Overclock.net › Forums › Benchmarks › Benchmarking Software and Discussion › Need ideas on how to benchmark PSU (please read before dismissing as weird idea).
New Posts  All Forums:Forum Nav:

Need ideas on how to benchmark PSU (please read before dismissing as weird idea).

post #1 of 7
Thread Starter 
I assembled my new PC with a old "Pentium 4 compatible" 500W PSU.

It has 2 12V rails, only 4 pins on the 12V connector, and the 12V rails summed can only output about 350W (or something like that, I forgot the precise calculation I did).

For most part the machine works fine, I even game with it, and got some awesome XTU scores, and good GPU scores too on local tests.

But when I run intentionally power-intensive tasks, for example OCCT PSU test, the PSU limits does show up, with the machine throttling to compensate VDROP caused by the PSU.

I ordered a RM1000x

I wanted to know how to make some itneresting benchmarks, games included, to see the effects of changing the PSU, in a way that is pretty and organized enough for me to post the data online for anyone curious.

Maybe even to dispell, or confirm, the "PSU doesn't increase FPS" myth.

So, ideas? What softwares I should use to bench and track the data?

And what games I should use, and how I track their data, and make them easy to repeat betwee ndifferent PSUs?

Also OC settings ideas, to see how much PSU affect OC stability and performance.


The system where I will do this:

Motherboard: ASRock z97 Extreme4 (it has a 8-pin 12V connector, and a molex connector, with the current PSU I only plugged a 4-pin cable on the 8-pin connector).
RAM: 32GB DDR3-2400 (4 sticks of 8gb).
CPU: 4690K
GPU: Sapphire Nitro 380X
storage: 3 HDD WD Black

air cooled, open bench. Highest temperature recorded with current cooler was 56C during a XTU run with processor OCed to 4530mhz (accidental, my setting on XTU was multiplier 44x, and cache 42x... I dunno why the CPU ran at 45).
post #2 of 7
Are you 100% sure the power supply is causing the vdrop? Is the 12 volt falling or the cpu voltage?

Since cpus run on a much lower voltage than the power supply puts out, the regulators deal with that. Even at 11 volts technically a regulator should have no issues(but by the time most power supplies get that low they have too much voltage ripple. Only so much the onboard regulators can clean up). The efficiency may drop and cause some temperature increase on the regulation system it self.

You can not actually see voltage ripple with software or even multi-meters you need something like an oscilloscope to actually see that.

Best I can see you doing is just having the system at the fastest stable overclock you can and running a benchmark x number of times and either taking an average or the middle result.

If the results are noticeably lower with one power supply, you have your result. You will want to use the exact same testing so if it is start the system and leave it idle for x number of minutes before running the test. All power supplies should get the same idle time.

You are pushing that little power supply pretty hard most likely.
Gaming
(15 items)
 
Media Center
(17 items)
 
 
CPUMotherboardGraphicsRAM
2600K @ 4.4 Maximus IV Gene-z Asus Dual 1070 OC with Artic Mono Plus G.SKILL F3-1866C9D-16GXM Ripjaws X 16GB 
Hard DriveHard DriveOptical DriveCooling
M4 256 MX 300 Liteon Slot load H80i 
OSMonitorKeyboardPower
Windows 7 HP 64 bit BenQ XL2420Z Logitech Wave Corsair 850HX 
CaseMouseAudio
Silverstone FT03 Logitech G9 X-fi Titanium Fatality 
CPUMotherboardGraphicsRAM
i5 750 h55n usb3 msi gtx 650 ti Corsair XMS3 CMX8GX3M2A1600C9 
Hard DriveHard DriveHard DriveOptical Drive
WDC Red 3tb M4 128 WDC scorpio Blue 2.5 Sony AD-7690H-01 
CoolingOSMonitorKeyboard
Geminii S Windows 7 HP 64 bit SyncMaster 245t Logitech Wave 
PowerCaseMouseAudio
FSP SFX 300 watt SG05 Logitech G9 Sound Blaster Omni 
Other
HD PVR 
  hide details  
Reply
Gaming
(15 items)
 
Media Center
(17 items)
 
 
CPUMotherboardGraphicsRAM
2600K @ 4.4 Maximus IV Gene-z Asus Dual 1070 OC with Artic Mono Plus G.SKILL F3-1866C9D-16GXM Ripjaws X 16GB 
Hard DriveHard DriveOptical DriveCooling
M4 256 MX 300 Liteon Slot load H80i 
OSMonitorKeyboardPower
Windows 7 HP 64 bit BenQ XL2420Z Logitech Wave Corsair 850HX 
CaseMouseAudio
Silverstone FT03 Logitech G9 X-fi Titanium Fatality 
CPUMotherboardGraphicsRAM
i5 750 h55n usb3 msi gtx 650 ti Corsair XMS3 CMX8GX3M2A1600C9 
Hard DriveHard DriveHard DriveOptical Drive
WDC Red 3tb M4 128 WDC scorpio Blue 2.5 Sony AD-7690H-01 
CoolingOSMonitorKeyboard
Geminii S Windows 7 HP 64 bit SyncMaster 245t Logitech Wave 
PowerCaseMouseAudio
FSP SFX 300 watt SG05 Logitech G9 Sound Blaster Omni 
Other
HD PVR 
  hide details  
Reply
post #3 of 7
Thread Starter 
Quote:
Originally Posted by Nukemaster View Post

Are you 100% sure the power supply is causing the vdrop? Is the 12 volt falling or the cpu voltage?

FPS on OCCT PSU test drops when the PSU 12V rails start (According to the mobo VRM and the GPU VRM) to output less than 11.1V

The lowest I saw was 10.6V
post #4 of 7
Quote:
Originally Posted by OrSpeeder View Post

I assembled my new PC with a old "Pentium 4 compatible" 500W PSU.

It has 2 12V rails, only 4 pins on the 12V connector, and the 12V rails summed can only output about 350W (or something like that, I forgot the precise calculation I did).

For most part the machine works fine, I even game with it, and got some awesome XTU scores, and good GPU scores too on local tests.

But when I run intentionally power-intensive tasks, for example OCCT PSU test, the PSU limits does show up, with the machine throttling to compensate VDROP caused by the PSU.

I ordered a RM1000x

I wanted to know how to make some itneresting benchmarks, games included, to see the effects of changing the PSU, in a way that is pretty and organized enough for me to post the data online for anyone curious.

Maybe even to dispell, or confirm, the "PSU doesn't increase FPS" myth.

So, ideas? What softwares I should use to bench and track the data?

And what games I should use, and how I track their data, and make them easy to repeat betwee ndifferent PSUs?

Also OC settings ideas, to see how much PSU affect OC stability and performance.


The system where I will do this:

Motherboard: ASRock z97 Extreme4 (it has a 8-pin 12V connector, and a molex connector, with the current PSU I only plugged a 4-pin cable on the 8-pin connector).
RAM: 32GB DDR3-2400 (4 sticks of 8gb).
CPU: 4690K
GPU: Sapphire Nitro 380X
storage: 3 HDD WD Black

air cooled, open bench. Highest temperature recorded with current cooler was 56C during a XTU run with processor OCed to 4530mhz (accidental, my setting on XTU was multiplier 44x, and cache 42x... I dunno why the CPU ran at 45).

A PSU can only limit your other hardware to a lower FPS by not supplying quality power. That one line made me not want to read the rest of the post no matter what your thread title says.
post #5 of 7
Thread Starter 
Quote:
Originally Posted by mmonnin View Post

A PSU can only limit your other hardware to a lower FPS by not supplying quality power. That one line made me not want to read the rest of the post no matter what your thread title says.

Then I don't understand what is the problem.

I will compare a 500W Pentium4 era PSU that doesn't have 8pin connector, to a extremely high quality PSU.

So, if my PSU maybe supplies not "quality power" as you say, and lower FPS, wouldn't you not like data on that?
post #6 of 7
Quote:
Originally Posted by OrSpeeder View Post

FPS on OCCT PSU test drops when the PSU 12V rails start (According to the mobo VRM and the GPU VRM) to output less than 11.1V

The lowest I saw was 10.6V
10.6 is very low. That poor power supply.

Also remember older systems did not load the 12 volt rail as heavily so they tend to have less current on that rail. In general most current units make one huge 12 volt rail and get the lower 5 and 3.3 volt rails off of that 12 volt rail with dc-dc conversion. This allows lots of 12 volt or decent amounts on the other rail when the 12 volt load is not too high(and anything in between).
Gaming
(15 items)
 
Media Center
(17 items)
 
 
CPUMotherboardGraphicsRAM
2600K @ 4.4 Maximus IV Gene-z Asus Dual 1070 OC with Artic Mono Plus G.SKILL F3-1866C9D-16GXM Ripjaws X 16GB 
Hard DriveHard DriveOptical DriveCooling
M4 256 MX 300 Liteon Slot load H80i 
OSMonitorKeyboardPower
Windows 7 HP 64 bit BenQ XL2420Z Logitech Wave Corsair 850HX 
CaseMouseAudio
Silverstone FT03 Logitech G9 X-fi Titanium Fatality 
CPUMotherboardGraphicsRAM
i5 750 h55n usb3 msi gtx 650 ti Corsair XMS3 CMX8GX3M2A1600C9 
Hard DriveHard DriveHard DriveOptical Drive
WDC Red 3tb M4 128 WDC scorpio Blue 2.5 Sony AD-7690H-01 
CoolingOSMonitorKeyboard
Geminii S Windows 7 HP 64 bit SyncMaster 245t Logitech Wave 
PowerCaseMouseAudio
FSP SFX 300 watt SG05 Logitech G9 Sound Blaster Omni 
Other
HD PVR 
  hide details  
Reply
Gaming
(15 items)
 
Media Center
(17 items)
 
 
CPUMotherboardGraphicsRAM
2600K @ 4.4 Maximus IV Gene-z Asus Dual 1070 OC with Artic Mono Plus G.SKILL F3-1866C9D-16GXM Ripjaws X 16GB 
Hard DriveHard DriveOptical DriveCooling
M4 256 MX 300 Liteon Slot load H80i 
OSMonitorKeyboardPower
Windows 7 HP 64 bit BenQ XL2420Z Logitech Wave Corsair 850HX 
CaseMouseAudio
Silverstone FT03 Logitech G9 X-fi Titanium Fatality 
CPUMotherboardGraphicsRAM
i5 750 h55n usb3 msi gtx 650 ti Corsair XMS3 CMX8GX3M2A1600C9 
Hard DriveHard DriveHard DriveOptical Drive
WDC Red 3tb M4 128 WDC scorpio Blue 2.5 Sony AD-7690H-01 
CoolingOSMonitorKeyboard
Geminii S Windows 7 HP 64 bit SyncMaster 245t Logitech Wave 
PowerCaseMouseAudio
FSP SFX 300 watt SG05 Logitech G9 Sound Blaster Omni 
Other
HD PVR 
  hide details  
Reply
post #7 of 7
Thread Starter 
PSU arrived, noone gave a decent reply yet.

Can someone at least suggest a microstutter recording software?
New Posts  All Forums:Forum Nav:
  Return Home
Overclock.net › Forums › Benchmarks › Benchmarking Software and Discussion › Need ideas on how to benchmark PSU (please read before dismissing as weird idea).