[SOLVED] ryzen voltages always maxed even at idle

[jordank_boro]

hey guys, im having a bit of a weird problem. i've searched online but every forum i've looked in people try to make sense of this issue but just end up trailing off into memory and other settings that have nothing to do with cpu voltage haha. the problem im having is that when i use ryzen master to oc my cpu the voltage stays maxed out even if the cpu is idle at 1 percent. my oc settings are 4.4 ghz all core on a ryzen 3600x at 1.45 volts and stable and never going over 72 Celsius under full load with a 240mm aio, temps sit at 32 while idle but voltage is still maxed out at 1.45. i know that 1.45 is pretty much the max i should go and that it will degrade slightly faster so please dont just hit me with that because it's not what im asking. i dont really mind if the cpu lasts 8 years instead of 10 haha. i want to know why at idle my voltage is still maxed out all the time. this seems like if anything it would degrade the cpu much faster if voltage is just stuck at max even though the cpu isnt under load at all. things i have tried are switching the windows power plan to balanced and ryzen balanced and neither did anything once the oc is applied in ryzen master again. i've also amd cool and quiet in bios and then set to pstate0 as well. stll changed nothing sadly. this is my first experience with ryzen so im not sure if this is normal or if this is bad. i've oc'd many systems before but never a ryzen. is there some sort of voltage wall built into the cpu where it wont let voltage throgh unless load goes up. this would make sense in the sense that you cant see the voltage change cause it's not the voltage changing but the wall being placed or removed to let the voltage into the cpu. maybe ryzen just functions differently? any help would be greatly appreciated.

 

[CountMike]

hey guys, im having a bit of a weird problem. i've searched online but every forum i've looked in people try to make sense of this issue but just end up trailing off into memory and other settings that have nothing to do with cpu voltage haha. the problem im having is that when i use ryzen master to oc my cpu the voltage stays maxed out even if the cpu is idle at 1 percent. my oc settings are 4.4 ghz all core on a ryzen 3600x at 1.45 volts and stable and never going over 72 Celsius under full load with a 240mm aio, temps sit at 32 while idle but voltage is still maxed out at 1.45. i know that 1.45 is pretty much the max i should go and that it will degrade slightly faster so please dont just hit me with that because it's not what im asking. i dont really mind if the cpu lasts 8 years instead of 10 haha. i want to know why at idle my voltage is still maxed out all the time. this seems like if anything it would degrade the cpu much faster if voltage is just stuck at max even though the cpu isnt under load at all. things i have tried are switching the windows power plan to balanced and ryzen balanced and neither did anything once the oc is applied in ryzen master again. i've also amd cool and quiet in bios and then set to pstate0 as well. stll changed nothing sadly. this is my first experience with ryzen so im not sure if this is normal or if this is bad. i've oc'd many systems before but never a ryzen. is there some sort of voltage wall built into the cpu where it wont let voltage throgh unless load goes up. this would make sense in the sense that you cant see the voltage change cause it's not the voltage changing but the wall being placed or removed to let the voltage into the cpu. maybe ryzen just functions differently? any help would be greatly appreciated.

That's pretty well how it functions with Ryzen. Seemingly higher idle voltages are not a problem because of light loads. One and most popular way to lower it is to set some negative voltage offset in BIOS, It mostly ranges from .0.05 to 0.1V. nut you'll have to try it yourself and see what works best.
More important are voltages under load, they should drop under 1.3v.

 

[dorsai]

"once the oc is applied in ryzen master"
There's little to be gained by overclocking a 3600x...you'd be better off running at stock settings, or even a slight undervolt, and focusing on cooling to max performance.

If the processor is not idling down the OC settings are interfering, something is running in the background, or the processor c-states have been altered through power plan settings.

Normal operation should see idle voltages well below .5v with the cores listed as "sleep" in Ryzen Master.

 

[jordank_boro]

"once the oc is applied in ryzen master"
There's little to be gained by overclocking a 3600x...you'd be better off running at stock settings, or even a slight undervolt, and focusing on cooling to max performance.

If the processor is not idling down the OC settings are interfering, something is running in the background, or the processor c-states have been altered through power plan settings.

Normal operation should see idle voltages well below .5v with the cores listed as "sleep" in Ryzen Master.

well i hate to break it to you but with the 3600x oc'd to 4.4 all core gives extreme performance gains in games and in benchmarks. for example with no oc and stock settings this cpu scores a much lower cpu score on 3dmark. the box says boosts to 4.4 so i can see why you'd think that. but thats not all core. that means one core will hit 4.4 for sure. in gtaV im getting a good 15 to 20 fps more with the oc and 4.4 all core. so theres a decent ammount of performance to be gained here.

that being said i'd like to keep this oc but i'd also like my cpu to lower the voltage while not in use or under light load. im using ryzen master simple view, is there cstate settings in ryzen master advanced somewhere? my cstate is set properly in the bios like i described above in the original post. but maybe ryzen master is just taking over those settings? i'll try a few things and report back.

 

[Karadjgne]

Run Cinebench R20, both cpu and single. Save your oc to a profile.
Goto guru3d.com and get ClockTuner2 and Dram Calculator, and grab Typhoon Burner.
Use TB to get your ram info and export the values into DC. Take a picture on your phone of that screen. Then goto bios, reset to default, and manually plug in the timings, subtimings, speeds and voltages for ram. Save that under a fresh profile.

Reboot and start up CT2, there's a video explaining the procedure. But basically what it does is OC by reducing VID voltages to a more comfortable level, which allows the Ryzen to boost higher stable. There's 3 profiles you can run, stock, light load, heavy load, with appropriate settings.

On my 3700x,
stock: 3.9GHz all core, 1.475v, 85°C, score 3723
OC: 4.4GHz all core, 1.425v, 85°C, score 5101
CT2 : 4.28GHz all core, 1.205v, 62°C, score 5000

What was really interesting was that the single core score for the CT2 was actually higher by 30 points than the 4.4GHz OC, even running at a lower speed, and thats pretty common with Ryzens and the way they boost with voltages and temps. Since most games rely on a single master thread, and have much to do with single core strengths, even if using multiple cores like GtaV does, a higher single core score can get you better results than a higher all core.

This can be verified using Timespy, look to see exactly what the scores are and when they start dropping, regardless of actual core speeds.

 

[Fiorezy]

Run Cinebench R20, both cpu and single. Save your oc to a profile.
Goto guru3d.com and get ClockTuner2 and Dram Calculator, and grab Typhoon Burner.
Use TB to get your ram info and export the values into DC. Take a picture on your phone of that screen. Then goto bios, reset to default, and manually plug in the timings, subtimings, speeds and voltages for ram. Save that under a fresh profile.

Reboot and start up CT2, there's a video explaining the procedure. But basically what it does is OC by reducing VID voltages to a more comfortable level, which allows the Ryzen to boost higher stable. There's 3 profiles you can run, stock, light load, heavy load, with appropriate settings.

On my 3700x,
stock: 3.9GHz all core, 1.475v, 85°C, score 3723
OC: 4.4GHz all core, 1.425v, 85°C, score 5101
CT2 : 4.28GHz all core, 1.205v, 62°C, score 5000

What was really interesting was that the single core score for the CT2 was actually higher by 30 points than the 4.4GHz OC, even running at a lower speed, and thats pretty common with Ryzens and the way they boost with voltages and temps. Since most games rely on a single master thread, and have much to do with single core strengths, even if using multiple cores like GtaV does, a higher single core score can get you better results than a higher all core.

This can be verified using Timespy, look to see exactly what the scores are and when they start dropping, regardless of actual core speeds.

OC all cores on Ryzen is just useless, even 1usmus the creator of CTR stated there won't be any noticeable difference between 4.1Ghz and 4.4Ghz in day to day usage, it just higher scores in benchmarks and that is it

 

[CountMike]

OC all cores on Ryzen is just useless, even 1usmus the creator of CTR stated there won't be any noticeable difference between 4.1Ghz and 4.4Ghz in day to day usage, it just higher scores in benchmarks and that is it

That is just about true for any OC, except for some rare cases when it's at a border of usability, 100-300MHz is difficult to detect, but........
Modern CPUs including Ryzen have variable frequency with substantial frequency boost on one and/or more cores when SW demands it. Transition to higher frequency happens really so fast and seamless that at times CPU seems to be "nervous" with frequencies jumping from "sleep" states to full boost frequency very often. That of course causes corresponding jumps in voltage and with it heat and is mostly seen as cooler fan(s) speeding up and down many find that pretty irritating.
One of the ways to "tame" it is to manually set frequency and voltage to all cores which doesn't necessarily mean top frequency which boost provides.
As Ryzen boost depends mostly on temperature, other good way is just to provide better (than stock) cooling which can keep max temps up to 70c (even better 65c) and let it rip on auto boost settings.

 

[Fiorezy]

That is just about true for any OC, except for some rare cases when it's at a border of usability, 100-300MHz is difficult to detect, but........
Modern CPUs including Ryzen have variable frequency with substantial frequency boost on one and/or more cores when SW demands it. Transition to higher frequency happens really so fast and seamless that at times CPU seems to be "nervous" with frequencies jumping from "sleep" states to full boost frequency very often. That of course causes corresponding jumps in voltage and with it heat and is mostly seen as cooler fan(s) speeding up and down many find that pretty irritating.
One of the ways to "tame" it is to manually set frequency and voltage to all cores which doesn't necessarily mean top frequency which boost provides.
As Ryzen boost depends mostly on temperature, other good way is just to provide better (than stock) cooling which can keep max temps up to 70c (even better 65c) and let it rip on auto boost settings.

It is how these chips were designed, the spikes in frequencies, voltages and temps are totally normal, and if the fans are ramping up every time you open a program, then you may have to adjust the fan curve to a less agressive settings.

I've been trying to have the best oc possible on my 3700x but I always revert back to stock settings, even PBO is useless, it is just extra heat and power for a negligible performance boost.

 

[CountMike]

It is how these chips were designed, the spikes in frequencies, voltages and temps are totally normal, and if the fans are ramping up every time you open a program, then you may have to adjust the fan curve to a less agressive settings.

I've been trying to have the best oc possible on my 3700x but I always revert back to stock settings, even PBO is useless, it is just extra heat and power for a negligible performance boost.

Without PBO, my 3700x stays close to stock speed which lowers performance by 1/3 and that's not negligible at all. "Secret" to good PBO boost is in temperature, anything under 70c which produces boost to 4.4GHz on single and about 4.25-4.3GHz on rest of cores at about 1.3v (load).
I don't see many reasons not to get performance I paid for. That's why I use a 240 AIO as that is fulfilling requirements.

 

[Karadjgne]

Ryzens start dropping MHz boost on a per core basis starting at 60°C.

Running fixed voltages at 1.3 or higher can be detrimental. Ryzen 3700x run a variable load voltage to the cores. With single core boost, they'll easily hit 1.5v and a 4.4GHz, but as you add cores, as you said, they downclock to 4.3, 4.2 etc until all core is running closer to 3.9-4.05GHz. But by that time you should be closer to 1.2v for extended use.

Many ppl were setting a max turbo locked OC in the 1 & 2 series at 1.325v or higher and burning out the cpu in 6 months.

The 'secret' to Ryzen performance is in letting it behave like a Ryzen, and not treating it like an intel.

For all intents and purposes, PBO on the higher grade cpus does only 1 thing, nothing. Setting a cpu to use power limits that it already does use, does nothing. Setting a 2600 to hit PBO limits turns it into a 2600x, but the 3700x already hits its 65w. Setting a 3800x to hit its 105w PPT is a gain, but setting a 3900x or 3950x is pointless.

So the real 'secret' to a 3700x is in doing the opposite. Chopping VID voltage to the cores, lowering voltage supply, dropping temps and fine tuning the differences so that you achieve max boost under the best scenarios. If you cut the wattage from 65w used, to 55w used, the cpu now has 10 extra watts to play with, which boosts 2-3 cores to 4.4GHz with 4.3 on the rest, instead of single 4.4GHz and 4.1 on the rest.

Treating a Ryzen like an intel doesn't work, use 3dMax Timespy and you'll see the results. It's highly sensitive, even dropping 0.1 on the score is a serious loss of performance. It's very easy to think that higher GHz actually gets you better performance, that's an intel trait, but with Ryzen and infinity fabric and other factors, Higher GHz doesn't always mean higher performance, it can also mean lowered performance, especially in games, as single thread performance can drop.

Think of it this way. You walk across a room, taking slow but long strides. Cover a lot of ground quickly, even as slow as your pace is. A Geisha walks across the same room, feet moving at a furious pace, easily twice as fast as yours, but the stride is so short she might as well be running in place, you end up waiting on her to get where you are. That's a Ryzen. You can slow the GHz and get higher fps with a stronger single thread performance, or raise the GHz and single thread suffers.

Using Timespy can maximize single thread and multi thread performance, no matter what the actual GHz is. I get higher fps from the CT2 4.28 multiplier than with a set 4.4GHz all core, because single core is best performance. Blender shows a slightly higher score on all core, but games aren't using 16 threads, and I'm happier running 62°C gaming than 85°C gaming.

And since I'm running a full custom loop with 2x 240mm rads, hitting 85°C for any reason means there's something seriously wrong with the voltages to get those temps.

 

[CountMike]

Not exactly in my experience Started with first Ryzen 1600x on a x370MB and upgraded to 1700x than to 2700x and now to 3700x on a x470 Asus MB. It was mostly as you said until 3rd gen but with last couple of BIOS versions with AGESA 1201 things changed considerably. Max boost temps went up from 62-65c limit up to 70c over which boost would start dropping by about 50-100MHz for every degree c. Max Voltages under load also dropped to just over 1.3v although idle voltages just touch 1.4v. which is quite acceptable.
My 3700x is one from first batch which CTR 2 characterizes as "Bronze". Newer 3rd gen in average have about 100MHz higher boost with 0.1v lower voltages.
One thing to keep in mind is that boost bursts may last just short time and are greatly influenced by temps which kinda screws up results in benchmarks but can be useful in real world applications because demand also doesn't last long.
My real concern is not actually boost on one one core but frequency on all cores in demand because that's where real world applications work

 

[Karadjgne]

but frequency on all cores in demand

And that's the magic words. With gaming you have a master thread that's continuously high boost, and then multiple others that'll boost up or down as demand waxes and wanes. There's never a constant value. Cpus act more like a pwm fan, boost, stop, boost, stop on a per core basis, not as a DC fan with just setting voltage and leaving it there. It's more efficient, resulting in lower core temps overall, which boosts efficiency.

And yeah, my 3700x is Silver. Ryzens are not Intel. Intel are static cpus, DC fans, set 1 voltage and it's good for any amount of cores, regardless of temps or loads. Ryzens are dynamic cpus, voltages, temps, boosts, loads are all interconnected and constantly changing, and Ryzens do that to give the best performance possible without going over self imposed limits. An intel will happily climb the heat ladder upto its TjMax and then throttle back/shutdown. A Ryzen tries hard not to let you get there in the first place. A game played at lower fps for an hour or more being better performance than a game played at higher fps for 10minhtes before crashing.

So I guess performance really depends on your point of view, I much prefer gaming at 62° and 100fps avg, than gaming at 85° with a 100fps average. Could care less that the first is at 4280MHz and the second at 4400MHz. That's just a number, to me it's more important what those numbers can actually do.

 

[CountMike]

And that's the magic words. With gaming you have a master thread that's continuously high boost, and then multiple others that'll boost up or down as demand waxes and wanes. There's never a constant value. Cpus act more like a pwm fan, boost, stop, boost, stop on a per core basis, not as a DC fan with just setting voltage and leaving it there. It's more efficient, resulting in lower core temps overall, which boosts efficiency.

And yeah, my 3700x is Silver. Ryzens are not Intel. Intel are static cpus, DC fans, set 1 voltage and it's good for any amount of cores, regardless of temps or loads. Ryzens are dynamic cpus, voltages, temps, boosts, loads are all interconnected and constantly changing, and Ryzens do that to give the best performance possible without going over self imposed limits. An intel will happily climb the heat ladder upto its TjMax and then throttle back/shutdown. A Ryzen tries hard not to let you get there in the first place. A game played at lower fps for an hour or more being better performance than a game played at higher fps for 10minhtes before crashing.

So I guess performance really depends on your point of view, I much prefer gaming at 62° and 100fps avg, than gaming at 85° with a 100fps average. Could care less that the first is at 4280MHz and the second at 4400MHz. That's just a number, to me it's more important what those numbers can actually do.

Yes, of course it's better to work or game with lower voltages and temps but not at the expense of performance when it actually matters. There are workloads and games that would work just as good and fast with half the performance or less of particular CPU but also ones that let's say 3700x is not enough, that's why I like to have some or as much as possible reserve. I have SW that can easily overpower 5950x but it's rare so additional expense is not warranted. Ok so I loose half or an hour to compile but if it had to be on daily or hourly basis I'd have to switch to more expensive equipment
If you are getting same results but have that much variance in temps, specially 80c+ something must be wrong as I with single 20AIO do neot go over 70c even during hours of use at full PBO..

 

[Karadjgne]

I don't use pbo, or level 1 or level 2 or level 3. Don't even use the added 200MHz OC.

But, it's a difference of wattage vs soak. On a full custom loop I've also got the upto 230w added in for the 2070Super, which is considerably higher output than the 3700x alone on an aio. And I don't use max fans or max pump, that little ddc gets a little noisy at 4400rpm, I prefer it around 1800-2000.

So I could drop temps quite a bit if I bumped up the rpms on the loop, especially at higher loads, at @ 550-600w capacity I have plenty to spare, about double what's actually a bare necessity.

But temps did drop 20°C, just with tweaks and the gaming performance is better on average with the tweaks than with a full OC, most likely due to the single core performance difference and my games not using more than 6 threads. So I get the 4.4GHz on most threads anyway.

Which is what I've been trying to tell Op, he says he 'paid for the performance' so that's what he wants, I'm just saying that performance is a relative term, not necessarily a MHz number. Higher core speeds on a Ryzen do not necessarily mean higher performance.

Like the HP curve in a motor. You might get Peak HP at 5500rpm, but from 5500-7000 that HP actually drops. So slamming the gas pedal and waiting to shift until you hit 7k on the tach, is robbing you of overall speeds, should be shifting at 6000rpm to maintain the peak HP range between gears.

https://community.amd.com/t5/red-te...-master-and-quot-performance-quot/td-p/117235

Interesting read. I got the exact same performance as that author at stock, but using CT2 I'm getting 5000 at 4.28 all core 1.232v vs his 4709 at 4.07GHz at 1.34v.
His single core is 494, mine is 536.

Tweaked performance and still better than 4.4GHz 5101 all core, 503 single core and 1.375v with pbo and 200MHz boosts.

Just a different way of skinning a cat...

 

[Fiorezy]

I don't use pbo, or level 1 or level 2 or level 3. Don't even use the added 200MHz OC.

But, it's a difference of wattage vs soak. On a full custom loop I've also got the upto 230w added in for the 2070Super, which is considerably higher output than the 3700x alone on an aio. And I don't use max fans or max pump, that little ddc gets a little noisy at 4400rpm, I prefer it around 1800-2000.

So I could drop temps quite a bit if I bumped up the rpms on the loop, especially at higher loads, at @ 550-600w capacity I have plenty to spare, about double what's actually a bare necessity.

But temps did drop 20°C, just with tweaks and the gaming performance is better on average with the tweaks than with a full OC, most likely due to the single core performance difference and my games not using more than 6 threads. So I get the 4.4GHz on most threads anyway.

Which is what I've been trying to tell Op, he says he 'paid for the performance' so that's what he wants, I'm just saying that performance is a relative term, not necessarily a MHz number. Higher core speeds on a Ryzen do not necessarily mean higher performance.

Like the HP curve in a motor. You might get Peak HP at 5500rpm, but from 5500-7000 that HP actually drops. So slamming the gas pedal and waiting to shift until you hit 7k on the tach, is robbing you of overall speeds, should be shifting at 6000rpm to maintain the peak HP range between gears.

https://community.amd.com/t5/red-te...-master-and-quot-performance-quot/td-p/117235

Interesting read. I got the exact same performance as that author at stock, but using CT2 I'm getting 5000 at 4.28 all core 1.232v vs his 4709 at 4.07GHz at 1.34v.
His single core is 494, mine is 536.

Tweaked performance and still better than 4.4GHz 5101 all core, 503 single core and 1.375v with pbo and 200MHz boosts.

Just a different way of skinning a cat...

Mind sharing your tweaks?

 

[Karadjgne]

I did. I used typhoon burner to pull up my ram info and exported that into dram calculator (saves a bunch of typing and everything is assigned correctly).
Dram calculator then gave me a bunch of timing adjustments for the subtimings on my ram. I took a picture of those by phone, then changed all the appropriate timings and subtimings in bios and saved a user profile. About 30 or so different settings, but the affect is getting the data in the door, through the ram, then out the other side much faster, without any collisions.

That sped up the ram a good amount, more than just dropping the main 5 timings down a couple notches. Which can be important for Ryzens especially, since infinity fabric is tied to ram.

Then ran ClockTuner2 several tkmes (its lengthy) with a couple different settings for clocks and voltages. The 3700x has 2 CCX, and CT2 adjusts both seperately, not as a single cpu, as often 1 chip can be slightly better at OC than the other. That got me the best performance at VID and temps for all core heavy usage. Saved P2. Did it again for lighter usage, with higher clocks. Saved P1.

So now I have 3 actual power states (if you want to call them that). The cpu runs bare stock for anything less than 18%. So most anything running 1-2 cores simple like web surfing gets max boost and or full idle as assigned by AMD. 19-54% loads get the higher clocks, max boost, minimum voltage, low temps, so I'm getting 4.3-4.4GHz on 4-5 cores and 4.1-4.2GHz on the rest. At 55% or higher, I'm getting 4.28-4.4 on all cores, depending on the amount of load.

I'm allowing the Ryzen to behave like a Ryzen, act like a Ryzen, so its still under governing itself, all I've done is stuck it on a leash and not allowing it to run amuck. It's now a gray hound chasing the rabbit on a track instead of a mutt chasing squirrels on the park.

With a standard OC, you set voltage and other settings, reboot, test for 10 minutes, reboot, make adjustments, reboot, test for another 10 minutes... rinse and repeat all that multiple times. With CT2, it does all that for you, set it up as youd want, start it going, and go grab a drink or a smoke. When you come back it has an answer. Might be better or slightly worse, but you'll get best stable speeds/vid in the range you specified.

 

[Fiorezy]

I did. I used typhoon burner to pull up my ram info and exported that into dram calculator (saves a bunch of typing and everything is assigned correctly).
Dram calculator then gave me a bunch of timing adjustments for the subtimings on my ram. I took a picture of those by phone, then changed all the appropriate timings and subtimings in bios and saved a user profile. About 30 or so different settings, but the affect is getting the data in the door, through the ram, then out the other side much faster, without any collisions.

That sped up the ram a good amount, more than just dropping the main 5 timings down a couple notches. Which can be important for Ryzens especially, since infinity fabric is tied to ram.

Then ran ClockTuner2 several tkmes (its lengthy) with a couple different settings for clocks and voltages. The 3700x has 2 CCX, and CT2 adjusts both seperately, not as a single cpu, as often 1 chip can be slightly better at OC than the other. That got me the best performance at VID and temps for all core heavy usage. Saved P2. Did it again for lighter usage, with higher clocks. Saved P1.

So now I have 3 actual power states (if you want to call them that). The cpu runs bare stock for anything less than 18%. So most anything running 1-2 cores simple like web surfing gets max boost and or full idle as assigned by AMD. 19-54% loads get the higher clocks, max boost, minimum voltage, low temps, so I'm getting 4.3-4.4GHz on 4-5 cores and 4.1-4.2GHz on the rest. At 55% or higher, I'm getting 4.28-4.4 on all cores, depending on the amount of load.

I'm allowing the Ryzen to behave like a Ryzen, act like a Ryzen, so its still under governing itself, all I've done is stuck it on a leash and not allowing it to run amuck. It's now a gray hound chasing the rabbit on a track instead of a mutt chasing squirrels on the park.

With a standard OC, you set voltage and other settings, reboot, test for 10 minutes, reboot, make adjustments, reboot, test for another 10 minutes... rinse and repeat all that multiple times. With CT2, it does all that for you, set it up as youd want, start it going, and go grab a drink or a smoke. When you come back it has an answer. Might be better or slightly worse, but you'll get best stable speeds/vid in the range you specified.

Thank you for sharing this.

I actually tried CTR since day 1, I thought that it just oc all cores at a fixed vid, not sure about what you said regarding acting like stock by having adaptive vid and clocks depending on loads.

Regarding RAM, I already got a stable oc from 3200Mhz CL16 to 3600Mhz CL16.

 

[Karadjgne]

Ahh. There's a section in CT2, profiles. When you get a profile you like, you save it as P1 or P2 (there's a 3rd for 5000cpus it looks like). Then you can run the different profiles automatically. It starts out with stock, then light load, then heavy loads. The light and heavy are fixed vid but the boost will be variable depending on loads and cores and temps.

My 3600 ram is at 3600, fclock and uclock and mclock are fixed at 1800, and the timings adjusted tighter. Didn't OC the speeds as its not necessary.

Timings are like entering a door, crossing a room, exiting a door. The subtimings are how fast you open the door, how long you wait for the door to open, how fast you actually go through, how long it takes to close the door. Your main timings are how fast your legs move (MHz speed), how many strides to cover the distance etc. 3600 at cl18 is slower than 3200 at cl14. Legs might move faster, but you take many more short strides vrs the slower MHz longer and less strides.

So if you can cut down on the delays without hitting the unopened door, speed up enters and closing doors etc then the whole process of moving data in, through and out is overall much faster. Often times considerably faster than just bumping up the MHz.

Basically my 3600/16 is acting closer to 3600/14 timings, somewhat faster than 3200/14 ram that costs more.

 

[Fiorezy]

Ahh. There's a section in CT2, profiles. When you get a profile you like, you save it as P1 or P2 (there's a 3rd for 5000cpus it looks like). Then you can run the different profiles automatically. It starts out with stock, then light load, then heavy loads. The light and heavy are fixed vid but the boost will be variable depending on loads and cores and temps.

My 3600 ram is at 3600, fclock and uclock and mclock are fixed at 1800, and the timings adjusted tighter. Didn't OC the speeds as its not necessary.

Timings are like entering a door, crossing a room, exiting a door. The subtimings are how fast you open the door, how long you wait for the door to open, how fast you actually go through, how long it takes to close the door. Your main timings are how fast your legs move (MHz speed), how many strides to cover the distance etc. 3600 at cl18 is slower than 3200 at cl14. Legs might move faster, but you take many more short strides vrs the slower MHz longer and less strides.

So if you can cut down on the delays without hitting the unopened door, speed up enters and closing doors etc then the whole process of moving data in, through and out is overall much faster. Often times considerably faster than just bumping up the MHz.

Basically my 3600/16 is acting closer to 3600/14 timings, somewhat faster than 3200/14 ram that costs more.

So I used CTR 2.0 RC 5, managed to get 4300Mhz@1.325mv P1 and 4150Mhz@1244mv P2 (Bronze simple), highest temp was 58c.
My only concern was the profiles not activating on Windows startup and I have to activate them manually, any tips?

 

[Karadjgne]

In CT2 there's a profile page, and the settings tab. Once you save the profiles and activate them they run automatically. There's a setting to enable CT2 to run the profile at windows startup.

 

[Fiorezy]

In CT2 there's a profile page, and the settings tab. Once you save the profiles and activate them they run automatically. There's a setting to enable CT2 to run the profile at windows startup.

I already enabled those settings but whenever I reboot my PC, CTR auto starts but the profiles doesn't activate and I need to activate them manually.

 

[Karadjgne]

There's 2 switches on the right side, are both enabled on both profiles? And are you checking HWInfo at the load amounts per profile? You'll run stock for a decent % first before the first profile kicks in.
In the center of that profile it'll show the load %. Enabling the profile manually is an override if done under that %.

 

[Fiorezy]

There's 2 switches on the right side, are both enabled on both profiles? And are you checking HWInfo at the load amounts per profile? You'll run stock for a decent % first before the first profile kicks in.
In the center of that profile it'll show the load %. Enabling the profile manually is an override if done under that %.

Am I missing something?