New SFF Rig

[corruption]

Sup buddy the new build looks great , i wish you were still local so I would get a chance to see it in person .
One thing I was thinking about is your Fans . Does that case have some type of built in PWM fan hub as your ITX board really only has 2 free headers . If it doesn't you might need to grab a Hub to power them all . Those Fans don't work properly if speed is set lower then %40 .

You'll have to come down for a visit. (I'd recommend the last week of July/beginning of August to catch the Cup Match festivities.)

I think the fans come with a y-cable so I should be able to get by with just the 2 connections on the MB. If they don't, I'll look into getting a fan hub or a y-cable.

---

Yes, I'd consider mATX but I don't think you want to go that way. That ITX case is 14.05" x 5.31" x 11.30" which is fairly large, but has very poor airflow that almost forces you to add that AIO water cooler. The 3700X doesn't need to be water cooled, that pricey case is making you do it. IMO the $500 price tag for the case, fans and AIO is insane. You should be able to get by with a $100 case like the Phanteks Enthoo Evolv Shift Air while using the AMD cooler and apply the $275 savings to a better GPU. It is a bit larger at 18.5" x 6.7" x 10.8" but has a similar look and doesn't stack the GPU on top of the motherboard which is what suffocates the CPU HS/Fan. Here's the manufacturer's info page on it. But it is your money.

I was looking to get an AIO cooler to allow for additional cooling capacity if I decide to upgrade to a more power-hungry CPU in the future. If I was going to stick with air cooling alone, I'd go for something closer to the Sliger SM550.

I agree that the cost of the case, AIO, and fans is steep. If I wasn't looking to reasonably minimize space, I'd get any sized board, GPU, CPU, and RAM and throw it into my Thermaltake Armor.

I looked at the Evolve Shift for a long time. I like the look of the case with the glass panels, but repeatedly saw people report issues with trying to keep temps down. An AIO is needed for the CPU, but the GPU often runs hot. The Evolve Shift X allows for a larger custom loop, or two AIO loops (120 rad and 240 rad), but I'd be worried that it wouldn't last long due to being tipped over. Unfortunately, I'm not as big a fan of the mesh side panels. The mesh may not help keep the GPU cool unless you can rotate it to allow the fans to face to outside of the case. The default GPU orientation is to face the fans to the center of the case. I've seen one reviewer cut the case to allow his GPU fans to face outwards. The I/O plate's position at the top of the case (I'd only run this standing up as it looks too good to lie it flat) throws me a bit, but I'm sure I'd get over it quickly.

---

+1 great looking case. You can also get it with a glass panel instead if you really need the bling. You can also get airflow front panels here: https://modmymods.com/catalogsearch/result/?q=evolv+shift

Some videos about the shift and shift x

I hadn't seen those front panels yet, thank you for the link!

---

I already posted that before but this is my ITX with Ryzen and 1660Ti:
View attachment 207687

CPU: 3700X@stock + Noctua U12A, can be U12S with 2 fans, max temps in a closed case with AVX ~86°C
Case: Raijintek Metis Plus, costs ~$65 link they have also something like this but I think that Sliger has a better design - link
RAM: 16GB Dominator Plat RGB but can be anything 16GB or more if it's required ~$100+
GFX card: MSI GTX1660Ti OC - ~$275 but EVGA should fit too and short 2060 also but I don't recommend more than ~130W cards in ITX cases or they get noisy or throttle
PSU: Enermax 650W SFX ... can be any SFX like Corsair SF600 is great and runs passive up to 50%+ load ... SFX saves space and is lighter than ATX
Mobo: there is ASUS CH8 Impact but can be any ITX/DTX mobo

It's just an example of what I made.

To run higher wattage graphics card without noise or overheating you need a larger case like mATX or something with good airflow next to the GFX card. That vented Sliger case looks great but most others in ITX format will cause issues.
My experience with the Evolv Shift ... big, problems with reasonable coolers, bad airflow, every 130W+ graphics card is noisy.

Btw. AMD stock cooler can't even handle 3k Ryzens at stock without throttling. It causes the CPU to hit 95°C and throttle down but the user can't see how the CPU clock drops.

Thanks for the pic, build specs, and case links! Are the max temps for the CPU obtained with a stress test or when gaming?

I think that with both upper and lower fans set to exhaust along with a vented panel on the GPU side, I can expect GFU temperatures close to what I would see on a bench/open air.

---

I think I'm going to go for the 5700XT over the 2060 Super. The lower power consumption definitely made it tempting. Several things pushed me onto that side among which includes the need for a "fruit-based" OS boot. Recent reports make it look like it will have native 5700XT drivers. (As a tech, I'm usually the go-to in the family for questions regarding that OS. I need to keep somewhat up-to-date on features and be able to provide offline backups.)

I'm still on the fence about the backup drive, but need to pull the trigger on the order soon. I'll most likely make a last-minute decision.

Thanks for the input guys, it's appreciated!

 

[||Console||]

The Gigabyte card you are looking at is shown in this review (

) It wasn't the best performer but it defiantly wasnt the worst only off a few FPS ( that im sure could be fixed with a small overvolt when gaming ) , It had lower then avg GPU temps and power draw ( both will be good for the temps you get in your area and the cost of power there ) .

For the storage drive IMO spend the extra $10 on a 7200 RPM drive If you are ok with losing that bottom fan . Better yet just put the storage drive in the 4670k system I know that armor has room for 3.5 drives , your case must be pretty empty as I have your WC gear =)

 

[Woomack]

Then they shouldn't include them in the package. It just makes AMD look bad to novice builders who just assume (quite correctly) that they should work. They look decent, much better than Intel stock HS/Fans. Intel doesn't provide HS/Fans with their "K" series overclockable CPUs and AMD shouldn't include them with their "X" series higher TDP CPUs just like with the original 1000 series Ryzens. A bonus is that AMD could sell them cheaper deleting both the cost of the cooler and the extra shipping cost due to the weight of the cooler. I only used one of the AMD coolers once, with a mini ITX Ryzen 5 1600 setup when they first came out. The AMD cooler didn't work well in that instance due to clearance issues between the top of the cooler fan and bottom of the power supply. Since I've used a 240mm AIO Corsair H100i and then my trusty old Hyper 212 EVO.

Then I would blame the case (and perhaps ambient) cooling, not the cooler. And of course, proper case cooling is the responsibility of the computer assembler/user. If it was the stock cooler's lack of capability, virtually every one of those Ryzen users would be complaining - and that is not happening.

I agree with DaveB - they shouldn't include them in the package if not capable. And to that, I say they wouldn't.

...

I was testing Ryzen 3600, 3600X, and 3700X. Because of the design, all of them are heating almost the same (even though 3700X has more cores). I was consulting my results with Noctua, and they confirmed everything based on their lab results.
Also, every reviewer who made proper tests will tell you that at auto settings, these CPUs reach 95°C and then throttle down to keep this temp as max during longer work.
I was testing more detailed Ryzen 3600 with the cooler from 3700X, and it couldn't handle the CPU. Every Ryzen 3000 is starting to throttle once it hits 95°C. At this temp, it drops voltage and clock so the user can't see that it actually does that without additional monitoring. So end-users don't complain about things they can't see, and those who are more advanced are not using stock AMD coolers.
Throttling is happening almost only under full CPU load, but if you close it in a small case without good airflow, then it's clear that it will appear faster.

For me, it's weird that most reviews are skipping the cooler issue. On the other hand, it's not hard to notice that most "professional" (not always written by professionals but people who got paid for that) reviews were affected by AMD. We can only start from the reviewers kit delivered by AMD with CPU+mobo+RAM and whatever else they wanted to give.

Hyper 212 will keep these CPUs at 92-95°C. At 95°C, the CPU will keep the voltage at 1.1-1.25V.
Just for comparison:
- Noctua D9L with one fan can keep them up to 94°C, without throttling - mackerel was mentioning about slightly better results on his D9L and 3600 CPU but not much, more like 1-2°C less. The same cooler with two fans can keep it at about 92°C.
- Scythe Fuma 2 so twin tower cooler with 2x 120mm fans gives about 87-88°C max on fully loaded 3600/3700X
- Scythe Big Shuriken 3 (reviewed some time ago also on Overclockers) gives about 88-89°C, but the fan spins at its max
- Noctua U12A keeps my 3700X in a closed ITX case at about 85-86°C
- Silentium PC 280 AIO (the same OEM as Corsair) had 84-85°C

Black series Noctua coolers arrived yesterday, but I won't have time to test them till next week.

You can't compare these chips with the 2k series. I was actually really disappointed with temps when I got my first 3700X as I thought I would use it in a smaller ITX, and there it was throttling (Kolink Rocket case). After some searching, I found mentioned Raijintek, which is probably the smallest case in which all is quiet and keeps reasonable temps.

Re my photo. Fans are set as exhaust. I had to remove a case fan or the one on the cooler wouldn't fit. Intake is next to the graphics card, and it seems enough. I'm actually surprised it handles the airflow so well. Started it as a "maybe it will work," but after tests, I just keep it this way. With some luck, maybe today I will get the 3900X, so I will try if I can use it in this ITX case.

@ corruption
Max temps were during the 3h+ AIDA64 CPU+FPU test. On these new AMD, it's like to run Prime95 with small FTT. While gaming, it's significantly less as most games can't use all CPU cores.
I made multiple ITX builds with higher series CPUs, and in most cases, the main problem was the graphics card, not the CPU. Usually, you can keep CPU cool enough with the average cooler. The graphics card has to be with a good cooler already and needs additional space or any other forced airflow so it won't overheat. New graphics cards will slightly throttle when they are too hot but usually not much more than 5-10%, so in games, you won't really see it. Worse is that fans will run at 100% and will be noisy if the card's design isn't the best.

I also recommend a bit larger case than the one I used. It was just an example. I find Sliger vented version pretty good, and it should handle a higher wattage graphics card.


You may consider high-end SFX PSU for your build like this -> https://www.corsair.com/eu/en/Categ...pply-Units-Advanced/SF-Series/p/CP-9020182-AR
Even if the case has space for ATX PSU then this SFX one runs passive up to 50%+, it's really quiet, has already braided cables and cables are quite short so in a small case it saves a lot of space + improves airflow.
I'm using the SF600 in the 7800X+Vega56 PC and it doesn't even spin the fan. The one on the photo is Enermax Revolution SFX 650W 80+ Gold and works about the same but is starting to spin the fan much faster ... and has worse finished cables.
In the case with space for ATX PSU you will need ATX to SFX bracket which costs maybe $5.

 

[Zerileous]

I think to be fair we need to consider the way Precision Boost is designed and whether or not we should use the term "throttling" to describe a reduction in boost in this case. My understanding is that Intel CPUs (I don't own an overclockable one) have a fixed duration which limits their boost. AMD CPUs using precision boost do not have a fixed duration, but rather will maintain the boost frequency until they encounter thermal limitations (in theory other limitations exist, but in practice it's always thermal for Zen2).

I have seen this behavior described by Gamers Nexus as very similar to how nVidia GPUs behave. I have seen in those cases the term throttle used to describe a rapid decrease in clocks and voltage due to temperature but I have not seen it used to describe the gradual modulation of clocks and frequencies based on cooler performance that we have learned to expect from said GPUs. Are you seeing a rapid or gradual decrease? I believe I have seen the terms "hard throttle" and "soft throttle" thrown around here to describe this.

Throttling has a negative connotation that the CPU is performing below spec. My question is whether this behavior is truly below "spec" if it is first operating within the parameters of precision boost (the stock boosting algorithm). If we don't look at why the clock is being decreased is this much different from a stock Intel CPU that decreases boost due to duration?

GamersNexus demonstrated that this behavior is almost infinite by using LN2 to measure clocks at steady temperatures all the way down to -80C. They found that performance scaled with temperature all the way down to -80C with stock settings. Does that mean that the CPU is "throttling" under LN2?

I'm not trying to fanboy too hard here, I do have a nostalgic bias as my first overclocking experience was on AMD, but I'm trying to be objective in understanding this. Please don't hesitate to educate me if I'm missing the point.

 

[trents]

I agree with Zerileous. We are using old terminology to describe the designed normal behavior of new technology that plays by new rules. We're still focusing on clocks in heavy load situations when we need to be evaluating these new chips on their performance across a broader range of usage scenarios for which the stock cooler is adequate.

 

[Bill_Bright]

It seems we (the industry) have gone full circle. This feels more to me like it is all marketing hype - not true technological features.

Back in the day, when a little extra "oomph" was needed, the processor would kick into "Turbo" mode. But many enthusiasts wanted their processors to always be in Turbo mode. Marketing weenies got wind of that so they tweaked the BIOS to keep the processors in Turbo mode. But of course, "Heat is the bane of all electronics" so when thermal thresholds were crossed, they had to slow down the processors. They had to come up with another marketing term. So they borrowed it from the notebook world where, due to case cooling restrictions, heat concerns were more readily accepted, and called it "throttling" hoping consumers would not see it as a bad thing.

It's like reducing the "standard" size 16 ounce bag of Lay's Potato Chips to 14 ounces and putting "New and Improved" on the bag and charging an extra 25 cents. Then to make consumers feel like they are getting a good deal, add back in 1 additional ounces of chips, change the labeling on the bag to "For a limited time only, get 1 free extra ounce".

 

[EarthDog]

You know what this sounds like, Woo, Trents, and Z...

"That depends on what the definition of "is" is..." - Bill Clinton





Being serious, even though it is normal behavior it is still 'throttling' to lower clocks due to how it is designed to work. If it didn't hit those thresholds power/temps/whatever, it wouldn't throttle/lower clocks and instead, maintain the specifications. But I still call that throttling as it is responding to limits and making adjustments.

 

[trents]

But what I'm driving at is I think it is inappropriate to, as some are doing, claim false advertising on the part of AMD when we are trying to hold them to power management standards that are defined differently than the manufacturer's design. I have no problem with using the term "throttling" if people quit using it in a critical way that implies dishonesty on the part of the manufacturer.

 

[DaveB]

"That depends on what the definition of "is" is..." - Bill Clinton

It is what it is. Throttling is a good thing, it protects the processor.

Delivering a CPU + HS/Fan that ALWAYS throttles if you run it at 100% is not a good thing.

Putting a Turbo speed on the box that the processor can't hit, also is not a good thing.

All the AMD fans need to suck it up and just admit that their beloved AMD, at a minimum, made some "errors in judgement". Now hopefully everyone can get over this, and either go back to discussing "New SFF Rig", which is the actual topic of this thread, or just move on to something else.

 

[Bill_Bright]

I have no problem with using the term "throttling" if people quit using it in a critical way that implies dishonesty on the part of the manufacturer.

I agree with this. Generally, when I think of processors throttling, I do think of it being a negative thing - but negative as in something is over-heating so the CPU is throttling out of self-preservation. And to that, I agree with DaveB - that's a good thing.

But the fact is, if something is over-heating, that typically means the user has failed to set up case cooling properly (to include using the wrong case), or they failed to keep the interior clean of heat trapping dust, or they over-pushed their clocks, or their ambient temps are too high or some combination thereof.

If the case interior is clean and is providing an adequate flow of cool air through the case, if the ambient (room) temperature is "normal" (depending on who you talk to is no more than 77°F/25°C), and all the voltages and clocks are at their defaults, no CPU should throttle with the OEM cooler or an aftermarket cooler that meets the CPU's makers requirements. That is a lot of "ifs" but IMO, perfectly acceptable expectations. So if the CPU throttles under those conditions, and the CPU is not otherwise faulty, then we have a problem with the CPU maker.

 

[Woomack]

No matter if it's AMD or Intel, these CPUs are acting the same for a good couple of years. Simply it's throttling when the CPU is overheating and various technologies implemented in the chip are starting to lower its clock and voltage to keep it at specified "safe" values.

For me it's not normal that the CPU which is designed to work at ~4.4GHz on 1-2 cores and ~3.8GHz on all cores, in real runs at 3.2GHz on all cores because any higher setting is causing it to overheat. This is the case with AMD and their stock coolers.
One of the problems which is related to both AMD and Intel chips is their rated TDP. If you maybe noticed (or not as barely anyone is checking specs all the time) then Noctua removed their TDP and compatibility list because latest gen CPUs TDP is nowhere near their recommended coolers TDP. It's just one example but I bet that some other cooler manufacturers made the same or at least changed requirements for each CPU.
When CPU that in theory has 65W TDP requires 120W TDP cooler so it won't overheat then something isn't right. How this CPU can work with a cooler designed for ~100W TDP?

Next thing. Ryzen 3000 series works at up to 1.45V. To keep it below 95°C throttling point, the CPU has to work at ~1.20-1.25V max. I know there is a voltage range and all that is specified but it simply means that the cooler is not designed to handle the CPU at all cores above ~1.2V.

Now the funny thing, at least for me and some who were comparing AMD CPUs on similar coolers can see the same. Take Noctua U9 series cooler. Similar cooler is for AMD and Intel, also for AMD TR. So my brother has 1920X CPU which using Noctua U9 series cooler runs at 4GHz on all cores and temps are around 70°C. Similar cooler on Ryzen 3600 can keep this CPU at ~90°C+ at stock settings so below 4GHz and at manual settings at 4GHz but 1.25V so less than designed for this clock. Thats the same clock and 100% cores difference.

I'm not saying that Ryzen 3000 CPUs are bad. They're great but are heating up a lot and AMD shouldn't sell them with that stock coolers like Intel does with their higher CPU series.

 

[Bill_Bright]

For me it's not normal that the CPU which is designed to work at ~4.4GHz on 1-2 cores and ~3.8GHz on all cores, in real runs at 3.2GHz on all cores because any higher setting is causing it to overheat. This is the case with AMD and their stock coolers.

It's NOT normal and for sure that should not happen AS LONG AS case cooling is properly setup and the cooler is properly mounted.

The only reason, IMO, it should throttle back is simply because the extra horsepower is not needed for the assigned task(s). If you are just typing text in a forum post, the CPU does need all cores running at full speed. That just needlessly wastes energy. No big deal on a PC (unless you are a treehugger) but sure would be if a notebook running on battery.

This is the case with AMD and their stock coolers.

Not in my experience. We use a lot of AMDs in our client builds. And again, in a properly cooled case, we are not seeing this problem. And for sure, I don't mean maxing out case cooling with 6 fans running at full speed. I'm talking about a typical mid-tower case with a nice 140mm fan in front pulling cool air in, and another in back exhausting heated air out. That's with the stock CPU cooler running at the default clock speeds.

Also for sure, if the client will be overclocking, we always push for additional cooling. But typically, if the client will be overclocking, they opt for processor that does not come with an OEM cooler.

 

[Woomack]

It's NOT normal and for sure that should not happen AS LONG AS case cooling is properly setup and the cooler is properly mounted.

The only reason, IMO, it should throttle back is simply because the extra horsepower is not needed for the assigned task(s). If you are just typing text in a forum post, the CPU does need all cores running at full speed. That just needlessly wastes energy. No big deal on a PC (unless you are a treehugger) but sure would be if a notebook running on battery.

Not in my experience. We use a lot of AMDs in our client builds. And again, in a properly cooled case, we are not seeing this problem. And for sure, I don't mean maxing out case cooling with 6 fans running at full speed. I'm talking about a typical mid-tower case with a nice 140mm fan in front pulling cool air in, and another in back exhausting heated air out. That's with the stock CPU cooler running at the default clock speeds.

Also for sure, if the client will be overclocking, we always push for additional cooling. But typically, if the client will be overclocking, they opt for processor that does not come with an OEM cooler.

I guess this is just a longer off topic and I thought it was already said but again ... I was testing everything on multiple open space rigs. Also, these results are repeatable and confirmed by other users from the OCF.
Power saving options work as designed. There is no problem with that but once the CPU is under full load then the stock cooler can't handle it.
Typical home/office user won't see that just because at auto, these CPUs balance clock, heat and power what I don't call throttling as it's normal behaviour under mixed load. Throttling as a lower than expected performance caused by overheating, will be visible only under full CPU load like long rendering etc.

I had no problems with 1000 and 2000 series Ryzens and no problems with 1000 series Threadripper. All were working great at low temps and were amazing for SFF builds. 3000 Ryzen is just something else.

 

[Zerileous]

Thank you for clarifying, what you're describing I would also describe as throttling. I think TDP needs to go away and manufacturers need to just state plain old power draw under stock settings. Neither want's to do it because then the other will be seen to have a massive advantage in efficiency because it will be compared to the false "TDP" of the competitor. Nerd Jesus chimed in on the issue recently:

.

Apologies to the OP for derailment: I think most of us agree that you'll be happier using an aftermarket cooler on your 3700X.

I'm happy to conclude this aside here, if anyone feels we need to discuss it further then we should probably spin this off into it's own thread.

 

[Bill_Bright]

Also, these results are repeatable and confirmed by other users from the OCF.

No doubt it happens. I am not denying that. I am just saying it's still not "normal". If it was "normal" the vast majority of the users of those processors would be experiencing those problems and surely there would be a HUGE, global uproar about it - perhaps even forcing AMD to have a product recall. But the fact is, the vast majority of users are not experiencing these problems.

 

[corruption]

Reading the temperature debate definitely makes me feel that a 240 AIO cooling solution is the way to go, especially if I want to have a more power-hungry CPU in the future. I have ordered the case and will be ordering the other parts later on today. It looks like the RAM is out of stock until Friday, so even if I get all the other parts by this weekend the build won't be able to be completed until the following week. I'll be sure to create a new post, and link to it here, with pics for the build once I have all the gear.

@ ||Console||
LOL, yeah the Armor is relatively bare in comparison to what it was with all my watercooling gear in it! The old rig will be used for business purposes, i.e. - data recovery, clones, database migration (essentially grunt-work). It won't be powered on all the time so I won't be able to reliably use it as a location for backup storage. I'm strongly resisting installing a 3.5" HD in the new case as I'd rather the additional airflow, at least at first. I'll cut off airflow to one bottom fan (unplug and physically block off) once I have it assembled and see what kind of impact a 3.5" drive would have on temperatures, minus the additional heat that it would put into the setup. If it's minimal, it would be a cheaper way to get faster bulk storage than a 2.5" drive.

 

[corruption]

Received the case yesterday and the bulk of the rest of the order today. I'm just waiting on the power supply, RAM, and SSD. I've decided to wait and think about the backup drive. I can use one of the old backup drives in my toaster while I make up my mind. (Not really what I want to stay with, but a good stop-gap.)

 

[habbajabba]

"Btw. AMD stock cooler can't even handle 3k Ryzens at stock without throttling. It causes the CPU to hit 95°C and throttle down but the user can't see how the CPU clock drops."

If you consider general latency in the system related to the actual hardware as it is configured off the pcb the only throttling here is from the manufacturers and not the hardware at all. Heat be damned.
I just bought a 6tb (wanted the 8) ironwolf pro for $200 off the egg. Really? A 5400 spinner in that? I actually plan on putting all 64GB of memory in my nas as I agree with billbright in that sometimes more is better than faster, at least in the short term.
The problem with airflow is the actual dust and debri in the air the box receives. I use some common filters for hvac registers that are only about 4"x12" though removed them recently after a hd change. Optimal airflow that is also quiet is actually larger fans running slower with lots of ventilation. Or, a water cooled setup, but that still needs fans...
I have a couple winix room air filters that have huge 8.5" fans. With the filters that sit in front (~1") if I were to put them on one side of the box and use the winix fan itself mounted to the other, it would be pretty awesome. You'd have to run it full speed but that is the point now aint it.

the pic kinda sucks but it is what it is-exhaust at the top btw. You can tell how good the filters work (~$20/per off ebay) by how clean that pic is and I smoke too.
What case config did you end up with btw?