Nazo
2[H]4U
- Joined
- Apr 2, 2002
- Messages
- 3,672
So I just wanted to make a quick check. Is there any reason I should limit any of the PBO settings as long as the cooler and everything is keeping up?
To explain more specifically, basically I have a pretty good third party cooler and my CPU is underclocked just a little (5600X with max boost set to -250 so it maxes at 4.4GHz -- this puts it on the good side of the voltage increase curve before it goes up exponentially from >4.4 up to the full boost of 4.65 so it stays super cool and hopefully will last me a very very long time.) In a worst case scenario (namely Prime95 small FFTs) it hits about 74C, maybe 75C peak, so this is well below the margin of 80C that is considered best if you want a modern CPU to last a really long time. Of course, temperature is not the only consideration and it still draws a LOT of power in that small FFT test of Prime95 (around 114 watts or so total.) Now, I basically generally want the CPU to perform at its best most of the time and I do rarely do a few things that can actually max it out (such as encoding -- yes I realize Prime95 is not a good real life example, but encoding in particular does push it just about as hard) and would rather it wasn't limited if at all reasonable to keep it this way. However, I note that I have very definitely had to exceed the stock PBO values for it to be able to max out like that at 4.4GHz. For reference, these are the PBO values:
Stock: PPT: 76 TDC: 60 EDC: 90
Motherboard: PPT: 1000 TDC: 160 EDC: 190
Obviously the motherboard limits are set ridiculously high. I presume the point there is to effectively disable precision boost caring about any sensor but temperature since no CPU existing today could ever hit those numbers with, I guess, a simple sanity check in regards of if something were ever made that could it would hopefully not fry the board instantly. Using this "unlimited" profile, I watched the actual usage values while Prime95 testing and thought I may as well at least set some "sane" limits in case something ever goes horribly wrong so the CPU can't just fry itself if somehow it could actually go over reasonable values and put them where each would hit apx 97-98% the maximum values I got in Prime95, so I ended up on these values:
<=98%: PPT: 117 TDC: 79 EDC: 108
This way with something pushing the CPU as hard as normally possible it never exceeds any of those values, but if something could theoretically push any of those values even higher it should start to limit things rather than going all out as the motherboard settings would allow. I'd be willing to bet good money if something could somehow push the numbers the mb limit sets it would fry this CPU. But is that even possible even if something somehow went wrong -- say an infinite loop or something? Will any of these ever limit me in some way while using the CPU in a sane manner even though they were set above what Prime95 going all out produced?
Anyway, my question here is fairly simple otherwise: should I be setting any of those down any from those values I came up with? I've tried to understand them and sort of have a basic idea, but still don't 100% understand what each truly means in regards to how it affects the performance boost. Should I be limiting EDC below the result I got when it was unlimited to reduce total amps, for example? I'm still unclear on exactly what TDC translates to as well. It's supposed to be "limiting current when thermally constricted," which would sort of seem to imply to limit current as temperature increases, but exactly what this even means I don't know given that the stock TDC value of 60 certainly didn't keep it from hitting 90C in Prime95 on all stock settings. In particular, these numbers are quite a bit above the stock values and I wonder if this can affect the CPU's lifetime greatly. I could potentially need this CPU to last me 5+ years (I hope it doesn't come to that, but I have to work with the possibility it could,) but at the same time, I still want it to perform well, so I really don't want to limit it any more than I have to and I do occasionally do things like encoding where it will actually push the CPU around the same sort of range that Prime95 did -- it's not just gaming. (Gaming, btw, uses far far less power and runs really cool around 41C or so in most, so I presume there is no issue there.) And yes, I'm aware that running anything over stock has potential to affect its lifetime, so I'm having to try to find the right balance here where I get performance in those heavy tasks but it won't die on me too soon.
BTW, my curve optimizer values are pretty low for all the cores but one (-30, -29, -30, -30, -21, -30.) I have read where people were saying that lower voltages do equal more current (I presume that the idea is that the net result is it needing roughly the same amount of power to perform the same, so lower voltage would, of course, mean more current to reach the same amount of power.) However, of course, higher voltages equal more temperatures and temperature goes up fast with voltage on the 5000 series CPUs, so I can't raise them a very large amount before it starts to hold temperature ranges I don't want it to in Prime95/encoding/etc. I have read, and I quote "amps kill" so I do sort of wonder if I might be better off with higher CO numbers (or I guess I should say "lower" to be less confusing) in terms of long term lifetime? Is there such a tradeoff -- or if so, is it enough even to matter? I don't relish the idea of running a bunch of tests and value checking to try to figure out the best balance of voltage temperature for current, but I suppose it would be worth it if it translates to the CPU lasting significantly longer. Would the current possibly be more important than the temperature at these sorts of ranges and would setting those CO values higher translate to actually being a lot better for the CPU or is that just not worth doing?
To explain more specifically, basically I have a pretty good third party cooler and my CPU is underclocked just a little (5600X with max boost set to -250 so it maxes at 4.4GHz -- this puts it on the good side of the voltage increase curve before it goes up exponentially from >4.4 up to the full boost of 4.65 so it stays super cool and hopefully will last me a very very long time.) In a worst case scenario (namely Prime95 small FFTs) it hits about 74C, maybe 75C peak, so this is well below the margin of 80C that is considered best if you want a modern CPU to last a really long time. Of course, temperature is not the only consideration and it still draws a LOT of power in that small FFT test of Prime95 (around 114 watts or so total.) Now, I basically generally want the CPU to perform at its best most of the time and I do rarely do a few things that can actually max it out (such as encoding -- yes I realize Prime95 is not a good real life example, but encoding in particular does push it just about as hard) and would rather it wasn't limited if at all reasonable to keep it this way. However, I note that I have very definitely had to exceed the stock PBO values for it to be able to max out like that at 4.4GHz. For reference, these are the PBO values:
Stock: PPT: 76 TDC: 60 EDC: 90
Motherboard: PPT: 1000 TDC: 160 EDC: 190
Obviously the motherboard limits are set ridiculously high. I presume the point there is to effectively disable precision boost caring about any sensor but temperature since no CPU existing today could ever hit those numbers with, I guess, a simple sanity check in regards of if something were ever made that could it would hopefully not fry the board instantly. Using this "unlimited" profile, I watched the actual usage values while Prime95 testing and thought I may as well at least set some "sane" limits in case something ever goes horribly wrong so the CPU can't just fry itself if somehow it could actually go over reasonable values and put them where each would hit apx 97-98% the maximum values I got in Prime95, so I ended up on these values:
<=98%: PPT: 117 TDC: 79 EDC: 108
This way with something pushing the CPU as hard as normally possible it never exceeds any of those values, but if something could theoretically push any of those values even higher it should start to limit things rather than going all out as the motherboard settings would allow. I'd be willing to bet good money if something could somehow push the numbers the mb limit sets it would fry this CPU. But is that even possible even if something somehow went wrong -- say an infinite loop or something? Will any of these ever limit me in some way while using the CPU in a sane manner even though they were set above what Prime95 going all out produced?
Anyway, my question here is fairly simple otherwise: should I be setting any of those down any from those values I came up with? I've tried to understand them and sort of have a basic idea, but still don't 100% understand what each truly means in regards to how it affects the performance boost. Should I be limiting EDC below the result I got when it was unlimited to reduce total amps, for example? I'm still unclear on exactly what TDC translates to as well. It's supposed to be "limiting current when thermally constricted," which would sort of seem to imply to limit current as temperature increases, but exactly what this even means I don't know given that the stock TDC value of 60 certainly didn't keep it from hitting 90C in Prime95 on all stock settings. In particular, these numbers are quite a bit above the stock values and I wonder if this can affect the CPU's lifetime greatly. I could potentially need this CPU to last me 5+ years (I hope it doesn't come to that, but I have to work with the possibility it could,) but at the same time, I still want it to perform well, so I really don't want to limit it any more than I have to and I do occasionally do things like encoding where it will actually push the CPU around the same sort of range that Prime95 did -- it's not just gaming. (Gaming, btw, uses far far less power and runs really cool around 41C or so in most, so I presume there is no issue there.) And yes, I'm aware that running anything over stock has potential to affect its lifetime, so I'm having to try to find the right balance here where I get performance in those heavy tasks but it won't die on me too soon.
BTW, my curve optimizer values are pretty low for all the cores but one (-30, -29, -30, -30, -21, -30.) I have read where people were saying that lower voltages do equal more current (I presume that the idea is that the net result is it needing roughly the same amount of power to perform the same, so lower voltage would, of course, mean more current to reach the same amount of power.) However, of course, higher voltages equal more temperatures and temperature goes up fast with voltage on the 5000 series CPUs, so I can't raise them a very large amount before it starts to hold temperature ranges I don't want it to in Prime95/encoding/etc. I have read, and I quote "amps kill" so I do sort of wonder if I might be better off with higher CO numbers (or I guess I should say "lower" to be less confusing) in terms of long term lifetime? Is there such a tradeoff -- or if so, is it enough even to matter? I don't relish the idea of running a bunch of tests and value checking to try to figure out the best balance of voltage temperature for current, but I suppose it would be worth it if it translates to the CPU lasting significantly longer. Would the current possibly be more important than the temperature at these sorts of ranges and would setting those CO values higher translate to actually being a lot better for the CPU or is that just not worth doing?
Last edited: