Need advanced OC advices.

[AcMaiden]

I've been trying to overclock my Pentium G3258 to at least 4.5Ghz without going over the safe limit of Voltage of 1.3V

The best I could do was 4.4Ghz at 1.3V. And even then, I tested it with Prime 95 this night and when I woke up it had failed.

 

I have seen somewhere that at a certain point, you need to tweak other settings like the Ring core and Ring voltage and other kind of voltages.

I've seen a ton of people saying that they got theirs to 4.7Ghz. But how???

 

 

My config : http://ca.pcpartpicker.com/p/ptZ8wP

[Taswell]

Those people probably are using slightly higher end motherboards than you.

 

If you bump it to 1.35 - 40  your still pretty safe

 

I would try 1.35 and see if it makes any difference

[AcMaiden]

Those people probably are using slightly higher end motherboards than you.

 

If you bump it to 1.35 - 40  your still pretty safe

 

I would try 1.35 and see if it makes any difference

1.350V -1.400V safe? I heard that 1.300V was pretty much the max to be safe.

I know that cpu is cheap but I don't wanna fry it after 3 day of use ^^

[vSyNd1c4t3]

more voltage.

[Queek]

I've been trying to overclock my Pentium G3258 to at least 4.5Ghz without going over the safe limit of Voltage of 1.3V

The best I could do was 4.4Ghz at 1.3V. And even then, I tested it with Prime 95 this night and when I woke up it had failed.

 

I have seen somewhere that at a certain point, you need to tweak other settings like the Ring core and Ring voltage and other kind of voltages.

I've seen a ton of people saying that they got theirs to 4.7Ghz. But how???

 

 

My config : http://ca.pcpartpicker.com/p/ptZ8wP

Ring voltage is another term for the cache voltage.  The CPU cache is overclocked in a similar fashion to the core multiplier.  You could try overclocking the cache, I have seem some reports that a large discrepancy between cache and core ratios can lead to instability, but haven't personally seen this.  

 

 

more {cpu vcore} voltage.

Do this ^.  As long as your temperatures are reasonable, you can definitely go higher.  

[AcMaiden]

Ring voltage is another term for the cache voltage.  The CPU cache is overclocked in a similar fashion to the core multiplier.  You could try overclocking the cache, I have seem some reports that a large discrepancy between cache and core ratios can lead to instability, but haven't personally seen this.  

Ok thanks!

 

Do this ^.  As long as your temperatures are reasonable, you can definitely go higher.  

Define reasonable  

Currently, while testing with Prime 95 I'm at 70C (according to MSI afterburner)

I'm using the Hyper 212 Evo but it's in a mini case so I can't add any fans on the side panel because of it.

[Queek]

Ok thanks!

 

Define reasonable

Currently, while testing with Prime 95 I'm at 70C (according to MSI afterburner)

I'm using the Hyper 212 Evo but it's in a mini case so I can't add any fans on the side panel because of it.

Well that's for you to decide.  Tj max is 100°C for the G3258, so practically speaking you can go just below that, but most people stay well below tjmax.  You cpu probably won't be anywhere near that temperature during regular use (Prim95 is beyond worst case scenario for that cpu).  Maybe go up to 80°C and see where that gets you.  

[AcMaiden]

 (Prim95 is beyond worst case scenario for that cpu)

If I botleneck, it will go up to 100% use. So aint that the same thing as Prime95?

I'll be using this rig for gaming btw and from what I've seen so far at 4.0Ghz, it will still botleneck in some game.

[MeDownYou]

Ring voltage is another term for the cache voltage.  The CPU cache is overclocked in a similar fashion to the core multiplier.  You could try overclocking the cache, I have seem some reports that a large discrepancy between cache and core ratios can lead to instability, but haven't personally seen this.  

 

 

Do this ^.  As long as your temperatures are reasonable, you can definitely go higher.  

 

Cache multiplier should be kept under core multiplier by 300mhz or so otherwise instability is a potential problem (so if you raise the core multi to 45 then your cache multi can be 40-42 for sure)

[AcMaiden]

Cache multiplier should be kept under core multiplier by 300mhz or so otherwise instability is a potential problem (so if you raise the core multi to 45 then your cache multi can be 40-42 for sure)

Thanks. I was wondering why I was always crashing when I was taking it to the same value of the Core!

I'm currently testing at 4.4Ghz 1.3V - Ring 4.0Ghz 1.2V

[MeDownYou]

Thanks. I was wondering why I was always crashing when I was taking it to the same value of the Core!

I'm currently testing at 4.4Ghz 1.3V - Ring 4.0Ghz 1.2V

Yeah, it use to be a lot worse on older chips most of these ones should get close to core, but I'd just leave it as you don't gain much from it anyway and realistically between 4.4 and 4.7 isn't that big of a jump on a dual core anyway

 

as for temps if your doing prim95 don't go past 90C on prime

[Queek]

If I botleneck, it will go up to 100% use. So aint that the same thing as Prime95?

I'll be using this rig for gaming btw and from what I've seen so far at 4.0Ghz, it will still botleneck in some game.

Not necessarily.  Prime95 is a really, really hot stress test.  Your bottleneck temperatures will probably be closer to the temperatures that you get with AIDA64, at a guess.  

[AcMaiden]

Yeah, it use to be a lot worse on older chips most of these ones should get close to core, but I'd just leave it as you don't gain much from it anyway and realistically between 4.4 and 4.7 isn't that big of a jump on a dual core anyway

 

as for temps if your doing prim95 don't go past 90C on prime

 

4.4Ghz to 4.7Ghz aint going to affect perfomances much? Well from 3.2Ghz to 4.0Ghz gained me almost 30Fps in Farcry3 and Metro Last Light.

I wonder, does gaining, say 0.4Ghz, will affect performances in the same way, no matter if it's 3.2Ghz to 3.6Ghz or 4.2Ghz to 4.6Ghz?

 

Not necessarily.  Prime95 is a really, really hot stress test.  Your bottleneck temperatures will probably be closer to the temperatures that you get with AIDA64, at a guess.  

 

I was wondering, will those Prime95 heavy test will damage the chip in a few  long tests.

Also, how long would you recommend testing? I was planning 24h.

[Queek]

I was wondering, will those Prime95 heavy test will damage the chip in a few  long tests.

Also, how long would you recommend testing? I was planning 24h.

It shouldn't.  Prime95 isn't really recommended for haswell or devil's canyon chips because it doesn't actually validate your processor.  As I understand it, the instruction set it too old, so it isn't really testing the instructions that your processor will use day to day [citation needed].  Grab the trial version of AIDA64 and use it for the week to validate your overclock.  

Personally I like 24h as a duration, but everyone has their own happy place.  24h should give you a good idea if it's stable or not, but the real test will be using it.  If it crashes later down the road, add a bit more voltage.  

[AcMaiden]

It shouldn't.  Prime95 isn't really recommended for haswell or devil's canyon chips because it doesn't actually validate your processor.  As I understand it, the instruction set it too old, so it isn't really testing the instructions that your processor will use day to day [citation needed].  Grab the trial version of AIDA64 and use it for the week to validate your overclock.  

Personally I like 24h as a duration, but everyone has their own happy place.  24h should give you a good idea if it's stable or not, but the real test will be using it.  If it crashes later down the road, add a bit more voltage.  

Thanks a lot for your help. Very informative and greatly appreciated!

[MeDownYou]

4.4Ghz to 4.7Ghz aint going to affect perfomances much? Well from 3.2Ghz to 4.0Ghz gained me almost 30Fps in Farcry3 and Metro Last Light.

I wonder, does gaining, say 0.4Ghz, will affect performances in the same way, no matter if it's 3.2Ghz to 3.6Ghz or 4.2Ghz to 4.6Ghz?

 

 

I was wondering, will those Prime95 heavy test will damage the chip in a few  long tests.

Also, how long would you recommend testing? I was planning 24h.

Minimum 24 hours burn in for stability testing.

 

3.2ghz to 4ghz is a big jump in proportional speed whereas 4.4 to 4.7 isn't near as large or significant %wise

[Queek]

Thanks a lot for your help. Very informative and greatly appreciated!

Cheers!  Good luck with the rest of your overclock 

Don't forget to like the posts that were helpful and mark as solved for the next person who comes along with your problem 

[AcMaiden]

Minimum 24 hours burn in for stability testing.

 

3.2ghz to 4ghz is a big jump in proportional speed whereas 4.4 to 4.7 isn't near as large or significant %wise

Thanks for the help! Really appreciated!

[AcMaiden]

Cheers!  Good luck with the rest of your overclock 

Don't forget to like the posts that were helpful and mark as solved for the next person who comes along with your problem 

I can't choose a best answer since it came in several posts. The others will have to read it all