Overclock.net › Forums › Benchmarks › Overclock.net HWBot Team › Super PI 32m tweaking guide
New Posts  All Forums:Forum Nav:

Super PI 32m tweaking guide - Page 2

post #11 of 101
Yep, nice guide there Rob. Should do well to help some people out. Subtimings are super important for 32m, that can't be stressed enough. Even more important than waza. Though using both will do well to shave many seconds off.

Here's a run I did at 4.5ghz a while back following the same principles outlined in the guide:
600x624px-LL-78e9e429_1.jpeg
Phenomenal
(12 items)
 
  
CPUMotherboardGraphicsRAM
PhenomII x4 965 @ 4.327GHz 1.44v Asus M5A99X-EVO Gigabyte HD69501GB 4GB SuperTalent @ 1822 CAS7 
Hard DriveHard DriveCoolingOS
2x WD 500GB Raid0 WDCB 640GB Corsair H100 W7 x64  
MonitorPowerCaseAudio
LG 23" IPS  Corsair HX750 Fractal Arc Midi Auzentech X-Raider 7.1 
  hide details  
Reply
Phenomenal
(12 items)
 
  
CPUMotherboardGraphicsRAM
PhenomII x4 965 @ 4.327GHz 1.44v Asus M5A99X-EVO Gigabyte HD69501GB 4GB SuperTalent @ 1822 CAS7 
Hard DriveHard DriveCoolingOS
2x WD 500GB Raid0 WDCB 640GB Corsair H100 W7 x64  
MonitorPowerCaseAudio
LG 23" IPS  Corsair HX750 Fractal Arc Midi Auzentech X-Raider 7.1 
  hide details  
Reply
post #12 of 101
Thread Starter 
Quote:
Originally Posted by GENiEBEN View Post

I dont think Wazza actually worked in your case, 2-3sec can be the error margin alone. Have you been doing it manually or OCXTweaker / BenchTec ? Can you also do a run with Command Rate set as 2T, I found it sometimes better for 32MB if used with tight timings.

I tested it many times and the times were always consistent. In my testing there was never once a 2-3 second 'error' tongue.gif
Saosin
(13 items)
 
  
CPUMotherboardGraphicsRAM
Core i7 2600K @ 4.7 Asus Gene-Z Gen3 EVGA GTX680 @ 1240/3500 16gb G.Skill Trident X @ 2133 9-11-9-28 
Hard DriveHard DriveCoolingOS
Samsung 128gb 830 - WD Velociraptor 300gb  2x Samsung 1tb F3 - 2x Samsung 2tb F4 Corsair H80 Win7 x64 
MonitorKeyboardPowerCase
Samsung 2443BW+ Leopold Cherry Browns  Corsair HX650 Fractal Design Define R3 
Audio
Asus Xonar D2X  
  hide details  
Reply
Saosin
(13 items)
 
  
CPUMotherboardGraphicsRAM
Core i7 2600K @ 4.7 Asus Gene-Z Gen3 EVGA GTX680 @ 1240/3500 16gb G.Skill Trident X @ 2133 9-11-9-28 
Hard DriveHard DriveCoolingOS
Samsung 128gb 830 - WD Velociraptor 300gb  2x Samsung 1tb F3 - 2x Samsung 2tb F4 Corsair H80 Win7 x64 
MonitorKeyboardPowerCase
Samsung 2443BW+ Leopold Cherry Browns  Corsair HX650 Fractal Design Define R3 
Audio
Asus Xonar D2X  
  hide details  
Reply
post #13 of 101
Quote:
Originally Posted by robbo2 View Post

Quote:
Originally Posted by GENiEBEN View Post

I dont think Wazza actually worked in your case, 2-3sec can be the error margin alone. Have you been doing it manually or OCXTweaker / BenchTec ? Can you also do a run with Command Rate set as 2T, I found it sometimes better for 32MB if used with tight timings.

I tested it many times and the times were always consistent. In my testing there was never once a 2-3 second 'error' tongue.gif

Were they on the same boot? Different boots can provide a 2-3 second difference just in the way the cache is managed. I'm more inclined to agree with GenieBen on this one, even with runs in the same boot (dependant on services enabled).

Just plucking numbers for arguments sake, for example, look at 1M - score can be +/- 0.05 secs between runs. Now extrapolate that from 7 secs to 7 minutes (x60), and the +/- error becomes 0.05*60 = 3 secs.
post #14 of 101
Thread Starter 
Lots of re-booting and the times were consistent. They could be out by .2 of a second at most with the same settings. If your getting 2-3 seconds differences then your OS is horrible.

This thread is not about if waza was working or not. I do not have to prove that and I'm not going to argue it. It's about improving efficiency to get lower times and more boints. I did a run today at under 4.5, now take a look here http://hwbot.org/submission/2277588_. Look at my score and my core speed then look at all the people around my time and there core speeds.

The guide is to help improve your efficiency. Simple as that.
Edited by robbo2 - 4/28/12 at 5:04am
Saosin
(13 items)
 
  
CPUMotherboardGraphicsRAM
Core i7 2600K @ 4.7 Asus Gene-Z Gen3 EVGA GTX680 @ 1240/3500 16gb G.Skill Trident X @ 2133 9-11-9-28 
Hard DriveHard DriveCoolingOS
Samsung 128gb 830 - WD Velociraptor 300gb  2x Samsung 1tb F3 - 2x Samsung 2tb F4 Corsair H80 Win7 x64 
MonitorKeyboardPowerCase
Samsung 2443BW+ Leopold Cherry Browns  Corsair HX650 Fractal Design Define R3 
Audio
Asus Xonar D2X  
  hide details  
Reply
Saosin
(13 items)
 
  
CPUMotherboardGraphicsRAM
Core i7 2600K @ 4.7 Asus Gene-Z Gen3 EVGA GTX680 @ 1240/3500 16gb G.Skill Trident X @ 2133 9-11-9-28 
Hard DriveHard DriveCoolingOS
Samsung 128gb 830 - WD Velociraptor 300gb  2x Samsung 1tb F3 - 2x Samsung 2tb F4 Corsair H80 Win7 x64 
MonitorKeyboardPowerCase
Samsung 2443BW+ Leopold Cherry Browns  Corsair HX650 Fractal Design Define R3 
Audio
Asus Xonar D2X  
  hide details  
Reply
post #15 of 101
Quote:
Originally Posted by borandi View Post

Were they on the same boot? Different boots can provide a 2-3 second difference just in the way the cache is managed. I'm more inclined to agree with GenieBen on this one, even with runs in the same boot (dependant on services enabled).
Just plucking numbers for arguments sake, for example, look at 1M - score can be +/- 0.05 secs between runs. Now extrapolate that from 7 secs to 7 minutes (x60), and the +/- error becomes 0.05*60 = 3 secs.

You can always, you know, stop speculating and give it a go for yourself. Never have I had a 32m run variate by 3s with the same settings and OS install. I think I've even struggled to see a .5s variation on 32m, the runs stay very close if they're on the same settings.

His waza method is working, though he may be able to squeeze a little more out of it and chop off another second or 2, but it's definitely not a variation due to spi itself.

We're both speaking from experience, and we're not wrong.
Phenomenal
(12 items)
 
  
CPUMotherboardGraphicsRAM
PhenomII x4 965 @ 4.327GHz 1.44v Asus M5A99X-EVO Gigabyte HD69501GB 4GB SuperTalent @ 1822 CAS7 
Hard DriveHard DriveCoolingOS
2x WD 500GB Raid0 WDCB 640GB Corsair H100 W7 x64  
MonitorPowerCaseAudio
LG 23" IPS  Corsair HX750 Fractal Arc Midi Auzentech X-Raider 7.1 
  hide details  
Reply
Phenomenal
(12 items)
 
  
CPUMotherboardGraphicsRAM
PhenomII x4 965 @ 4.327GHz 1.44v Asus M5A99X-EVO Gigabyte HD69501GB 4GB SuperTalent @ 1822 CAS7 
Hard DriveHard DriveCoolingOS
2x WD 500GB Raid0 WDCB 640GB Corsair H100 W7 x64  
MonitorPowerCaseAudio
LG 23" IPS  Corsair HX750 Fractal Arc Midi Auzentech X-Raider 7.1 
  hide details  
Reply
post #16 of 101
Remember, waza size is very important and is also platform specific.
Daily
(16 items)
 
  
CPUMotherboardGraphicsRAM
i7 2600K @ 4.8GHz, 1.4V Maximus IV Extreme GTX 560 DCII TOP 4x4GB Kingston HyperX 1866MHz CL11 
Hard DriveHard DriveCoolingCooling
Kingston HyperX 3K 120GB Samsung 640GB Thermochill PA120.2 Jingway DB-1 Pump 
CoolingCoolingOSMonitor
2x Kaze Jyuni 1900RPM EK Supreme HF Copper plexi Arch Linux x64 Acer X223HQ 1920x1080 
KeyboardPowerCaseMouse
Microsoft Sidewinder X4 Corsair HX750 Modded Corsair Carbide 500R Logitech M500 
  hide details  
Reply
Daily
(16 items)
 
  
CPUMotherboardGraphicsRAM
i7 2600K @ 4.8GHz, 1.4V Maximus IV Extreme GTX 560 DCII TOP 4x4GB Kingston HyperX 1866MHz CL11 
Hard DriveHard DriveCoolingCooling
Kingston HyperX 3K 120GB Samsung 640GB Thermochill PA120.2 Jingway DB-1 Pump 
CoolingCoolingOSMonitor
2x Kaze Jyuni 1900RPM EK Supreme HF Copper plexi Arch Linux x64 Acer X223HQ 1920x1080 
KeyboardPowerCaseMouse
Microsoft Sidewinder X4 Corsair HX750 Modded Corsair Carbide 500R Logitech M500 
  hide details  
Reply
post #17 of 101
Quote:
Originally Posted by jjjc_93 View Post

You can always, you know, stop speculating and give it a go for yourself. Never have I had a 32m run variate by 3s with the same settings and OS install. I think I've even struggled to see a .5s variation on 32m, the runs stay very close if they're on the same settings.

His waza method is working, though he may be able to squeeze a little more out of it and chop off another second or 2, but it's definitely not a variation due to spi itself.

We're both speaking from experience, and we're not wrong.

You can always, you know, realise that people can't quote real numbers or screenshots out of their ass any time of day. If you have a direct problem with me, then come out and say it, lets not beat around the bush here. We've all had experience, and these are mine. If they're different to yours, be methodical and courteous.

I'll admit, I don't run much 32m, only for competitions or when a new rare CPU crosses my desk, but those are my experiences nonetheless.
post #18 of 101
Thread Starter 
Quote:
Originally Posted by ARandomOWl View Post

Remember, waza size is very important and is also platform specific.

That's why I left it out. You can't just give settings and they work. To give you an idea, when I first ram spi 32 at the same settings I got 9:03 with simple tweaks it came down to the 8:49 run. That's 14 seconds doing very little. No waza.
Saosin
(13 items)
 
  
CPUMotherboardGraphicsRAM
Core i7 2600K @ 4.7 Asus Gene-Z Gen3 EVGA GTX680 @ 1240/3500 16gb G.Skill Trident X @ 2133 9-11-9-28 
Hard DriveHard DriveCoolingOS
Samsung 128gb 830 - WD Velociraptor 300gb  2x Samsung 1tb F3 - 2x Samsung 2tb F4 Corsair H80 Win7 x64 
MonitorKeyboardPowerCase
Samsung 2443BW+ Leopold Cherry Browns  Corsair HX650 Fractal Design Define R3 
Audio
Asus Xonar D2X  
  hide details  
Reply
Saosin
(13 items)
 
  
CPUMotherboardGraphicsRAM
Core i7 2600K @ 4.7 Asus Gene-Z Gen3 EVGA GTX680 @ 1240/3500 16gb G.Skill Trident X @ 2133 9-11-9-28 
Hard DriveHard DriveCoolingOS
Samsung 128gb 830 - WD Velociraptor 300gb  2x Samsung 1tb F3 - 2x Samsung 2tb F4 Corsair H80 Win7 x64 
MonitorKeyboardPowerCase
Samsung 2443BW+ Leopold Cherry Browns  Corsair HX650 Fractal Design Define R3 
Audio
Asus Xonar D2X  
  hide details  
Reply
post #19 of 101
Quote:
Originally Posted by borandi View Post

You can always, you know, realise that people can't quote real numbers or screenshots out of their ass any time of day. If you have a direct problem with me, then come out and say it, lets not beat around the bush here. We've all had experience, and these are mine. If they're different to yours, be methodical and courteous.
I'll admit, I don't run much 32m, only for competitions or when a new rare CPU crosses my desk, but those are my experiences nonetheless.

What are you on about? I have no personal issue with you, I take issue to what you've said because there are two of us here with exactly the same "opinion" and experiences to back that up. I have played with 32m heaaps, it is one of the most tweakable and consistent benchmarks that are awarded points on benchmark, definitely the pick of 32m stuff.

Your severe lack of experience with 32m shows, anybody who has taken the time of day to show just a little efficiency in their 32m runs will agree with what I'm saying here. You will NEVER see a 3s variance on the same OS and settings. It just doesn't happen.

Rob has a screenshot up with and without waza showing a 3s difference between them and it has led to this crap. I feel sorry for him and sorry for OCN. He's put the time and effort into what really is a great guide and it's tarnished with your inexperience and lack of want to try and replicate the bs claims you're making. I'll be very surprised if he bothers with another guide again, and that will be a shame for OCN because Rob has a lot to share.

EDIT: Reading over I guess I do have a little bit of a problem with you, but that's directly tied to what you've said in here.
Phenomenal
(12 items)
 
  
CPUMotherboardGraphicsRAM
PhenomII x4 965 @ 4.327GHz 1.44v Asus M5A99X-EVO Gigabyte HD69501GB 4GB SuperTalent @ 1822 CAS7 
Hard DriveHard DriveCoolingOS
2x WD 500GB Raid0 WDCB 640GB Corsair H100 W7 x64  
MonitorPowerCaseAudio
LG 23" IPS  Corsair HX750 Fractal Arc Midi Auzentech X-Raider 7.1 
  hide details  
Reply
Phenomenal
(12 items)
 
  
CPUMotherboardGraphicsRAM
PhenomII x4 965 @ 4.327GHz 1.44v Asus M5A99X-EVO Gigabyte HD69501GB 4GB SuperTalent @ 1822 CAS7 
Hard DriveHard DriveCoolingOS
2x WD 500GB Raid0 WDCB 640GB Corsair H100 W7 x64  
MonitorPowerCaseAudio
LG 23" IPS  Corsair HX750 Fractal Arc Midi Auzentech X-Raider 7.1 
  hide details  
Reply
post #20 of 101
C'mon guys, chill.

Some great efficiency from jjjc and robbo biggrin.gif Thanks for the guide, we needed one.
One gear, GO!
(14 items)
 
  
CPUMotherboardRAMHard Drive
FX8150 Crosshair V Formula G.Skill PIS 2200 cl7 Acard ans9010 raid 0+Corsair Force 3 
Cooling
OCN Marksman 
  hide details  
Reply
One gear, GO!
(14 items)
 
  
CPUMotherboardRAMHard Drive
FX8150 Crosshair V Formula G.Skill PIS 2200 cl7 Acard ans9010 raid 0+Corsair Force 3 
Cooling
OCN Marksman 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Overclock.net HWBot Team
Overclock.net › Forums › Benchmarks › Overclock.net HWBot Team › Super PI 32m tweaking guide