Motherboard Incoming AM4 Mini-ITX boards

[szymon247]

Yeah, after double-checking the L9a should work. But it's hard to say 100% since I've yet to come across an official spec for the keepout zone on either AM3 or AM4.

That would be great.
What makes me wonder is whether it will provide sufficient cooling to the higher clocked chips.

It is rated at 65W for AMD yet still works with 95W Intel chips, yet despite AMD's reputation apparently even the 1600X (95W) doesn't get as hot under stress as the 7700K, which apparently can be kept in its 70s on the L9i. Probably at max it could keep a stock 1700X at just below throttling temps under sustained load. Surely overclocking wouldn't work though, not like it's much of a thing on current Ryzens. It seems like a perfect cooler for the stock 1700, 1600 and anything below though.

 

[cadiguno]

That's exactly my problem and it's surprising how tall they made it. It's about 0.5cm taller than the SG13 can fit. Otherwise it would be a perfect low profile cooler actually.

If I'm not an SFF enthusiast (which, let's be honest, we're quite a minority) I would really appreciate how effective the Wraith Coolers perform. Way better than an Intel stock, adequate for mediocre overclocking, and look pretty cool as well on top of that.

 

[szymon247]

If I'm not an SFF enthusiast (which, let's be honest, we're quite a minority) I would really appreciate how effective the Wraith Coolers perform. Way better than an Intel stock, adequate for mediocre overclocking, and look pretty cool as well on top of that.

Yeah I agree, although SFF is on the path to move into mainstream judging by how things are and despite some manufacturers (and sometimes even AMD themselves) being a little slow to see that, showcasing huge, mostly empty cases on hardware events. Everything is down-sizing, my rig included, and there are not many reasons these days for a full blown ATX tower. Well, heck, I couldn't find any downsides to moving everything I have from a 50 liter case to 11.5 and it made me wonder why they even still make huge cases, and I always had them before.

Since we moved this discussion off topic already, I checked your game in the sig - do you do your own artwork? It has a nice, Final Fantasy-ish style but is really nice.

And back on topic, did anyone by any chance managed to successfully reach Asrock or Gigabyte about how's the progress on their ITX boards? ASRock claimed April and then just rumor had it it moved to June, but no official updates. Gigabyte gave no official updates whatsoever. Maybe any lurkers?

 

[Thehack]

That would be great.
What makes me wonder is whether it will provide sufficient cooling to the higher clocked chips.

It is rated at 65W for AMD yet still works with 95W Intel chips, yet despite AMD's reputation apparently even the 1600X (95W) doesn't get as hot under stress as the 7700K, which apparently can be kept in its 70s on the L9i. Probably at max it could keep a stock 1700X at just below throttling temps under sustained load. Surely overclocking wouldn't work though, not like it's much of a thing on current Ryzens. It seems like a perfect cooler for the stock 1700, 1600 and anything below though.

You can't compare temperatures across platforms. Ryzen has a lower max thermal shut off temperature than Intel, and uses a soldered IHS.

What is cool or hot to either chips is completely relative.

 

[aquelito]

Do you know if there are X370 motherboards supporting PCIe bifurcation ?
I've read this is one of the chipsest feature (although the Biostar motherboard doesn't offer this option).

 

[cadiguno]

Since we moved this discussion off topic already, I checked your game in the sig - do you do your own artwork? It has a nice, Final Fantasy-ish style but is really nice.

Yeah we do art stuffs in-house, thank you so much for the praise!

And back on topic, did anyone by any chance managed to successfully reach Asrock or Gigabyte about how's the progress on their ITX boards? ASRock claimed April and then just rumor had it it moved to June, but no official updates. Gigabyte gave no official updates whatsoever. Maybe any lurkers?

Somebody on Reddit said ASRock rep confirmed it'll come May. No idea how far it's gone since though...

 

[szymon247]

You can't compare temperatures across platforms. Ryzen has a lower max thermal shut off temperature than Intel, and uses a soldered IHS.

What is cool or hot to either chips is completely relative.

Yeah, sure thing, I understand that. My logic was that:
1. When compared directly using the same cooler, the highest clocked Ryzens are cooler than the 7700k, with the 1800X hitting 65C against 7700K's 74C, and 1700X being some 20% cooler here: http://www.tomshardware.com/reviews/amd-ryzen-7-1700x-review,4987-8.html
2. The stock 7700k was said to be kept at 70-80-ish C by the NH-L9i, with the highest any review got being 92C during a stress test http://www.funkykit.com/wp-content/uploads/2016/08/NF9i_res1.jpg
3. I believe throttling for Zen kicks in at 75C (AMD claims 95C but the offset makes it confusing). Therefore the lower clocked Ryzen chips might do completely fine with that heatsink as they are significantly cooler than the 1800X, which is cooler than the 7700k. Going by those calculations of the 1700X being 20% cooler than the 7700k (IRL probably more if counted over ambient) all things equal and extrapolating those results, even the 1700X would max out at ~73C on that cooler. The 1600 and 1700, for instance, run significantly cooler than the 1700X so a ballpark estimation is they'd be completely safe. 1700X could be pushing it under sustained load if you factor in higher ambient temps, inferior airflow, unlucky silicon etc.

It's just my speculation at this point as I'm not sure if anyone had a reason to test Ryzen with low profile coolers yet apart from the Wraith Stealth, which is taller.

I just found this, however, which is I guess the best thing for estimating cooler clearance for the Biostar ITX board at the moment:
http://www.biostar.com.tw/app/en/mb/introduction.php?S_ID=878#specification
"Cooler Installation Notice :
AMD Wraith Stealth CPU cooler only compatible with DDR4 modules without heatsink(The depth of memory within 45-70mm can be installed, but it may touch the cooler.)
AMD Wraith Spire CPU cooler compatible with most DDR4 modules with heastink(86mm)
AMD Wraith Max CPU cooler only compatible with DDR4 modules without heatsink(The depth of memory within 45-70mm can be installed, but it may touch the cooler.)"

No info regarding messing with the VRMs, but the L9a is just 5mm wider than the Spire and smaller in all other dimensions so it should be ok I guess? (Biostar suggests that depth is the problem, as width of all 3 AMD coolers is the same).
http://noctua.at/en/nh-l9a/specification
http://cdn.wccftech.com/wp-content/uploads/2017/02/AMD-Ryzen-1800x-1700x-1700-boxes-coolers.jpg (wrong height, does not include fan)

 

[Thehack]

Yeah, sure thing, I understand that. My logic was that:
1. When compared directly using the same cooler, the highest clocked Ryzens are cooler than the 7700k, with the 1800X hitting 65C against 7700K's 74C, and 1700X being some 20% cooler here: http://www.tomshardware.com/reviews/amd-ryzen-7-1700x-review,4987-8.html
2. The stock 7700k was said to be kept at 70-80-ish C by the NH-L9i, with the highest any review got being 92C during a stress test http://www.funkykit.com/wp-content/uploads/2016/08/NF9i_res1.jpg
3. I believe throttling for Zen kicks in at 75C (AMD claims 95C but the offset makes it confusing). Therefore the lower clocked Ryzen chips might do completely fine with that heatsink as they are significantly cooler than the 1800X, which is cooler than the 7700k. Going by those calculations of the 1700X being 20% cooler than the 7700k all things equal, even the 1700X would max out at 73.6C going by the max temp the 7700k achieved during a stress test.

It's just speculation at this point as I'm not sure if anyone had a reason to test Ryzen with low profile coolers yet apart from the Wraith Stealth, which is taller.

I just found this, however, which is I guess the best thing for estimating cooler clearance for the Biostar ITX board at the moment:
http://www.biostar.com.tw/app/en/mb/introduction.php?S_ID=878#specification
"Cooler Installation Notice :
AMD Wraith Stealth CPU cooler only compatible with DDR4 modules without heatsink(The depth of memory within 45-70mm can be installed, but it may touch the cooler.)
AMD Wraith Spire CPU cooler compatible with most DDR4 modules with heastink(86mm)
AMD Wraith Max CPU cooler only compatible with DDR4 modules without heatsink(The depth of memory within 45-70mm can be installed, but it may touch the cooler.)"

No info regarding messing with the VRMs, but the L9a is just 5mm wider than the Spire and smaller in all other dimensions so it should be ok I guess? (Biostar suggests that depth is the problem, as width of all 3 AMD coolers is the same).
http://noctua.at/en/nh-l9a/specification
http://cdn.wccftech.com/wp-content/uploads/2017/02/AMD-Ryzen-1800x-1700x-1700-boxes-coolers.jpg (wrong height, does not include fan)

What matters is distance to TjMAX. Not temperature of the chip itself.

I think the best way to measure performance when comparing these two platforms is % to TJMAX:

(Current Temp - Ambient) / (TjMAX - Ambient).

Ryzen = ( 65 - 25) / (75 - 25) = 80%
Intel = (75 - 25 ) / (100-25) = 67%

 

[Dak]

I have completed my build using this motherboard

Here are a couple pics:
OC: http://i.imgur.com/DZgT78f.png

 

[lhl]

The cooling is obviously more efficient, it is hitting 95C (throttling temp) only after a full hour, and its only two radiators involved! This looks acceptable indeed, can Linux now without worrying of turning the DB4 into a stove. Now if only I could slightly undervolt the thing in UEFI BIOS, it would be a perfect fanless PC.

Hmm, too bad the BIOS doesn't allow undervolting, that'd lower temps significantly. For Linux, one thing that you can do to lower temps for day-to-day use is to switch from the default ondemand/schedutil governor to powersave. Phoronix did some testing early on that showed a not-too-bad performance hit for a bunch of use-cases with a significant power savings. When I run my system w/ powersave, it pretty much parks the cores at 1.55GHz w/ no noticeable perf hit when I'm just browsing/typing/etc. Switching governors back to performance or schedutil is a simple cpupower command in the terminal. Not as nice as having Ryzen Master/software OC options, but fingers crossed that'll come in a future kernel update. (Honestly, I'd just like if AMD would publish the BIOS and Kernel Developer's Guide for Ryzen so that CPU lm_sensors support can be added.)

BTW, for motherboard sensors, the MSI boards use a Nuvoton chip and can be supported w/ the nct6775 module. Gigabyte boards seem to use it87 modules.

 

[Thehack]

Hopefully Biostar can provide it via an update. I too would like so undervolt.

 

[Zgembo]

Hmm, too bad the BIOS doesn't allow undervolting, that'd lower temps significantly. For Linux, one thing that you can do to lower temps for day-to-day use is to switch from the default ondemand/schedutil governor to powersave. Phoronix did some testing early on that showed a not-too-bad performance hit for a bunch of use-cases with a significant power savings. When I run my system w/ powersave, it pretty much parks the cores at 1.55GHz w/ no noticeable perf hit when I'm just browsing/typing/etc. Switching governors back to performance or schedutil is a simple cpupower command in the terminal. Not as nice as having Ryzen Master/software OC options, but fingers crossed that'll come in a future kernel update. (Honestly, I'd just like if AMD would publish the BIOS and Kernel Developer's Guide for Ryzen so that CPU lm_sensors support can be added.)

BTW, for motherboard sensors, the MSI boards use a Nuvoton chip and can be supported w/ the nct6775 module. Gigabyte boards seem to use it87 modules.

Glad to hear there are some fellow Phoronix followers in here Yes, messing up manually with CPU power states is always a nice option to have. According to AIDA64, the sensor @ B350GTN is ITE IT8613E (ISA A30h), and lm-sensors is mentioning it @ https://github.com/groeck/it87/issues/22, commit https://github.com/groeck/lm-sensors/commit/8aa117ffd7256351fb6d661a1605ac84db71f07d.

Most importantly, kernel support for the chip itself is brewing at https://github.com/groeck/it87

 

[Zgembo]

Just confirmed, it is the ITE IT8613E. This worked, after forcing acpi_enforce_resources=lax in /etc/default/grub:

root@B350GTN:~# modprobe it87 force_id=0x8623
root@B350GTN:~# sensors
it8603-isa-0a30
Adapter: ISA adapter
in0:          +0.85 V  (min =  +0.00 V, max =  +3.06 V)
in1:          +1.30 V  (min =  +0.00 V, max =  +3.06 V)
in2:          +2.23 V  (min =  +0.00 V, max =  +3.06 V)
in3:          +2.23 V  (min =  +0.95 V, max =  +2.94 V)
in4:          +2.26 V  (min =  +0.00 V, max =  +3.06 V)
3VSB:         +3.58 V  (min =  +0.00 V, max =  +6.12 V)
Vbat:         +3.43 V
+3.3V:        +3.55 V
fan2:           0 RPM  (min =   14 RPM)  ALARM
fan3:           0 RPM  (min =   12 RPM)  ALARM
temp1:        +53.0°C  (low  = +88.0°C, high = +93.0°C)  ALARM  sensor = thermistor
temp2:        +48.0°C  (low  = +127.0°C, high = +127.0°C)  sensor = thermistor
temp3:        +38.0°C  (low  = +127.0°C, high = +127.0°C)  sensor = thermal diode
intrusion0:  ALARM

acpitz-virtual-0
Adapter: Virtual device
temp1:        +38.0°C  (crit = +127.0°C)

in0 matches CPU core voltage pretty well
temp1 matches AIDA64 "Aux" reading
temp2 matches chipset temp
temp3 is the CPU thermal diode (exposed by ACPI as well)

 

[dbjungle]

Just got all of my hardware except my CPU cooler AM4 bracket. I had a few issues last night, but all except 1 thing is sorted out now.

At first my fans were blowing full speed the entire time. I'm using a Y cable on the CPU header and a powered PWM splitter on the SYS header. In order to fix this I had to attach just 1 fan to each header and calibrate the fans in the BIOS. The manual fan control is different than what I'm used to so I will tinker with that again very soon.

My CPU was running very hot. Idling at 50-65 degrees. I'm assuming I didn't tighten the Wraith Spire down enough and it didn't make contact. I re-tightened all of the screws then did the fan calibration and this seems to have resolved it.

The final thing is that my board POSTs very slow. It takes around 20 seconds, then the OS boots in about 2 seconds. I haven't been able to dive into this yet. I think it's related to the enter setup delay and possibly the fact that I'm using a port on my GPU that I don't normally use because it's not at my main desk yet.

Haven't got a chance to test performance yet, but I will soon. I will be overclocking when I get the AM4 bracket for my cooler. I'm a little disappointed in the options. I was hoping to set a base of 3.4-3.6 and a boost of 3.8 or whatever my max is. It seems this isn't possible in the BIOS because there is only 1 state. I'll dig deeper later. Hope this helps someone!

EDIT: Forgot RAM! I'm on G.Skill Trident Z 3200 16GBx2 (32GB total). It recognizes XMP profiles, but won't boot over 2400 atleast not with XMP. Maybe I can get it higher with manual overclocking.

 

[Zgembo]

. . .

I'm a little disappointed in the options. I was hoping to set a base of 3.4-3.6 and a boost of 3.8 or whatever my max is. It seems this isn't possible in the BIOS because there is only 1 state. I'll dig deeper later. Hope this helps someone!

EDIT: Forgot RAM! I'm on G.Skill Trident Z 3200 16GBx2 (32GB total). It recognizes XMP profiles, but won't boot over 2400 atleast not with XMP. Maybe I can get it higher with manual overclocking.

This seems identical as B350GTN, both BIOS-options-wise and RAM-clocks-wise. Can you please check if there are any under-volting options in the BIOS?

 

[BongoUnicorns]

X370GTN in my hand rn. I'm just a teeny bit excited for my Dr. Zaber Sentry...

 

[dbjungle]

This seems identical as B350GTN, both BIOS-options-wise and RAM-clocks-wise. Can you please check if there are any under-volting options in the BIOS?

There is undervolting in hex under p state control. I think someone else mentioned the lowest setting is 1.2 or something. I'm not at home to see ATM.

 

[Jin]

X370GTN in my hand rn. I'm just a teeny bit excited for my Dr. Zaber Sentry...

waiting on my AB350M ITX =[

 

[mrchowderclam]

Just completed my build with the biostar x370 board and a Ryzen 1700 (my first SFF build! I'm using this as a video encoding machine for my twitch channel)

Everything is in the Ncase M1 and I'm using a Noctua NH-L12. The cooler fits no problem with the heat pipes facing toward the I/O ports. On stock clocks I'm seeing about 29C on idle and 39C under load while encoding video.


As an aside, Ryzen is insane for video encoding. It doesn't break a sweat encoding 900p60 video @ x264 slow preset and 1080p60 @ medium preset. I bet with some overclocking I could get it to encode at 1080p60 slow.

edit: better pics

 

[dbjungle]

Just completed my build with the biostar x370 board and a Ryzen 1700 (my first SFF build! I'm using this as a video encoding machine for my twitch channel)

Everything is in the Ncase M1 and I'm using a Noctua NH-L12. The cooler fits no problem with the heat pipes facing toward the I/O ports. On stock clocks I'm seeing about 29C on idle and 39C under load while encoding video.


As an aside, Ryzen is insane for video encoding. It doesn't break a sweat encoding 900p60 video @ x264 slow preset and 1080p60 @ medium preset. I bet with some overclocking I could get it to encode at 1080p60 slow

Does your build take a long time to POST? Mine is taking around 20 seconds.