Upgrade to a 4790k now to save money in the future? (Update: Got it!)

[Techhog]

Bought it. Now I need to decide if I want to use the game code now or wait and see if the game is actually good/well optimized.

 

[Techhog]

I'm asking this a bit late and probably won't get a response in time, but do I need to reset my UEFI settings before swapping the CPU, or will it just detect the CPU change and adjust?

 

[Yuriman]

You shouldn't need to reset.

 

[Techhog]

You shouldn't need to reset.

Even though my CPU is overclocked and my RAM is using the XMP profile?

 

[Techhog]

You shouldn't need to reset.

Even though my CPU is overclocked and my RAM is using the XMP profile?

I should have waited for an answer to this question lol. It took a lot of tries to get it to boot properly.

Stock voltage is 1.25V? That seems really high. I might not get much out of this one...

 

[Techhog]

So, now that I know what that 1.25V actually meant...

From the looks of things, I'm stable at 4.6GHz @1.25V. That seems pretty good to me. Temperatures could be a little better though, so I'm probably going to buy a second fan for the cooler though (and be fore anyone says it, if I had picked the NH-D15 I would not have had enough clearance for my RAM; I need a 120mm fan).

 

[Techhog]

So, OCCT killed my overclock lol. It shut down my computer within a minute of trying it. Same happened at 4.5. There wasn't a bluescreen though, so it's as if the PSU was tripped or something. Is it possible that a voltage spike caused an issue somehow?

 

[jkauff]

I have the same CPU. I don't know what settings you've changed, but if you're only doing XMP and changing the multiplier, you're probably going to have to move to a fixed voltage higher than 1.25V to pass OCCT, or play around with adaptive mode settings.

My Asus Z97 Pro can do a stable 4.6 at 1.25V using adaptive mode, but that's only at idle. For anything intensive, the voltage goes well above 1.3V. The Asus board has an OC utility in the UEFI/BIOS that does a good job of optimizing the adaptive mode settings, but I don't know if ASRock has a similar utility. If not, you're going to have to do some research and experimenting.

Running at higher voltages will probably also require better cooling than you currently have.

I've found that overclocking beyond the stock 4.4 turbo doesn't make a big difference in the apps I use, so after playing around to see how far I could push it, I reverted to stock turbo--less wear and tear, and a lower electric bill.

 

[GAO]

Mine is stable on a Z87, 4.6 @1.205 VID in BIOS (adaptive), including AVX stressing. You might have got a below average chip or just need to do more work.

 

[Techhog]

I have the same CPU. I don't know what settings you've changed, but if you're only doing XMP and changing the multiplier, you're probably going to have to move to a fixed voltage higher than 1.25V to pass OCCT, or play around with adaptive mode settings.

My Asus Z97 Pro can do a stable 4.6 at 1.25V using adaptive mode, but that's only at idle. For anything intensive, the voltage goes well above 1.3V. The Asus board has an OC utility in the UEFI/BIOS that does a good job of optimizing the adaptive mode settings, but I don't know if ASRock has a similar utility. If not, you're going to have to do some research and experimenting.

Running at higher voltages will probably also require better cooling than you currently have.

I've found that overclocking beyond the stock 4.4 turbo doesn't make a big difference in the apps I use, so after playing around to see how far I could push it, I reverted to stock turbo--less wear and tear, and a lower electric bill.

I tried 1.275V, and that made no difference. I'm also seeing 88C in IBT at 1.25V (though no other stress test goes above 80C). It's ridiculous if you need a custom loop to hit 1.3V. It's possible that I used too much paste though.

Also, OCCT reports the voltage that i set as the VID and the Vcore in it is listed as ~0.9V no matter that I do. What's going on there?

 

[RussianSensation]

I tried 1.275V, and that made no difference. I'm also seeing 88C in IBT at 1.25V (though no other stress test goes above 80C). It's ridiculous if you need a custom loop to hit 1.3V. It's possible that I used too much paste though.

Why don't you test your overclock in the most demanding apps you currently run? There is a Russian review site that tested Corsair H110i GT on i7 6700K and they could only manage 4.3Ghz overclock in stress-tests using OCCT but in regular applications such as games, office apps, etc. it hits 4.5-4.6Ghz easily on a $40 air cooler.

There is absolutely no need to use synthetic power virus programs like OCCT or IBT unless you run distributed computing/rendering apps 24/7 since those actually use 99% of each CPU core/thread you have. Still even a x264 test isn't anywhere near as stressful as IBT and OCCT.

 

[Arachnotronic]

Why don't you test your overclock in the most demanding apps you currently run? There is a Russian review site that tested Corsair H110i GT on i7 6700K and they could only manage 4.3Ghz overclock in stress-tests using OCCT but in regular applications such as games, office apps, etc. it hits 4.5-4.6Ghz easily on a $40 air cooler.

There is absolutely no need to use synthetic power virus programs like OCCT or IBT unless you run distributed computing/rendering apps 24/7 since those actually use 99% of each CPU core/thread you have. Still even a x264 test isn't anywhere near as stressful as IBT and OCCT.

I don't use an overclock unless it passes the craziest stress tests at acceptable temperatures. The extra 100-200MHz isn't worth it if there's any hint of instability, IMO.

 

[Techhog]

So, it looks like my thermal paste was applied badly, so I migh have a little more voltage headroom. Not sure if I'm going to use it or not, though.

Why don't you test your overclock in the most demanding apps you currently run? There is a Russian review site that tested Corsair H110i GT on i7 6700K and they could only manage 4.3Ghz overclock in stress-tests using OCCT but in regular applications such as games, office apps, etc. it hits 4.5-4.6Ghz easily on a $40 air cooler.

There is absolutely no need to use synthetic power virus programs like OCCT or IBT unless you run distributed computing/rendering apps 24/7 since those actually use 99% of each CPU core/thread you have. Still even a x264 test isn't anywhere near as stressful as IBT and OCCT.

That sounds nice, but after I tried OCCT and the computer hard rebooted, I tried again with fixed voltage, and it hard-rebooted as OCCT was opening. That's pretty bad, so more voltage is definitely needed. I just don't know it's it's worth it to go all the way up to 1.3V to get 100MHz, 200Mhz if I'm lucky.

 

[Techhog]

So, it looks like the issue was the CPU cache ratio and the fact that LLC was disabled (since I couldn't find the setting). Toying with those, I'm back to 4.6GHz @ 1.25V. I might even be able to push it a bit more than that.

 

[RussianSensation]

I don't use an overclock unless it passes the craziest stress tests at acceptable temperatures. The extra 100-200MHz isn't worth it if there's any hint of instability, IMO.

OCCT and IBT are not real world programs. This is akin to stress testing a GPU back in the days of Furmark before thermal GPU clock throttling was added to the drivers on purpose.

Going back to Athlon XP/P4 when I got big into overclocking, I would test all my CPUs with Prime95, OCCT, etc. Over the years I've monitored real world apps and none even comes close to any of these programs. What is the point of stress testing a CPU/GPU in a power virus when no real world application is coded like that?

If it makes you feel better, keep using them, but it's not particularly logical to use these programs for measuring real world overclocking limits imo.

Over the years my opinion of a stable OC changed - stable OC for my system is the max my components can run in my apps without stability issues. I don't care if some synthetic program stresses them more because I'll never run any real world app like that.

AMD is even catching on to the times and allowing PER game profile overclocking. This makes a lot of sense since stable overclocking can be adjusted on a PER application basis.

As I said, a site achieved just a 4.3Ghz overclock on a 6700K with a $100+ AIO but in games the same CPU hits 4.6Ghz on a $40 air cooler. For most PC users, it's simply impractical/irrelevant to use OCCT for CPU testing.

 

[jkauff]

I agree. Encoding an HD video with Handbrake is a perfectly good stress test for an OC. Runs for more than an hour and uses close to 100% CPU.

If my OC works with Handbrake, it's stable.

 

[LTC8K6]

For intel chips, I use the stress tests in XTU. If it passes those, it's good to go, imo.

 

[Techhog]

It doesn't make that huge of a difference at this point tbh. I don't want to push any farther due to heat. I'm seeing as high as 68C in video rendering. IBT went up to 92C, so I'm thinking about repasting again. :/

 

[Kenmitch]

Try Asus RealBench app. I just use that and real world use testing.

 

[VirtualLarry]

OCCT and IBT are not real world programs. This is akin to stress testing a GPU back in the days of Furmark before thermal GPU clock throttling was added to the drivers on purpose.

Going back to Athlon XP/P4 when I got big into overclocking, I would test all my CPUs with Prime95, OCCT, etc. Over the years I've monitored real world apps and none even comes close to any of these programs. What is the point of stress testing a CPU/GPU in a power virus when no real world application is coded like that?

If it makes you feel better, keep using them, but it's not particularly logical to use these programs for measuring real world overclocking limits imo.

Well, some of us do DC, so we run programs akin to Prime95 24/7. For those of us that do that, I still recommend doing "extreme" stress-testing, INCLUDING "real-world" stress-testing. (Running your DC app of choice for a week or more.)

That said, my G3258 @ 4.0, 1.200v, isn't 100% stable. I get reboots, around once every two weeks. I've been running F@H lately, on both one core of the CPU, and my 7950 (98% GPU usage, I think it cleverly avoids 100% GPU utilization on purpose, so as not to starve the Windows' GUI of GPU cycles.)