so daily max loads are kept at 60's).

    The ryzen memory calculator thread is the support thread for that soft intended to optimize timings. Memory speed maters on Ryzen because it also dictates the speed for the Infinity Fabric, reducing latency in between cores (so 3000MHz is enough to have some gains from it). So if auto is setting you memory speed lower than rated, you could try to either use DOCP, or set main timings/speed/voltage manually (that's what DOCP does according to XMP). Beware that the higher the memory speed, the higher the SOC voltage will be auto set and you might not need it so high. On the linked thread, there should be people with a lot of knowledge to help you if you wish to go above rated, or tighten secondary timings.

    Yes, PE set on Auto is a form of overclocking (it sets lvl1 or 2, in the same way of changing PBO settings). Having it on "Default" disables it, and allows you to run cooler. Hwmonitor isn't the best for checking cpu speeds, you'll see more precise clocks under hwinfo. Those clocks are the expected for a 2700X. The small extra above 4.3Ghz is set by XFR2 according to (good) thermals and available tdp; non X CPUs don't do that above the regular PBO2 range.

    Standby/Sleep usually refers to the same. I just pointed the "Standby" instruction on that post, because it has a workaround that permits to execute a regular non-hybrid "sleep" via command line without permanently disabling hibernation beforehand.
    Thanks thats interesting, I wonder why the max temp for the 2700X is lower- because it's an octocore CPU? Do you know what the max temp for the 2700 is? Also if you are shooting for a max temp of 75C with your 2600X should I shoot for a max temp of 65C with my 2700X since its rated max temp is 10C lower? Or should I shoot for similar max and shutdown temps as you do?

    I'm very cautious with monitoring software now because of the monitor crashes, I actually dont use either HWMonitor or HWiNFO anymore because they were both causing them and in the HWiNFO forums I saw a thread where the developer said they had worked with Asus to develop a solution but Asus had yet to implement it in its bios. So I completely uninstalled HWiNFO and only run HWMonitor for very short periods of time since leaving it on for hours also brings about the issue. From when I was running them together they seemed to display the same values. Right now the only monitoring software that I run from start up are Core Temp and Afterburner. I got rid of AI Suite and AI Charger too, since I never use them and wanted to be cautious. Core Temp and Afterburner seem to work fine and display the same values as the other software I had, and yet they do not cause problems with this motherboard.

    Thanks for the help with the memory I was going to experiment with the settings a bit to see how far they can go, from what I understand OC'ing the memory also increases the voltage to the CPU. How does Auto determine what speed to set the memory at- does it automatically adjust it based on temperatures or does it function the same way PE Auto does?

    I like how the CPU is so power efficient, the cores that are getting the usage are the ones being boosted while the others are being held back. My CPU temps in the 40s and 50s and GPU temps in the 50s are nice to see, and I have copper heat pipes going through both (4 through the CPU cooler and 2 through the GPU.) Most of the time the GPU twin fans aren't even needed. I was wondering had I gone the 2700 route could I have just had it passively cooled, but the price difference was so small that I couldn't resist the 2700X.

    Thanks for the sleep mode links, so is the recommendation to use sleep mode vs hibernation a good one to reduce the number of writes to the SSD drive? I moved my page file to the mechanical hard drive for the same reason. Only other thing I need to figure out with regard to this is how to cut off the lights to my RGB keyboard during sleep mode, it doesn't have that option in the software (neither does my mouse for that matter.) The workaround I am using is to put it into reactive ripple mode so the keys only light up when I press one.
    Last edited by Alex0915; 11-10-2018 at 06:00 AM.

  7. #457
    ROG Enthusiast Array
    Join Date
    Nov 2018
    Reputation
    10
    Posts
    55

    On amd.com each CPU page has max temps stated on the info at the end. 95ºC for 2700.
    Manufacturing process on both CPUs is the same, I have no idea why max temps are different. Could be because the 2700x has more area of the dies in use, so it'll be harder to dissipate heat at the hottest spots that could be reaching a higher temp than reported (vs a 2600x), or due to pulling higher voltage/amps by default at >4300, so the cap is lower to avoid damage. Same vs a 2700, at higher temps a X could be using more voltage and drawing more amps due to higher pb2/xfr limits, so a lesser temp is required to prevent damage. Still, I'd shoot for under 70ºC max for all these parts in regular use, also to ensure proper boosting. The example I gave of 72-73ºC in my cpu is an extreme consumption situation (all thread avx2 under prime95 with PBO/PE2, not to mention the very old 12cm tower type cooler), it doesn't reflect regular workloads, but only stress testing.

    Beware if your hwmonitor or hwinfo sensor crashes were happening because AB was running at the same time. Hmonitor and hwinfo display different values for mhz on some boards I own, including the strix f (but voltages and temperatures are similar). On these hwmonitor usually has a lot of fluctuation on the base clock, affecting all other reported locks that will sometimes show abnormally high values.

    Auto settings are usually a mystery, as they rely on some formula Asus uses to ensure stability, usually with a conservative approach (too much voltage) to account for the worst behaving hw. As an example, if I set DOCP that will place my memory at 3200, cpsu VSOC will be pushed to 1.15V. If I set that field to auto (instead of DOCP) and memory to 3200, it'll be at 1.10, and if it's set to manual, auto voltage on SOC will be 1.15V. When soc voltage set to offset, without any value set then or a very short value, you'll see that for 3200 default is 1.1V instead of 1.15V. Makes sense? It doesn't to me. So the best is to observe how values left on auto react to other manual settings and make adjustments if required. Something like 1.0-1.05V vsoc should the be enough at 3200 on a ryzen+, maybe 1.1V at most, only testing will tell.
    I don't believe the main cpu vcore voltage on auto is changed with the memory speed setting, only the cpu's vsoc voltage.

    To know what Auto does to your memory speeds, the only way is experimenting. It'll probably use its higher SPD value and/or up to 2933 (the max officially supported by your cpu), that will be a lower speed for compatibility under low voltage when compared to the XMP profile. When using DOCP, memory speed, voltage, and main timings will be set according (or close) to the XMP profile (the one your memory is rated at by the manufacturer/model). You can also set those exact values manually. Forgot to link also this thread for help on memory settings.

    If you're worried about reducing drive writes, both hibernate and hybrid sleep will write to the hiberfil.sys on the root of c: (it's not on the pagefile). Setting powercfg -hibernate off on your command prompt will disable both, and then "sleep" will be the regular and faster sleep mode available on older OS'es without extra writing. The disadvantage to sleep vs hybrid sleep, is that hybrid sleep (as with hibernate) will still store to your drive some ram data, so that if it isn't possible to resume from sleep as expected, some data loss can be prevented. I have no clue about what's stored or possible to recover, so you'll have to search about it for more info.
    Unless you use your pagefile a lot and heavily due to being out of ram (reaching constant high bandwidth writes and loads during extended periods on it), the regular writes it does aren't nothing to worry about.
    Last edited by AlleyViper; 11-11-2018 at 02:07 AM.

  8. #458
    ROG Guru: Yellow Belt Array
    Join Date
    Jul 2018
    Reputation
    10
    Posts
    218

    I have just been using it moderately so far and my temps peak at around 65C but are mostly between the low 40s and upper 50s. Is that normal? I'm wondering if I need to adjust my fan curves. So if the cpu reaches 85C my computer will automatically shut down? If I dont like the peak temps around 65C with just moderate usage what else can I do to keep temps in the 50s? Edit the Balanced Power Plan maybe and reduce max percentage to 90%?


    I am going to experiment on different memory settings and timings and see how they affect the cpu and gpu, I downloaded CPU-Z and GPU-Z to look at that. Is GPU-Z's video memory usage monitoring accurate?

    As far as reducing SSD writes is concerned, when in Windows Balanced Mode after a certain period of inactivity does the computer go into regular sleep mode or is that hybrid sleep? I'm wondering if I need to use that command to disable hibernation and hybrid mode. I dont think my pagefile is an issue because I have 16 GB of RAM so that should mean the page file gets used very little? I have it set to a constant size of 10 GB on my other HD.
    Last edited by Alex0915; 11-13-2018 at 08:03 PM.

  9. #459
    ROG Enthusiast Array
    Join Date
    Nov 2018
    Reputation
    10
    Posts
    55

    Your temps are more than fine for a 8 core, no need to make it underperform. Regarding exceedingly high temperatures, at some point you CPU should throttle (cut frequency and voltage to run very slow and contain temperature) and up to the max limit it'll cause a shutdown.

    GPU-Z's video memory usage is related to the dedicated gddr memory usage by your GPU, not your DDR4. If a Shared memory field shows, that is the value for graphics that are temporarily kept on system memory (when not everything fits on the card), so they are swapped to the video card when needed.

    If you want to make the best out of your sticks, you'll find specialized help on the threads I've linked -- specially for going above rated specs, which requires knowing IC behavior. Under the Memory tab on GPU-Z, you can check your current memory speed and primary timings, but for all secondaries a more complete program is ryzen timing checker (link). If current settings are considerably worse than your sticks are rated for (on the cpu-z's SPD tab you can check their XMP profile), you have to make adjustments on your bios to set them along with rated voltage. Setting DOCP and the correct memory profile (under Ai OC tuner) on your bios should take care of that automatically. Then comes testing, p.e. ex. running 16 instances of HCI memtest, each one testing something like 856MB for a combined 13696 MB (for 16GB) to leave just enough memory free for you OS to idle while testing without constant swapping to pagefile, until they all complete at least 400% coverage without errors.

    If hibernation isn't disabled via that command or if you disable it inside the advanced properties of your power profile, your sleep mode will always be hybrid sleep mode by default. I wouldn't worry much about it, though.
    Last edited by AlleyViper; 11-14-2018 at 02:40 AM.

  10. #460
    ROG Guru: Yellow Belt Array
    Join Date
    Jul 2018
    Reputation
    10
    Posts
    218

    Quote Originally Posted by AlleyViper View Post
    Your temps are more than fine for a 8 core, no need to make it underperform. Regarding exceedingly high temperatures, at some point you CPU should throttle (cut frequency and voltage to run very slow and contain temperature) and up to the max limit it'll cause a shutdown.

    GPU-Z's video memory usage is related to the dedicated gddr memory usage by your GPU, not your DDR4. If a Shared memory field shows, that is the value for graphics that are temporarily kept on system memory (when not everything fits on the card), so they are swapped to the video card when needed.

    If you want to make the best out of your sticks, you'll find specialized help on the threads I've linked -- specially for going above rated specs, which requires knowing IC behavior. Under the Memory tab on GPU-Z, you can check your current memory speed and primary timings, but for all secondaries a more complete program is ryzen timing checker (link). If current settings are considerably worse than your sticks are rated for (on the cpu-z's SPD tab you can check their XMP profile), you have to make adjustments on your bios to set them along with rated voltage. Setting DOCP and the correct memory profile (under Ai OC tuner) on your bios should take care of that automatically. Then comes testing, p.e. ex. running 16 instances of HCI memtest, each one testing something like 856MB for a combined 13696 MB (for 16GB) to leave just enough memory free for you OS to idle while testing without constant swapping to pagefile, until they all complete at least 400% coverage without errors.

    If hibernation isn't disabled via that command or if you disable it inside the advanced properties of your power profile, your sleep mode will always be hybrid sleep mode by default. I wouldn't worry much about it, though.
    I dont see a shared memory field but I see a separate field for system memory and one for video memory. The system memory is at 4.5 GB and the video memory one is at 1.5 GB. Video card usage is around 25% and video card temp is around 57C even without the fans spinning. The fans only spin occasionally unless the GPU is over 60C. All these copper pipes are doing a world of difference.
    I've never had a computer stay this cool before I just need to make sure too much dust doesn't get inside since I live in a very dusty area.

    Thanks for that memory program I am getting that now. I also use Ryzen Master but that's primarily for cpu speeds.

    I have so much running on here that neither sleep nor hibernation is happening, but the monitors do power off. If neither hibernation or sleep get activated will cpu usage/speeds/voltage automatically be reduced if there is no keyboard or mouse input from me?
    Last edited by Alex0915; 11-14-2018 at 05:57 AM.

Page 46 of 191 FirstFirst ... 36 44 45 46 47 48 56 96 146 ... LastLast

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •