Intel Haswell Overclocking Clubhouse.

[EarthDog]

End voltage is too ambiguous due to its dependency on the workload. I think everything posted to the chart should be static voltage settings required for stability. For me, 1.35v is the static voltage needed for a stable 4.7GHz, including AVX workloads. 1.325v on a 4.6GHz cache, although I don't have a cache voltage reading to see what that manual setting actually yields.

So not knowing at all is less ambiguous ? The difference between 1.325 and 1.35 is a lot different than an unknown value in stock voltage (I have seen .95 load to 1.175 personally) +.XXv.

Give the actual load voltage on what is STABLE across all 'workloads'.

 

[Womper]

So not knowing at all is less ambiguous ? The difference between 1.325 and 1.35 is a lot different than an unknown value in stock voltage (I have seen .95 load to 1.175 personally) +.XXv.

Give the actual load voltage on what is STABLE across all 'workloads'.

Let me rephrase what I said: end voltage is ambiguous, therefore report your static voltage setting because it is the least ambiguous. I originally thought reporting offset voltages under load would be comparable, but as I learned:

Nope, but you're using an offset voltage instead of a consistent one. For initial testing, it's better to test with a set voltage, find stability, and then take that knowledge to apply the offset.


When the multi changes, the voltage changes, as each multi has it's own VID associated with it. As well, there are other conditions that can adjust that VID, so rather than dealing with that for testing, simply avoiding that is best.

That's why I didn't include your OC voltage in the chart. It really could be anything, and as you can see, as load changes, so does the voltage, so who's to say what you are really running?

So, for the chart on the front page, I say:
I am stable at 4.7GHz core/4.6GHz cache using the following core/cache static voltage settings: 1.35v static core voltage, 1.325v static cache voltage.

 

[PolRoger]

Wow, thats a lot

What codes on post led?

If it stuck @ "78" or "71" mean cpu half dead with pci or pch lane burnt.

"00" means totally rip.

Hey Dumo,

What does half dead cpu really mean??... Your chip will try and post, attempt to go through memory training but will still hang with a "78" or a "71" code? Where as totally dead chip means no posting... Just straight to a "00" code? What happens to half dead cpu? Do they ever come back?

 

[dumo]

Hey Dumo,

What does half dead cpu really mean??... Your chip will try and post, attempt to go through memory training but will still hang with a "78" or a "71" code? Where as totally dead chip means no posting... Just straight to a "00" code? What happens to half dead cpu? Do they ever come back?

It won't come back, just stay "78" stuck @ pci initialization. Tested on 2 boards same results.

 

[PolRoger]

It seems that my chip will no longer run/boot at 2666 memory speed using 4x4gb (full bank) with CR 1 (1T)? Recently it always seem to get stuck in memory training boot loop or hang with a "78" code or something similar even when running at stock settings. :shadedshu

However, if I drop back to CR 2 it will still boot/run 2666 4x4gb and it does seem to run 2x4gb 2666 with CR 1. I think this pattern may now also hold true at 2600 speeds but when dropping back down to 2400/2133... the chip still runs/maintains CR 1 @ full bank.

 

[TheHunter]

^
Yeah that's what I've read for G.Skill Trident CL10 2400mhz.

It will oc to ~ 2520mhz anything higher needs 2T.

 

[Ponteral]

bacth

Does anybody experience of L312B350? It seems as awesome chip on other forums. Thx

 

[EarthDog]

Get off the batch thing people... it really is a crap shoot.

 

[Grnfinger]

Does anybody experience of L312B350? It seems as awesome chip on other forums. Thx

I have a 312B341, it's seems decent so far but I haven't pushed it beyond 4500MHz

Get off the batch thing people... it really is a crap shoot.

That was not true for the Q6600 or the E8400
certain batches were proven better clockers than others

 

[EarthDog]

Also true. But I wasnt talking about those CPUs in a Haswell thread.

But since Sandybridge and especially Ivybridge, it has meant little to nothing.

 

[Grnfinger]

yeah you make a valid point

 

[PolRoger]

Previously I hadn't spent much time testing the two working bclk straps (125/167) but I'm liking this current overclock at 4.66Ghz DDR3-2666. Not a single random BSOD once dialed in either. Makes me think that some of my random BSOD's that occur when dialing in and running 2133/2400 full bank @1T under long term stress might be due to BIOS/platform immaturity for the 1T command rate?

~21hrs. "crunching" Rosetta 100% load:

 

[Womper]

Previously I hadn't spent much time testing the two working bclk straps (125/167) but I'm liking this current overclock at 4.66Ghz DDR3-2666. Not a single random BSOD once dialed in either. Makes me think that some of my random BSOD's that occur when dialing in and running 2133/2400 full bank @1T under long term stress might be due to BIOS/platform immaturity for the 1T command rate?

~21hrs. "crunching" Rosetta 100% load:

I still get the random BSOD 0x124 every now and then under game loads or stress loads. I set VRM to "Current Balance" and added a 200mm fan on the side of my case to put a breeze over the motherboard, but still hit the BSOD. I'm going to revisit additional VCC SA, DIO, AIO, and Vin voltages, I believe the sandforce-related BSOD invalidates my previous trials.

I'm seeing a loose correlation between this BSOD and higher ambient temps; are there any other temperature sensitive power delivery components I can tweak from the BIOS? Maybe this is just platform immaturity, or maybe my RAM wasn't properly qualified by Asus.

 

[PolRoger]

^^^ Hah! It was wishful thinking on my part...

It froze/hanged around ~24hr mark and since then I've actually had a few random BSOD. I've been tweaking the settings and I'm now looking to try and pass 32/40/48hrs.

 

[MxPhenom 216]

I have a 312B341, it's seems decent so far but I haven't pushed it beyond 4500MHz



That was not true for the Q6600 or the E8400
certain batches were proven better clockers than others

the 4770k that is coming today is the same batch number as yours. But yeah since Sandy Bridge, batch numbers met literally nothing. It is just a big silicon lottery at this point. Back in the day though, batch numbers meant a lot.

 

[TheHunter]

I still get the random BSOD 0x124 every now and then under game loads or stress loads. I set VRM to "Current Balance" and added a 200mm fan on the side of my case to put a breeze over the motherboard, but still hit the BSOD. I'm going to revisit additional VCC SA, DIO, AIO, and Vin voltages, I believe the sandforce-related BSOD invalidates my previous trials.

I'm seeing a loose correlation between this BSOD and higher ambient temps; are there any other temperature sensitive power delivery components I can tweak from the BIOS? Maybe this is just platform immaturity, or maybe my RAM wasn't properly qualified by Asus.

In digi+ keep all at auto, and maybe manually select LLC to 5 or 6.

try lower ring cache bus. And if you use offest turn off adv c-states (c3,c6,c7)

I had similar bsod in cpu bound games (Bf3 64player, Rift, MaxPyne3, Grid2,..), it kept happening at adaptive cpu offset voltage (max so far cpu +0.085v, turbo +0.004v and still).

Im now at standard offset (+0.090v) and so far so good, but this additional boost looks bad.. I mean it can boost to 1.25v for longer times by GPU caps viewer, using cpu openCL 1.2 cpu-postFX test,. Otherwise its at 1.176v

But I didnt notice this extra voltage boost in games yet, hopefully it stays this way

 

[Grnfinger]

the 4770k that is coming today is the same batch number as yours. But yeah since Sandy Bridge, batch numbers met literally nothing. It is just a big silicon lottery at this point. Back in the day though, batch numbers meant a lot.

Would be interested in seeing how your chip performs

 

[radrok]

batch numbers meant a lot.

I remember that waaaaay before even the manufacturing site meant (costa rica etc) a ton

 

[Womper]

In digi+ keep all at auto, and maybe manually select LLC to 5 or 6.

try lower ring cache bus. And if you use offest turn off adv c-states (c3,c6,c7)

I had similar bsod in cpu bound games (Bf3 64player, Rift, MaxPyne3, Grid2,..), it kept happening at adaptive cpu offset voltage (max so far cpu +0.085v, turbo +0.004v and still).

Im now at standard offset (+0.090v) and so far so good, but this additional boost looks bad.. I mean it can boost to 1.25v for longer times by GPU caps viewer, using cpu openCL 1.2 cpu-postFX test,. Otherwise its at 1.176v

But I didnt notice this extra voltage boost in games yet, hopefully it stays this way

So...I just crapped a brick after getting to the bottom of this Adaptive Voltage mode. I've been wondering all this time why the "Additional Turbo Voltage" field would bluescreen even if I set the field's value the same as I would the "Offset" field.

Here it is: http://rog.asus.com/forum/showthread.php?33680-Beginner-Adaptive-Voltage-4pin-CPU-power/page2

"I feel so dumb right now. It was a language barrier that caused all my trouble. The way the Adaptive mode is worded in the UEFI confused me. It said "Additional Turbo voltage" so I thought it meant I had to ADD voltage to the default VID, eg: It was 0.992V so in the "Additional Turbo" box I wrote 0.183 --- and that limited the VCORE to 0.183 lol."

Am I the only one in this thread who feels similarly dumb/soiled?

I know what I'm trying when I get home...

 

[TheHunter]

So...I just crapped a brick after getting to the bottom of this Adaptive Voltage mode. I've been wondering all this time why the "Additional Turbo Voltage" field would bluescreen even if I set the field's value the same as I would the "Offset" field.

Here it is: http://rog.asus.com/forum/showthread.php?33680-Beginner-Adaptive-Voltage-4pin-CPU-power/page2

"I feel so dumb right now. It was a language barrier that caused all my trouble. The way the Adaptive mode is worded in the UEFI confused me. It said "Additional Turbo voltage" so I thought it meant I had to ADD voltage to the default VID, eg: It was 0.992V so in the "Additional Turbo" box I wrote 0.183 --- and that limited the VCORE to 0.183 lol."

Am I the only one in this thread who feels similarly dumb/soiled?

I know what I'm trying when I get home...

Hm, so lets say I want 1.178v
normal
cpu offset +0.090v

adaptive
cpu offset +0.085v
turbo offset +0.005v
total offset +0.090v

In both cases its ~1.178v, but for some reason adaptive isn't stable. Although I didnt try cpu offset +0.090v and turbo +0.005v yet


btw is this turbo offset that boosts to 1.25v in normal offset mode? I think by adaptive it didnt go so high

 

[MxPhenom 216]

Would be interested in seeing how your chip performs

Dave seems to believe it's on the upper end of average.

 

[Shou Miko]

Hm, so lets say I want 1.178v
normal
cpu offset +0.090v

adaptive
cpu offset +0.085v
turbo offset +0.005v
total offset +0.090v

In both cases its ~1.178v, but for some reason adaptive isn't stable. Although I didnt try cpu offset +0.090v and turbo +0.005v yet


btw is this turbo offset that boosts to 1.25v in normal offset mode? I think by adaptive it didnt go so high

nice voltage, but when i touce offset mode no matter if it's offset or adaptive mode i get over 1.2xxv

 

[Womper]

Hm, so lets say I want 1.178v
normal
cpu offset +0.090v

adaptive
cpu offset +0.085v
turbo offset +0.005v
total offset +0.090v

In both cases its ~1.178v, but for some reason adaptive isn't stable. Although I didnt try cpu offset +0.090v and turbo +0.005v yet


btw is this turbo offset that boosts to 1.25v in normal offset mode? I think by adaptive it didnt go so high

For Asus boards, "Additional turbo mode voltage" isn't an offset. It's the absolute load voltage you want to see. If you want 1.178v, then for adaptive mode you should set:

Cpu offset +0.000v (or +0.001v if that's the lowest option)
Additional turbo mode 1.178v
total 1.178v

With this setting, you will see 1.178v for non-AVX loads, and ~1.278v for AVX loads (whenever the CPU is above 3.5GHz in these cases). When you aren't running anything, your CPU will be at 3.5GHz or less and use stock voltages.

Last night a had a little time to convert my offset voltage to adaptive. For my 4.7GHz core speed, I changed from a +0.16v offset voltage to a +0.001 offset and 1.264v Turbo mode voltage. For the 4.6GHz cache speed, I changed from a +0.3v offset to a +0.001 offset and 1.325v Turbo mode voltage. Here's what I noticed:
* Peak non-AVX load voltage is virtually identical between +0.16v offset and Adaptive
* Peak AVX load voltage might be slightly higher- I'm not sure, but I think it was +0.006 higher using Adaptive. Not really a big deal.
* Using adaptive, the CPU maintains the ~1.264v load voltage for all frequency bins between 3.5GHz and 4.7GHz. This differs from using a +0.16v offset, where the voltage decreases for frequencies less than the 4.7GHz max, or for lighter workloads at 4.7GHz. In theory, this could increase stability in the scenarios where an intermediate bin like 4.4GHz gets too low a voltage and crashes, or when 4.7GHz needs the full 1.264v even though the workload is light.

So far, Adaptive mode seems just as stable as using an Offset. It's nice seeing ~0.7v when idling at 800MHz.


And I'm still looking for someone to find out whether Cache voltage jumps up by +0.1v during AVX loads like the Core voltage does when using Offset or Adaptive.

 

[cadaveca]

I'll check for you this afternoon. Thanks for adding that excellent info.


Been swamped here lately(literally, under water), many things I wanted to get done got dropped completely. Today is the first day I'm finally back on track. I got a G.Skill kit in my memory testing rig, once it's done the current Memtest runs, I'll check about cache voltage for you.

 

[petedread]

Hi Guys. I wanna tell you my experience. I had 4770k BATCH L313B846, and it was total crap..

At load in stock clocks it had 1.169V. my max was 4.4 GHz at 1.344V.

I loaded in win at 4.5Ghz, 4.6Ghz, nope. So I decided to send it back, and look for another. I'm looking now batch L310. I mean those - L310B487, L310B488, L310B491 and L310B492, which seem to be the best.

Please if you know about other good batches, let me know, Thank you.




What's the batch of this CPU? If I had something like that....

My chip is almost exactly the same. Except my max OC is 4.3x100mhz @1.224v (CPU-Z) cache 4.3 @1.100v, no mem OC. (testing 2133 up from 1600 rated ram).
1.35v for 4.4 means I hit a thermal wall instantly.

Something strange going on, I can't game or run Vally with out PC crash. With or with out OC.