Ryzen Owners Zen Garden

[Gmr_Chick]

In all seriousness, should I take the top cover off and see what's going on, leave it be, or RMA as a last resort?

 

[freeagent]

I would just leave it be.. but to me it looks normal? But I don't know what normal looks like..

I think I have my daily settings hammered out.. I'm sure stock is fine too..

Haven't run it yet.. First boot as fast as she goes lol..

So when running PBO +200 -30 I noticed a CCD was getting hammered pretty hard. In the end I saw 13c difference between the two. It wasn't completely stable that high, but +150 -30 is working good with no crashing or unexpected behavior so far. Single core up to 5100.. I bet the new gen will do x55 no problem.. maybe.

 

[wolf]

Got my hands on a 5900X finally, only 225 days after ordering!

My Noctua NH-L12 can't really keep up with it with stock power limits, so it's currently a smidge above ECO modes 87w max, sitting limited to 95w. Temps are well in check now, gaming seems totally unaffected, only all-core loads are lower @ about 3.5ghz.

Any good guides to follow to tune up PBO2 per core curve optimiser? will this work in conjunction with the power limit and claw back some speed? is it worth it? or should I be putting a better cooler on so I can let it run at 142w+ and then try PBO2?

Will the 1usmus Ryzen clock tuner work with the power limit set so low? I also forget where to even go to see if I got a bronze/silver/gold etc sample, and will that be affected by my power limiting?

Any advice on getting the most out of it without spending extra money appreciated!

 

[Space Lynx]

Got my hands on a 5900X finally, only 225 days after ordering!

My Noctua NH-L12 can't really keep up with it with stock power limits, so it's currently a smidge above ECO modes 87w max, sitting limited to 95w. Temps are well in check now, gaming seems totally unaffected, only all-core loads are lower @ about 3.5ghz.

Any good guides to follow to tune up PBO2 per core curve optimiser? will this work in conjunction with the power limit and claw back some speed? is it worth it? or should I be putting a better cooler on so I can let it run at 142w+ and then try PBO2?

Will the 1usmus Ryzen clock tuner work with the power limit set so low? I also forget where to even go to see if I got a bronze/silver/gold etc sample, and will that be affected by my power limiting?

Any advice on getting the most out of it without spending extra money appreciated!

i know it seems obvious but sometimes people do forget. increase fan curves in mobo bios for entire system just a slightly more aggressive curve than stock can do wonders.

 

[tabascosauz]

Got my hands on a 5900X finally, only 225 days after ordering!

My Noctua NH-L12 can't really keep up with it with stock power limits, so it's currently a smidge above ECO modes 87w max, sitting limited to 95w. Temps are well in check now, gaming seems totally unaffected, only all-core loads are lower @ about 3.5ghz.

Any good guides to follow to tune up PBO2 per core curve optimiser? will this work in conjunction with the power limit and claw back some speed? is it worth it? or should I be putting a better cooler on so I can let it run at 142w+ and then try PBO2?

Will the 1usmus Ryzen clock tuner work with the power limit set so low? I also forget where to even go to see if I got a bronze/silver/gold etc sample, and will that be affected by my power limiting?

Any advice on getting the most out of it without spending extra money appreciated!

Congrats! Not too sure about CTR, if you run all-core on the L12S I suspect you're looking at just 4.0GHz or so to keep thermals in check.

Based on my experience with the 5900X+C14S in the Cerberus and someone I know with a 5900X+L12S in the Ghost S1, you're probably better off selling the L12S and getting either a C14S or U12A. Which one is better depends on how you want to set up your NR200 and how important GPU thermals are to you.

You've got plenty of clearance so unless you just can't spare the money, not much reason to take the Fuma 2 over the U12A.

But basically the consensus is that the L12S isn't really enough for the 5900X. It's not really the CPU's fault either, it wasn't exactly sufficient for a 3900X either. Only thing that changed was that the 5000s make more aggressive use of the thermal envelope they're allowed to use (90C) so it feels like they run hotter.

Curve Optimizer doesn't seem to reduce temps much or at all, it's more to move the V-F curve to get more clock out of the same Vcore. Anyways, there's a Curve Optimizer thread on TPU:

AMD Curve Optimizer any guides / experience

I'm testing my curve optimizer undervolts with CoreCycler, and it turns out my new 5600x's Core 1 returns rounding errors even with no offset (negative offset set to 0). Does this mean this core is defective (and if so, is it worth it to activate my warranty just for this)? Or should I attempt...

www.techpowerup.com

I'm running -10 on my two priority cores, -12 on the other CCD1 cores, and -15 on the rest. It's not too hard to work out, just test with the corecycler script for a couple of iterations. That's the limit of the 2 top cores, but the rest of the corws I haven't gone to see what the limit is.

GitHub - sp00n/corecycler: Script to test single core stability, e.g. for PBO & Curve Optimizer on AMD Ryzen or overclocking/undervolting on Intel processors

Script to test single core stability, e.g. for PBO & Curve Optimizer on AMD Ryzen or overclocking/undervolting on Intel processors - sp00n/corecycler

github.com

 

[Final_Fighter]

Got my hands on a 5900X finally, only 225 days after ordering!

My Noctua NH-L12 can't really keep up with it with stock power limits, so it's currently a smidge above ECO modes 87w max, sitting limited to 95w. Temps are well in check now, gaming seems totally unaffected, only all-core loads are lower @ about 3.5ghz.

Any good guides to follow to tune up PBO2 per core curve optimiser? will this work in conjunction with the power limit and claw back some speed? is it worth it? or should I be putting a better cooler on so I can let it run at 142w+ and then try PBO2?

Will the 1usmus Ryzen clock tuner work with the power limit set so low? I also forget where to even go to see if I got a bronze/silver/gold etc sample, and will that be affected by my power limiting?

Any advice on getting the most out of it without spending extra money appreciated!

225 days!! holly... i just ran into a best buy that had a bunch of them in stock, along with the 5800 and 5600. looks like stock is almost back to normal.

 

[wolf]

i know it seems obvious but sometimes people do forget. increase fan curves in mobo bios for entire system just a slightly more aggressive curve than stock can do wonders.

Yeah I run Argus with custom fan curves, mostly because with the de-shrouded card, the case fans cool it so I needed case fans to run their curve based on GPU temp, so I already made the CPU curve quite agressive.

Congrats! Not too sure about CTR, if you run all-core on the L12S I suspect you're looking at just 4.0GHz or so to keep thermals in check.

Based on my experience with the 5900X+C14S in the Cerberus and someone I know with a 5900X+L12S in the Ghost S1, you're probably better off selling the L12S and getting either a C14S or U12A. Which one is better depends on how you want to set up your NR200 and how important GPU thermals are to you.

You've got plenty of clearance so unless you just can't spare the money, not much reason to take the Fuma 2 over the U12A.

But basically the consensus is that the L12S isn't really enough for the 5900X. It's not really the CPU's fault either, it wasn't exactly sufficient for a 3900X either. Only thing that changed was that the 5000s make more aggressive use of the thermal envelope they're allowed to use (90C) so it feels like they run hotter.

Curve Optimizer doesn't seem to reduce temps much or at all, it's more to move the V-F curve to get more clock out of the same Vcore. Anyways, there's a Curve Optimizer thread on TPU:

AMD Curve Optimizer any guides / experience

I'm testing my curve optimizer undervolts with CoreCycler, and it turns out my new 5600x's Core 1 returns rounding errors even with no offset (negative offset set to 0). Does this mean this core is defective (and if so, is it worth it to activate my warranty just for this)? Or should I attempt...

www.techpowerup.com

I'm running -10 on my two priority cores, -12 on the other CCD1 cores, and -15 on the rest. It's not too hard to work out, just test with the corecycler script for a couple of iterations. That's the limit of the 2 top cores, but the rest of the corws I haven't gone to see what the limit is.

GitHub - sp00n/corecycler: Script to test single core stability, e.g. for PBO & Curve Optimizer on AMD Ryzen or overclocking/undervolting on Intel processors

Script to test single core stability, e.g. for PBO & Curve Optimizer on AMD Ryzen or overclocking/undervolting on Intel processors - sp00n/corecycler

github.com

Cheers for this! all-core doesn't phase me too-too much, even @ 3.5ghz this will still be significantly faster in my video editing loads than my 3700X which topped at 3.9-4ghz all core. If I get a better cooler, however, I will aim to push that higher.

My main focus is game performance, which seems unaffected by lowering the wattage the CPU can draw, but I would be all about trying to squeeze more out of it still, would you say optimising the curve will help allow individual cores to boost higher/long etc within the power envelope I've established?

225 days!! holly... i just ran into a best buy that had a bunch of them in stock, along with the 5800 and 5600. looks like stock is almost back to normal.

Yeah mine was a day 3 order from a local store here in Aus, so right around filling all the day 1 orders the next batch covered me and now the store is in surplus, along with the 5800X and 5600X tho, albeit at what I'd call inflated pricing still.

 

[tabascosauz]

Yeah I run Argus with custom fan curves, mostly because with the de-shrouded card, the case fans cool it so I needed case fans to run their curve based on GPU temp, so I already made the CPU curve quite agressive.

Cheers for this! all-core doesn't phase me too-too much, even @ 3.5ghz this will still be significantly faster in my video editing loads than my 3700X which topped at 3.9-4ghz all core. If I get a better cooler, however, I will aim to push that higher.

My main focus is game performance, which seems unaffected by lowering the wattage the CPU can draw, but I would be all about trying to squeeze more out of it still, would you say optimising the curve will help allow individual cores to boost higher/long etc within the power envelope I've established?

Yeah mine was a day 3 order from a local store here in Aus, so right around filling all the day 1 orders the next batch covered me and now the store is in surplus, along with the 5800X and 5600X tho, albeit at what I'd call inflated pricing still.

You definitely should use Curve Optimizer, it's pretty much free performance at no drawback except time spent in stability testing. I run -10 on my best 2 cores and see about up to ~100MHz better effective clocks on the regular, at same power and same temps. What's the core ranking order on your chip?

95W PPT should be good for most games, I can think of a handful of games that max out two cores though and probably draw 100W or just north of that. Might affect clocks a little bit, but nothing too much and again it depends on what you play. On these 2CCD chips there seems to be about 15W of power that just disappears into a black hole (probably estimates from some minor rails), not showing up in the cores or SOC domains but still part of package power draw.

Go in increments on 5 (negative offset) and test at every step until the corecycler script tells you which core fails first (probably will be one of your good CCD1 cores), then leave that core there and keep going on the others. Best to run at least 2 iterations (each iteration cycling through all the cores once) to determine whether a specific setting is stable.

But yeah, definitely compared to an 8-core MT perf will be stronger (I had the same 3700X>5900X upgrade), but defs consider a better cooler since you will have no trouble fitting one. Honestly, all-core temps aren't bad at all at the stock 142W PPT (70-75C on air), 2-6 core gaming loads get quite a bit hotter.

 

[wolf]

You definitely should use Curve Optimizer, it's pretty much free performance at no drawback except time spent in stability testing. I run -10 on my best 2 cores and see about up to ~100MHz better effective clocks on the regular, at same power and same temps. What's the core ranking order on your chip?

The core ranking I believe is in Ryzen Master? If so it shows;

Core 3 and 8 with a gold star
Core 1 and 11 with a silver circle
All the rest ..n ormal?

95W PPT should be good for most games, I can think of a handful of games that max out two cores though and probably draw 100W or just north of that. Might affect clocks a little bit, but nothing too much and again it depends on what you play. On these 2CCD chips there seems to be about 15W of power that just disappears into a black hole (probably estimates from some minor rails), not showing up in the cores or SOC domains but still part of package power draw.

I could probably stretch this cooler to 105w quite easily, maybe even ~120w and it'll still cope.

Go in increments on 5 (negative offset) and test at every step until the corecycler script tells you which core fails first (probably will be one of your good CCD1 cores), then leave that core there and keep going on the others. Best to run at least 2 iterations (each iteration cycling through all the cores once) to determine whether a specific setting is stable.

But yeah, definitely compared to an 8-core MT perf will be stronger (I had the same 3700X>5900X upgrade), but defs consider a better cooler since you will have no trouble fitting one. Honestly, all-core temps aren't bad at all at the stock 142W PPT (70-75C on air), 2-6 core gaming loads get quite a bit hotter.

Cool I'll look into that cheers! and indeed with the temps, they seem much more directly linked to voltage rather than power draw.

All 12 cores loaded @ 3.5 GHz ~1.02v @ 95w flat is ~65-70c
One/two cores loaded @ 4.7 - 4.95GHz ~ 1.48v @ 95w is 70c+

 

[tabascosauz]

The core ranking I believe is in Ryzen Master? If so it shows;

Core 3 and 8 with a gold star
Core 1 and 11 with a silver circle
All the rest ..n ormal?

HWInfo shows you the complete CPPC/hardware order. Ryzen Master just identifies the best 2 cores in each CCX - not very useful since CCD2 is usually lower quality.

I could probably stretch this cooler to 105w quite easily, maybe even ~120w and it'll still cope.

Cool I'll look into that cheers! and indeed with the temps, they seem much more directly linked to voltage rather than power draw.

All 12 cores loaded @ 3.5 GHz ~1.02v @ 95w flat is ~65-70c
One/two cores loaded @ 4.7 - 4.95GHz ~ 1.48v @ 95w is 70c+

I actually thought the L12S would be a little bit cooler. I had one for a little bit, it's a great performer on a 65W chip and probably on par or better than the U9S, and the U9S was used by a few SFF people on 3900X with relatively reasonable temps. But that other 5900X+L12S owner I mentioned couldn't run his chip without throttling, without resorting to eco mode, so that put it into perspective for me.

I hit about the same 60-80ish temps in heavy 1/2-core loads depending on game, seems to just be by design. I'm still logging and looking at the temp data. If you're just looking for the minimum 142W+CO performance maybe a U12S or U12S Redux could even manage? My C14S should be around there in performance, but with the airflow setup I have and the fan it's probably closer to the U14S right now.

 

[ratirt]

HWInfo shows you the complete CPPC/hardware order. Ryzen Master just identifies the best 2 cores in each CCX - not very useful since CCD2 is usually lower quality.

View attachment 202513



I actually thought the L12S would be a little bit cooler. I had one for a little bit, it's a great performer on a 65W chip and probably on par or better than the U9S, and the U9S was used by a few SFF people on 3900X with relatively reasonable temps. But that other 5900X+L12S owner I mentioned couldn't run his chip without throttling, without resorting to eco mode, so that put it into perspective for me.

I hit about the same 60-80ish temps in heavy 1/2-core loads depending on game, seems to just be by design. I'm still logging and looking at the temp data. If you're just looking for the minimum 142W+CO performance maybe a U12S or U12S Redux could even manage? My C14S should be around there in performance, but with the airflow setup I have and the fan it's probably closer to the U14S right now.

Which game makes your CPU hit 60-80? Just curious.

 

[Space Lynx]

225 days!! holly... i just ran into a best buy that had a bunch of them in stock, along with the 5800 and 5600. looks like stock is almost back to normal.

I think things are coming back to normal yeah, I have had my rx 6800 and 5600x rig on sale for tpu for a few days now. no offers. i guess i was a little too late in selling. oh well. i can just keep it if worst comes to worst. im not willing to take a loss over what i paid

 

[tabascosauz]

Which game makes your CPU hit 60-80? Just curious.

The same MW19 and Genshin Impact I've been babbling on about too much lately unfortunately, turns out none of the other games I have actually work the CPU much at all, so it's just these two that hit CCD1 this hard.

MW19 hits 2 cores with AVX, resulting in the unpredictable fast spikes in temp between 60-80C. Genshin loads one core moderately (4.3-4.7) most of the time at about 50-70C, unless loading in or loading into a coop world where it'll actually max out two cores (4.7-4.9) for about 90 seconds precisely, steady at 75-78C for that duration.

I heard BFV also is similar to MW19 on Ryzen 5000 thanks to AVX, but I don't have the game. Others on forums running even bigger air coolers (DRP4, D15) corroborate almost the same temps and behaviour in MW19 but no one has chimed in here yet.

I can't replicate any of this outside of games in any benchmarks.

The most telling part is that max fan speed, an obnoxious 2000rpm on mine, makes zero difference to the MW19 temp behaviour, even though max fan speed lowers 2-3C in every synthetic benchmark.

 

[ratirt]

The same MW19 and Genshin Impact I've been babbling on about too much lately unfortunately, turns out none of the other games I have actually work the CPU much at all, so it's just these two that hit CCD1 this hard.

MW19 hits 2 cores with AVX, resulting in the unpredictable fast spikes in temp between 60-80C. Genshin loads one core moderately (4.3-4.7) most of the time at about 50-70C, unless loading in or loading into a coop world where it'll actually max out two cores (4.7-4.9) for about 90 seconds precisely, steady at 75-78C for that duration.

I heard BFV also is similar to MW19 on Ryzen 5000 thanks to AVX, but I don't have the game. Others on forums running even bigger air coolers (DRP4, D15) corroborate almost the same temps and behaviour in MW19 but no one has chimed in here yet.

I can't replicate any of this outside of games in any benchmarks.

The most telling part is that max fan speed, an obnoxious 2000rpm on mine, makes zero difference to the MW19 temp behaviour, even though max fan speed lowers 2-3C in every synthetic benchmark.

Just wanted to ask about the games but unfortunately, I don't own any of these you have pointed out.
So far with the games I've tested the CPU, it has never shown a temperature above 55 degrees Celsius.
I'll keep looking though.
Need to try Horizon Zero Dawn since that might be more demanding title.

 

[tabascosauz]

So far with the games I've tested the CPU, it has never shown a temperature above 55 degrees Celsius.
I'll keep looking though.
Need to try Horizon Zero Dawn since that might be more demanding title.

You do have an AIO right? Not magical, but water is certainly better at dealing with Ryzen heat density.

I mean, that is understandable. Most games won't be CPU-bound, and most games load 1 core. Anything else is like that. CSGO, Siege, War Thunder, ROR2, Minecraft (except with shaders), SE4, MCC, etc. 50-65C, perfectly logical. In ST benchmarks that max out 1 core, temps are steady around 60C, perfectly logical.

Genshin is only that way because the game is hardcoded to 60fps, GPU is basically idle.

And MW19 doesn't seem to care about "CPU-bound". Regardless of framecap, refresh rate, or whether the GPU is at 100%, makes zero difference to how CPU behaves. The behaviour doesn't really extend to Ryzen 3000.

Maybe someone will be able to do some HWInfo logging in BFV if they don't have COD.

The frustrating thing about games that don't have consistent 2-core load is that data from 1-core loads are impossible to plot. Ryzen constantly juggles single core load between the #1 and #2 ranked cores every few seconds, you can imagine what that load/freq/power/temp graph looks like.

 

[freeagent]

Some interesting results with PBO off while using Curve Optimzer.. playing with it right now. I haven’t checked on how these are clocking because I wanted a surprise lol.. but if I were to guess it’s not a stellar example. So I kind of gave up on PBO +200 and a full negative 30. It will do 2K fclk 1:1 but no more and it gets a bit fussy. It will do 1900 1:1 with 4 sticks with ease.. so a little weird but I haven’t tested thoroughly because I have been trying to dial in clocks like a 5600. Only problem with that is it isn’t a 5600 and I can’t really treat it like one. So now I am testing stock clock range with a -30 curve and PBO on auto which I assume is off.. only thing left to do really is per core clocking which I am not sure I will do.. at least right now.

Outside of that it is a monster that doesn’t need to be clocked. She rips just fine @ all stock all around. Hit 99 with it once with an all core oc.. I don’t think it likes manual control at all if I’m honest. Oops.

The numbers in some of the Aida tests puts me as strong or stronger than threadrippers in those particular tests which is crazy to me. I see all of my old CPUs waaay down there lol.. like a speck with their puny numbers

 

[Gmr_Chick]

Don't shoot the noob but, can someone help me learn how to use Ryzen Master? It's billed as being easy to use, even for beginners but...yeah, I'm lost. Doesn't help that my board's BIOS defaults the PPT, TDC and EDC values to max settings - 1000, 520, and 540 respectively. Basically, I want to do a nice little auto OC or enable PBO for a gaming profile (and keep the CPU at stock all other times) but I don't know where to start and what values to tweak.

PS: try to explain it in simple terms, please. I'm smart, but not so much when it comes to PPT's and EDC's, lol

 

[freeagent]

I am by no means a pro, I have only used it two or three times months ago.. but try putting it in game mode

 

[Gmr_Chick]

I am by no means a pro, I have only used it two or three times months ago.. but try putting it in game mode

Same thing happens in Game Mode. All mentioned values in my post are pegged

 

[tabascosauz]

Only problem with that is it isn’t a 5600 and I can’t really treat it like one.

Told you nevertheless, very impressive quality 5900X you've got

So now I am testing stock clock range with a -30 curve and PBO on auto which I assume is off.. only thing left to do really is per core clocking which I am not sure I will do.. at least right now.

As long as you have time to run the script and whatnot I think you'll be surprised how much performance out of per-core alone you can gain, without any extra heat or power draw
GitHub - sp00n/corecycler: Stability test script for PBO & Curve Optimizer stability testing on AMD Ryzen processors

It'll be quite obvious from the script output if a core doesn't like your settings

 

[Space Lynx]

Same thing happens in Game Mode. All mentioned values in my post are pegged

I find its best to just use Auto OC mode with Ryzen master, honestly its pretty good. and the gains you would get doing it manually are very minimal over what Auto OC achieves.

 

[freeagent]

Told you nevertheless, very impressive quality 5900X you've got



As long as you have time to run the script and whatnot I think you'll be surprised how much performance out of per-core alone you can gain, without any extra heat or power draw
GitHub - sp00n/corecycler: Stability test script for PBO & Curve Optimizer stability testing on AMD Ryzen processors

It'll be quite obvious from the script output if a core doesn't like your settings

Thanks man! Ok I will play around with it when I get a chance, thanks for the link!

So it is a decent one? Ok I thought I was doing it wrong Either way like I said it is an absolute beast, and it’s not that bad to cool either I thought it would have been worse. We had a couple of moments between us when things got really hot though. I thought maybe I hurt it after a 99c all core 4750MHz 1.45v R20 run.. oops. But I think she’s ok.. my 5600 hit 103 once for a sec and she still rips. Didn’t sell it btw.. I think you were right about her so she’s snapped up tight in her clamshell.

I just set PBO and all things related to it in the bios. You can achieve the same thing as Ryzen master but without doing it in windows..

 

[tabascosauz]

Don't shoot the noob but, can someone help me learn how to use Ryzen Master? It's billed as being easy to use, even for beginners but...yeah, I'm lost. Doesn't help that my board's BIOS defaults the PPT, TDC and EDC values to max settings - 1000, 520, and 540 respectively. Basically, I want to do a nice little auto OC or enable PBO for a gaming profile (and keep the CPU at stock all other times) but I don't know where to start and what values to tweak.

PS: try to explain it in simple terms, please. I'm smart, but not so much when it comes to PPT's and EDC's, lol

Honestly probably easier/safer/more consistent just to use the PBO menu in BIOS. ASRock BIOS should have a dedicated PBO menu on the main tweaker page right (not AMD OC menu)? Ryzen Master doesn't work that great sometimes, it said my absolute turd 3700X could do 4.3 all-core (which it clearly cannot) then its built-in questionable stress test purported to show that it was "stable"

I had good results with experimenting with tightening just EDC on my 3700X. 90A stock, 83A was my PBO profile. I gave it a little extra PPT headroom so I think my final settings were 115W/60A/83A or something. ST and MT performance were both equivalent or better than all-out PBO. With an AIO you can probably crank it up to 2x scalar or more, but whether scalar higher than 1x is long-term safe is a bit contentious.

How's the chip quality? Can you manage say, all-core 4.3 or 4.4 @ 1.2V or so? If CTR says your sample is a silver or gold, 3600 stock clocks are low enough that it might be worth just going all-core in the 1.2V range, especially under AIO.

Thanks man! Ok I will play around with it when I get a chance, thanks for the link!

So it is a decent one? Ok I thought I was doing it wrong Either way like I said it is an absolute beast, and it’s not that bad to cool either I thought it would have been worse. We had a couple of moments between us when things got really hot though. I thought maybe I hurt it after a 99c all core 4750MHz 1.45v R20 run.. oops. But I think she’s ok.. my 5600 hit 103 once for a sec and she still rips. Didn’t sell it btw.. I think you were right about her so she’s snapped up tight in her clamshell.

I just set PBO and all things related to it in the bios. You can achieve the same thing as Ryzen master but without doing it in windows..

Hey, any chip that can manage 5075MHz effective on air is pretty damn "decent" what did CTR diagnostic say as to the quality?

1.45V 99C just one CB run, probably fine

 

[freeagent]

I haven't run core cycler on this CPU, I will try before the weekend though. I did try it on my last CPU but it kind of took over my system and was hard to shut the program down. I was probably looking at something wrong though..

Man this thing is spicy lol.. maybe those liquid cooled guys are on to something

 

[freeagent]

I was able to play with the power a little bit. I got +200 -30 stable I think.. no errors or crashing so far like I was getting before. Temps are a little better too

Edit:

I was getting reboots running pi32m usually after the first 1/3 pass.

Edit:

No more crashes and seems to run everything fine, no error reports yet. Hit 103f in the backyard, upstairs is 76 and the basement is 21.5c