hey guys im doing some thorough OC'ing on my 3770k to find the min vcore for each speed, i saw a pattern where it would pas ibt and prime, and when the volts got too low it would fail ibt but run prime. i got up to 4.5ghz at 1.195v and failed ibt(if i fail ibt prime will fail also above 4.5) so i went back to 1.195, saved profile and went to 4.6, 1.24. it passed ibt and failed prime, 1.235 same. should i go back to 1.24 and step up until both test pass, or assume its stable enough for gaming and lesser usage and bump down till i fail ibt? im only doing 10 min large in place ftf etc. for prime and standard 10 passes for ibt cuz when it passes it says thats the lowest level to be considered "completely stable"... board is z77 mpower cooling is rs240 kit so temps dont worry me till 4.7+. ppl say prime is tougher than ibt but ibt seems to run about 9-12c hotter than prime. should i choose one or both, and what do i do if 1 fails and the other passes?
Ibt and prime
- Thread starter sfvirgo86
- Start date
Rustyy117
Splendid
- Oct 10, 2008
- 5,076
- 0
- 26,460
Hi,
My advice would be to stress test for at least 8 hours with Prime and 20 passes of IBT if you want to consider your overclock stable.
Even a slightly unstable overclock will cause games to crash or your PC to restart so you might aswell do it properly once and make sure it's stable.
My advice would be to stress test for at least 8 hours with Prime and 20 passes of IBT if you want to consider your overclock stable.
Even a slightly unstable overclock will cause games to crash or your PC to restart so you might aswell do it properly once and make sure it's stable.
ericjohn004
Honorable
- Oct 26, 2012
- 651
- 0
- 11,010
Your not stressing you CPU nearly enough especially being that your running such low voltages. You CPU probably isn't really stable at the voltages you specify. If you fail Intel Burn In Test then your not stable, period. There's no such thing as being stable enough or a little bit stable. Your either stable or your unstable.
What I do to check if I'm really stable is I run Intel Burn In Test on very high, and while I'm doing that, I play a demanding game for about 15 minutes. Then I continue running IBT until it's finished. After you pass this test, then try to get your voltage lower and run it again. After finding your lowest Vcore while still being fully stable, run IBT for 3-4 more hours. If the game crashes or you get a blue screen while running IBT then your not fully stable. This is the only full proof way I know of. To me if you don't do this you'll never know if your stable or not. And one of these days you'll get a blue screen while doing something important.
I've managed to get to 4.7Ghz at only 1.28Vcore. And I can run Intel Burn In Test for hours. But as soon as I play a game my PC crashes. So if I wanted to I could just run IBT and say "I'm at 4.7Ghz with only 1.28Vcore and I'm stable, wow" but I know that's not reality. In reality I need 1.3875Vcore for a 4.7Ghz TRUE overclock. This overclock can pass that test I mentioned easily. 25% of the overclocks I hear of, I'd bet any money they are not truly stable. And if they would run that test I mentioned, they would realize they weren't stable too.
What I do to check if I'm really stable is I run Intel Burn In Test on very high, and while I'm doing that, I play a demanding game for about 15 minutes. Then I continue running IBT until it's finished. After you pass this test, then try to get your voltage lower and run it again. After finding your lowest Vcore while still being fully stable, run IBT for 3-4 more hours. If the game crashes or you get a blue screen while running IBT then your not fully stable. This is the only full proof way I know of. To me if you don't do this you'll never know if your stable or not. And one of these days you'll get a blue screen while doing something important.
I've managed to get to 4.7Ghz at only 1.28Vcore. And I can run Intel Burn In Test for hours. But as soon as I play a game my PC crashes. So if I wanted to I could just run IBT and say "I'm at 4.7Ghz with only 1.28Vcore and I'm stable, wow" but I know that's not reality. In reality I need 1.3875Vcore for a 4.7Ghz TRUE overclock. This overclock can pass that test I mentioned easily. 25% of the overclocks I hear of, I'd bet any money they are not truly stable. And if they would run that test I mentioned, they would realize they weren't stable too.
I just primed 4.6 1.245v, over night after passing 20 very high on ibt, no errors no throttling. That's plenty stable for me. Those numbers u stated for 4.7 are very close to what I was seeing when I tried 4.7. I completely trust atleast 8 hrs prime and no less than 20 passes ibt. Ill try ur techniques when I'm off and have time to pick and prod, right now i can only ibt while getting rdy for wrk and prime while im at work or overnight ater i get home. Also what's the highest you've attempted and what's ur cooling im on a rs240 kit?
ericjohn004
Honorable
- Oct 26, 2012
- 651
- 0
- 11,010
Oh you passed the test I spoke of. Great, that means that your 100% stable for SURE. I'm glad you tried that. I get terrible FPS while doing this test with my 660 Ti OCed. You get bad FPS because your CPU is all tied up with IBT. So your CPU really is an important factor in graphics performance. So yeah, that's completely normal. I really wish more people would try this technique because then they would probably realize they need a little more voltage because as I said, I got to 4.7Ghz with as little as a 1.280 Vcore and passed a quick prime95 test at that. But it really wasn't stable at all. BTW, I've tried running prime95 and IBT at the same time, but it doesn't work. They cancel eachother out and your temps end up being really low compared to what they would be if you'd only use one program. So don't try that.
Let me ask you, how many GFLOPS are you getting while testing with IBT?
ericjohn004
Honorable
- Oct 26, 2012
- 651
- 0
- 11,010
My hardware is below in my sig. But if you can't read it I have 3570k@4.7Ghz 1.3875 Vcore (I readjust it a lot), EVGA GTX 660Ti OCed 1200Mhz core clock, Gigabyte Z77X-UD5H MoBo, Scythe Ninja 3 CPU Cooler (which I think works great considering it's a 40$ air cooler), Antec 902 v3 Case, Arctic Silver 5 Thermal Compound, Rosewill 850w Extreme Edition PSU. And my temps are about 90C average during IBT and 79C average during Prime95. So I figure I'm pretty safe. My MAX temp on my hottest core during IBT is 96C. This seems pretty hot but during normal use I never even get above 65C.
I'm curious, What are your MAX and average temps during IBT when your at 4.7Ghz? I'd kinda like to compare your cooler with mine. I know my coolers pretty good because a lot of people can't even consider going up to 4.7-4.8Ghz like I can. I've also reached a 4.8Ghz overclock, but it takes 1.425 V core which puts me at 102C on the hottest core during IBT. I just had to replace my MoBo because I accidentally broke it. So I'm moving up from a Gigabyte Z77 UD3H to a Gigabyte Z77 UD5H. It costs about 40$ extra but it's suppose to be the top of the line MoBo. So when I get my PC back this week I will see if the MoBo makes a difference at all in the temps and in overclocking in general.
i get 85-91, i've seen ppl get much higher. i think my ram is the culprit there, i have the g skill value 1x 8gb that came with the board and it will only run at 1333. my 4.7 took around 1.35 i think(cant remember, have to look at bios prof). im now priming(since 4am) 4.8 at 1.395, and ill try some fc3 after posting. my temps spiked to 84 when i started prime and settled at 74 after about 10 secs
ericjohn004
Honorable
- Oct 26, 2012
- 651
- 0
- 11,010
I'm more in the neighborhood of 89-96C. So Wow those are some pretty good temps. What type of CPU cooler are you using and how much did you pay for it. Your chip does seem like it uses less voltage than mine so that's probably the difference. Your only using 1.35 for 4.7Ghz, that's really good. The lowest I can get it to still pass my test is 1.370 Vcore. But I like to keep it a little higher than whats needed in order to be more stable as long as it doesn't affect temps too much.
How many GFLOPS are you getting at 4.7Ghz while running IBT?
ericjohn004
Honorable
- Oct 26, 2012
- 651
- 0
- 11,010
Are you sure that it's only 85-91GFLOPS? You gotta test using VERY HIGH to get the numbers I'm getting. I get anywhere between 128Gflops and 130Gflops and the test is normally 59 seconds long. A guy I talked to had a 2500k@5.0Ghz and he got around 126Gflops on very high with IBT. You should be getting right around that area. What do you use to benchmark your CPU? We can compare that way. I like to use Cinebench 11.5, PassMark8, 3DMark11 Physics test and SuperPI 1M. In Cinebench 11.5 I score an 8.0, my passmark8 CPU score is 9,700, my 3DMark11 Physics score is 9450, and my SuperPI 1M score is 7.82 seconds. I'm just curious how close these chips are to one and other. So far, on Passmark8, I haven't found a single 3570k that can top my score. If your looking for a great full system benchmark, I love Passmark 8.0. It's free for 30 days and after 30 days just delete it and re-download it. And superPI and Cinebench are free.
Don't use PCMark7. It's absolutely no good. A 3570k with HD4000 graphics has a lot higher score than I get with a GTX 660 Ti overclocked. Because the test gives absurdly high scrores when using quick sync, which HD4000 has. So a 3570k with HD4000 graphics will beat a 3930k with a GTX 680. Weird right?
ericjohn004
Honorable
- Oct 26, 2012
- 651
- 0
- 11,010
Hey sorry, I lost track of this thread for a while. Running Cinebench I get like a 7.97 or 8.0 usually. Ima give you some other benchmark numbers so we can compare. 3DMark11 physics test I get 9550. PassMark 8.0 CPU Mark I get a 9700 total. A 1M run of SuperPI takes 7.72 seconds for 1 thread. I think that's all the main benchmarks. Let me know what you score.
ericjohn004
Honorable
- Oct 26, 2012
- 651
- 0
- 11,010
Wow how the hell are you getting a 9.71 cinebench. That's 3770k territory. You'd have to be overclocked to 5.5Ghz to get that score. Are you using Cinebench 11.5? or 10? BTW I have some G. Skill Ares 2133Mhz RAM and my benchmarks went up a fair amount when switching from 1600Mhz to 2133Mhz. Like Cinebench went up from 7.79 to 8.0. And PassMark8 went up from 9450 to 9700 just on the CPU test. My 3DMark11 physics score went up from 9200 to 9550. So you can't tell me RAM doesn't help. Some people say it doesn't help you at all. But I beg to differ. I can only imagine how much your scores will go up going from some cheapo RAM to some G. SKill Ares. Check out the G. SKill Ares 2133Mhz for real. They are only 129.99 for 16GB of it. To me that's more than a fair price. If you only need 8GB it may only cost you 65 bucks. The reviews for the Ares RAM are all great. I've read every one of them. They are right up there with the best of the best. Trust me on this.
ericjohn004
Honorable
- Oct 26, 2012
- 651
- 0
- 11,010
Oh I thought you had a 3570k. Well that makes more sense. Definately save up and get you 8GB of some decent RAM. I didnt mean to offend you or anything when I called your RAM cheapo. I thought I was just pointing out the obvious. Some really good cheaper RAM is the Kingston HyperX Genisis 1600Mhz. You can get 8GB for around 40 bucks. And if you want 16GB later you can just buy another 2 sticks. But personally I would go for the 2133Mhz. It's almost the same price but you can get a whole lot faster speed. I know for sure you can find 8GB of 2133 RAM for 50 bucks. You don't have one of those big air coolers either so you can get any kind of RAM you want.
Facebook
Twitter
Instagram