Intel Haswell Overclocking Clubhouse.

[cadaveca]

Still working my way through memory clocking, I just wanted to thank you guys for the continued input. I am seeing similar stuff to you guys, overall, will have an update in a few days, busy working on review stuff right now.

 

[Womper]

Just ran Cinebench 11.5 and yes cpu voltage didnt pass 1.155v (adaptive offset +0.060v, turbo +0.005v) ring bus voltage also stayed ~ 1.117v max


Although before when I edited this post I got another 0x124 (cpu adaptive offset +0.055), i was uploading photos and checking my imgur album bammm bsod, now that i raised to 0.060v its ok! I really hope this is it., gonna play some RIFT to see what's up


EDIT2: nope 0x124 in Rift, I now raised to similar 1.17v with offset +0.070v (turbo +0.005v) >> 1.166v.. But by the looks of it i will probably need anther 0.005v by cpu so it makes 1.17v (btw its stable at fixed 1.17v).

45x for me needs somewhere between +0.050 and +0.075 offset, so it sounds reasonable. Is the turbo voltage actually helping with your stability? I tried decreasing offset and compensating with turbo voltage, but it always ends up crashing at some point until I'm back to the "known good" offset anyways. Turbo voltage has been useless so far.

I'd love to know how adaptive really works. If it requires the same offset for stability, then is Adaptive with 0v turbo the same as normal "Offset" mode? Sure looks like it from my testing.

Most guides seem to indicate that Adaptive only affects overclocking regions. I interpret that as "only affects clock speed bins that are higher than stock." I'm guessing this is not the case; rather it only affects the highest speed bin (Turbo, duh). Which means you need an offset to maintain stability at speed bins between stock max and your overclock turbo. This offset, at least for me, exceeds the manual voltage I need for load anyways.

Is there a way to specify the speed bins allowed? At 4.8GHz, I see core speed all the way up and down the 4GHz range. Adaptive would work great if I could tell the CPU to run at 3.5 or 3.9GHz, and turbo only to 4.8GHz with no steps in between that require all of that offset voltage.

 

[TheHunter]

Yeah I ended up at +0.075v which makes exactly 1.170v.

Idk about turbo though, it looks like it doesnt make any diff., i lowered it to +0.003v for now, this is what intel says about offset vs adaptive.

Offset

Adaptive

Fixed

 

[HammerON]

New MB:

 

[Womper]

After a couple of emails back and forth with ASUS tech support, they finally said that the lack of cache voltage reading is a chipset limitation. And the only reason I didn't buy the Deluxe was because I already had a wireless card, and they seemed identical otherwise. Looks like I'll just have to hope that my cache doesn't melt with a +0.25v offset.

I also reported the known issue with my Vertex LE (Sandforce 1500), and they forwarded it the engineers. No news there I guess.

Anyways, at +0.25v cache offset, 4.6GHz cache speed seems stable. I did an Aida CPU/Cache/Memory stress test for an hour with 4.6GHz core speeds (at high voltage to ensure cores are stable). Also saw no issues with a few hours of gaming. +0.20v cache offset will freeze pretty quickly, so that's my lower bound (probably around 1.3v, which will freeze quickly also under load). Note that Aida temps are low since I skipped the FPU stress test.

 

[cadaveca]

After a couple of emails back and forth with ASUS tech support, they finally said that the lack of cache voltage reading is a chipset limitation. And the only reason I didn't buy the Deluxe was because I already had a wireless card, and they seemed identical otherwise. Looks like I'll just have to hope that my cache doesn't melt with a +0.25v offset.

I also reported the known issue with my Vertex LE (Sandforce 1500), and they forwarded it the engineers. No news there I guess.

Anyways, at +0.25v cache offset, 4.6GHz cache speed seems stable. I did an Aida CPU/Cache/Memory stress test for an hour with 4.6GHz core speeds (at high voltage to ensure cores are stable). Also saw no issues with a few hours of gaming. +0.20v cache offset will freeze pretty quickly, so that's my lower bound (probably around 1.3v, which will freeze quickly also under load). Note that Aida temps are low since I skipped the FPU stress test.

View attachment 51629

0.25V offset should give you around 1.3V, given data about known CPUs and were Cache voltage sits in relation to CPU voltage(roughly +/- 0.050V). from the info I have, that's relatively safe, but border-line on the high side.

 

[MxPhenom 216]

New MB:
http://img.techpowerup.org/130625/ROG2.jpg
http://img.techpowerup.org/130625/ROG.jpg

You must have money coming from every orifice of your body. Just got 4770k and then the MSI GD65 board, and now this board lol.

 

[EarthDog]

Or more money than sense... I hope he is into extreme cooling!

 

[TheHunter]

Did anyone test adaptive voltage like so;

For example my scenario

cpu offset +0.075v
turbo offset auto (or +0.002v) >> btw I tried at auto and its the same idle voltage like by +0.002.

now

cpu offset auto (or lets say +0.002v so it doesnt over volt like crazy)
turbo offset +0.075v


would this be safe to try?


EDIT: I think i found my answer
Newegg TV: ASUS Z87 4770K Overclocking Tutorial & ...




EDIT2: nah not working, only cpu offset at 0.075 and turbo at min 0.001v works and boots. Other way around no go.

 

[Kursah]

So is setting to adaptive even useful then? I was really hoping so after watching the newegg vids and buying an Asus Z87-Pro and 4770k..that's not the only reason, but from how I understood the explanation that it'll use stock voltage at stock speeds...or in reality is it just idle? Or does it scale?

Then it seems turbo offset seems useless to everyone here, which sucks...I was hoping it'd take over after 3.5Ghz for the OC voltage. I guess if that was the intention it'd be nice.

 

[cadaveca]

Then it seems turbo offset seems useless to everyone here

I think that might be because everyone is running all cores at the same speed. This is modifying Turbo profile even further than just upping the multis, so I would understand why there might be issues. This is one thing no OC guide covers, either. But perhaps I missed it? Did anyone see anything relevant?

 

[springs113]

Hey Dave a lil off topic but any 8gb/16gb memory kit suggestion

 

[cadaveca]

G.Skill 2666 MHz 4x4 GB C11. Many boards incorporate T-Topology stuff that makes using 4 DIMMs better than 2, and this kit offers lots of flexibility for OC.

That said, many other 2666 MHZ C11 kits will offer the same flexibility, but without having them in my hands to play with, there's not much I can say directly.

I also have Dominator Platinum 2666 MHz 4x4 GB C10 sticks, and like them a lot as well.

Each kit uses different ICs, and the same thing applies, other 2666 MHz C10 kits can offer the same, but 2666 C10 does come at a price premium.

I have several other kits on-hand that I am testing with now, reviews will be coming very soon.

typically, anything memory-wise affecting performance...still investigating. I'm not done yet, have all dividers to test, different stick densities, etc...and I am testing about 20 different apps. Until I finish and get my graphs made, there's not much I can say about performance yet. It takes a lot of time to run benchmarks properly, I started benching this motherboard @ about 9 am today, still testing now, and that's just basic benchmarks for me.

 

[springs113]

G.Skill 2666 MHz 4x4 GB C11. Many boards incorporate T-Topology stuff that makes using 4 DIMMs better than 2, and this kit offers lots of flexibility for OC.

That said, many other 2666 MHZ C11 kits will offer the same flexibility, but without having them in my hands to play with, there's not much I can say directly.

I also have Dominator Platinum 2666 MHz 4x4 GB C10 sticks, and like them a lot as well.

Each kit uses different ICs, and the same thing applies, other 2666 MHz C10 kits can offer the same, but 2666 C10 does come at a price premium.

I have several other kits on-hand that I am testing with now, reviews will be coming very soon.

typically, anything memory-wise affecting performance...still investigating. I'm not done yet, have all dividers to test, different stick densities, etc...and I am testing about 20 different apps. Until I finish and get my graphs made, there's not much I can say about performance yet. It takes a lot of time to run benchmarks properly, I started benching this motherboard @ about 9 am today, still testing now, and that's just basic benchmarks for me.

my discount ends in about 23 hours, i can only use 2 slots due to my aio blocking slot 1 so 2x kits please max budget 150(still haven't decided on gpu yet).

 

[Jstn7477]

Just wanted to stop by and mention that I have finally gotten my system 99% stable at 43x/42x 1.16v fixed voltage. 43x ring is unstable regardless of the voltage and 44x core needs 1.22+ volts to achieve and that is quite a waste. My chip reached about 170 BCLK (must use LC PLL for PCIe, SB PLL option doesn't work at all) but it was unstable at the same clock frequencies even at 126.5 BCLK so it's back at 100MHz. Maximum core temp is 92c under WCG with my TPC-812 push-pull and MX4 paste. Forget about IBT, hits 102c almost right away.

I also exchanged my RAM from the two sets of mid-late 2011 CL11 2x4GB Ripjaws for some late 2012 Sniper 2x8GB CL10 1866 RAM running at 2000MHz 10-11-10-30/2T and 180 tRFC for now (default was 243). I felt the newer and higher density RAM might be a little better for my system and the speed/timing difference seemed moot.

Anyway, I think my next upgrade is going to be this Fractal Design Define XL for $99 (29% off) and free shipping. Fractal Design Define XL R2 FD-CA-DEF-XL-R2-BL Bla... After using "under $50" cases for years, I think it's about time I treated myself to something awesome.

 

[cadaveca]

my discount ends in about 23 hours, i can only use 2 slots due to my aio blocking slot 1 so 2x kits please max budget 150(still haven't decided on gpu yet).

$150 should get you 16 GB of 2133 or 2400 MHz. Check out G.SKill, they seem to have lower prices than most others, often. Last time I bought ram I spent $525 on 4 sticks. I have no problems paying for high-end memory.

 

[PolRoger]

After a couple of emails back and forth with ASUS tech support, they finally said that the lack of cache voltage reading is a chipset limitation. And the only reason I didn't buy the Deluxe was because I already had a wireless card, and they seemed identical otherwise. Looks like I'll just have to hope that my cache doesn't melt with a +0.25v offset.

Anyways, at +0.25v cache offset, 4.6GHz cache speed seems stable. I did an Aida CPU/Cache/Memory stress test for an hour with 4.6GHz core speeds (at high voltage to ensure cores are stable). Also saw no issues with a few hours of gaming. +0.20v cache offset will freeze pretty quickly, so that's my lower bound (probably around 1.3v, which will freeze quickly also under load). Note that Aida temps are low since I skipped the FPU stress test.

View attachment 51629

Is overclocking your cpu multi/cache with a 1:1 ratio really a necessary requirement for daily type usage? I find running higher cache speeds also require greater cache voltages to stabilize... What about dropping cache multi down some? I'm currently testing (dialing in) a 46x/43x overclock for possible use while "crunching" Rosetta/WCG... etc. I might even consider going lower to 41x/42x. I believe that the performance hit for running lower cache speeds is rather minimal?

 

[EarthDog]

my discount ends in about 23 hours, i can only use 2 slots due to my aio blocking slot 1 so 2x kits please max budget 150(still haven't decided on gpu yet).

Unless you are benchmarking, DDR3 1600/1866 CL9. Anything else is, for the most part, wasting money as there is little to no real world performance gains going higher with a discrete GPU. Only in rare bandwidth intensive apps do you see a performance increase worth the cost difference.

 

[radrok]

$150 should get you 16 GB of 2133 or 2400 MHz. Check out G.SKill, they seem to have lower prices than most others, often. Last time I bought ram I spent $525 on 4 sticks. I have no problems paying for high-end memory.

Quick hijacking advice then, asking you memory nuts

I need two kits of 32GB memory which one would you pick between these three?

http://www.gskill.com/en/product/f3-2400c10q-32gtx

http://www.corsair.com/en/memory-by...400mhz-c10-memory-kit-cmd32gx3m4a2400c10.html

http://www.corsair.com/en/vengeance...400mhz-c9-memory-kit-cmy32gx3m4a2400c10r.html

Don't mind the prices, I have good sources where I pay much less than retail.

I can use up to 40-50 GB of RAM when loading and rendering big scenes in 3Ds. Key is big and fast.

 

[TheHunter]

My new OC 4.5Ghz @ 1.170v

Cpu multi 45x, cache 41x @ 1.115v (overvolted with LLC? to 1.125v)

http://valid.canardpc.com/2844801



Also to add missing data in clubhouse chart, default cache voltage is 1.052v


A little tighter ram latency boosted my score from 9.87 to 9.91

9.89 was with default ram latency (9-11-10-27-2T), but cache at 42 or 43x (idk atm)


@Cadaveca
Can you post your ram bandwidth in Aida64 3.0? Im really curious @ 2666mhz bandwidth, thanks

 

[EarthDog]

Good luck getting that 3930K to those memory speeds... not happening in all likely hood.

That said, the Vengy CL9 stuff would be my choice. Cheapest option with the lowest CL, same speed.

 

[cadaveca]

Quick hijacking advice then, asking you memory nuts

I need two kits of 32GB memory which one would you pick between these three?

http://www.gskill.com/en/product/f3-2400c10q-32gtx

http://www.corsair.com/en/memory-by...400mhz-c10-memory-kit-cmd32gx3m4a2400c10.html

http://www.corsair.com/en/vengeance...400mhz-c9-memory-kit-cmy32gx3m4a2400c10r.html

Don't mind the prices, I have good sources where I pay much less than retail.

I can use up to 40-50 GB of RAM when loading and rendering big scenes in 3Ds. Key is big and fast.

As EarthDog said...but...


Personally, because I like to OC ram, when looking at Vengeance, the "C10R" is what I would look at. Unfortunately, I don't have these sticks so I don't know how they behave exactly, but the entire Vengeance line-up sees the best sticks for OCing DRAM use that "R" in the SKU. All the other kits are fine, of course, it just seems to be what Corsair does.

I don't talk to Corsair too much ,so I haven't confirmed this, just something I've noticed since they launched that line-up. the 1866 C9R were the same, the Vengeance Extreme have "R", and there are other RED sticks without the "R"( although most red sticks do have it. there's just a few sets that do not, or have typos ).

I had hoped Corsair would send me a set, but I guess not. They seem to ignore me now. So buy those G.Skills, man, and support the companies that support TPU. XD.

Can you post your ram bandwidth in Aida64 3.0? Im really curious @ 2666mhz bandwidth, thanks

Check out my most recent board review, numbers are there. Working on a memory review (two, actually) right this moment. That's wit hGTX670 SLI, AMD cards get different numbers in some benchmarks, and memory-testing rig is using ASUS HD 7970 Matrix Platinum.

 

[cadaveca]

So is this Aida64 3.0 here?
http://www.techpowerup.com/reviews/Gigabyte/Z87X-OC/13.html

Yessir. I actually went and rebuilt rigs with the previously-tested boards to get those numbers, and also added in PCMark8 while I was at it. Also gave me the chance to see if BIOSes have improved int hat short time for MSI and Gigabyte, and they have.


I just finished re-testing PCMark8 too (yesterday), numbers will be different in next board review. I think I screwed up the initial runs.

 

[TheHunter]

Ok cool, some crazy numbers ^^


Although my read speed looks just a tinny bit faster ;P

 

[cadaveca]

This is v3.0 right?
http://www.techpowerup.com/reviews/Gigabyte/Z87X-OC/13.html

Edit: nvm , I see it has very high speeds ^^

What's interesting is the huge differences between boards. Not sure what's up with that, actually, might be the T-Topology stuff on the faster boards paying off. I finished testing the ASUS Maximus VI Hero last night and it put up quite similar numbers as the Gigabyte Z87X-OC.

Ok cool, some crazy numbers ^^


Although my read speed looks a little faster ;P

X79 reports avg 60k MB/s, slightly higher latency. The bench is very different now. Sub-timing tweaks make for the huge differences with the Z87 you and I see, IMHO. What I report in board reviews for memory performance is "out-of-the-box" performance, with just XMP enabled. Cache may play a role as well, although the impact is minimal that I have seen. CPU speed also seems to have little impact on Z87, where as with SNB and IVB the differences between stock CPU and OC to 4.6 GHz were rather pronounced.