Overclock.net › Forums › Intel › Intel CPUs › How to run LinPack stress test(LinX/IBT)properly-an explanation(maybe a guide)
New Posts  All Forums:Forum Nav:

How to run LinPack stress test(LinX/IBT)properly-an explanation(maybe a guide) - Page 5

post #41 of 128
Quote:
Originally Posted by JimmyHart84 View Post
Hi PERSPOLIS, can you give me some advice if you don't mind. A few months ago when I built my new pc, I did an overclock for the first time and got it up to 4.11GHz. I've just flashed my bios today which reset all the settings, so I need to do the oc again, but I can remember what all the settings were. However, last time I just used prime95 but this time, I'd like to use linx to give me more peace of mind that I'm stable.

So most of your guide is excellent, and thanks for taking the time to create it. I understand most of it about not running apps, leaving 10 mins, checking free ram, etc, but I'm confused about one thing, what value of gigaflops I should be expecting to see in linx. I've got an i7 920 and 6GB of memory, would you be able to work this out for me, as I'm confused by the calculation to do it myself.
The i7 is a quad core, so you take your core speed and multiply it by 16.
4.11GHz x 16 = 65.76

So you should get something between 55-65 GFlops
Echo
(8 items)
 
   
CPUMotherboardGraphicsRAM
Intel Core i5-3570K Asus Maximus V Gene Asus Radeon HD 7970 3GB Mushkin Blackline 16GB 
Hard DriveHard DriveCaseOther
2x Samsung 830 256GB Seagate Barracuda 3TB Silverstone Temjin TJ08-E BeQuiet E9-700w 
CPUMotherboardRAMHard Drive
Intel i3-3225 ASUS P8B75-MLX Corsair Ballistix Tactical Western Digital Red 
OSPowerCaseOther
Ubuntu Corsair CX500M Norco RPC-4224 LSI SAS 9207-4i4e HBA 
Other
Chenbro CK22804 24-Port SAS Expander Card 
  hide details  
Reply
Echo
(8 items)
 
   
CPUMotherboardGraphicsRAM
Intel Core i5-3570K Asus Maximus V Gene Asus Radeon HD 7970 3GB Mushkin Blackline 16GB 
Hard DriveHard DriveCaseOther
2x Samsung 830 256GB Seagate Barracuda 3TB Silverstone Temjin TJ08-E BeQuiet E9-700w 
CPUMotherboardRAMHard Drive
Intel i3-3225 ASUS P8B75-MLX Corsair Ballistix Tactical Western Digital Red 
OSPowerCaseOther
Ubuntu Corsair CX500M Norco RPC-4224 LSI SAS 9207-4i4e HBA 
Other
Chenbro CK22804 24-Port SAS Expander Card 
  hide details  
Reply
post #42 of 128
Quote:
Originally Posted by [Adz] View Post
The i7 is a quad core, so you take your core speed and multiply it by 16.
4.11GHz x 16 = 65.76

So you should get something between 55-65 GFlops
Oh is that all it is? I thought the calculation was more complex than that and took about of ram into account too. Ah well, thanks!

One question though, how did you work out the 55 to 65 value?
Desktop
(16 items)
 
  
CPUMotherboardGraphicsRAM
Core i7 920 D0 ASUS P6T GeForce GTX 560 Ti Corsair XMS3 
Hard DriveOptical DriveOptical DriveCooling
WD Caviar Black LG DVD Writer Samsung Bluray Reader 4 case fans, PSU fan, 2 heatsink fans 
OSMonitorKeyboardPower
Windows 7 Pro 64-bit Samsung T220 22" TFT Logitech MX 5500 Coolermaster Silent Pro 700W Modular 
CaseMouseMouse PadAudio
Antec 300 Logitech MX Revolution Half Life 2 Auzentech X-Fi Forte 
  hide details  
Reply
Desktop
(16 items)
 
  
CPUMotherboardGraphicsRAM
Core i7 920 D0 ASUS P6T GeForce GTX 560 Ti Corsair XMS3 
Hard DriveOptical DriveOptical DriveCooling
WD Caviar Black LG DVD Writer Samsung Bluray Reader 4 case fans, PSU fan, 2 heatsink fans 
OSMonitorKeyboardPower
Windows 7 Pro 64-bit Samsung T220 22" TFT Logitech MX 5500 Coolermaster Silent Pro 700W Modular 
CaseMouseMouse PadAudio
Antec 300 Logitech MX Revolution Half Life 2 Auzentech X-Fi Forte 
  hide details  
Reply
post #43 of 128
Quote:
Originally Posted by JimmyHart84 View Post
Oh is that all it is? I thought the calculation was more complex than that and took about of ram into account too. Ah well, thanks!

One question though, how did you work out the 55 to 65 value?
LinX isn't able to sustain 100% usage constantly, which causes the speed to drop slightly. Also, there is the Operating System to account for as that will be using CPU cycles too, so it's just a rough estimate.
Echo
(8 items)
 
   
CPUMotherboardGraphicsRAM
Intel Core i5-3570K Asus Maximus V Gene Asus Radeon HD 7970 3GB Mushkin Blackline 16GB 
Hard DriveHard DriveCaseOther
2x Samsung 830 256GB Seagate Barracuda 3TB Silverstone Temjin TJ08-E BeQuiet E9-700w 
CPUMotherboardRAMHard Drive
Intel i3-3225 ASUS P8B75-MLX Corsair Ballistix Tactical Western Digital Red 
OSPowerCaseOther
Ubuntu Corsair CX500M Norco RPC-4224 LSI SAS 9207-4i4e HBA 
Other
Chenbro CK22804 24-Port SAS Expander Card 
  hide details  
Reply
Echo
(8 items)
 
   
CPUMotherboardGraphicsRAM
Intel Core i5-3570K Asus Maximus V Gene Asus Radeon HD 7970 3GB Mushkin Blackline 16GB 
Hard DriveHard DriveCaseOther
2x Samsung 830 256GB Seagate Barracuda 3TB Silverstone Temjin TJ08-E BeQuiet E9-700w 
CPUMotherboardRAMHard Drive
Intel i3-3225 ASUS P8B75-MLX Corsair Ballistix Tactical Western Digital Red 
OSPowerCaseOther
Ubuntu Corsair CX500M Norco RPC-4224 LSI SAS 9207-4i4e HBA 
Other
Chenbro CK22804 24-Port SAS Expander Card 
  hide details  
Reply
post #44 of 128
Quote:
Originally Posted by [Adz] View Post
LinX isn't able to sustain 100% usage constantly, which causes the speed to drop slightly. Also, there is the Operating System to account for as that will be using CPU cycles too, so it's just a rough estimate.
OK brill, thanks a lot! Once I'm done, I'll post a screenshot to see if you think it's alright.
Desktop
(16 items)
 
  
CPUMotherboardGraphicsRAM
Core i7 920 D0 ASUS P6T GeForce GTX 560 Ti Corsair XMS3 
Hard DriveOptical DriveOptical DriveCooling
WD Caviar Black LG DVD Writer Samsung Bluray Reader 4 case fans, PSU fan, 2 heatsink fans 
OSMonitorKeyboardPower
Windows 7 Pro 64-bit Samsung T220 22" TFT Logitech MX 5500 Coolermaster Silent Pro 700W Modular 
CaseMouseMouse PadAudio
Antec 300 Logitech MX Revolution Half Life 2 Auzentech X-Fi Forte 
  hide details  
Reply
Desktop
(16 items)
 
  
CPUMotherboardGraphicsRAM
Core i7 920 D0 ASUS P6T GeForce GTX 560 Ti Corsair XMS3 
Hard DriveOptical DriveOptical DriveCooling
WD Caviar Black LG DVD Writer Samsung Bluray Reader 4 case fans, PSU fan, 2 heatsink fans 
OSMonitorKeyboardPower
Windows 7 Pro 64-bit Samsung T220 22" TFT Logitech MX 5500 Coolermaster Silent Pro 700W Modular 
CaseMouseMouse PadAudio
Antec 300 Logitech MX Revolution Half Life 2 Auzentech X-Fi Forte 
  hide details  
Reply
post #45 of 128
Thread Starter 
@ [Adz]
Let's analyze your situation:
CPU @ stock
You are getting 27.5 gigaflops instead of say,32-33,which means 5 gigaflops less than expected.
I can think of 2 reasons for that;one-your ram latencies are high & two you are using 2800MB for the test,which is not all physial ram.These 2 factors may account for the lower value.

Your oc
Although the difference between the expected & actual value is more,I still think the same 2 factors are responsible.You have overclocked your cpu(and L1 & L2 cache)to 3.38 GHz,your FSB to 450,but your ram is running @ 840 MHz/6-7-7-19 which is less than 1:1.
On INTEL chipsets DRAM:FSB less than 1:1 is not possible,may be because the performance hit is large.The bios setting for performance level(aka tRD) could also impact performance.

So I suggest you use 2400 MB for now.Also if higher ram clocks or lower latencies are not an option,then lower command rate from 2 to 1.Evidently this works on Nvidia chipsets...sometimes.
Dont close explorer.exe(is it even possible?)and dont use any 3rd party app to free ram.
Use the method I explained a few posts earlier in answer to liberalelephant.Post back your results.

Good luck!

@ JimmyHart84

Post your results & we discuss your gigaflops value.[Adz] answered your other questions correctly.
Edited by PERSPOLIS - 2/12/10 at 3:31pm
Pebble
(13 items)
 
  
CPUMotherboardGraphicsRAM
E5200-lapped 3.5GHz/Vcore@1.24 volt ASUS P5QL PRO 9600GT@700/1750/2200 2X2 Gb GEIL@1066 
Hard DriveOptical DriveOSMonitor
250+120 gigs dvd rw+cd/dvd combo winxp32 sp3/vista64 sp2 LG crt 
PowerCase
330w cont./480w peak HEC 6A34/220X30 mm YATE LOON 135 CFM side fan 
  hide details  
Reply
Pebble
(13 items)
 
  
CPUMotherboardGraphicsRAM
E5200-lapped 3.5GHz/Vcore@1.24 volt ASUS P5QL PRO 9600GT@700/1750/2200 2X2 Gb GEIL@1066 
Hard DriveOptical DriveOSMonitor
250+120 gigs dvd rw+cd/dvd combo winxp32 sp3/vista64 sp2 LG crt 
PowerCase
330w cont./480w peak HEC 6A34/220X30 mm YATE LOON 135 CFM side fan 
  hide details  
Reply
post #46 of 128
Hi guys, attached are 2 screenshots, 1 after starting the run and 1 after it finished. I'll change it to 20 runs next time to ensure stability, but I did 10 this time just to get the gigaflop values for you and to make sure I'm running it correctly first.

Desktop
(16 items)
 
  
CPUMotherboardGraphicsRAM
Core i7 920 D0 ASUS P6T GeForce GTX 560 Ti Corsair XMS3 
Hard DriveOptical DriveOptical DriveCooling
WD Caviar Black LG DVD Writer Samsung Bluray Reader 4 case fans, PSU fan, 2 heatsink fans 
OSMonitorKeyboardPower
Windows 7 Pro 64-bit Samsung T220 22" TFT Logitech MX 5500 Coolermaster Silent Pro 700W Modular 
CaseMouseMouse PadAudio
Antec 300 Logitech MX Revolution Half Life 2 Auzentech X-Fi Forte 
  hide details  
Reply
Desktop
(16 items)
 
  
CPUMotherboardGraphicsRAM
Core i7 920 D0 ASUS P6T GeForce GTX 560 Ti Corsair XMS3 
Hard DriveOptical DriveOptical DriveCooling
WD Caviar Black LG DVD Writer Samsung Bluray Reader 4 case fans, PSU fan, 2 heatsink fans 
OSMonitorKeyboardPower
Windows 7 Pro 64-bit Samsung T220 22" TFT Logitech MX 5500 Coolermaster Silent Pro 700W Modular 
CaseMouseMouse PadAudio
Antec 300 Logitech MX Revolution Half Life 2 Auzentech X-Fi Forte 
  hide details  
Reply
post #47 of 128
Quote:
Originally Posted by PERSPOLIS View Post
@ [Adz]
Let's analyze your situation:
CPU @ stock
You are getting 27.5 gigaflops instead of say,32-33,which means 5 gigaflops less than expected.
I can think of 2 reasons for that;one-your ram latencies are high & two you are using 2800MB for the test,which is not all physial ram.These 2 factors may account for the lower value.
With stock settings, I should be getting nearer to 40GFlops, not 32-33? I mean, in your first post, you were getting very close to your calculated, whereas this is a massive drop.
I thought you said that RAM timing doesn't impact very much, in the first post?
Nevertheless, I've put the brought the OC down so that it's 1:1 and here are my current settings:
CPU: 3.0GHz
FSB: 400 (1600QDR) * 7.5
RAM: 800MHz (400*2) 6-6-6-18 2T (Stock)
FSB-DRAM: 1:1

I'm running the test now. I've stopped a lot of unneeded services, disabled antivirus and all that stuff and closed background apps. Task manager and LinX are both reporting total available physical memory as 3201MB, but I've set it to use 2400MB.
Open applications: LinX 0.6.4, RealTemp 3.40.
Will post back results soon.
EDIT: Still getting just over 30GFlops out of a calculated maximum 48.


About to run the test again with the command rate set to 1. Will post back.
Task manager and LinX are both reporting total available physical memory as 3300MB, but I've set it to use 2400MB.
Open applications: LinX 0.6.4, RealTemp 3.40.
EDIT: Peaks at 33GFlops, out of a calculated maximum of 38, BUT is unstable - LinX stopped with errors in under 20 minutes, three attempts.

Quote:
Originally Posted by PERSPOLIS View Post
Your oc
Although the difference between the expected & actual value is more,I still think the same 2 factors are responsible.You have overclocked your cpu(and L1 & L2 cache)to 3.38 GHz,your FSB to 450,but your ram is running @ 840 MHz/6-7-7-19 which is less than 1:1.
On INTEL chipsets DRAM:FSB less than 1:1 is not possible,may be because the performance hit is large.The bios setting for performance level(aka tRD) could also impact performance.

So I suggest you use 2400 MB for now.Also if higher ram clocks or lower latencies are not an option,then lower command rate from 2 to 1.Evidently this works on Nvidia chipsets...sometimes.
Dont close explorer.exe(is it even possible?)and dont use any 3rd party app to free ram.
Use the method I explained a few posts earlier in answer to liberalelephant.Post back your results.

Good luck!
This board has an NVidia chipset, so the non-1:1 timings shouldn't cause that much of a performance loss, right?
Explorer.exe can be killed via task manager, or by Ctrl+Shift+RightClick on a clear area of the start menu.
The third party app just runs a windows script to clear the RAM and then closes so it's no longer running. However, I didn't use it in the current test at all.
Wouldn't reducing the command rate decrease performance?

Quote:
Originally Posted by JimmyHart84 View Post
Hi guys, attached are 2 screenshots, 1 after starting the run and 1 after it finished. I'll change it to 20 runs next time to ensure stability, but I did 10 this time just to get the gigaflop values for you and to make sure I'm running it correctly first.
Those values seem fine to me, very close to the maximum. Just out of curiosity though, try running it in safe mode where the system has a smaller footprint - you may be able to hit 60GFlops
Edited by [Adz] - 2/13/10 at 6:45am
Echo
(8 items)
 
   
CPUMotherboardGraphicsRAM
Intel Core i5-3570K Asus Maximus V Gene Asus Radeon HD 7970 3GB Mushkin Blackline 16GB 
Hard DriveHard DriveCaseOther
2x Samsung 830 256GB Seagate Barracuda 3TB Silverstone Temjin TJ08-E BeQuiet E9-700w 
CPUMotherboardRAMHard Drive
Intel i3-3225 ASUS P8B75-MLX Corsair Ballistix Tactical Western Digital Red 
OSPowerCaseOther
Ubuntu Corsair CX500M Norco RPC-4224 LSI SAS 9207-4i4e HBA 
Other
Chenbro CK22804 24-Port SAS Expander Card 
  hide details  
Reply
Echo
(8 items)
 
   
CPUMotherboardGraphicsRAM
Intel Core i5-3570K Asus Maximus V Gene Asus Radeon HD 7970 3GB Mushkin Blackline 16GB 
Hard DriveHard DriveCaseOther
2x Samsung 830 256GB Seagate Barracuda 3TB Silverstone Temjin TJ08-E BeQuiet E9-700w 
CPUMotherboardRAMHard Drive
Intel i3-3225 ASUS P8B75-MLX Corsair Ballistix Tactical Western Digital Red 
OSPowerCaseOther
Ubuntu Corsair CX500M Norco RPC-4224 LSI SAS 9207-4i4e HBA 
Other
Chenbro CK22804 24-Port SAS Expander Card 
  hide details  
Reply
post #48 of 128
Am i'm thinking right that if you want to get your result close to what you have calculated before you have to close ALL your apps & basically your os to achive that? I'm getting 4.3-4.4 GFlops on my 955 @ 3.6 but i should get 16*3.6= 57.6. I think the result depends on the CPU it's using (Intel gets higher GFlops).
Baby Dragon
(13 items)
 
  
CPUMotherboardGraphicsRAM
Phenom II X4 955BE Gigabyte MA790FXT-UD5P Palit 3850 CORSAIR TW3X4G1600C9DHX 
Hard DriveOSMonitorPower
WD Caviar GP Windows 7 LG W2243S CHIEFTEC 750W 
Case
ANTEC P183 
  hide details  
Reply
Baby Dragon
(13 items)
 
  
CPUMotherboardGraphicsRAM
Phenom II X4 955BE Gigabyte MA790FXT-UD5P Palit 3850 CORSAIR TW3X4G1600C9DHX 
Hard DriveOSMonitorPower
WD Caviar GP Windows 7 LG W2243S CHIEFTEC 750W 
Case
ANTEC P183 
  hide details  
Reply
post #49 of 128
Quote:
Originally Posted by mindthecap View Post
Am i'm thinking right that if you want to get your result close to what you have calculated before you have to close ALL your apps & basically your os to achive that? I'm getting 4.3-4.4 GFlops on my 955 @ 3.6 but i should get 16*3.6= 57.6. I think the result depends on the CPU it's using (Intel gets higher GFlops).
Yep, because everything that's running uses CPU cycles. The less there is that's running, the more cycles there are available for LINPACK to use.

And do you mean 43-44 GFlops, or 4.3-4.4? There's a massive difference
Echo
(8 items)
 
   
CPUMotherboardGraphicsRAM
Intel Core i5-3570K Asus Maximus V Gene Asus Radeon HD 7970 3GB Mushkin Blackline 16GB 
Hard DriveHard DriveCaseOther
2x Samsung 830 256GB Seagate Barracuda 3TB Silverstone Temjin TJ08-E BeQuiet E9-700w 
CPUMotherboardRAMHard Drive
Intel i3-3225 ASUS P8B75-MLX Corsair Ballistix Tactical Western Digital Red 
OSPowerCaseOther
Ubuntu Corsair CX500M Norco RPC-4224 LSI SAS 9207-4i4e HBA 
Other
Chenbro CK22804 24-Port SAS Expander Card 
  hide details  
Reply
Echo
(8 items)
 
   
CPUMotherboardGraphicsRAM
Intel Core i5-3570K Asus Maximus V Gene Asus Radeon HD 7970 3GB Mushkin Blackline 16GB 
Hard DriveHard DriveCaseOther
2x Samsung 830 256GB Seagate Barracuda 3TB Silverstone Temjin TJ08-E BeQuiet E9-700w 
CPUMotherboardRAMHard Drive
Intel i3-3225 ASUS P8B75-MLX Corsair Ballistix Tactical Western Digital Red 
OSPowerCaseOther
Ubuntu Corsair CX500M Norco RPC-4224 LSI SAS 9207-4i4e HBA 
Other
Chenbro CK22804 24-Port SAS Expander Card 
  hide details  
Reply
post #50 of 128
Quote:
And do you mean 43-44 GFlops, or 4.3-4.4? There's a massive difference
Yeah i mean 43-44, sry for that. But still why such a big difference? Is it normal? And when i test the computer in safe mode and it BSODs or reboots will it make more damage to the computer than testing it in windows? As i understand it doesn't matter where you test your computer, clock cycles are being used either way.
Baby Dragon
(13 items)
 
  
CPUMotherboardGraphicsRAM
Phenom II X4 955BE Gigabyte MA790FXT-UD5P Palit 3850 CORSAIR TW3X4G1600C9DHX 
Hard DriveOSMonitorPower
WD Caviar GP Windows 7 LG W2243S CHIEFTEC 750W 
Case
ANTEC P183 
  hide details  
Reply
Baby Dragon
(13 items)
 
  
CPUMotherboardGraphicsRAM
Phenom II X4 955BE Gigabyte MA790FXT-UD5P Palit 3850 CORSAIR TW3X4G1600C9DHX 
Hard DriveOSMonitorPower
WD Caviar GP Windows 7 LG W2243S CHIEFTEC 750W 
Case
ANTEC P183 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel CPUs
Overclock.net › Forums › Intel › Intel CPUs › How to run LinPack stress test(LinX/IBT)properly-an explanation(maybe a guide)