enough cooling?

What temp does 3970X start throttling? I know some of these big base Threadrippers start throttling at about 67-68c. What they do is slow down processing speed to keep temps from going too high.

Edit: did a little searching and seems yours is throttling, but seems AMD isn't saying exactly what temp it starts throttling at.

"I will add this ... the design temperature at stock for the 60/70x is 68c. Your CPU should not surpass this temperature under the heaviest load if left at stock. "
https://hardforum.com/threads/amd-threadripper-3970x-temps.1990566/post-1044434912
 
Last edited:
What temp does 3970X start throttling? I know some of these big base Threadrippers start throttling at about 67-68c. What they do is slow down processing speed to keep temps from going too high.

Edit: did a little searching and seems yours is throttling, but seems AMD isn't saying exactly what temp it starts throttling at.

"I will add this ... the design temperature at stock for the 60/70x is 68c. Your CPU should not surpass this temperature under the heaviest load if left at stock. "
https://hardforum.com/threads/amd-threadripper-3970x-temps.1990566/post-1044434912
Thank you doyll. I do not know what temp it starts throttling. I have read lots of different places and get different information. Seems AMD originally said max 69 at first but then later revised that to a much higher number like 90 max. I can't remember exactly. What temp it actually needs to be at is a great question I would like to know.

I did a 90 hr render with all cores staying at 100% load and around approx 3840-3866MHz. Is it slowing down? It's a 3.7GHz processor right?
The CPU temp stays 73-74c almost all the time occassionally hitting 75 and 76. Although according to CoreTemp now says it hit 86c at some point.

My brain is tired at the moment but I do not recall reading about anyone being able to keep one at 67-68 under full load. Maybe with extensive custom water cooling?
If I remember correctly the guys using Big Air coolers run in the mid 70s.

Another long render in progress now but I'm going to bed so I'll check it in the morning.
 
AMD says base clock is 3.7GHz with boost clock up to 4.5Ghz. My understanding is boost clock is how fast each core can run if it stays cool. I think the original 68c temp means if temp is 68c or below cores can run at speeds up to 4.5GHz .. probalby not that fast, but faster than 3.7GH base clock spec. I do know that when Kyle was testing coolers on his Threadripper that when temps went above 68c clock speeds were lower. You can kinda see what I'm talking about in below bar graphs:
HardOCP TR4 cooler testing.png
 
Thank you doyll. I do not know what temp it starts throttling. I have read lots of different places and get different information. Seems AMD originally said max 69 at first but then later revised that to a much higher number like 90 max. I can't remember exactly. What temp it actually needs to be at is a great question I would like to know.

I did a 90 hr render with all cores staying at 100% load and around approx 3840-3866MHz. Is it slowing down? It's a 3.7GHz processor right?
The CPU temp stays 73-74c almost all the time occassionally hitting 75 and 76. Although according to CoreTemp now says it hit 86c at some point.

My brain is tired at the moment but I do not recall reading about anyone being able to keep one at 67-68 under full load. Maybe with extensive custom water cooling?
If I remember correctly the guys using Big Air coolers run in the mid 70s.

Another long render in progress now but I'm going to bed so I'll check it in the morning.

From what I recall the TR's like to stay under 70C to get peak performance, cores boosting. I know that even my R9 3900x does better under 70C all core than above and by the time it hits 80C it is down at least 250-300Mhz all core compared to under 70C. By the time it hits mid 80's it is near stock clock speeds all core. I imagine the TR's with more cores probably prefer the same if not lower temps.
 
Thanks guys.
Been watching mine some more and seems to stay around 3866-3891MHz much of the time during these long 100% load renders. Only time I see it hit 4000MHz or more is when a frame render finishes and the next frame starts the temp drops to 71 and then the next frame render starts and it's back up to 73 or 74.
I have 6 fans installed in the top 360 radiator of my case (push/pull) but it doesn't seem like they are spinning very fast. None of my fans seem to be spinning very fast. I wonder if I could crank up my fans speed somehow and drop the temp a bit?
 
Thanks guys.
Been watching mine some more and seems to stay around 3866-3891MHz much of the time during these long 100% load renders. Only time I see it hit 4000MHz or more is when a frame render finishes and the next frame starts the temp drops to 71 and then the next frame render starts and it's back up to 73 or 74.
I have 6 fans installed in the top 360 radiator of my case (push/pull) but it doesn't seem like they are spinning very fast. None of my fans seem to be spinning very fast. I wonder if I could crank up my fans speed somehow and drop the temp a bit?
umm yeah, do it. make a custom fan curve that is flat lined at 25% until 50c then ramp to full at 80c. see what that gets ya.
 
umm yeah, do it. make a custom fan curve that is flat lined at 25% until 50c then ramp to full at 80c. see what that gets ya.
Thank you. I do not know how to do that. Is that adjusted via cooling hardware or motherboard or ?

Here is what my PC came with:

CYBERPOWERPC DEEPCOOL CASTLE 360EX ARGB AIO LIQUID CPU COOLER 1


THERMALTAKE RIING RGB 120MM HIGH STATIC PRESSURE FAN TRIPLE PACK W/ CONTROLLER 1


APEVIA DUAL ADDRESSABLE DIGITAL RGB 120MM FAN MUST FAN CONROL 3


ADDRESSABLE DIGITAL ARGB ADAPTER + REMOTE 1
 
Asus has their own software to control fan speed on the motherboard. If you have AI Suite 3 installed on the computer, and when you open it up, you can go into Fan Xpert 4 and control each fan header on the motherboard.

If you dont have it installed on your computer, go ahead and download and install it here.
 
I would want fans to be full speed before temps reach 68c because CPU is already stating to throttle at 68c. I would want to keep temps belwo 68c so there is no throttling. Ramping could be slow up to about 60c, but something like 70%@60c, 80%@64c, 90%@65-6c, and 100%@66-7c. Performance / speed seem to be higher with temps below 68c, so probably better to speed up cooling and airflow fans so temp of air into cooler stays at room ambient and thus gives cooler lower temp air so we end up with lower CPU temp and higher boost speed.
 
Last edited:
I would want fans to be full speed by 68c. I would want to keep temps belwo 68c so there is no throttling. Ramping could be slow up to about 60c, but something like 70%@60c, 80%@64c, 90%@66c, and 100%@67c. Performance may be even better if fans speed up a little earllier.
its not throttling, its just not boosting as high as it could if cooler.
 
its not throttling, its just not boosting as high as it could if cooler.
CPU lowers boost below maximum rated GHz to keep from running too hot.
CPU throttles below maximum rated GHz to keep from running too hot.
Both doing the same thing for same reasons.

Thanks guys. It was set to ramp fans up to 100 around 70c so I changed the curve from standard to "turbo" which gets fans up to 100% in the upper 60s.
Are lower temps allowing you to get higher clock speeds?
 
CPU lowers boost below maximum rated GHz to keep from running too hot.
CPU throttles below maximum rated GHz to keep from running too hot.
Both doing the same thing for same reasons.


Are lower temps allowing you to get higher clock speeds?
do you not understand what throttling is? if it goes below base clock then it is throttling. his cpu is still boosting, just not as high as it could if it was cooler.
 
do you not understand what throttling is? if it goes below base clock then it is throttling. his cpu is still boosting, just not as high as it could if it was cooler.
https://www.pcmag.com/encyclopedia/term/cpu-throttling
"Adjusting the clock speed of the CPU. Also called "dynamic frequency scaling," CPU throttling is commonly used to automatically slow down the computer when possible to use less energy and conserve battery, especially in laptops. CPU throttling can also be adjusted manually to make the system quieter, because the fan can then run slower. Contrast with overclock. See thermal throttling."

https://www.pcmag.com/encyclopedia/term/thermal-throttling
"Adjusting the clock speed of the CPU based on the amount of heat it is currently generating. Thermal throttling helps cool the chip when it gets too hot by lowering the speed."

It doesn't matter if it is base speed and is boost speed, it is still thermal throttling, the same thing we are calling throttling in this thread.

If you have difinitions that say I am wrong please post them with source links so we can see them.
 
https://www.pcmag.com/encyclopedia/term/cpu-throttling
"Adjusting the clock speed of the CPU. Also called "dynamic frequency scaling," CPU throttling is commonly used to automatically slow down the computer when possible to use less energy and conserve battery, especially in laptops. CPU throttling can also be adjusted manually to make the system quieter, because the fan can then run slower. Contrast with overclock. See thermal throttling."

https://www.pcmag.com/encyclopedia/term/thermal-throttling
"Adjusting the clock speed of the CPU based on the amount of heat it is currently generating. Thermal throttling helps cool the chip when it gets too hot by lowering the speed."

It doesn't matter if it is base speed and is boost speed, it is still thermal throttling, the same thing we are calling throttling in this thread.

If you have difinitions that say I am wrong please post them with source links so we can see them.
any clock speed between base and boost is normal. when it drops below base its being throttled. pcmag is not the end all answer.
 
Thanks guys.
So I'm a little over 9 hrs into another long render and it's staying at 68c and 3899.20MHz on all 32 cores at 100% load.
I realize you are all smarter than I am but here are my thoughts. See if you agree or please correct me.

Prior to today my PC was typically running 73/74c at 100% load with the MHz varying a little 3866-3891MHz and sometimes a little above 4000MHz briefly.
Last night I made the "mistake" (misunderstood trial) of changing my BIOS from Normal to Optimal, which resulted in a 5% Overclock, according to the boot screen now. So my PC became a 3.9GHz instead of 3.7. It won't change back. I changed the fans to ramp to 100 sooner than before. So I am guessing that the Optimal setting tweaks my system to run at whatever consistant frequency it can sustain hence now it stays exactly at 3899.20MHz instead of varying like it did before?
So while it could boost to 4.5GHz it wont' because it can't cool it so the system just decided to stay at 3899.20?
Or it's as if the system is now just set to a constant 5% OC and no longer boosts, or drops for that matter?
So this makes me wonder if somehow I could get it back to the Normal setting instead of Optimal, with my cooling adjusted maybe it would boost higher than before? Or is it faster to just stay at 3899.20 and be consistant?

Perhaps I've reached some type of optimal setting given my current hardware and should be happy? Or should I adjust something?
I feel like an idiot because I don't know much but I am thankful for the helpful people here and that the system is running as good as it is. It's exciting to finally be able to do these animations I've been wanting to do for years!
 
Thanks guys.
So I'm a little over 9 hrs into another long render and it's staying at 68c and 3899.20MHz on all 32 cores at 100% load.
I realize you are all smarter than I am but here are my thoughts. See if you agree or please correct me.

Prior to today my PC was typically running 73/74c at 100% load with the MHz varying a little 3866-3891MHz and sometimes a little above 4000MHz briefly.
Last night I made the "mistake" (misunderstood trial) of changing my BIOS from Normal to Optimal, which resulted in a 5% Overclock, according to the boot screen now. So my PC became a 3.9GHz instead of 3.7. It won't change back. I changed the fans to ramp to 100 sooner than before. So I am guessing that the Optimal setting tweaks my system to run at whatever consistant frequency it can sustain hence now it stays exactly at 3899.20MHz instead of varying like it did before?
So while it could boost to 4.5GHz it wont' because it can't cool it so the system just decided to stay at 3899.20?
Or it's as if the system is now just set to a constant 5% OC and no longer boosts, or drops for that matter?
So this makes me wonder if somehow I could get it back to the Normal setting instead of Optimal, with my cooling adjusted maybe it would boost higher than before? Or is it faster to just stay at 3899.20 and be consistant?

Perhaps I've reached some type of optimal setting given my current hardware and should be happy? Or should I adjust something?
I feel like an idiot because I don't know much but I am thankful for the helpful people here and that the system is running as good as it is. It's exciting to finally be able to do these animations I've been wanting to do for years!

Set it back to normal, optimal is a static overclock. You want normal so it will boost properly and give you the best single core clocks when needed.
 
any clock speed between base and boost is normal. when it drops below base its being throttled. pcmag is not the end all answer.
I'll accept the definition given by a known and respected computer magazine, but I will notl believe someone postig personal opinions with noth supporting their claims.
Thanks guys.
So I'm a little over 9 hrs into another long render and it's staying at 68c and 3899.20MHz on all 32 cores at 100% load.
I realize you are all smarter than I am but here are my thoughts. See if you agree or please correct me.

Prior to today my PC was typically running 73/74c at 100% load with the MHz varying a little 3866-3891MHz and sometimes a little above 4000MHz briefly.
Last night I made the "mistake" (misunderstood trial) of changing my BIOS from Normal to Optimal, which resulted in a 5% Overclock, according to the boot screen now. So my PC became a 3.9GHz instead of 3.7. It won't change back. I changed the fans to ramp to 100 sooner than before. So I am guessing that the Optimal setting tweaks my system to run at whatever consistant frequency it can sustain hence now it stays exactly at 3899.20MHz instead of varying like it did before?
So while it could boost to 4.5GHz it wont' because it can't cool it so the system just decided to stay at 3899.20?
Or it's as if the system is now just set to a constant 5% OC and no longer boosts, or drops for that matter?
So this makes me wonder if somehow I could get it back to the Normal setting instead of Optimal, with my cooling adjusted maybe it would boost higher than before? Or is it faster to just stay at 3899.20 and be consistant?

Perhaps I've reached some type of optimal setting given my current hardware and should be happy? Or should I adjust something?
I feel like an idiot because I don't know much but I am thankful for the helpful people here and that the system is running as good as it is. It's exciting to finally be able to do these animations I've been wanting to do for years!
What Jamie Marsala said, but I have no to make the change. Sorry.l
 
Set it back to normal, optimal is a static overclock. You want normal so it will boost properly and give you the best single core clocks when needed.
Static Overclock. That is what I was trying to say. :) Thank you. I set it back to Normal mode and rebooted but it still says 5% Overclock and 3.9GHz. So I rebooted again and went to BIOS and it shows it's on Normal mode but it's still operating at the 5% OC, 3.9GHz. It won't change back even though it shows to have changed back to Normal setting. It's still really running at the OC setting that occurred when I went to Optimal mode
 
Static Overclock. That is what I was trying to say. :) Thank you. I set it back to Normal mode and rebooted but it still says 5% Overclock and 3.9GHz. So I rebooted again and went to BIOS and it shows it's on Normal mode but it's still operating at the 5% OC, 3.9GHz. It won't change back even though it shows to have changed back to Normal setting. It's still really running at the OC setting that occurred when I went to Optimal mode
clear the bios and leave it at auto.
 
Static Overclock. That is what I was trying to say. :) Thank you. I set it back to Normal mode and rebooted but it still says 5% Overclock and 3.9GHz. So I rebooted again and went to BIOS and it shows it's on Normal mode but it's still operating at the 5% OC, 3.9GHz. It won't change back even though it shows to have changed back to Normal setting. It's still really running at the OC setting that occurred when I went to Optimal mode

Interesting. You do not have some OC software running do you, like the motherboard manufacturers? I have a MSI board and their crap software does that, reason why I removed it. You had to click a certain plan twice to get it to go back to BIOS controlled settings.

Edit: On a side note I am sitting here folding@home and my chip is sitting at all core boost of 4.275Ghz and under 75C. It has raised my coolant temp 6C though.
 
Interesting. You do not have some OC software running do you, like the motherboard manufacturers? I have a MSI board and their crap software does that, reason why I removed it. You had to click a certain plan twice to get it to go back to BIOS controlled settings.

Edit: On a side note I am sitting here folding@home and my chip is sitting at all core boost of 4.275Ghz and under 75C. It has raised my coolant temp 6C though.
No I do not have any extra OC software running. When the computer is booting if you press DEL or F2 then it boots to BIOS screen. And from there you can change settings. For years ASUS has had this 3 mode option of "power saving", "normal", "optimal" and each show a little tachometer-like gauge you can click on. On my i7 I clicked on optimal in the past and it changed itself back to normal. But when I clicked on optimal on my Ryzen PC it OC'd and remains OC'd even though I rebooted several times and changed it normal. It's now stuck on OC / Optimal.

As mentioned above I guess I'll clear the BIOS and then go from there. Not sure how to do that? Click DEFAULT settings? I also thought about clicking "power saving" mode to see if that gets it off OC and then try to go back to normal or change settings myself.
 
No I do not have any extra OC software running. When the computer is booting if you press DEL or F2 then it boots to BIOS screen. And from there you can change settings. For years ASUS has had this 3 mode option of "power saving", "normal", "optimal" and each show a little tachometer-like gauge you can click on. On my i7 I clicked on optimal in the past and it changed itself back to normal. But when I clicked on optimal on my Ryzen PC it OC'd and remains OC'd even though I rebooted several times and changed it normal. It's now stuck on OC / Optimal.

As mentioned above I guess I'll clear the BIOS and then go from there. Not sure how to do that? Click DEFAULT settings? I also thought about clicking "power saving" mode to see if that gets it off OC and then try to go back to normal or change settings myself.

To clear your BIOS/CMOS the best way is to short or jumper the reset pins on the motherboard or you can remove the battery from the board with the power cable unplugged and wait like 5 or 10 minutes then put it back in. I got lucky and my motherboard actually has a clear the CMOS button right on the back panel.
 
No problem. Glad I can help.

No, twin tower coolers with only middle fan perform within a couple degrees of 2x fans in push / push. In fact single middle fan on twin towers cool the same at same noise level as 2x fans in push / push. Reason is while 2x fans do flow a little more air though cooler, single fan makes about 3dB less noise, and if ran faster at same dB noise level it flow same amout of air as 2x fans at same dB, so same cooling.


[H]ard|OCP tested Silver Arrow TR4 when first out, but it was combined with Liqtech TR4 and a couple others, but that review was pulled I think becuase of the Liqtech TR4 problems. Initial testing Only ran TY-143 at half speed (1300rpm) and temps were not great, but when they re-tested and ran it full speed it actually cooled within 9.5c of fancy custom loop and better than Liqutech TR4.

Here's detailed dimensional drawing:
View attachment 226207


Below link is to member dpriest here using Silver Arrow TR4
https://hardforum.com/threads/noctu...oler-roundup-h.1969417/page-2#post-1043930057
I swapped out the TY-143 fan for the Noctua NF-A15 Chromax 140mm fan and added a second one. It cools my TR 1950x very effectively however I am concerned it will effectively cool the TR 3960x which I am thinking about upgrading to. I am also thinking about the 3950x and the only air cooler I would consider is the Thermalright Le Grand Macho RT but not sure this will effectively cool the 3950x. I have not considered a custom water loop because it seems like a headache with having to periodically change the coolant plus heard many stories about leaking issues at some point in time. Somebody once said water and computers don't mix. What are your thoughts?
 
I swapped out the TY-143 fan for the Noctua NF-A15 Chromax 140mm fan and added a second one. It cools my TR 1950x very effectively however I am concerned it will effectively cool the TR 3960x which I am thinking about upgrading to. I am also thinking about the 3950x and the only air cooler I would consider is the Thermalright Le Grand Macho RT but not sure this will effectively cool the 3950x. I have not considered a custom water loop because it seems like a headache with having to periodically change the coolant plus heard many stories about leaking issues at some point in time. Somebody once said water and computers don't mix. What are your thoughts?
I assume single NF-A15 Chromax gives same performance as Ty-143 .. and even with 2x NF-A15 they cool the same at same noise level as single TY-143? Did you changed to NF-A15 Chromax because you wanted blck fans? Being a 1500rpm fan vs the TY-143's 2500rpm means while A15 performs basically the same as TY-143 at same speeds up to it's 1500rpm, limit, the TY-143 being able to go on up to 2500rpm can move way more air thus cooler better .. but obviously making much more noise doing it.

I don't know if LGM RT will cool as well as as Silver Arrow TR4. LGM RT also doesn't have near as large a base like Silver Arrow TR4 has. But on 3950X it should work just fine. If it was me I would be looking at TRUE Spirit 140 Power. It has more cooling ability with much lower cost of about $450 compared LGM RT costing about $70. Just keep in mind TRUE Spirit 140 Power is 171mm tall and 155mm wide but only 53.4mm deep compared to LGM RT 159mm tall, 159mm tall and 124.5mm deep. without fan.
 
AMD says base clock is 3.7GHz with boost clock up to 4.5Ghz. My understanding is boost clock is how fast each core can run if it stays cool.

Negative. The boost clock is the maximum speed achievable by the CPU in single-threaded or lightly threaded workloads. Only one or two cores on any Ryzen 3000 series CPU is capable of achieving the maximum boost clock. There is a separate, all core boost clock for multi-threaded workloads. Generally, this is going to be around 4.0-4.15GHz, best case scenario.
 
Negative. The boost clock is the maximum speed achievable by the CPU in single-threaded or lightly threaded workloads. Only one or two cores on any Ryzen 3000 series CPU is capable of achieving the maximum boost clock. There is a separate, all core boost clock for multi-threaded workloads. Generally, this is going to be around 4.0-4.15GHz, best case scenario.
Are you saying a single core will boost up to 4.5GHz even if it's temp is above throttle point?
And that only one or two f threads can achieve those speeds even if temps ae below throttle point?
 
Are you saying a single core will boost up to 4.5GHz even if it's temp is above throttle point?
And that only one or two f threads can achieve those speeds even if temps ae below throttle point?

I said absolutely nothing about CPU throttling. However, to answer your questions directly, I'd have to say: "No." A Ryzen 3000 series CPU will not boost a single core up to 4.5GHz if it is above the throttle point. Their thermal protection won't allow it. Secondly, the answer is: "Yes." Only one or two individual CPU cores can achieve the maximum rated boost clocks even if temperatures are well below the thermal throttling threshold. Taking that a step further, only one or two CPU cores are physically capable of clocking to 4.5GHz or higher in any given Ryzen 3000 series CPU.

I don't think you understand how Ryzen boost clocks work at all. These are not like Intel CPU's. AMD's advertised boost clocks are not achievable by all of the individual cores in the CPU and at no point can these CPU's ever clock all of these cores at once to such speeds. The advertised boost clock only applies to single threaded workloads. AMD hesitates to call it a single-core boost clock, but that's what it is. That is to say you will only see the advertised maximum boost clock for any Ryzen CPU on a single CPU core when the CPU is processing single-threaded or lightly threaded workloads. Period. End of story.

Using a Ryzen 9 3900X as an example, only one or two cores are physically capable of hitting 4.6GHz. The reason why you will only ever see one or two CPU cores achieve the maximum rated boost clocks comes down to the quality of the individual CPU cores. Unlike Intel CPU's, these are not created equal. A Core i9 9900K has a max turbo boost rating of 5.0GHz. However, each and every core in that CPU can achieve that maximum rated speed. This is NOT true of AMD CPU's. Quality varies considerably from CCD to CCD, CCX to CCX and even down to individual cores. In short, most of the Ryzen cores on any given processor physically can't clock that high. Some will max out at 4.3GHz and others 4.4GHz. Some people report having CPU's with cores that can't do more than 4.2GHz reliably.

In order to hit the advertised value of 4.6GHz on a 3900X, AMD only has to have one core in a given CPU that's capable of achieving such speeds. This increases yields and allows AMD to advertise a higher clock speed than the CPU's are truly capable of. It's a misstep in marketing in my opinion as its confusing and misleading. AMD doesn't even disclose the maximum rated multi-threaded workload these CPUs will achieve in their specifications because it varies due to a multitude of factors. However, AMD has always forthcoming with data on the matter showing that under multi-threaded workloads, a CPU like the 3900X will start out at around 4.3GHz (or higher) on all cores and then slowly throttle down to around 4.0GHz or so over a prolonged period of time. Though, some people will experience lower clocks than that depending on their cooling and ambient temperatures. Other elements factor into your actual clock speeds such as workload, motherboard VRM implementation and quality, and so on.
 
Dan_D
Thank you for the detailed explaination. I was aware of it, but you cleared up a lot I wasn't sure of. I'm going to re-read it a couple times to get it all sorted in my head.

What speeds will other cores in your 3900X AMD example have? Do you know how slow the slowest cores are?

Thanks again for detailed explaination.
 
  • Like
Reactions: Dan_D
like this
Dan_D
Thank you for the detailed explaination. It cleared up a lot of questions. I'm going to re-read it a couple times to get it all sorted in my head.

What speeds will other cores in your 3900X AMD example have? Do you know how slow the slowest cores are?

It's a literal silicon lottery. The 3900X has two CCDs (Chiplet Core Dies) in them. The running joke is that you get one "Chiplet" and one "Shitlet." One of those CCD's is always considerably weaker than the other one. Even within a given CCD, one of the two CCX complexes is often weaker than the other. Generally speaking, Ryzen CCD's or CCX's generally max out at 4.3-4.4GHz. Silicon lottery tested a bunch of them and found that many won't do more than 4.2GHz. In multi-threaded applications, they don't clock even that high for very long. Load up Cinebench R20, Handbrake, Gooseberry, or the Blender Benchmark and they'll start out high on all cores but downclock almost immediately to 4.0GHz to 4.1GHz.

On my test chip, I haven't tested each individual CCX to see what they clock at. That said, I know at least one of them does 4.4GHz from messing around with it. All of the cores in my 3900X example can achieve a 4.3GHz manual overclock with 100% stability. I do it on every motherboard I review and I did it in my initial review of the CPU.
 
Just curious, are your chips from free samples groups or general sales? I've long suspected items given away as review review might be better than run of the mill retail, but have never been able to prove it.

Again, thanks for sharing your knowledge.
 
Just curious, are your chips from free samples groups or general sales? I've long suspected items given away as review review might be better than run of the mill retail, but have never been able to prove it.

Again, thanks for sharing your knowledge.

I've long since suspected that the CPU's I get from AMD and Intel are cherry picked. I've got no proof of this either beyond seeing slightly better results than the general populace often does. The AMD 3600X, 3700X (one of them), and 3900X's were all sent to us at the same time by AMD and were in sealed, retail packaging which was marked "Not for Resale." I can't say for sure that these are different than retail CPU's or cherry picked but it seems likely. Silicon Lottery's testing said something about most Ryzen 3000 series CPU's being incapable of 4.2GHz all core overclocks. That's below the 4.3GHz all core overclock I've seen all of our sample chips achieve.

Our 3950X was a retail CPU. I sent that one back to David after I was done reviewing it. I believe he sold it. I did purchase my own retail 3950X which is in my system now. AMD sampled the 3950X and Threadripper CPU's to reviewers at the same time and we weren't selected to receive them. In fact, only the biggest and most popular sites and Youtubers got them. This suggests that their stock of those was very small and this appears to have been true given their availability in the weeks after launch or even to this day.

Having said all of that, personally, I've generally been able to achieve the same results on both retail and engineering or press kit CPU's. Due to Kyle's sometimes antagonistic relationship to Intel, as well as my own, most of my Intel CPU samples are retail chips. I have had some engineering samples procured through various sources, so I have experience with them, but the majority are retail. With AMD, it's oddly the opposite, with most of them being from press kits.
 
Thanks Dan_D

My guess is you are able to achieve similar results and maybe sometimes even better results is because of your skillset as much or more than chip quality. Kyle and I sometimes butted heads, but I respect him (and you) trying to give us readers the truth and not what marketing hype wants. I did cooler testing/reviewing and know all too well how hard it is to get samples when reviewers tell the truth about products instead of keeping industry happy.
 
I assume single NF-A15 Chromax gives same performance as Ty-143 .. and even with 2x NF-A15 they cool the same at same noise level as single TY-143? Did you changed to NF-A15 Chromax because you wanted blck fans? Being a 1500rpm fan vs the TY-143's 2500rpm means while A15 performs basically the same as TY-143 at same speeds up to it's 1500rpm, limit, the TY-143 being able to go on up to 2500rpm can move way more air thus cooler better .. but obviously making much more noise doing it.

I don't know if LGM RT will cool as well as as Silver Arrow TR4. LGM RT also doesn't have near as large a base like Silver Arrow TR4 has. But on 3950X it should work just fine. If it was me I would be looking at TRUE Spirit 140 Power. It has more cooling ability with much lower cost of about $450 compared LGM RT costing about $70. Just keep in mind TRUE Spirit 140 Power is 171mm tall and 155mm wide but only 53.4mm deep compared to LGM RT 159mm tall, 159mm tall and 124.5mm deep. without fan.
I did switch the TY-143 for the NF-A15 Chromax because of the color. I also control all of my fans manually with 3.5" bay fan controllers. All of these Noctua fans that I have are LOUD. When not encoding with Vidcoder, I adjust the fans to the lowest setting they can handle before not spinning which is around 1300 RPM. When encoding, I turn them up to max settings, running at 2800 RPM. When they are turned up this high my rig sounds like a turbine jet engine. I usually leave the room when encoding but I can hear these fans in other rooms in the house.

I know most people connect their case fans directly to the mobo but I never understood how you can have full manual control of the fans this way. The reason I chose the Phanteks Enthoo Pro case was because of the three 3.5" bays. Two of the bays have fan controllers and the third bay has a Bluray burner which I can put in an enclosure and connect it via USB since I rarely use it. I would like to know if there is a way to control the fans the same way without a fan controller and then maybe I would consider another case with better air flow. Don't get me wrong, I love this case and it seems to have very good air flow but I am somewhat of a noobie at all this stuff compared to others on the forum. I would love your thoughts about what I am pondering with controlling fans and a case with great air flow. I like the full size ATX case. Makes it easier to do builds.
 
Thanks for the info about Chromax fans. I didn't realize they wouldn't idle belwo 1300rpm. I assume that is voltage control, not PWM control? I know TY-143 will idle down to 500-550rpm on PWM control.

I personally prefer manually setup temp controlled fan curves because fans spin at speed needed to keep things cool .. and in my own systems quiet because I limit OC to heat loads that when at full load fans can supply airflow needed to keep components at safe temps at noise level just loud enough so I know system is working extremely hard. This translates to about 35dB at 1 meter from case.

Enthoo Pro is a good case. Big enough to hold all most people need but not too big to fit under desk, with plenty of room for big coolers and GPUs and has good airflow. Wtth 2x 140mm front and 1x 140mm bottom intakes with matching case venting I can't think of a better airflow case off the top of my head. There are a few now out that will take 3x 140mm front and 2x 140mm bottom, but the bottom front fan and front bottom fan both try to push air into same space.

Removing all PCIe back slot covers increases rear vent area around GPU for better front to back airflow thus lower temps and noise. Removing bottom and back grills also greatly improves airflow and lowers noise too. Raising case on an open center caster base gives bottom intake vents much better airflow. I make mine out of 9mm Baltic birch 7-ply plywood cut in strips about 35mm wide with lap-joints on corners to conserve material. I use threaded stud twin wheel casters and tap threads into base for studs, then sand, prime, fill and sand until smooth and paint to match case. Link below is the design drawing and a couple images of one under Enthoo Primo case.
http://phanteks.com/forum/showthread.php?510-Tips-amp-Guides
 
Thanks for the info about Chromax fans. I didn't realize they wouldn't idle belwo 1300rpm. I assume that is voltage control, not PWM control? I know TY-143 will idle down to 500-550rpm on PWM control.

I personally prefer manually setup temp controlled fan curves because fans spin at speed needed to keep things cool .. and in my own systems quiet because I limit OC to heat loads that when at full load fans can supply airflow needed to keep components at safe temps at noise level just loud enough so I know system is working extremely hard. This translates to about 35dB at 1 meter from case.

Enthoo Pro is a good case. Big enough to hold all most people need but not too big to fit under desk, with plenty of room for big coolers and GPUs and has good airflow. Wtth 2x 140mm front and 1x 140mm bottom intakes with matching case venting I can't think of a better airflow case off the top of my head. There are a few now out that will take 3x 140mm front and 2x 140mm bottom, but the bottom front fan and front bottom fan both try to push air into same space.

Removing all PCIe back slot covers increases rear vent area around GPU for better front to back airflow thus lower temps and noise. Removing bottom and back grills also greatly improves airflow and lowers noise too. Raising case on an open center caster base gives bottom intake vents much better airflow. I make mine out of 9mm Baltic birch 7-ply plywood cut in strips about 35mm wide with lap-joints on corners to conserve material. I use threaded stud twin wheel casters and tap threads into base for studs, then sand, prime, fill and sand until smooth and paint to match case. Link below is the design drawing and a couple images of one under Enthoo Primo case.
http://phanteks.com/forum/showthread.php?510-Tips-amp-Guides
Correction...These fans are the 3000 RPM Chromax and will not run below 1500 RPM because they need a minimum of 6 volts in order to keep spinning. I have two Thermaltake Commander controllers. How do you manually setup temp controlled fan curves? This case has a PWM hub for powering case fans that connect to the motherboard. What grill on the bottom are you referring to? The PSU cover? This is my setup: Any feedback is greatly appreciated.
Front.jpg
Inside.jpg
Top View.jpg
 
Nice looking rig!

Hate to burst your bubble, but the intake & exhaust side by side in I assume top of case are moving air in a circle from one to the other.
05064c1b_Intake-Exhauxtairlfowloop.jpg
Motherboard fan control software/bios use compoent temp to regulate fan speed. Most can be setup in windows, but to get many to keep curve settings they need to be set in BIOS. They usually have 3-6 temperature / rpm points in fan curve from idle to full speed. You will probably also need a couple PWM splitters hubs with PSU SATA or Molex plug for 12v power. They use the PWM signal from motherboard PWM fan header to control all PWM fans plugged into hub. A single PWM signal can control up to about 6-8 fans, and that is how many headers are on them. Basically you would need at least one PWM splitter hubs for CPU and intake fans supplying it as well as others. I usually use 2 hubs, one for CPU and it's supply / exhasut fans, and 2nd for GPU and it's related fans and/or it's supply / exhasut fans.
Links below are to just a few of the many PWM splitter hubs. Just make sure it has the PSU power plug (SATA / Molex) you want for the 12v power to fans and it is a PWM splitter sending PWM signal to each fan.
https://www.amazon.co.uk/SilverStone-System-Cables-Black-CPF04/dp/B00VNW556I
https://www.aquatuning.co.uk/air-co...MIgP7OnPb06AIVAu7tCh0E2QK4EAQYCSABEgJP_fD_BwE
https://www.banggood.com/Computer-C...tCh1lowLkEAkYCyABEgLVjPD_BwE&cur_warehouse=CN


You probably know it, but just in case, the Phanteks PWM fan hub is not a PWM fan hub at all, but a PWM controlled converting to variable voltage controlled headers on hub. Below is PWM% signal to fan number vs voltage hub supplies.
attachment.php

Below is link to article if you want more info.
http://phanteks.com/forum/showthread.php?557-PWM-Fan-hub-Test-and-Review&p=2540#post2540
 
As an Amazon Associate, HardForum may earn from qualifying purchases.
Nice looking rig!

Hate to burst your bubble, but the intake & exhaust side by side in I assume top of case are moving air in a circle from one to the other.
View attachment 238936
Motherboard fan control software/bios use compoent temp to regulate fan speed. Most can be setup in windows, but to get many to keep curve settings they need to be set in BIOS. They usually have 3-6 temperature / rpm points in fan curve from idle to full speed. You will probably also need a couple PWM splitters hubs with PSU SATA or Molex plug for 12v power. They use the PWM signal from motherboard PWM fan header to control all PWM fans plugged into hub. A single PWM signal can control up to about 6-8 fans, and that is how many headers are on them. Basically you would need at least one PWM splitter hubs for CPU and intake fans supplying it as well as others. I usually use 2 hubs, one for CPU and it's supply / exhasut fans, and 2nd for GPU and it's related fans and/or it's supply / exhasut fans.
Links below are to just a few of the many PWM splitter hubs. Just make sure it has the PSU power plug (SATA / Molex) you want for the 12v power to fans and it is a PWM splitter sending PWM signal to each fan.
https://www.amazon.co.uk/SilverStone-System-Cables-Black-CPF04/dp/B00VNW556I
https://www.aquatuning.co.uk/air-co...MIgP7OnPb06AIVAu7tCh0E2QK4EAQYCSABEgJP_fD_BwE
https://www.banggood.com/Computer-C...tCh1lowLkEAkYCyABEgLVjPD_BwE&cur_warehouse=CN


You probably know it, but just in case, the Phanteks PWM fan hub is not a PWM fan hub at all, but a PWM controlled converting to variable voltage controlled headers on hub. Below is PWM% signal to fan number vs voltage hub supplies.
attachment.php

Below is link to article if you want more info.
http://phanteks.com/forum/showthread.php?557-PWM-Fan-hub-Test-and-Review&p=2540#post2540
Thanks! Here is my thought process for the top of the case with the 3 fans, one exhaust and two intake. Please don't laugh: The two fans closest to the front of the case will bring a lot of air onto the heat sink and the fan in the rear will exhaust the warmer air. What do you recommend for the top fans? What grill on the bottom are you referring to with removing? As for the splitter hubs, my GPU (Nvidia GTX 1050) never uses its fans because this rig is used exclusively for video encoding and taxes the processor so it seems like I would only need one splitter hub? For fans, I have two 150mm on the Silver Arrow, Two 140mm front intake, one 12mm intake on the bottom intake which I THINK IS A WASTE BECAUSE OF THE PSU COVER, and three 140mm on the top.

I would assume I connect one of the splitter hubs to the motherboard (Asrock X399 Taichi.) I have a few options and not sure it matters which one I connect it to:

1.) Chassis Fan/Waterpump fan connector
2.) CPU Fan connector
3.) CPU/ Waterpump Fan connector
4.) Chassis Fan connectors (x2)
 
As an Amazon Associate, HardForum may earn from qualifying purchases.
Back
Top