Ring Ratio? Huh

[AMD Lover]

So doe's ring ratio affect the performance of the PC that much? Do i need an exact one to one ratio if im trying to get it one to one and if i am and its not stable do i increase the ring voltage or the cpu voltage?

[X1XNobleX1X]

So doe's ring ratio affect the performance of the PC that much? Do i need an exact one to one ratio if im trying to get it one to one and if i am and its not stable do i increase the ring voltage or the cpu voltage?

Whats the ring ratio?

[Vitalius]

wat

No really. Normally I try to post constructively, but .... wat

[AMD Lover]

wat

No really. Normally I try to post constructively, but .... wat

Ring bus jog any memory's, idk, its new to Haswell, i can't even find anything about it but Linus mentions it in his Haswell OC Guide.

[Glenwing]

The ring bus is just part of the theory on how that architecture operates, it's not a "feature" or "spec".  Don't worry about it.

[AMD Lover]

The ring bus is just part of the theory on how that architecture operates, it's not a "feature" or "spec".  Don't worry about it.

Thanks

[deathjester]

If you mean the Cache? Yes it helps. Raising VCCIN, rinbgus also increases temps. Ring bus people usually run around .10 lower then their vcore when dialing in their chips.  VCCIN is different on every chip. Some need 1.9 to even do 3900 cache and a stable oc on the clock as well. My friend needed 1.9 VCCIN on the same board and I needed 1.776. Haswell's have huge swings EVERYWHERE.

 

Clock>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>Memory>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>Cache

 

4.5/4.0 will kick the crap out of 4.3/4.3

 

If you can clock the cache higher and still keep temps down on the max clock of the chip? Do it. If not? Who cares. 3900 4000 cache is plenty, especially on air.  My cinebench scores stomp on 1:1's with higher clock rates from oc'ing my memory to 2133 from 1600. I saw way more performance in benchmarks OC'ing memory then I did trying to get 1:1 cache or 100mhz cache or "within 300" crap. I actually saw about a 1.5 fps increase in a cpu bound game (Guild Wars 2) with a memory oc as well.

 

I have no idea where people came up with this 1:1, within 300mhz garbage (prob the same people who say adaptive voltage is awesome on air and say stress test with it, and 1.4-1.5 volts is good to go, and 90C temps are ok) but it sure doesn't hold to form in benchmarks or actual performance. No I am not saying you have to OC memory. I am saying that clock is everything. Memory is "ok" in benchmarks. Cache is nothing special. 

 

To test the cache OC use aida 64 and just click cache. Other ways of testing are way too lengthy and you will go crazy trying to dial it in over days. Crashing on cache 9 hours in? Just sucks. Many people just set cache to 3500 when dialing in a clock, because otherwise it makes stability testing a nightmare.

 

You might also find (and every chip is different) that by just raising V core and getting higher core clocks, you can run higher cache. For me at 1.150 V @ 4.3 I could only run 3900 cache. With just 1.776 VCCIN and 1.22 cpu voltage @ 4.5? I could raise my cache to 4000.

 

If your motherboard set it to 3900 (auto) and you don't want to mess with it? That is fine to. Cache is a gigantic pain in the butt, and offers the least performance of all the OC's. 

 

If all this wasn't confusing enough? You have this as well. You want your VCCIN to be within a certain range of your CPU voltage. If you aren't in that range? You can harm the chip. So these people advising "just go to 1.9 vccin blah blah"? Are wrong.

 

http://rog.asus.com/244672013/labels/featured/introduction-to-fully-integrated-voltage-regulators-fivr-on-maximus-vi/

 

So attempting 1:1 and setting some voltages to high in relation to other voltages? Can F your chip up.