LTT Forums CPU Overclocking Database!

[Jack.EXE]

I posted mine, but i could not use prime 95, i keep getting buggy versions (they always crash when I open them, Let alone start the test) so I used overdrive's stability test, which uses the same algorithm.

[MeltyMoon]

I posted mine, but i could not use prime 95, i keep getting buggy versions (they always crash when I open them, Let alone start the test) so I used overdrive's stability test, which uses the same algorithm.

Don't worry about it, as long as it is some sort of stability verification, I'm fine with it :)

[Jack L. Samuelsson]

One more Haswell for ya! Submitted.

[a1015n]

I think that an entry to be valid the PC has to be 24/7 stable. Most people run PRIME95/AIDA64 for 2-3 minutes thinking its stable. Big mistake!

For me an entry to be valid must run PRIME95/AIDA64 for at least 6 hours without any WHEA-Logger or PRIME95/AIDA64 errors then take the screen shot.

 

 

PC Enthusiast

[MeltyMoon]

I think that an entry to be valid the PC has to be 24/7 stable. Most people run PRIME95/AIDA64 for 2-3 minutes thinking its stable. Big mistake!

For me an entry to be valid must run PRIME95/AIDA64 for at least 6 hours without any WHEA-Logger or PRIME95/AIDA64 errors then take the screen shot.

 

 

PC Enthusiast

While this is true, I personally think that would deter a lot of people from making entries into the DB itself. So I'm going to assume if they are only priming it for 3 minutes now to show stability, they've primed it longer in the past. 

[Jack.EXE]

I allready tested my stability for 10 hours, so I decided to enter it in to the database, so when I saw the verification thing I set it to 1 minute just to take the screenshot.

[MeltyMoon]

I allready tested my stability for 10 hours, so I decided to enter it in to the database, so when I saw the verification thing I set it to 1 minute just to take the screenshot.

My point exactly :)

[Facety]

Lets go guys! I wanna see some haswell entries!  :lol:

 

Here is another one for you  :D

[rainbrodash666]

I will be posting a new entry here soon, I just ordered a FX6350 and ecs a990fxm-a, will shoot for 4.5ghz+, hell maybe I can beat the top 5.2 in this thread so far.

[Anult]

I got my new H100i cooler in the mail today, got it so I could reduce the temps of my overclock.

 

I'll have a form submitted soon.

[Jack.EXE]

I will be posting a new entry here soon, I just ordered a FX6350 and ecs a990fxm-a, will shoot for 4.5ghz+, hell maybe I can beat the top 5.2 in this thread so far.

 

ECS motherboards are good overclockers for the price, so go for it!

[rainbrodash666]

ECS motherboards are good overclockers for the price, so go for it!

my last ecs mobo had a amd athlon xp in it lol, it had 3 gigs of ram and 4 15gb quantam fireball hard drives lol

[ProKoN]

couple things i think should be changed when submitting a Valid result.

 

in my opinion screenshots should include

 

stress test for a minimum of 8 hours using ONE of the following

-aida64

-prime95 blend test

-Intel Extreme tuning utility (when testing intel cpus)

 

I really dont think prime95 small ffts is really relevant.

 

The goal of stress testing is to Validate system stability 

 

NOT to see how much heat you can generate under artificial load.

 

A temp monitoring tool (coretemp or many others) to display your temps under load

 

3CPUz tabs (3 instances running) CPU, Memory, Mainboard

-1 showing cpu frequency

-1 showing motherboard information including bios

-1 showing memory timings and speed

 

It would be nice to include two benchmarks so we can compare system performance.

-3d mark06 @1280x1024

-cinebench

 

Both these benchmarking tools can be found for free. 3d mark is universal and a dx9 benchmark. offers better backwards compatibility for people with old gear. I also think its better than newer iterations at  measuring real raw gpu performance.

 

These were all requirements on one of the original overclocking forums which had the largest, reliable and validated data base.. DFI-street.com. (13 years ago lol)

[MeltyMoon]

couple things i think should be changed when submitting a Valid result.

 

in my opinion screenshots should include

 

stress test for a minimum of 8 hours using ONE of the following

-aida64

-prime95 blend test

-Intel Extreme tuning utility (when testing intel cpus)

 

I really dont think prime95 small ffts is really relevant.

 

The goal of stress testing is to Validate system stability 

 

NOT to see how much heat you can generate under artificial load.

 

A temp monitoring tool (coretemp or many others) to display your temps under load

 

3CPUz tabs (3 instances running) CPU, Memory, Mainboard

-1 showing cpu frequency

-1 showing motherboard information including bios

-1 showing memory timings and speed

 

It would be nice to include two benchmarks so we can compare system performance.

-3d mark06 @1280x1024

-cinebench

 

Both these benchmarking tools can be found for free. 3d mark is universal and a dx9 benchmark. offers better backwards compatibility for people with old gear. I also think its better than newer iterations at  measuring real raw gpu performance.

 

These were all requirements on one of the original overclocking forums which had the largest, reliable and validated data base.. DFI-street.com. (13 years ago lol)

Small FFT is just what Linus would always use for stability testing in any of his overclocking videos (and I DO have a max temp tab that is mandatory, so personally I think it is the best thing to use in a CPU OC database like this), and its what I got used to as well. I no longer care what you stress test with, as long as it is a program I already know of, or can verify the effectiveness of it. Also, 8 hours + of stress testing will most likely turn away some entries to the DB due to people's time restraints, and general laziness. On top of the fact that everyone should know that not all results are guaranteed on something like a DB anyways, so dialing the exact same settings would be an AWFUL idea. This just gives you a general idea, and a goal to work towards ;) . As to your next point, ~85%+ post screenshots with cpu-z on the main tab already. And I have mandatory entry slots for your mainboard tab section. I have bios versions, motherboard model, etc. Finally I personally see the RAM tab irrelevant due to the fact that all RAM is going to end up differing with each other. All are going to have different speeds and timings. However I do see your point for having that tab, in case people lowered their speeds or raised their timings. I also like the idea of using cinebench, that may be an optional tab in the near future.  

[ProKoN]

Small FFT is just what Linus would always use for stability testing in any of his overclocking videos

check out this linus video, skip to 8:05 keep in mind haswell is amongst us.

 

 

 

"On top of the fact that everyone should know that not all results are guaranteed on something like a DB anyways, so dialing the exact same settings would be an AWFUL idea."

 

I agree 100%. best you can ensure are valid user results. All individual use cases will always vary.

 

"As to your next point, ~85%+ post screenshots with cpu-z on the main tab already. And I have mandatory entry slots for your mainboard tab section.I have bios versions, motherboard model, etc. Finally I personally see the RAM tab irrelevant due to the fact that all RAM is going to end up differing with each other"

 

Haswell "ring ratio" is displayed in the mem tab as NB frequency.  this ratio is extremely relevant to overclocking haswell, i have been fighting it for a week.

 

I think the ram tab would be nice :) Anyone else agree?

 

good for baseline reference when comparing a 3dmark or cinebench score.

 

 

"However I do see your point for having that tab, in case people lowered their speeds or raised their timings. I also like the idea of using cinebench"

 

 

 

 

 

thank you

 

 

just trying to offer some suggestions. can wait to post my results

[MeltyMoon]

check out this linus video, skip to 8:05 keep in mind haswell is amongst us.

 

 

 

"On top of the fact that everyone should know that not all results are guaranteed on something like a DB anyways, so dialing the exact same settings would be an AWFUL idea."

 

I agree 100%. best you can ensure are valid user results. All individual use cases will always vary.

 

"As to your next point, ~85%+ post screenshots with cpu-z on the main tab already. And I have mandatory entry slots for your mainboard tab section.I have bios versions, motherboard model, etc. Finally I personally see the RAM tab irrelevant due to the fact that all RAM is going to end up differing with each other"

 

Haswell "ring ratio" is displayed in the mem tab as NB frequency.  this ratio is extremely relevant to overclocking haswell, i have been fighting it for a week.

 

I think the ram tab would be nice :) Anyone else agree?

 

good for baseline reference when comparing a 3dmark or cinebench score.

 

 

"However I do see your point for having that tab, in case people lowered their speeds or raised their timings. I also like the idea of using cinebench"

 

 

 

 

 

thank you

 

 

just trying to offer some suggestions. can wait to post my results

In my defense the DB did launch before Haswell dropped :P However, I did not know so many things changed with Haswell, please forgive me for my lack of knowledge! I tend to pay more attention to AMD as is and I found the Haswell launch and lineup to be disappointing and uninteresting, so I never really followed it. I will edit the form and require all new Haswell entries to require a main tab and RAM tab in cpu-z. Users that don't comply will be flagged with an orange background. I will leave the stress testing suites as is however, being prime95 has been my go to for over-all stability and has never really let me down. I also don't want to force new software on users as well. 

[ProKoN]

how bout p95 8 hour blend test. ? its very realistic.

 

 haswell burns up under small fft's (gets smoking hot even at stock)

[MeltyMoon]

how bout p95 8 hour blend test. ? its very realistic.

 

 haswell burns up under small fft's (gets smoking hot even at stock)

I did remove small FFT from the form itself, I will edit it again and change it to blend :)

[ProKoN]

cool thanks alot. i really think it will slightly improve a  really good thing.

 

maybe do a poll on what benchmarks to include.

[Matu20]

My 3570k:

 

OC: 4.5 GHz

Volts: 1.21

Temps: Gaming -60, Prime95 75-85

[orangecat]

4.4 @1.27v prime 85c

[Lays]

I don't have a 3570k but I'll post anyways just coz I want to XD

 

Recently been doing some OC'ing.

 

 

3930k:

 

 

I'm still stability testing

 

OC 4.5GHZ

 

Vcore: 1.36 V

 

LLC: High

 

Temps:  Gaming: Haven't tested yet   Prime95: 74-77 C

[Matu20]

I don't have a 3570k but I'll post anyways just coz I want to XD

 

Recently been doing some OC'ing.

 

 

3930k:

 

 

I'm still stability testing

 

OC 4.5GHZ

 

Vcore: 1.36 V

 

LLC: High

 

Temps:  Gaming: Haven't tested yet   Prime95: 74-77 C

 

Thats a gorgeous overclock, good job.

[Lays]

Thats a gorgeous overclock, good job.

 

 

I was stable-ish with "Auto" LLC with 1.36 V, but prime95 crashed so I'm re-testing now, gonna spend some time today fine tuning it to see what I can really get lol. XD

[JordanMac]

4.3ghz at 1.32 volts, i have a pretty bad chip but oh well thats the silicon lottery.

temps: 25C at idle and 60C after prime95 for an hour.