Overclock.net › Forums › Intel › Intel CPUs › Limits for 2500k on air? 5Ghz? What do we know in 2013?
New Posts  All Forums:Forum Nav:

Limits for 2500k on air? 5Ghz? What do we know in 2013?

post #1 of 9
Thread Starter 
2500k, Noctua C14, Asrock Extreme 3 Gen 3, CM 690 with a handful of Noctuas

Basically I have a bunch of questions about overclocking 2500k to the extreme limits of air cooling.

After many months of data, what do we know about the absolute max vcore and temps? Not just for benchmarking, but for day to day use. Perhaps 8 hours a day, with relatively low load most of that time. What do we know from 2500Ks that have died before?

If I were trying to go to 1.5 vcore, is that safe for a 8 hour prime run? I'd also be leaving at 1.5 day to day use.

Is there anything wrong with using LLC level 1? I've always used LLC1. I heard that the issue was voltage spikes, but long term monitoring with lower overclocks with both CPU-Z and Asrock's tuning utility have always reported a max vcore of parity to what I set in bios, never higher. It fluctuates to as low as -.03 of bios setting.

I use fixed voltage. Anything wrong with that?

Is it safe to prime for 8 hours with peak temps at 85-89 C? Every time I see my temps go over 85C in a prime run I always just abort, but some people feel that Sandys have fairly high limits: http://www.overclock.net/t/1198504/complete-overclocking-guide-sandy-bridge-ivy-bridge-asrock-edition#

I usually use small FFT in prime. Whats the difference between that and large FFT? 1792 test? Blend?

What volts do I need to play around with, and what are their upper limits? PLL? VTT? Northbridge? Which ones are increase only? Which ones are increase or decrease?

Do constant BSODs damage my system? Corrupt data? I'd probably have to get a good 30-50 BSODs before I get anywhere. What about hanging while trying to POST and being forced to reset cmos?

Any updates on BSOD error codes?

Anything about C states and other settings?

Either way, I don't honestly expect to get anywhere, because I BSOD (code 101) within minutes at 4.5 GHZ, 1.400 vcore. Feels like a total dud cpu.


Sorry for the deluge of questions, and thanks in advance!
Edited by Kanashimu - 1/29/13 at 5:52pm
s
(12 items)
 
  
CPUMotherboardGraphicsRAM
1600X 4.025 @1.45v Gigabyte K7 1080 Ti 2050/5800 16 GB Trident Z 3200-12-14-14-30-1T-44-267 1.53v 
Hard DriveHard DriveCoolingOS
Samsung 960 Evo 500 GB Samsung 840 Pro 256 GB Noctua C14 Windows 10 x64 
MonitorPowerCase
Viotek GN34C UW 1440p 100Hz Corsair HX750 Phanteks Enthoo Evolv ATX 
  hide details  
Reply
s
(12 items)
 
  
CPUMotherboardGraphicsRAM
1600X 4.025 @1.45v Gigabyte K7 1080 Ti 2050/5800 16 GB Trident Z 3200-12-14-14-30-1T-44-267 1.53v 
Hard DriveHard DriveCoolingOS
Samsung 960 Evo 500 GB Samsung 840 Pro 256 GB Noctua C14 Windows 10 x64 
MonitorPowerCase
Viotek GN34C UW 1440p 100Hz Corsair HX750 Phanteks Enthoo Evolv ATX 
  hide details  
Reply
post #2 of 9
Quote:
Originally Posted by Kanashimu View Post

2500k, Noctua C14, Asrock Extreme 3 Gen 3, CM 690 with a handful of Noctuas

Basically I have a bunch of questions about overclocking 2500k to the extreme limits of air cooling.

After many months of data, what do we know about the absolute max vcore and temps? Not just for benchmarking, but for day to day use. Perhaps 8 hours a day, with relatively low load most of that time. What do we know from 2500Ks that have died before?

 

From what I've seen so far here on OCN, these CPUs are very tough.  This is the worst I've seen so far (posted August 12th, 2012):

 

Quote:
Originally Posted by $ilent View Post

Honestly id stick with whatever is stable under 1.40v at full load. Its not worth the extra .1v strain just for the extra 100mhz overclock.

Just for info also, since december 2011 my cpu at 4.9ghz and 1.408v has degraded to the point where it now needs 1.45v to remain stale at 4.9ghz.

Thats being at full load for about 7 months out of 9 but still not the best news considering temps never really went over 80C.

 

So he had degradation, but it was due to being at full load for 7 months straight.  I didn't ask, but I'm assuming he was Folding 24/7 because he's a Chimp Challenge participant, and he is a Folding Millionaire with 20 Million Points.

 

So yeah, these CPUs are tough.

 

 

Quote:
Originally Posted by Kanashimu View Post

If I were trying to go to 1.5 vcore, is that safe for a 8 hour prime run? I'd also be leaving at 1.5 day to day use.

 

Well, I guess if you use an Offset and if your multiplier goes down to x16 while idling, then it would be relatively safe because your voltage wouldn't sit that high 24/7.  Even so, I personally wouldn't do it with air cooling.

 

 

Quote:
Originally Posted by Kanashimu View Post

Is there anything wrong with using LLC level 1? I've always used LLC1. I heard that the issue was voltage spikes, but long term monitoring with lower overclocks with both CPU-Z and Asrock's tuning utility have always reported a max vcore of parity to what I set in bios, never higher. It fluctuates to as low as -.03 of bios setting.

 

The only issue I know of is higher temperatures of the Voltage Regulator Monitor, and if it gets too hot, then it can result in instability.  Another issue is if you are using a fixed voltage, then the idle voltage is lower than the load voltage due to vRise (the voltage goes up under load).  This can be bad if the idle voltage is too low for maintaining stability while idling.  So, I recommend using level 2 or 3.

 

Regarding the voltage spike issue:  as far as I know, the better your Voltage Regulator Module is, the less of a problem this is.  With today's boards (including yours), the VRM is so good that this is now a non-issue unless you have a fairly low-quality motherboard that's known to have a low-quality VRM.  In the earlier days when motherboards equipped with LLC controls were few and far between, this voltage spike issue was common because motherboards back then were not designed well enough to protect the CPU.  So, I would say that they didn't have VRMs that were able to stop or suppress these spikes.

 

Regarding the voltage spike itself:  there is no BIOS or software that can show the spike because the spike doesn't last long enough.  It's like pressing down a spring and then letting it go.  If you were to view this in very slow motion, then you'd see the spring expand quite a bit and stretch way out, then it would contract, stretch, contract, stretch, etc. over and over becoming less and less each time until finally settling to become motionless.  That's a little bit how the voltage spike works:  every time there's load placed on the PSU or taken off of it, there's that quick "boing" during the transition from idle to load or load to idle.  It's kind of like a little surge.  It happens so quickly that you'd need special equipment to detect it.

 

 

Quote:
Originally Posted by Kanashimu View Post

I use fixed voltage. Anything wrong with that?

 

I feel that it may be a matter of opinion.  I don't like having my full voltage powering my CPU 24/7, even if it's just idling or under extremely light loads.  One school of thought says that if it's idling, then it's harmless.  Well, I don't want to buy into that because if my CPU needs over 1.4V to be stable under load, then I absolutely do not want that powering my CPU 24/7.  That just seems harmful to me, especially in the long run.  So, I use an offset so that it can sit very low while it's idling.  For example, it's sitting at only ~1.000V while I'm sitting here typing.  The voltage adjusts dynamically based on how much load there is on the CPU.

 

 

Quote:
Originally Posted by Kanashimu View Post

Is it safe to prime for 8 hours with peak temps at 85-89 C? Every time I see my temps go over 85C in a prime run I always just abort, but some people feel that Sandys have fairly high limits: http://www.overclock.net/t/1198504/complete-overclocking-guide-sandy-bridge-ivy-bridge-asrock-edition#

 

I feel that it can depend on your voltage, especially after seeing $ilent's experience.  Mine needs 1.376V under load for 4.7 GHz, and the max core temp is around 70°C.  So, that should be extremely safe even for a 24-hour Prime95 run.  If mine were needing close to 1.5V, then I don't know if I would feel safe running Prime95 with an air cooler.

 

 

Quote:
Originally Posted by Kanashimu View Post

I usually use small FFT in prime. Whats the difference between that and large FFT? 1792 test? Blend?

 

The Small FFTs is useless on Sandy Bridge because it has an Integrated Memory Controller.  This means that the Small FFTs test is not using the entire CPU and is therefore a waste of time.  The Large FFTs can be just as much of a waste of time for the same reason because it's still not stressing only the CPU.

 

1792K FFT test is only to be used if you know your CPU is not passing that FFT size.  The same is true for 1344K and all other commonly-recommended FFTs.  Only focus on specific FFTs if you know your CPU is not passing them.

 

The default Blend test is better than the above because it uses the entire CPU.  However, the default Blend test only uses about 1600 MB of your memory.  Therefore, the Custom Blend test is the best one because then you get to tell Prime95 to use about 90% of your memory.  Except, if you have any version of Prime95 that's earlier than v27.7 build 2, then you'd be wasting your time because that's the first version to use the AVX instruction set.  Going further, Windows 7 SP1 is required because SP1 provides AVX support.

 

 

Quote:
Originally Posted by Kanashimu View Post

What volts do I need to play around with, and what are their upper limits? PLL? VTT? Northbridge? Which ones are increase only? Which ones are increase or decrease?

 

I'm not sure how to answer this because it depends.  If you decide to only go to about 4.7 GHz, then all you really need to adjust is the core voltage because on average, anything up to about 4.7 GHz is easy - so this includes 4.5 GHz.  If you go way beyond, then it can benefit you to adjust others as recommended by the experts who have explored higher overclocks.  The highest I tried is 4.8 GHz, but it needed 1.408V under load, so I put it back down to 4.7 GHz.  When I did that, I experimented with the CPU PLL voltage and I had it at approximately 1.71V as recommended by others and it seemed to help make it easier to achieve stability.  Still, 1.408V is too high for me.  With my limited income, I probably shouldn't be overclocking at all, but hey...  :)

 

 

Quote:
Originally Posted by Kanashimu View Post

Do constant BSODs damage my system? Corrupt data? I'd probably have to get a good 30-50 BSODs before I get anywhere.

 

No, but only as long as you do this for EVERY BSOD or system lock-up without exception:

 

  1. You get the BSOD or lock-up
  2. You press the Reset button on your computer case (that is, if you have Windows set so that your computer isn't automatically restarted upon a BSOD)
  3. You wait for your system to reboot and for Windows to start back up
  4. When Windows is finished starting up, choose Restart from the Start Menu
  5. Wait for it to fully restart
  6. When you get back to the desktop, choose Restart again
  7. Go into the BIOS
  8. Tweak some more in hopes of avoiding the BSOD or lock-up in the future

 

In other words, if you don't give it a proper restart in between each BSOD or lock-up like this, then you run the risk of not being able to start Windows and instead seeing one of those messages on a black screen.  They can vary, and one of them does indeed talking about something being "missing or corrupt" (I'm having a hard time remembering the details).  Another is "NTDLR is missing".  One time, I saw "Operating system not found".  Luckily and by some kind of miracle, I always figured out how to fix it and continue without having to lose anything.  I don't remember how I did that because it was almost 5 years ago now.

 

So, be careful to always ensure that Windows gets a proper restart before continuing.

 

 

Quote:
Originally Posted by Kanashimu View Post

What about hanging while trying to POST and being forced to reset cmos?

 

I don't know because I haven't heard of anyone having to do this too many times.  The only problem with resetting CMOS is you usually have to go back to factory default settings.

 

 

Quote:
Originally Posted by Kanashimu View Post

Any updates on BSOD error codes?

 

I personally don't pay attention to BSOD codes.  To me, a BSOD just means "oops, it's still not stable. lol".  I don't get into the codes like most people do, but I'm not proud of it.

 

 

Quote:
Originally Posted by Kanashimu View Post

Anything about C states and other settings?

 

If you switch to an Offset voltage, then keep C3 and C6 disabled.

 

If your motherboard has Internal PLL Overvoltage, then only enable it if Windows won't start at higher multipliers.  So if it's disabled and you're at x50 and Windows is still starting, then you don't need it.

 

I can't think of anything else because I would need you to be more specific.  :)

 

 

Quote:
Originally Posted by Kanashimu View Post

Either way, I don't honestly expect to get anywhere, because I BSOD (code 101) within minutes at 4.5 GHZ, 1.400 vcore. Feels like a total dud cpu.

 

What does CPU-Z say the core voltage is while running Prime95?

 

I don't think it's a total dud.  There are 2500Ks here on OCN that need a lot more voltage than that for only 4.5 GHz.  ;)

 

 

Quote:
Originally Posted by Kanashimu View Post

Sorry for the deluge of questions, and thanks in advance!

 

Nah, that's what OCN's for.  It makes me very happy to have any ability to help!

It's a computer!
(18 items)
 
  
CPUMotherboardGraphicsRAM
i5-2500K @ 4.5GHz (1.368-1.384V fixed voltage) ASUS P8P67 EVO B3 (UEFI ver. 1850) GTX 780 ASUS DirectCU II (1228 / 6300, 1.180V) G.SKILL Ripjaws X 8GB (2 x 4GB) 1866MHz, CL9 
Hard DriveHard DriveOptical DriveOptical Drive
250 GB Samsung 840 EVO (OS) 3 TB Toshiba P300 (storage) Samsung SH-S243N 24x DVD Burner Samsung SH-S203N 20X DVD Burner 
CoolingOSMonitorKeyboard
Thermaltake Frio Win 7 Home Premium x64 SP1 Retail AOC G2460PG (24" 1920 x 1080 144Hz G-SYNC) Filco Majestouch 104-key Cherry MX Blues w/NKRO 
PowerCaseMouseMouse Pad
Corsair HX650 (Bronze, ordered on 12-12-2009) CM 690 Intellimouse Optical (1.1A) 1000Hz polling rate Basic, but premium round 
AudioAudio
X-Fi Titanium HD Klipsch ProMedia 2.1 (with 16 AWG Monster Cable... 
  hide details  
Reply
It's a computer!
(18 items)
 
  
CPUMotherboardGraphicsRAM
i5-2500K @ 4.5GHz (1.368-1.384V fixed voltage) ASUS P8P67 EVO B3 (UEFI ver. 1850) GTX 780 ASUS DirectCU II (1228 / 6300, 1.180V) G.SKILL Ripjaws X 8GB (2 x 4GB) 1866MHz, CL9 
Hard DriveHard DriveOptical DriveOptical Drive
250 GB Samsung 840 EVO (OS) 3 TB Toshiba P300 (storage) Samsung SH-S243N 24x DVD Burner Samsung SH-S203N 20X DVD Burner 
CoolingOSMonitorKeyboard
Thermaltake Frio Win 7 Home Premium x64 SP1 Retail AOC G2460PG (24" 1920 x 1080 144Hz G-SYNC) Filco Majestouch 104-key Cherry MX Blues w/NKRO 
PowerCaseMouseMouse Pad
Corsair HX650 (Bronze, ordered on 12-12-2009) CM 690 Intellimouse Optical (1.1A) 1000Hz polling rate Basic, but premium round 
AudioAudio
X-Fi Titanium HD Klipsch ProMedia 2.1 (with 16 AWG Monster Cable... 
  hide details  
Reply
post #3 of 9
Quote:
Well, I guess if you use an Offset and if your multiplier goes down to x16 while idling, then it would be relatively safe because your voltage wouldn't sit that high 24/7. Even so, I personally wouldn't do it with air cooling.
How about on water cooling? Electromigration is based on both temperature and voltage.
Quote:
Either way, I don't honestly expect to get anywhere, because I BSOD (code 101) within minutes at 4.5 GHZ, 1.400 vcore. Feels like a total dud cpu.
It's the silicon lottery; chips can vary a lot. I got a decent 2600k and managed 4.6 GHz at 1.335V, though I only tested in prime for 10 minutes. It's down at 4.4GHz until I finish exams. frown.gif
Edited by Art Vanelay - 1/29/13 at 8:13pm
post #4 of 9
Quote:
Originally Posted by Art Vanelay View Post


How about on water cooling? Electromigration is based on both temperature and voltage.

 

I regret to say that I honest don't know.  Still, it's easy to set up an offset voltage, so I see no good reason to use a fixed one.

 

 

Quote:
Originally Posted by Art Vanelay View Post

It's the silicon lottery; chips can vary a lot. I got a decent 2600k and managed 4.6 GHz at 1.335V, though I only tested in prime for 10 minutes. It's down at 4.4GHz until I finish exams. frown.gif

 

This.

It's a computer!
(18 items)
 
  
CPUMotherboardGraphicsRAM
i5-2500K @ 4.5GHz (1.368-1.384V fixed voltage) ASUS P8P67 EVO B3 (UEFI ver. 1850) GTX 780 ASUS DirectCU II (1228 / 6300, 1.180V) G.SKILL Ripjaws X 8GB (2 x 4GB) 1866MHz, CL9 
Hard DriveHard DriveOptical DriveOptical Drive
250 GB Samsung 840 EVO (OS) 3 TB Toshiba P300 (storage) Samsung SH-S243N 24x DVD Burner Samsung SH-S203N 20X DVD Burner 
CoolingOSMonitorKeyboard
Thermaltake Frio Win 7 Home Premium x64 SP1 Retail AOC G2460PG (24" 1920 x 1080 144Hz G-SYNC) Filco Majestouch 104-key Cherry MX Blues w/NKRO 
PowerCaseMouseMouse Pad
Corsair HX650 (Bronze, ordered on 12-12-2009) CM 690 Intellimouse Optical (1.1A) 1000Hz polling rate Basic, but premium round 
AudioAudio
X-Fi Titanium HD Klipsch ProMedia 2.1 (with 16 AWG Monster Cable... 
  hide details  
Reply
It's a computer!
(18 items)
 
  
CPUMotherboardGraphicsRAM
i5-2500K @ 4.5GHz (1.368-1.384V fixed voltage) ASUS P8P67 EVO B3 (UEFI ver. 1850) GTX 780 ASUS DirectCU II (1228 / 6300, 1.180V) G.SKILL Ripjaws X 8GB (2 x 4GB) 1866MHz, CL9 
Hard DriveHard DriveOptical DriveOptical Drive
250 GB Samsung 840 EVO (OS) 3 TB Toshiba P300 (storage) Samsung SH-S243N 24x DVD Burner Samsung SH-S203N 20X DVD Burner 
CoolingOSMonitorKeyboard
Thermaltake Frio Win 7 Home Premium x64 SP1 Retail AOC G2460PG (24" 1920 x 1080 144Hz G-SYNC) Filco Majestouch 104-key Cherry MX Blues w/NKRO 
PowerCaseMouseMouse Pad
Corsair HX650 (Bronze, ordered on 12-12-2009) CM 690 Intellimouse Optical (1.1A) 1000Hz polling rate Basic, but premium round 
AudioAudio
X-Fi Titanium HD Klipsch ProMedia 2.1 (with 16 AWG Monster Cable... 
  hide details  
Reply
post #5 of 9
Thread Starter 
Thanks for the replies, they helped a lot.

I just realized I forgot to disable C3 and C6 after switching to offset voltage. Hours of testing wasted..

Should I leave package C state support on or off? Thanks.
s
(12 items)
 
  
CPUMotherboardGraphicsRAM
1600X 4.025 @1.45v Gigabyte K7 1080 Ti 2050/5800 16 GB Trident Z 3200-12-14-14-30-1T-44-267 1.53v 
Hard DriveHard DriveCoolingOS
Samsung 960 Evo 500 GB Samsung 840 Pro 256 GB Noctua C14 Windows 10 x64 
MonitorPowerCase
Viotek GN34C UW 1440p 100Hz Corsair HX750 Phanteks Enthoo Evolv ATX 
  hide details  
Reply
s
(12 items)
 
  
CPUMotherboardGraphicsRAM
1600X 4.025 @1.45v Gigabyte K7 1080 Ti 2050/5800 16 GB Trident Z 3200-12-14-14-30-1T-44-267 1.53v 
Hard DriveHard DriveCoolingOS
Samsung 960 Evo 500 GB Samsung 840 Pro 256 GB Noctua C14 Windows 10 x64 
MonitorPowerCase
Viotek GN34C UW 1440p 100Hz Corsair HX750 Phanteks Enthoo Evolv ATX 
  hide details  
Reply
post #6 of 9
Quote:
Originally Posted by Kanashimu View Post

Thanks for the replies, they helped a lot.

I just realized I forgot to disable C3 and C6 after switching to offset voltage. Hours of testing wasted..

Should I leave package C state support on or off? Thanks.

 

Nah, those hours weren't wasted.  The only benefit of disabling C3 and C6 is so that the idle voltage isn't too low for maintaining stability.  So disabling them doesn't change anything except for causing the idle voltage to be slightly higher.

 

I don't know what to do with the package C state support setting.  I have always guessed that it should be left enabled so that C1E works, but I haven't tested it.

It's a computer!
(18 items)
 
  
CPUMotherboardGraphicsRAM
i5-2500K @ 4.5GHz (1.368-1.384V fixed voltage) ASUS P8P67 EVO B3 (UEFI ver. 1850) GTX 780 ASUS DirectCU II (1228 / 6300, 1.180V) G.SKILL Ripjaws X 8GB (2 x 4GB) 1866MHz, CL9 
Hard DriveHard DriveOptical DriveOptical Drive
250 GB Samsung 840 EVO (OS) 3 TB Toshiba P300 (storage) Samsung SH-S243N 24x DVD Burner Samsung SH-S203N 20X DVD Burner 
CoolingOSMonitorKeyboard
Thermaltake Frio Win 7 Home Premium x64 SP1 Retail AOC G2460PG (24" 1920 x 1080 144Hz G-SYNC) Filco Majestouch 104-key Cherry MX Blues w/NKRO 
PowerCaseMouseMouse Pad
Corsair HX650 (Bronze, ordered on 12-12-2009) CM 690 Intellimouse Optical (1.1A) 1000Hz polling rate Basic, but premium round 
AudioAudio
X-Fi Titanium HD Klipsch ProMedia 2.1 (with 16 AWG Monster Cable... 
  hide details  
Reply
It's a computer!
(18 items)
 
  
CPUMotherboardGraphicsRAM
i5-2500K @ 4.5GHz (1.368-1.384V fixed voltage) ASUS P8P67 EVO B3 (UEFI ver. 1850) GTX 780 ASUS DirectCU II (1228 / 6300, 1.180V) G.SKILL Ripjaws X 8GB (2 x 4GB) 1866MHz, CL9 
Hard DriveHard DriveOptical DriveOptical Drive
250 GB Samsung 840 EVO (OS) 3 TB Toshiba P300 (storage) Samsung SH-S243N 24x DVD Burner Samsung SH-S203N 20X DVD Burner 
CoolingOSMonitorKeyboard
Thermaltake Frio Win 7 Home Premium x64 SP1 Retail AOC G2460PG (24" 1920 x 1080 144Hz G-SYNC) Filco Majestouch 104-key Cherry MX Blues w/NKRO 
PowerCaseMouseMouse Pad
Corsair HX650 (Bronze, ordered on 12-12-2009) CM 690 Intellimouse Optical (1.1A) 1000Hz polling rate Basic, but premium round 
AudioAudio
X-Fi Titanium HD Klipsch ProMedia 2.1 (with 16 AWG Monster Cable... 
  hide details  
Reply
post #7 of 9
Thread Starter 
Does the 1792 test have a higher fail rate than other prime95 tests? How is it different from blend? (assume both are custom with 90% of available memory used)

What is the 1792 test more equivalent to in terms of blend? 1 hr = ?
s
(12 items)
 
  
CPUMotherboardGraphicsRAM
1600X 4.025 @1.45v Gigabyte K7 1080 Ti 2050/5800 16 GB Trident Z 3200-12-14-14-30-1T-44-267 1.53v 
Hard DriveHard DriveCoolingOS
Samsung 960 Evo 500 GB Samsung 840 Pro 256 GB Noctua C14 Windows 10 x64 
MonitorPowerCase
Viotek GN34C UW 1440p 100Hz Corsair HX750 Phanteks Enthoo Evolv ATX 
  hide details  
Reply
s
(12 items)
 
  
CPUMotherboardGraphicsRAM
1600X 4.025 @1.45v Gigabyte K7 1080 Ti 2050/5800 16 GB Trident Z 3200-12-14-14-30-1T-44-267 1.53v 
Hard DriveHard DriveCoolingOS
Samsung 960 Evo 500 GB Samsung 840 Pro 256 GB Noctua C14 Windows 10 x64 
MonitorPowerCase
Viotek GN34C UW 1440p 100Hz Corsair HX750 Phanteks Enthoo Evolv ATX 
  hide details  
Reply
post #8 of 9
it is an fft size. that size fft is run in custom blend to make sure you can pass it rather than going roughly 8(?12?) hours to find you fail it. just a pre-emptive strike against that fft size.
w00t
(18 items)
 
   
CPUMotherboardGraphicsRAM
ph2 x20be asus m4a785-m sapphire vapor-x 4850  geil black dragon 1066 5-5-5-15-20 
Hard DriveOptical DriveCoolingOS
wd  asus dvd multi-write arctic cooling freezer 64 pro win 7 ultimate x64 
MonitorKeyboardPowerCase
samsung 42" lcd logitech wireless antec basic 500watt old dell crap 
MouseAudio
logitech wireless pioneer 7.1 
  hide details  
Reply
w00t
(18 items)
 
   
CPUMotherboardGraphicsRAM
ph2 x20be asus m4a785-m sapphire vapor-x 4850  geil black dragon 1066 5-5-5-15-20 
Hard DriveOptical DriveCoolingOS
wd  asus dvd multi-write arctic cooling freezer 64 pro win 7 ultimate x64 
MonitorKeyboardPowerCase
samsung 42" lcd logitech wireless antec basic 500watt old dell crap 
MouseAudio
logitech wireless pioneer 7.1 
  hide details  
Reply
post #9 of 9
Quote:
Originally Posted by Kanashimu View Post

Does the 1792 test have a higher fail rate than other prime95 tests? How is it different from blend? (assume both are custom with 90% of available memory used)

What is the 1792 test more equivalent to in terms of blend? 1 hr = ?

 

For Sandy Bridge (and probably also Ivy Bridge), the 1792K FFT size is one of the more common FFT sizes that fail.  However, it should only be done when you know that it failed.  The same holds true for any of the 82 FFT sizes (yeah, there are 82).

 

So, it's not comparable to Blend, Small FFTs, Large FFTs or anything.  It's not even really a test, per se.  I feel that the 1792K FFT gets recommended way too often.  It's as simple as this:  if an FFT size fails, then select Custom Blend, put that FFT size that failed in both of the boxes for "Min FFT size" and "Max FFT size", enter about 90% of your memory, and let it go for maybe 35-40 minutes.  This way you don't have to wait several hours to reach that FFT size in the Blend test.

It's a computer!
(18 items)
 
  
CPUMotherboardGraphicsRAM
i5-2500K @ 4.5GHz (1.368-1.384V fixed voltage) ASUS P8P67 EVO B3 (UEFI ver. 1850) GTX 780 ASUS DirectCU II (1228 / 6300, 1.180V) G.SKILL Ripjaws X 8GB (2 x 4GB) 1866MHz, CL9 
Hard DriveHard DriveOptical DriveOptical Drive
250 GB Samsung 840 EVO (OS) 3 TB Toshiba P300 (storage) Samsung SH-S243N 24x DVD Burner Samsung SH-S203N 20X DVD Burner 
CoolingOSMonitorKeyboard
Thermaltake Frio Win 7 Home Premium x64 SP1 Retail AOC G2460PG (24" 1920 x 1080 144Hz G-SYNC) Filco Majestouch 104-key Cherry MX Blues w/NKRO 
PowerCaseMouseMouse Pad
Corsair HX650 (Bronze, ordered on 12-12-2009) CM 690 Intellimouse Optical (1.1A) 1000Hz polling rate Basic, but premium round 
AudioAudio
X-Fi Titanium HD Klipsch ProMedia 2.1 (with 16 AWG Monster Cable... 
  hide details  
Reply
It's a computer!
(18 items)
 
  
CPUMotherboardGraphicsRAM
i5-2500K @ 4.5GHz (1.368-1.384V fixed voltage) ASUS P8P67 EVO B3 (UEFI ver. 1850) GTX 780 ASUS DirectCU II (1228 / 6300, 1.180V) G.SKILL Ripjaws X 8GB (2 x 4GB) 1866MHz, CL9 
Hard DriveHard DriveOptical DriveOptical Drive
250 GB Samsung 840 EVO (OS) 3 TB Toshiba P300 (storage) Samsung SH-S243N 24x DVD Burner Samsung SH-S203N 20X DVD Burner 
CoolingOSMonitorKeyboard
Thermaltake Frio Win 7 Home Premium x64 SP1 Retail AOC G2460PG (24" 1920 x 1080 144Hz G-SYNC) Filco Majestouch 104-key Cherry MX Blues w/NKRO 
PowerCaseMouseMouse Pad
Corsair HX650 (Bronze, ordered on 12-12-2009) CM 690 Intellimouse Optical (1.1A) 1000Hz polling rate Basic, but premium round 
AudioAudio
X-Fi Titanium HD Klipsch ProMedia 2.1 (with 16 AWG Monster Cable... 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel CPUs
Overclock.net › Forums › Intel › Intel CPUs › Limits for 2500k on air? 5Ghz? What do we know in 2013?