Overclock.net › Forums › AMD › AMD - General › VRM on the new AM4 motherboards
New Posts  All Forums:Forum Nav:

VRM on the new AM4 motherboards - Page 148

post #1471 of 2005
Quote:
Originally Posted by realtomatoes View Post

ah, so this is the 5050 fun zone. lol
looks like the X370GTN has it too.


folks, any thoughts about the vrm on the X370GTN?

4+3 phase.
It does have what looks like a Vishay trenchfet.

Edit: Too much hope...

The SoC has some nikos, maybe PZ0903BK.



Controller is ILS 95712


Something like SM4377or SM4337 on the memory vrm,



R7 1700@3.5Ghz Vcore 1.186V
Cryorig C7
Mem 2400Mhz
OCCT for 30 minutes
Unknown ambient...




Edited by virpz - 5/3/17 at 11:12pm
post #1472 of 2005
Quote:
Originally Posted by superstition222 View Post

Watercooling is a gimmick. If you can't cool it with a Noctua D15 then you're in gimmick territory.

Overlocking is a gimmick. If you can't get the performance you want at stock then you bought the wrong part.

It's a gimmick to not want to mess around with jury-rigging fans with hanging zip ties or some other cumbersome solution, or trying to get fans wedged into tiny spaces around big tower coolers. It's a gimmick to not want to dust those fans and deal with the extra noise and airflow complexity..

That's not the point and you know it. Water cooling something spitting out 150w+ makes sense. Water cooling something that can otherwise be designed to not run hot enough to need active cooling at all does not, that is why I call it a gimmick. Most of the overkill boards can cool themselves adequately with convection flow at this point, unless you are using very specific chips that are not on either platform being discussed here.

If you rely on "WC compatible" parts, you are doing it wrong.
If you have to jury rig fans in your case to compensate for your horrid motherboard decision that is YOUR problem.
If you are water cooling to get around "airflow complexity"... I can't even. Air is pretty simple, and easy to keep clean with less planning than water cooling.

Water cooling as an excuse for silence is pretty played out at this point as well. With large enough heatsinks designed for the purpose you can have a PC that is almost 100% silent, with a bit of a CPU OC to boot (had a Z77/i5 2500k set up like this for a long time). There are high end GPUs with fans that stop when not under load to go with this kind of system too.
Edited by KarathKasun - 5/4/17 at 12:08am
μRyzen
(12 items)
 
Mini Box
(4 items)
 
 
CPUMotherboardGraphicsRAM
Ryzen R5 1400 MSI B350M Gaming Pro Zotac GTX 670 4GB G.SKILL FORTIS Series 8GB (2 x 4GB) 
Hard DriveCoolingOSOS
WD Green 3tb Wraith Stealth Windows 10 Debian 8.7 
MonitorKeyboardPowerMouse
ViewSonic VX-2257-8 Chinese backlit mechanical Kingwin 850w Chinese laser optical 
CPUMotherboardGraphicsRAM
Athlon 5350 Asus AM1I-A EVGA GTX 750 Ti SC 2x4GB DDR 3 1333 
  hide details  
Reply
μRyzen
(12 items)
 
Mini Box
(4 items)
 
 
CPUMotherboardGraphicsRAM
Ryzen R5 1400 MSI B350M Gaming Pro Zotac GTX 670 4GB G.SKILL FORTIS Series 8GB (2 x 4GB) 
Hard DriveCoolingOSOS
WD Green 3tb Wraith Stealth Windows 10 Debian 8.7 
MonitorKeyboardPowerMouse
ViewSonic VX-2257-8 Chinese backlit mechanical Kingwin 850w Chinese laser optical 
CPUMotherboardGraphicsRAM
Athlon 5350 Asus AM1I-A EVGA GTX 750 Ti SC 2x4GB DDR 3 1333 
  hide details  
Reply
post #1473 of 2005
Quote:
Originally Posted by cssorkinman View Post

I grabbed the Raytek Ranger ST and took some temps during an IBT AVX run at 4050 mhz at 1.424 volts +/-
HWINFO is pretty darn close on its VRM temps from what I see. 20 C ambient.
Warning: Spoiler! (Click to show)

Choke temps 37 was the hottest where it wasn't getting direct ariflow the ones under the fan were much cooler.

Hottest spot on the back side of motherboard I could find ( that I could get to ). ( no airflow on it) 48 C

Top of SOC VRM heatsink ( black area)

Top of cpu vrm heatsink ( black area )

Ambient temp

Emissivity setting @ 95 - within 1C of ambient probe temp while scanning matte black surface of P5

@bardacuda I made a video, but my upload is 768k.... it would finish uploading about the time AM5 was released.tongue.gif

This is nice, but IBT would be a bad stress test to use for an apples to apples comparison. If you don't use exactly the right amount of RAM you can end up stressing the CPU different amounts, like you see below where it was only putting out ~155 GFlops. Since you don't have a screenshot of IBT in the above post at all it's impossible to tell what kind of stress the CPU was under. Again, I'm not suggesting you're fudging anything, but you could be, so it doesn't settle anything.
Quote:
Originally Posted by cssorkinman View Post

Warning: Spoiler! (Click to show)
Removed the fan that was blowing on the soc vrm and the upper half of the cpu vrm and had a go at ibt very high.


Prime would have the same issues if one was to use blend or some variation on it, since capturing readings at different FFT sizes would still be apples and oranges. I think a good test would be something that is close-ish to worst case that Joe noob OCer might encounter. Maybe OCCT with default settings (should be a constant, reproducable, yet high load) after 1/2 hour or more, inside a case, with no fan on the VRM, @ 1.4+ VCore. Frequency at 3850 or less since I probably can't go any higher.

It shouldn't be unreasonable to expect MSI's flagship to perform well in that kind of scenario...and given its price point, putting it up against the Prime Pro is going easy on it. The argument that I have been seeing is that the VRM is good enough for Ryzen...and any shortcomings it might have in mosfet quality is made up for by exceptional design for heat dissipation...so let's put that to the test.

If you have slow upload speed, or don't have a case to put your system in or whatever, I understand and I'm not asking you to go out of the way to do it, or even asking you specifically to do anything.
I'm just throwing it out there that if anyone does have a Titanium, and wants to do a test using the conditions I mentioned in order to put an end to the arguing with some hard data, then I'll be glad to do the same test for comparison purposes.
It has to be something that is hard to fudge though, be apples to apples, and show that the board can do what the price point suggests it should be able to do. Otherwise, it would be a waste of time.
post #1474 of 2005
Quote:
Originally Posted by bardacuda View Post

This is nice, but IBT would be a bad stress test to use for an apples to apples comparison. If you don't use exactly the right amount of RAM you can end up stressing the CPU different amounts, like you see below where it was only putting out ~155 GFlops. Since you don't have a screenshot of IBT in the above post at all it's impossible to tell what kind of stress the CPU was under. Again, I'm not suggesting you're fudging anything, but you could be, so it doesn't settle anything.
Prime would have the same issues if one was to use blend or some variation on it, since capturing readings at different FFT sizes would still be apples and oranges. I think a good test would be something that is close-ish to worst case that Joe noob OCer might encounter. Maybe OCCT with default settings (should be a constant, reproducable, yet high load) after 1/2 hour or more, inside a case, with no fan on the VRM, @ 1.4+ VCore. Frequency at 3850 or less since I probably can't go any higher.

It shouldn't be unreasonable to expect MSI's flagship to perform well in that kind of scenario...and given its price point, putting it up against the Prime Pro is going easy on it. The argument that I have been seeing is that the VRM is good enough for Ryzen...and any shortcomings it might have in mosfet quality is made up for by exceptional design for heat dissipation...so let's put that to the test.

If you have slow upload speed, or don't have a case to put your system in or whatever, I understand and I'm not asking you to go out of the way to do it, or even asking you specifically to do anything.
I'm just throwing it out there that if anyone does have a Titanium, and wants to do a test using the conditions I mentioned in order to put an end to the arguing with some hard data, then I'll be glad to do the same test for comparison purposes.
It has to be something that is hard to fudge though, be apples to apples, and show that the board can do what the price point suggests it should be able to do. Otherwise, it would be a waste of time.

What I'm seeing here is that the vrm temperatures on the titanium are quite lower than on the prime ibt tests you posted, despite the processor using more power on the titanium tests.

Still, I didn't expect that 10c difference in fet temperatures under the circumstances. The TI packages should be able for more. I'm afraid that it's the coils dragging the vrm down on the prime (just look at their temps, ouch). If there's an option on the prime to decrease the vrm switching frequency, it might help quite a bit.
post #1475 of 2005
Quote:
Originally Posted by bardacuda View Post

Quote:
Originally Posted by cssorkinman View Post

I grabbed the Raytek Ranger ST and took some temps during an IBT AVX run at 4050 mhz at 1.424 volts +/-
HWINFO is pretty darn close on its VRM temps from what I see. 20 C ambient.
Warning: Spoiler! (Click to show)

Choke temps 37 was the hottest where it wasn't getting direct ariflow the ones under the fan were much cooler.

Hottest spot on the back side of motherboard I could find ( that I could get to ). ( no airflow on it) 48 C

Top of SOC VRM heatsink ( black area)

Top of cpu vrm heatsink ( black area )

Ambient temp

Emissivity setting @ 95 - within 1C of ambient probe temp while scanning matte black surface of P5

@bardacuda I made a video, but my upload is 768k.... it would finish uploading about the time AM5 was released.tongue.gif

This is nice, but IBT would be a bad stress test to use for an apples to apples comparison. If you don't use exactly the right amount of RAM you can end up stressing the CPU different amounts, like you see below where it was only putting out ~155 GFlops. Since you don't have a screenshot of IBT in the above post at all it's impossible to tell what kind of stress the CPU was under. Again, I'm not suggesting you're fudging anything, but you could be, so it doesn't settle anything.
Quote:
Originally Posted by cssorkinman View Post

Warning: Spoiler! (Click to show)
Removed the fan that was blowing on the soc vrm and the upper half of the cpu vrm and had a go at ibt very high.


Prime would have the same issues if one was to use blend or some variation on it, since capturing readings at different FFT sizes would still be apples and oranges. I think a good test would be something that is close-ish to worst case that Joe noob OCer might encounter. Maybe OCCT with default settings (should be a constant, reproducable, yet high load) after 1/2 hour or more, inside a case, with no fan on the VRM, @ 1.4+ VCore. Frequency at 3850 or less since I probably can't go any higher.

It shouldn't be unreasonable to expect MSI's flagship to perform well in that kind of scenario...and given its price point, putting it up against the Prime Pro is going easy on it. The argument that I have been seeing is that the VRM is good enough for Ryzen...and any shortcomings it might have in mosfet quality is made up for by exceptional design for heat dissipation...so let's put that to the test.

If you have slow upload speed, or don't have a case to put your system in or whatever, I understand and I'm not asking you to go out of the way to do it, or even asking you specifically to do anything.
I'm just throwing it out there that if anyone does have a Titanium, and wants to do a test using the conditions I mentioned in order to put an end to the arguing with some hard data, then I'll be glad to do the same test for comparison purposes.
It has to be something that is hard to fudge though, be apples to apples, and show that the board can do what the price point suggests it should be able to do. Otherwise, it would be a waste of time.

A purely scientific apples to apples comparison would be impossible due to differences in cpu, psu, cpu cooling etc.

I was trying to get a handle on HWINFO's VRM temperature reporting accuracy regarding the Titanium ( VR -1 VR -2 ) - from what I was able to tell it's very close.
Ryzen Shine!
(14 items)
 
  
CPUMotherboardGraphicsRAM
Ryzen 1800X @ 4125mhz 1.432 Volts MSI X370 Titanium   Fury  G.SKILL TridentZ Series 32GB (4 x 8GB) 288-Pin... 
Hard DriveOptical DriveCoolingOS
Samsung 850 pro Hp 1260 I liquid unobtanium Koolance 480mm radiator, 39... Win 7 HP/Winspy 10 64bit 
MonitorKeyboardPowerCase
24" hp @ 1900x1200 Logitech G19 PC power and cooling 910 watt silencer Thermaltake P5 
MouseMouse Pad
Logitech MX 518 fUnc industries 
  hide details  
Reply
Ryzen Shine!
(14 items)
 
  
CPUMotherboardGraphicsRAM
Ryzen 1800X @ 4125mhz 1.432 Volts MSI X370 Titanium   Fury  G.SKILL TridentZ Series 32GB (4 x 8GB) 288-Pin... 
Hard DriveOptical DriveCoolingOS
Samsung 850 pro Hp 1260 I liquid unobtanium Koolance 480mm radiator, 39... Win 7 HP/Winspy 10 64bit 
MonitorKeyboardPowerCase
24" hp @ 1900x1200 Logitech G19 PC power and cooling 910 watt silencer Thermaltake P5 
MouseMouse Pad
Logitech MX 518 fUnc industries 
  hide details  
Reply
post #1476 of 2005
Doesn't need to be purely scientific since that would be basically impossible to set up over a forum. Not to mention I don't have the training or equipment to do one.

Just in a case, no fan, 1.4v, with a consistent high load.

I think that would give a good idea. I agree CPU cooling can make a difference but the other two seem negligible. If anyone has a better test that could reasonably be expected to be set up then I'm all ears.

Maybe later I'll just go ahead and make a video to get the ball rolling. Anyone that wants to try and reproduce the test as best they can with a different motherboard can share their results...doesn't have to be a Titanium. Maybe in that way we'll start getting some way to compare boards, instead just slinging around those same few FLIR images along with some poo and words.
post #1477 of 2005
Quote:
Originally Posted by bardacuda View Post

Doesn't need to be purely scientific since that would be basically impossible to set up over a forum. Not to mention I don't have the training or equipment to do one.

Just in a case, no fan, 1.4v, with a consistent high load.

I think that would give a good idea. I agree CPU cooling can make a difference but the other two seem negligible. If anyone has a better test that could reasonably be expected to be set up then I'm all ears.

Maybe later I'll just go ahead and make a video to get the ball rolling. Anyone that wants to try and reproduce the test as best they can with a different motherboard can share their results...doesn't have to be a Titanium. Maybe in that way we'll start getting some way to compare boards, instead just slinging around those same few FLIR images along with some poo and words.

With no way of quantifying the amount of airflow over the VRM heatsink I think we should compare best temperatures possible at a given voltage/clockspeed on a standardized load?

If you'd pop the sides off the case and point a house fan at both sides that would be about the same as what I'm running.
Ryzen Shine!
(14 items)
 
  
CPUMotherboardGraphicsRAM
Ryzen 1800X @ 4125mhz 1.432 Volts MSI X370 Titanium   Fury  G.SKILL TridentZ Series 32GB (4 x 8GB) 288-Pin... 
Hard DriveOptical DriveCoolingOS
Samsung 850 pro Hp 1260 I liquid unobtanium Koolance 480mm radiator, 39... Win 7 HP/Winspy 10 64bit 
MonitorKeyboardPowerCase
24" hp @ 1900x1200 Logitech G19 PC power and cooling 910 watt silencer Thermaltake P5 
MouseMouse Pad
Logitech MX 518 fUnc industries 
  hide details  
Reply
Ryzen Shine!
(14 items)
 
  
CPUMotherboardGraphicsRAM
Ryzen 1800X @ 4125mhz 1.432 Volts MSI X370 Titanium   Fury  G.SKILL TridentZ Series 32GB (4 x 8GB) 288-Pin... 
Hard DriveOptical DriveCoolingOS
Samsung 850 pro Hp 1260 I liquid unobtanium Koolance 480mm radiator, 39... Win 7 HP/Winspy 10 64bit 
MonitorKeyboardPowerCase
24" hp @ 1900x1200 Logitech G19 PC power and cooling 910 watt silencer Thermaltake P5 
MouseMouse Pad
Logitech MX 518 fUnc industries 
  hide details  
Reply
post #1478 of 2005
Ok I'll do that in the video after I take the other readings. I'll just do the one side though because I don't have two floor fans.
post #1479 of 2005
As long as you guys verify with ir at some point on vrm is similar to software...no reason you can not use the software to record peak and average like i did...that solves prime 95 iteration issues.

We saw 68c on back of giga and 69-70 in software. Top of the fet i have no acess to is more than realistic to be 2c higher than my ir.

So the 73c avg and 81c peak is realistic imo with no fans.
Edited by chew* - 5/4/17 at 8:41am
post #1480 of 2005
I had to split the video in 2 for this upload site but here you go. The OCCT settings didn't show up in the first part of the video apparently, but you can catch them at about 20 or 25 seconds into the second part. Remind me to speak more clearly and not cover the microphone either if I ever do another one of these. It feels like I'm talking to myself while recording it tongue.gif

First 9.5 mins:

https://streamable.com/1fq70

Last 3 mins where I basically just reboot to show BIOS settings:

https://streamable.com/gyahq

Long story short here...@1.4V the VRM heatsink is ~37 - 38°C with a ~22°C ambient when the case is closed with no extra fans.
I'm not sure anymore what Temp 3 - 6 are in HWiNFO. It could be the VRM package, or it could be CPU socket or something else. ¯\_(ツ)_/¯
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: AMD - General
Overclock.net › Forums › AMD › AMD - General › VRM on the new AM4 motherboards