AMD Ryzen 7 5800X3D

[DAPUNISHER]

I forgot this thread was around so I will just post it here.

I just installed my 5800X3D the other day for myself for Christmas on my current PC with the X470 MSI motherboard and I have a few questions.

First of all I havent updated the BIOS in forever! so I updated it to the one that was required for the CPU and I had already installed a few weeks before a brand new Noctua tower cooler the NH-U9S ready for the X3D. What I noticed with the BIOS update the 2700X which has been in the system since day one ran even better! after the update. Everything ran smoother, desktop, games etc and even the performance of the 2700X had climbed to 4.1GHz all core which was a good 100-150MHz faster all core then it used to do, I was surprised how good it ran!.

After installing the 5800X3D the BIOS got reset completely which I found a bit odd but whatever! and put in the XMP profile, checked temps, did the normal stuff, then let it boot into windows but that didnt happen it just loop over and over again into the BIOS and which I found it was trying to boot the older SSD which has no OS on it for some reason and the NVME drive wasnt there to boot from, it was there in the BIOS but not when I wanted to select it as 1st Boot drive. I figured out that I had to set it to Legcey for it to boot into W10 on the NVME Drive which would then show up on the boot list once that was select. It refused to do it on the UEFI BIOS setting. Anyway it booted and everything ran fine! but I wanted to then turn on Resizable Bar in the BIOS. I found the option turned it on but then wouldnt boot into Windows again, I found it would turn it to UEFI once turned on and you HAVE to have UEFI when using this option.........so I had to turn it off and leave it off so I could select Legacy so it would boot into W10. Anyone know how to fix this weird issue? so I can boot into Windows but also have Resizable turned on. without formatting.

Secondly this thing runs HOT! I knew it was going to be hotter then my 2700X but i wasnt expecting it to be that much hotter.....With the new cooler it was hitting 90c under a full load test (cinebench) and it would hit this temp pretty quickly....within 1 pass! and to me this was a concern. SO i removed the cooler cleaned off the thermal paste that I always use and used the Noctua paste just alot more of it and changed where the fan sits slightly and then tested it again and it did make a difference it would still hit 90c but it would take three times as long to do so and sits around 60c when gaming. I think by adding a second fan to the cooler it might keep it from hitting that 90c, we will see i guess.

EDIT: THat been said about the high temps I touched the cooler, I would feel the air coming off the cooler even touched the heat pipes and they would not be anywhere near the so called 90c the PC was telling me, do people think that the temps its showing is false? or?

The reason you can't use rebar is the NVME is using MBR instead of GPT for the file system. Here is a PC World guide to converting the drive without losing data. https://www.pcworld.com/article/2047620/convert-mbr-to-gpt-without-losing-data-in-windows.html

Then you'll be able to use rebar. Remember you have to enable above 4G decoding and set it to on or auto, whichever you UEFI uses.

For the temp issue, you can use the curve optimizer to set a negative value of -20 to -30 depending on how your 3D did in the silicon lottery. My 5800X3D with a -25 offset never breaks 70c in a 24c ambient room even while playing CPU heavy stuff like Spiderman with RT at 1080p. I am using a Deep Cool 240mm AIO I picked up on Newegg for $40 a couple years ago, with Arctic MX4, so nothing special. The fans are in silent mode and the CPU still boosts to a bit over 4.5GHz.

I have a 5600 and a 5600X both with the +200MHz boost and they can't touch the 3D for smooth frame pacing. Back to the Spiderman games with RT; they will have some noticeable frame drops in demanding areas with max crowds and traffic. The 3D is silky smooth. Same goes for Gotham Knights, even my 5800X boosting to 5050MHz had some pacing issues, the 3D yet again is buttery smooth. I tested all of them with either an RX 7800 XT or RX 6800 to minimize driver overhead. You are using near flagship RDNA2 so it's all good on that front. Avoid RT entirely and the 6 cores will do a better job.

Hope this helps, and let us know how things work out.

 

[Melvis]

1. The BIOS resetting itself upon installation of a new CPU or RAM is normal. This is how it prevents settings being used that were fine with the old part, but could potentially damage the new one.

2. You probably didn't find the UEFI boot option because Windows was installed with legacy mode (if that's even possible, I'm not sure). You might need to reinstall Windows with proper UEFI support, which is also needed for ReBar, I suppose.

3. What cooler are you using? I would suggest forgetting anything that's AMD stock, and getting something decent.

4. Exhaust air being way colder than it was with the old CPU is normal. Chiplets have a much smaller surface area, so the quantity of heat dissipated by your cooler is much smaller as well, despite the CPU running much hotter. The same way boiling a litre of water won't produce as much heat/steam as boiling 1,000 litres.

1. Ok thats fair and makes sense! just never seen that before!
2. Yeah this is what i figured! but im not sure I can even install W10 on a NVME without it been on UEFI.......So
3. No its not the stock AMD Cooler as said above its the Noctua tower cooler the NH-U9S which is the biggest tower cooler I can fit in this case.
4. Im not sure if I totally agree on this I get what your saying! but im sure if you put your hand over a kettle thats boiling with water and steam coming out of it you still be able to feel the heat maybe even more so then 1000litres honestly, but anyway im just surprised I can hold or touch any part of the U9S and it was never hot to the touch, warm but not 90c.

Most possibly the difference between CPU's temp and cooler's temp when touching it, means that the contact between the CPU and its cooler isn't optimal. Most older and many newer coolers aren't made with the Ryzen's chiplet design in mind but mostly for Intel's one.

Is the one in the review link below your cooler? If so, one point in the conclusion might help you a bit:

• Mounting procedure can be fiddly and has a lot of stages

https://www.eteknix.com/scythe-big-shuriken-2-rev-b-low-profile-cpu-cooler-review/

Well thats what I thought as well and I even took the cooler back off took pictures of how it was touching and it looked good! its a good cooler perfect base on it! but I just changed the paste, changed the fan slightly and temps are better but still will hit 90c under full load in hmm lets say 3mins time.

No no its the Noctua tower cooler the NH-U9S

Temps above 45°C will damage the skin. Even on a crappy cooler temps should be high enough to burn you if you touch them while the CPU is at 90°C! That's if you touch the heatpipes or the fin stack. Air could be colder, that depends.

As others mentioned above, check if the cooler is making good contact or just change it with something decent. Top end paste like TG is also worth considering.

I personally never had issues with the cooling of AMD chips, coming from the shape of their thermal plate. Just rub it good with IPA and then re-apply good paste. And be sure to put enough of it. Less is worse in this case.

Thats exactly what I thought as well! and was very surprised when i touched the pipes, the fins etc and they where warm at best......which just didnt make sense to me, this si why I was thinking it must be reading it wrong! as there was no way it was 90c.

I updated the stock cooler weeks before i got the 5800X3D as I knew it would need better cooling over stock so I got the Noctua tower cooler the NH-U9S as it was the best cooler I could get that would fit in the case. I also changed the paste to the Noctua paste, moved the fan slightly and the temps did improve as I mentioned above.

You and me both I have run more AMD CPU's then Intel for the past 20+ yrs and they all run cool compared to most of the Intel stuff ive used so this was a bit of a surprise!

The reason you can't use rebar is the NVME is using MBR instead of GPT for the file system. Here is a PC World guide to converting the drive without losing data. https://www.pcworld.com/article/2047620/convert-mbr-to-gpt-without-losing-data-in-windows.html

Then you'll be able to use rebar. Remember you have to enable above 4G decoding and set it to on or auto, whichever you UEFI uses.

For the temp issue, you can use the curve optimizer to set a negative value of -20 to -30 depending on how your 3D did in the silicon lottery. My 5800X3D with a -25 offset never breaks 70c in a 24c ambient room even while playing CPU heavy stuff like Spiderman with RT at 1080p. I am using a Deep Cool 240mm AIO I picked up on Newegg for $40 a couple years ago, with Arctic MX4, so nothing special. The fans are in silent mode and the CPU still boosts to a bit over 4.5GHz.

I have a 5600 and a 5600X both with the +200MHz boost and they can't touch the 3D for smooth frame pacing. Back to the Spiderman games with RT; they will have some noticeable frame drops in demanding areas with max crowds and traffic. The 3D is silky smooth. Same goes for Gotham Knights, even my 5800X boosting to 5050MHz had some pacing issues, the 3D yet again is buttery smooth. I tested all of them with either an RX 7800 XT or RX 6800 to minimize driver overhead. You are using near flagship RDNA2 so it's all good on that front. Avoid RT entirely and the 6 cores will do a better job.

Hope this helps, and let us know how things work out.

Ahhh your probably right! I most likely set it to MBR and not GPT when I installed W10 on this drive a yr or so ago, that would make sense, so thank you for that! and the link!

Yep it turns the 4G Decoding on by default once you turn on Rebar.

Im not sure if I can do that in the BIOS honestly and you might have to explain this more to me, im not up on the whole -20/30 offset thing as I never had to deal with that before but this would make sense as it doesnt feel like its getting anywhere near the so called 90c the programs are saying.....and I did set up a basic fan speed settings in the BIOS for the CPU and the fan runs at 100% when it thinks its hit 65c.

Yeah so far in the games I have tried it runs very well and smooth for the most part and going by the temps never gets over mid 60c while gaming. Ive built a 7800X3D for a client and a 5600X machine and they both ran very well with no fuss at all and the 5600X was using stock cooler. I have seen a boost in some games and none in others but it be nice to see what rebar can offer when I get that sorted.

Many thanks for this reply as very helpful!

 

[AusWolf]

1. Ok thats fair and makes sense! just never seen that before!
2. Yeah this is what i figured! but im not sure I can even install W10 on a NVME without it been on UEFI.......So
3. No its not the stock AMD Cooler as said above its the Noctua tower cooler the NH-U9S which is the biggest tower cooler I can fit in this case.
4. Im not sure if I totally agree on this I get what your saying! but im sure if you put your hand over a kettle thats boiling with water and steam coming out of it you still be able to feel the heat maybe even more so then 1000litres honestly, but anyway im just surprised I can hold or touch any part of the U9S and it was never hot to the touch, warm but not 90c.

2. I'm not sure either, but I don't have any other idea.
3. That's a shame. The U9S isn't a very large cooler with its 92 mm fan. Are you sure you can't fit a 120 mm tower in it, or maybe a larger down-draft cooler, like the be quiet! Dark Rock TR2?
4. Agree or not, this is how it works, unfortunately. I've had maybe 5-6 chiplet CPUs so far, so I can confirm. The exhaust air was much colder with each of them than with the Core i7-11700 that I had in my main rig before. My water boiling example was probably not the best, but the point stands: smaller, offset die surface area -> smaller contact with the heat spreader -> smaller amount of heat transferred. Another, possibly better example would be forging a nail or forging a sword. When you throw the hot nail into a bucket of water, the temperature of the water will only slightly change, whereas inserting the hot sword will make it boil.

 

[neatfeatguy]

1. Ok thats fair and makes sense! just never seen that before!
2. Yeah this is what i figured! but im not sure I can even install W10 on a NVME without it been on UEFI.......So
3. No its not the stock AMD Cooler as said above its the Noctua tower cooler the NH-U9S which is the biggest tower cooler I can fit in this case.
4. Im not sure if I totally agree on this I get what your saying! but im sure if you put your hand over a kettle thats boiling with water and steam coming out of it you still be able to feel the heat maybe even more so then 1000litres honestly, but anyway im just surprised I can hold or touch any part of the U9S and it was never hot to the touch, warm but not 90c.


Well thats what I thought as well and I even took the cooler back off took pictures of how it was touching and it looked good! its a good cooler perfect base on it! but I just changed the paste, changed the fan slightly and temps are better but still will hit 90c under full load in hmm lets say 3mins time.

No no its the Noctua tower cooler the NH-U9S

Just kind of skimming through the post here and notice you mentioned the U9S cooler. I ran that cooler on my 5900X for a handful of months (was the only cooler I had that would fit in my case to allow me to fit the top 200mm fan, the one I wanted to use - which escapes my memory - was too tall for me to mount the 200mm fan at the top and that is my only exhaust fan for the case so I need it) until I moved to an AIO when funds were available. Gaming wise, the U9S kept the CPU around 70s. Under full load the temps would be in the low 90s. It worked, but barely for my 5900X.

That cooler just isn't cutting it for your 5800X3D if TPU's temp chart on it is correct (W1zzard using the U14S) and shows the 5900X runs 10C cooler than the 5800X3D under full load.

You just need a better cooler if temps are a problem for you.

 

[DAPUNISHER]

@Melvis

I set the curve optimizer through the UEFI but you can just use AMD's Ryzen Master software to do it https://www.amd.com/en/technologies/ryzen-master

Here's the FAQ for it https://www.amd.com/system/files/documents/faq-curve-optimizer.pdf

There are many guides on using it too. I'd try -25 first and see if it's solid or you experience instability. If it's good you can gamble on -30, if not back down to -20. I think in the end that @neatfeatguy is probably right that you will need to use a better cooler.

 

[zlobby]

1. Ok thats fair and makes sense! just never seen that before!
2. Yeah this is what i figured! but im not sure I can even install W10 on a NVME without it been on UEFI.......So
3. No its not the stock AMD Cooler as said above its the Noctua tower cooler the NH-U9S which is the biggest tower cooler I can fit in this case.
4. Im not sure if I totally agree on this I get what your saying! but im sure if you put your hand over a kettle thats boiling with water and steam coming out of it you still be able to feel the heat maybe even more so then 1000litres honestly, but anyway im just surprised I can hold or touch any part of the U9S and it was never hot to the touch, warm but not 90c.



Well thats what I thought as well and I even took the cooler back off took pictures of how it was touching and it looked good! its a good cooler perfect base on it! but I just changed the paste, changed the fan slightly and temps are better but still will hit 90c under full load in hmm lets say 3mins time.

No no its the Noctua tower cooler the NH-U9S


Thats exactly what I thought as well! and was very surprised when i touched the pipes, the fins etc and they where warm at best......which just didnt make sense to me, this si why I was thinking it must be reading it wrong! as there was no way it was 90c.

I updated the stock cooler weeks before i got the 5800X3D as I knew it would need better cooling over stock so I got the Noctua tower cooler the NH-U9S as it was the best cooler I could get that would fit in the case. I also changed the paste to the Noctua paste, moved the fan slightly and the temps did improve as I mentioned above.

You and me both I have run more AMD CPU's then Intel for the past 20+ yrs and they all run cool compared to most of the Intel stuff ive used so this was a bit of a surprise!



Ahhh your probably right! I most likely set it to MBR and not GPT when I installed W10 on this drive a yr or so ago, that would make sense, so thank you for that! and the link!

Yep it turns the 4G Decoding on by default once you turn on Rebar.

Im not sure if I can do that in the BIOS honestly and you might have to explain this more to me, im not up on the whole -20/30 offset thing as I never had to deal with that before but this would make sense as it doesnt feel like its getting anywhere near the so called 90c the programs are saying.....and I did set up a basic fan speed settings in the BIOS for the CPU and the fan runs at 100% when it thinks its hit 65c.

Yeah so far in the games I have tried it runs very well and smooth for the most part and going by the temps never gets over mid 60c while gaming. Ive built a 7800X3D for a client and a 5600X machine and they both ran very well with no fuss at all and the 5600X was using stock cooler. I have seen a boost in some games and none in others but it be nice to see what rebar can offer when I get that sorted.

Many thanks for this reply as very helpful!

May the smotth frame-pacing be ever in your favor!

 

[Melvis]

2. I'm not sure either, but I don't have any other idea.
3. That's a shame. The U9S isn't a very large cooler with its 92 mm fan. Are you sure you can't fit a 120 mm tower in it, or maybe a larger down-draft cooler, like the be quiet! Dark Rock TR2?
4. Agree or not, this is how it works, unfortunately. I've had maybe 5-6 chiplet CPUs so far, so I can confirm. The exhaust air was much colder with each of them than with the Core i7-11700 that I had in my main rig before. My water boiling example was probably not the best, but the point stands: smaller, offset die surface area -> smaller contact with the heat spreader -> smaller amount of heat transferred. Another, possibly better example would be forging a nail or forging a sword. When you throw the hot nail into a bucket of water, the temperature of the water will only slightly change, whereas inserting the hot sword will make it boil.

Indeeed it isnt large cooler but for a tower cooler its the biggest best cooler I can get and its brand new! I think if I add a second fan it might do the trick honestly, only time will tell! But if it doesnt do a good enough job for my liking ill have a look at the Dark Rock TR2. I gamed all day with it and it never got over 70c. No offset used at this time.

I would agree with them been cooler then your i7-11700 thats for sure but thats a no brainer really but Ive had many chiplet CPU's as well, 2700X, 1600X, 2200G, 2400G, 2600X, 3200G, 3400G, 4750G, 7800X3D, 3600, and from memory I could always feel from the coolers that where used the warmth or heat coming from the CPUs especially the 7800X3D as I only build it 6weeks ago and under load could easily feel the heat from the cooler, a Deepcool AK400 and it didnt run that hot compared to what this 5800X3D is "saying" this is why I think the temps that are showing arnt true at all.

Just kind of skimming through the post here and notice you mentioned the U9S cooler. I ran that cooler on my 5900X for a handful of months (was the only cooler I had that would fit in my case to allow me to fit the top 200mm fan, the one I wanted to use - which escapes my memory - was too tall for me to mount the 200mm fan at the top and that is my only exhaust fan for the case so I need it) until I moved to an AIO when funds were available. Gaming wise, the U9S kept the CPU around 70s. Under full load the temps would be in the low 90s. It worked, but barely for my 5900X.

That cooler just isn't cutting it for your 5800X3D if TPU's temp chart on it is correct (W1zzard using the U14S) and shows the 5900X runs 10C cooler than the 5800X3D under full load.

You just need a better cooler if temps are a problem for you.

I can understand that and those temps with a 5900X been a high clocked more cored CPU i wouldnt run a cooler lesser then a 120mm tower cooler on a CPU like that but yeah true it looks like the 5800X3D is a hotter running CPU then the 5900X which is surprising but I can understand why with the large amount of Cache it has but I still think the readings are wrong but when I get more time I will do more testing on this and get a temp gun and another cooler in the future and see whats going on because the cooler isnt getting hot compared to the temps the programs are saying.

 

[neatfeatguy]

Indeeed it isnt large cooler but for a tower cooler its the biggest best cooler I can get and its brand new! I think if I add a second fan it might do the trick honestly, only time will tell! But if it doesnt do a good enough job for my liking ill have a look at the Dark Rock TR2. I gamed all day with it and it never got over 70c. No offset used at this time.

I would agree with them been cooler then your i7-11700 thats for sure but thats a no brainer really but Ive had many chiplet CPU's as well, 2700X, 1600X, 2200G, 2400G, 2600X, 3200G, 3400G, 4750G, 7800X3D, 3600, and from memory I could always feel from the coolers that where used the warmth or heat coming from the CPUs especially the 7800X3D as I only build it 6weeks ago and under load could easily feel the heat from the cooler, a Deepcool AK400 and it didnt run that hot compared to what this 5800X3D is "saying" this is why I think the temps that are showing arnt true at all.

I can understand that and those temps with a 5900X been a high clocked more cored CPU i wouldnt run a cooler lesser then a 120mm tower cooler on a CPU like that but yeah true it looks like the 5800X3D is a hotter running CPU then the 5900X which is surprising but I can understand why with the large amount of Cache it has but I still think the readings are wrong but when I get more time I will do more testing on this and get a temp gun and another cooler in the future and see whats going on because the cooler isnt getting hot compared to the temps the programs are saying.

Think what you'd like, but even with two fans it won't resolve your situation for that little cooler keeping the 5800X3D or in my situation the 5900X as cool as you need it to. I had two fans for my U9S and the low 70s for gaming and low 90s for full load is what I experienced. At best, the U9S works for letting these CPUs operate, but it's not designed to keep them cool when running at full load.

TDP for the U9S is listed on Noctua's site at 93. The 5800X3D is listed at 105 on AMD's site, which is the same TDP that AMD lists the 5900X at.

 

[AusWolf]

Just kind of skimming through the post here and notice you mentioned the U9S cooler. I ran that cooler on my 5900X for a handful of months (was the only cooler I had that would fit in my case to allow me to fit the top 200mm fan, the one I wanted to use - which escapes my memory - was too tall for me to mount the 200mm fan at the top and that is my only exhaust fan for the case so I need it) until I moved to an AIO when funds were available. Gaming wise, the U9S kept the CPU around 70s. Under full load the temps would be in the low 90s. It worked, but barely for my 5900X.

That cooler just isn't cutting it for your 5800X3D if TPU's temp chart on it is correct (W1zzard using the U14S) and shows the 5900X runs 10C cooler than the 5800X3D under full load.

You just need a better cooler if temps are a problem for you.

Agreed. Single-CCD Ryzen is not very good for small form factor builds, unfortunately. I learned this the hard way.

Indeeed it isnt large cooler but for a tower cooler its the biggest best cooler I can get and its brand new! I think if I add a second fan it might do the trick honestly, only time will tell! But if it doesnt do a good enough job for my liking ill have a look at the Dark Rock TR2. I gamed all day with it and it never got over 70c. No offset used at this time.

I would agree with them been cooler then your i7-11700 thats for sure but thats a no brainer really but Ive had many chiplet CPU's as well, 2700X, 1600X, 2200G, 2400G, 2600X, 3200G, 3400G, 4750G, 7800X3D, 3600, and from memory I could always feel from the coolers that where used the warmth or heat coming from the CPUs especially the 7800X3D as I only build it 6weeks ago and under load could easily feel the heat from the cooler, a Deepcool AK400 and it didnt run that hot compared to what this 5800X3D is "saying" this is why I think the temps that are showing arnt true at all.

Out of all those, only the 7800X3D and 3600 are chiplet CPUs, the rest aren't. The 7800 is very conservative with power, so it can do with sub-optimal cooling, and so is the 3600 more or less.

So far, I've owned the 3100, 3600, 5950X, 7700X and 7800X3D, and I can confidently say that my system's heat output has been considerably lower with all of them compared to any Intel CPU running at the same core temperature. That's why I'm saying, heat and temperature are not the same thing.

I can understand that and those temps with a 5900X been a high clocked more cored CPU i wouldnt run a cooler lesser then a 120mm tower cooler on a CPU like that but yeah true it looks like the 5800X3D is a hotter running CPU then the 5900X which is surprising but I can understand why with the large amount of Cache it has but I still think the readings are wrong but when I get more time I will do more testing on this and get a temp gun and another cooler in the future and see whats going on because the cooler isnt getting hot compared to the temps the programs are saying.

It's not surprising. The 5800X3D has a single CCD, while the 5900 has two, so it dissipates its heat through a much larger surface area, also without an extra cache layer obstructing heat transfer.

X Watts pushed through 8 cores on a single chip will make it hotter than X Watts pushed through 12 cores shared across two chips.

 

[Melvis]

Think what you'd like, but even with two fans it won't resolve your situation for that little cooler keeping the 5800X3D or in my situation the 5900X as cool as you need it to. I had two fans for my U9S and the low 70s for gaming and low 90s for full load is what I experienced. At best, the U9S works for letting these CPUs operate, but it's not designed to keep them cool when running at full load.

TDP for the U9S is listed on Noctua's site at 93. The 5800X3D is listed at 105 on AMD's site, which is the same TDP that AMD lists the 5900X at.

Wait what? 93 TDP? that cant be right! I swear when I bought it from NZ online shop it said something more like 140W TDP it was able to handle. That makes no sense to me as then this cooler would be worse! then AMD stock cooler.........that cant be right! as the AMD Wraith stock cooler handled the 2700X 105W CPU and the U9S was able to easily keep the 2700X cool for a few weeks before I changed out CPU's. This also makes me think the readings are wrong! and ill investigate this more in the near future as this just isnt adding up to me.

Agreed. Single-CCD Ryzen is not very good for small form factor builds, unfortunately. I learned this the hard way.


Out of all those, only the 7800X3D and 3600 are chiplet CPUs, the rest aren't. The 7800 is very conservative with power, so it can do with sub-optimal cooling, and so is the 3600 more or less.

So far, I've owned the 3100, 3600, 5950X, 7700X and 7800X3D, and I can confidently say that my system's heat output has been considerably lower with all of them compared to any Intel CPU running at the same core temperature. That's why I'm saying, heat and temperature are not the same thing.


It's not surprising. The 5800X3D has a single CCD, while the 5900 has two, so it dissipates its heat through a much larger surface area, also without an extra cache layer obstructing heat transfer.

X Watts pushed through 8 cores on a single chip will make it hotter than X Watts pushed through 12 cores shared across two chips.

Ummm im pretty sure the 2700X, 1600, well most of AMD's CPU's are chiplet/CCX design just back then they where not 8cores per CCX but only 4.

Yeah I can understand that and it would make sense that it would dissipate heat more across the heat spreader but i think its the amount of extra Cache thats laid ontop of the CCX is what makes this CPU hot! but again I have a sneaky feelings its just not reading the temps right but ill investigate this more in the future. For now while gaming the temps are fine around 60c and its summer here at the moment.

 

[AusWolf]

Wait what? 93 TDP? that cant be right! I swear when I bought it from NZ online shop it said something more like 140W TDP it was able to handle. That makes no sense to me as then this cooler would be worse! then AMD stock cooler.........that cant be right! as the AMD Wraith stock cooler handled the 2700X 105W CPU and the U9S was able to easily keep the 2700X cool for a few weeks before I changed out CPU's. This also makes me think the readings are wrong! and ill investigate this more in the near future as this just isnt adding up to me.

Chiplet CPUs are harder to cool. If cooler X can cool a monolithic (Intel) CPU with a 120 W power consumption, it won't do the same with AMD chiplets. It will go as far as 80-ish W.

Ummm im pretty sure the 2700X, 1600, well most of AMD's CPU's are chiplet/CCX design just back then they where not 8cores per CCX but only 4.

No, they're not: source (2700X review). AMD started the chiplet design with the Ryzen 3000 series (not including 3000G or 5000G which are monolithic designs).

Yeah I can understand that and it would make sense that it would dissipate heat more across the heat spreader but i think its the amount of extra Cache thats laid ontop of the CCX is what makes this CPU hot! but again I have a sneaky feelings its just not reading the temps right but ill investigate this more in the future. For now while gaming the temps are fine around 60c and its summer here at the moment.

That's the other thing: the extra cache is an insulating layer on top of the CPU die. That's why X3D CPUs operate with lower clocks and voltages than their non-X3D counterparts.

 

[Count von Schwalbe]

Ummm im pretty sure the 2700X, 1600, well most of AMD's CPU's are chiplet/CCX design just back then they where not 8cores per CCX but only 4.

Kinda. 2 4-core CCX and a dual-channel memory interface per CCD.

2700X had 1 CCD, and the dedicated IOD wasn't introduced until Zen 2.

It let AMD glue a bunch of 2700X together to make things like Threadripper.

Edit: ninja'd.

 

[neatfeatguy]

Wait what? 93 TDP? that cant be right! I swear when I bought it from NZ online shop it said something more like 140W TDP it was able to handle. That makes no sense to me as then this cooler would be worse! then AMD stock cooler.........that cant be right! as the AMD Wraith stock cooler handled the 2700X 105W CPU and the U9S was able to easily keep the 2700X cool for a few weeks before I changed out CPU's. This also makes me think the readings are wrong! and ill investigate this more in the near future as this just isnt adding up to me.

Sorry, I was mistaken. It has been a long while since I looked at Noctua's listing of their TDP NSRP listing for their coolers. It's been so long that they used to have it listed as TDP, but now they use their own rating of NSRP (higher the number the better it cools) and that's where I saw 93 listed for the U9S.

Trying to find an actual TDP listing for the U9S is kind of a pain. I see it listed at spots as 115W and even in a few spots as 140W. I know for a fact that 140W can't be right because that little cooler couldn't keep my 5900X from throttling when at full load due to the temps getting over 90C.

I'd have to say that 115W TDP for the U9S sounds more accurate, but even then with the 5900X and 5800X3D being listed by AMD at 105W doesn't seem right because clearly the U9S struggles to keep these CPUs cool enough to prevent them from throttling.

Long story short, the cooler isn't good enough for the 5900X or 5800X3D when put under full load. It works just fine if you're strictly gaming, temps stay cool enough to not cause any throttling, but if you need to run either of these CPUs under a very heavy load for long periods of time it's just not going to keep them cool enough to prevent throttling.

 

[Melvis]

Well after all that I decided to do the -30 offset thing which was easy but also wasnt as the BIOS on this board makes it a little tricky to figure it out but after looking at another thread and giving the whole kombo Strike thing a go it made a big difference! Temps have dropped (going by the program) 7c so now its maxing out around 83c all core and the clocks are even higher now at almost 4.4Ghz all core, big difference! was very surprised.

Im glad the Kombo Strike thingy was there or id be a bit lost as I couldnt change the value I couldnt move it away from Auto.......I could change the others but not the actual value I wanted to put in like 0.0500.

 

[Mussels]

(1) 5600x vs 5700x vs 5800X3D Benchmarked with an RTX 4080 - Is there a difference? - YouTube

Pretty good video here. I'm not sure how interested I am in future GPU upgrades for them. Maybe maybe not. My youngest boy is 7 and my oldest is 10. The amount of living off me in terms of gaming builds will only occur for a little while longer.

Although I normally pass down my GPUs so there's that...

The thing people forget is that it's settings dependent: At ultra settings it matters far less.

Basically, you choose your target framerate and pick a CPU that matches based on the 1% lows, then lower GPU settings til you're in that range. Resolution is simple, but misleading because high on 4K can have higher FPS than ultra on 1080p.

Trying to find an actual TDP listing

Dead metric - TDP doesn't work well with modern CPUs due to heat density issues and intels PL1/PL2 stuff.

What cools one 250W CPU may not cool a 65W CPU on a smaller process.

 

[Melvis]

Second Fan installed on the U9S and after a good 5 mins of R23 it maxed out at 80c, mostly sat at 78-79c most of the time, im very happy with that!