I will not change from my 3930K until it feels like a Pentium MMX running in my system and unless software gets tech upgrades from space aliens that is going to be quite a long time. Not to mention games are driven by the console industry and unless PS4 is some Kraken of a badass Kratos smashing GOD hammer machine there will be no games that can truly push this 6 core for quite some time.
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
Retail 4770k from microcenter OC results
- Thread starter WinMan_x2000
- Start date
If we want to talk about stock speed for stock speed that's fair. I was just saying, for the price of the 3570k bundle deal, you can build for less and do the same things.
But you forgot about those synthetic benchmarks.
albertyouwrong
n00b
- Joined
- May 9, 2012
- Messages
- 12
Sounds like some weren't too lucky with their CPU's... Was able to get my 4770K stable 4.6 ghz @ 1.25v with a Gigabyte UD5H board. Didn't do any deep tweaking, just multiplier and vcore.
ShuttleLuv
Supreme [H]ardness
- Joined
- Apr 12, 2003
- Messages
- 7,295
^^^ vcore? Cooling? Stability test used?
WhoBeDaPlaya
2[H]4U
- Joined
- Dec 16, 2002
- Messages
- 2,642
Also, fixed vcore or adaptive?
albertyouwrong
n00b
- Joined
- May 9, 2012
- Messages
- 12
I kinda have to go back on what I said, it's not stable. =\ Ran Prime95 blend overnight and was good this morning, but earlier today she crashed in BF3. What worries me though is it's not hard locking/BSOD, just the 670's display drivers are crashing out.. So I dropped her down to 4ghz @ 1.20 vcore and it's still doing it. GPU is stock speed. It's starting to worry me.
To answer the other questions though, 1.25 vcore adaptive and a Rasa 240 water kit, no extras.
To answer the other questions though, 1.25 vcore adaptive and a Rasa 240 water kit, no extras.
WhoBeDaPlaya
2[H]4U
- Joined
- Dec 16, 2002
- Messages
- 2,642
IINM, doesn't the voltage shoot past 1.25V under actual load when using adaptive voltage? Think it was mentioned in a review or two.
It is ironic when you mention BF3 crashing. When I was trying my same overclock I have now with lower voltage, Prime passed, yet both BF3 and Crysis 3 crashed. I never just use prime as a test of stability, if my games don't run, its not correct.
WhoBeDaPlaya
2[H]4U
- Joined
- Dec 16, 2002
- Messages
- 2,642
Got done with another two chips, also from Microcenter and same batch # (L307B239, which is all they seem to have).
All the chips below were OCed with Speedstep disabled, fixed / non-adaptive Vcore, automatic/default voltages (except Vcore), DDR3 1866, 3.9GHz uncore and 100 - 101MHz BCLK.
Voltages reported are values under IBT load. No memtest86+ errors. Considered stable if it makes it through 100 passes of IBT (will follow up with 24 hour Prime95 runs later).
Chip 1 : 4.55GHz 1.24V (CPU-Z), 1.247V (measured)
Chip 2 : 4.55GHz 1.241V (CPU-Z), 1.249V (measured)
Chip 3 : 4.6GHz 1.25V (CPU-Z), 1.256V (measured)
Chip 4 : 4.35GHz 1.242V (CPU-Z), 1.253V (measured)
Not sure about ASUS' recommendation for using AIDA64. Settings that were AIDA64 stable would flake out during short 10-pass IBT runs, so I'll be sticking with IBT as the OC gatekeeper,
while keeping Prime95 and AIDA64 as follow up tests.
All the chips below were OCed with Speedstep disabled, fixed / non-adaptive Vcore, automatic/default voltages (except Vcore), DDR3 1866, 3.9GHz uncore and 100 - 101MHz BCLK.
Voltages reported are values under IBT load. No memtest86+ errors. Considered stable if it makes it through 100 passes of IBT (will follow up with 24 hour Prime95 runs later).
Chip 1 : 4.55GHz 1.24V (CPU-Z), 1.247V (measured)
Chip 2 : 4.55GHz 1.241V (CPU-Z), 1.249V (measured)
Chip 3 : 4.6GHz 1.25V (CPU-Z), 1.256V (measured)
Chip 4 : 4.35GHz 1.242V (CPU-Z), 1.253V (measured)
Not sure about ASUS' recommendation for using AIDA64. Settings that were AIDA64 stable would flake out during short 10-pass IBT runs, so I'll be sticking with IBT as the OC gatekeeper,
while keeping Prime95 and AIDA64 as follow up tests.
Last edited:
What asus' stated is relevant if you're using Adaptive VID. If you're using manual voltage it's a non issue, voltage applied will be static.
However, if you're using Adaptive VID (which is actually the best type of voltage to use), then AVX type of instructions will put another .1V on top of the turbo boost voltage. So if, for instance, you have a 1.25V turbo boost voltage with adaptive VID - that then becomes 1.35V when using adaptive VID while using Prime 95 and only prime 95. Keep in mind that other workloads are not affected by this, because Prime95 uses AVX instructions in a way that basically no other program in existence will - this is why asus is suggesting AIDA64 to stress test when using Adaptive VID. Again, not an issue if you're using manual or offset voltage.
Check out JJs Z87 overclocking video @ newegg TV if you can, it has some fantastic information in it.
However, if you're using Adaptive VID (which is actually the best type of voltage to use), then AVX type of instructions will put another .1V on top of the turbo boost voltage. So if, for instance, you have a 1.25V turbo boost voltage with adaptive VID - that then becomes 1.35V when using adaptive VID while using Prime 95 and only prime 95. Keep in mind that other workloads are not affected by this, because Prime95 uses AVX instructions in a way that basically no other program in existence will - this is why asus is suggesting AIDA64 to stress test when using Adaptive VID. Again, not an issue if you're using manual or offset voltage.
Check out JJs Z87 overclocking video @ newegg TV if you can, it has some fantastic information in it.
Michaelius
Supreme [H]ardness
- Joined
- Sep 8, 2003
- Messages
- 4,684
Ehh time to start collecting money for Ivy Bridge-E (and pray they don't pull same shit with tim under lid)
SixFootDuo
Supreme [H]ardness
- Joined
- Oct 5, 2004
- Messages
- 5,825
You guys are making me fall in love all over again with my Sandy-E 3930k that I have running at 4.8ghz.
I'm not really up on the the new Haswell K but if it allows you guys to run a 125 boot strap, do it. You can lower your multi and your volts. If you can run boot straps, you guys might end up being very surprised with the OC you can get.
I'm not really up on the the new Haswell K but if it allows you guys to run a 125 boot strap, do it. You can lower your multi and your volts. If you can run boot straps, you guys might end up being very surprised with the OC you can get.
YOU are comparing Ivy i5-K to Haswell I7-k; I was comparing Ivy i5-K to Haswell i5-K. The price difference (at MicroCenter, mind) is no *worse* than $55 - and that is due to price cuts on the Ivy motherboard end. With a price difference less than $100 (which is itself partially ameliorated due to features and feature improvements in Z87 v. Z77), OC aside, it's almost a draw.
How large overclocks can they get with 1.2 V and 1.1 V? What is theirs default voltage on stock?
WhoBeDaPlaya
2[H]4U
- Joined
- Dec 16, 2002
- Messages
- 2,642
Don't recall default Vcore offhand, but most of them got 4.4GHz stable at 1.2V.
were all the cpus from the same mc? if so, what mc? i will check mine when i get home.
WhoBeDaPlaya
2[H]4U
- Joined
- Dec 16, 2002
- Messages
- 2,642
Yes, they were all from the Dallas MC. they did not seem to have different batches. would be interesting to see how Kyle's three chips turn out.
i have a cambridge ma chip at home. i will check the batch after work. hopefully i will be installing this weekend if corsair get my h100 back to me.
Also curious about this - it seems nearly everyone is overclocking Haswell in the traditional sense (for mainstream platforms, that is) by merely upping the multiplier. Really curious as to whether bclk would produce better results..
WinMan_x2000
Limp Gawd
- Joined
- Oct 3, 2003
- Messages
- 183
This is quite the thought here. I am going to give it a shot when I get some time. Anyone know if the it can be changed freely or is it set steps of change?
I have batch Batch# L307B240 coming from micro center
I put together my 4770k + Asus Z87-Plus last night. Noctua NT-H1 under a Zalman CNPS10x Quiet, Corsair Dominator 2133 @ CL9 and 1.5V.
I'll have to get exact values written down, but 4.6GHz @ 1.20v could not do Prime95 Large FFTs. Bumping up into the 1.3v range I was able to hit 4.7GHz, although temperatures were approaching 100. 4.8GHz @ 1.4v-ish ran just fine for half a minute before I killed it due to temps reaching 100.
Something that might help those shopping for parts for their haswell build- I noticed a 10C difference in CPU temp between memory @ 1333MHz with RAM voltage 1.45 and 2133MHz w. RAM @ 1.5v. I am going to try aggressively reducing RAM voltage and keep DDR speeds low in order to control CPU temps.
Furthermore, core # 0 is always the hottest, and temps cool off as you go down the list. I noticed up to a 15C delta between core # 3 and core # 0. I checked several reviews (PCPer, xbit) and their screenshots indicate similar behavior, so I'm not going to remount my cooler. Luckily, with haswell you can bump the multiplier individually, so I plan to run the cooler cores 2 & 3 a bit faster.
TL;DR
My haswell chip is stable even at high voltages and temps, so the name of the game will be reducing temps. DDR voltages and speed may be a major player in this.
I'll have to get exact values written down, but 4.6GHz @ 1.20v could not do Prime95 Large FFTs. Bumping up into the 1.3v range I was able to hit 4.7GHz, although temperatures were approaching 100. 4.8GHz @ 1.4v-ish ran just fine for half a minute before I killed it due to temps reaching 100.
Something that might help those shopping for parts for their haswell build- I noticed a 10C difference in CPU temp between memory @ 1333MHz with RAM voltage 1.45 and 2133MHz w. RAM @ 1.5v. I am going to try aggressively reducing RAM voltage and keep DDR speeds low in order to control CPU temps.
Furthermore, core # 0 is always the hottest, and temps cool off as you go down the list. I noticed up to a 15C delta between core # 3 and core # 0. I checked several reviews (PCPer, xbit) and their screenshots indicate similar behavior, so I'm not going to remount my cooler. Luckily, with haswell you can bump the multiplier individually, so I plan to run the cooler cores 2 & 3 a bit faster.
TL;DR
My haswell chip is stable even at high voltages and temps, so the name of the game will be reducing temps. DDR voltages and speed may be a major player in this.
WhoBeDaPlaya
2[H]4U
- Joined
- Dec 16, 2002
- Messages
- 2,642
Welp, two things to try tonight - undervolting / underclocking RAM, and BCLK straps (although I don't see why a higher BCLK + lower multi would help)
So delidding will be popular if it's only heat the problem? I can't wait to try it out. I was wondering what are the "safe spot" for voltage, I've heard it's 1.35v soo....
Can't wait to receive my mobo, nothing more frustrating then seeing my 4770k sitting beside, looking me in the eyes like "heuheuhe you can't do shit"
If you have such a huge delta between cores, sounds like a problem delidding will fix easily.
I think heat is pretty much always going to be your ceiling with these, unless you have a poopy chip that needs 1.4v at 4.4GHz. But we'll see how the first brave delidders fare.
If the temp delta is because core # 0 is up by the IMC and other stuff, would delidding even out the core temps, or just reduce everyone by 10C or w/e? Is that what people see with Ivy?
I've seen a few results between hardforum and anandtech forums where delidding will cause not only a temperature drop, but a delta of only 2-3 degrees between each core instead of 10-15+ for IVB.
Here's a video of someone doing the procedure listing his temps. You can skip ahead to 20:50 for a review of the temps before and after.
http://www.youtube.com/watch?v=XXs0I5kuoX4
Last edited:
WhoBeDaPlaya
2[H]4U
- Joined
- Dec 16, 2002
- Messages
- 2,642
A 15C spread is a bit much. I only saw 4-7C spread on all 4 chips.
Edit : Update 1 [Memory Speed / Voltage]
Used the same chip, 4.4GHz @ 1.2V (fixed, not adaptive), Antec 920, IC Diamond, Crucial Ballistix Elite 1866 9-9-9-27-2T 1.5V
All runs were using the XMP timings (fixed at 9-9-9-27-2T) - only the speed and voltage were changed. Verified to be error free in memtest86+ 5 RC1.
Highest core temperature during a 25-pass IBT run is reported, averaged across 2 runs.
1333 - 1.4V : 75C
1333 - 1.5V : 75C
1866 - 1.4V : 79C
1866 - 1.5V : 79C
Conclusion : Memory voltage doesn't seem to make a difference, but speed definitely does.
Edit : Update 2 [BCLK/multi]
Jack sh*t. Can't seem to get the Z87-GD65 to take the 1.25 BCLK ratio/strap.
Edit : Update 1 [Memory Speed / Voltage]
Used the same chip, 4.4GHz @ 1.2V (fixed, not adaptive), Antec 920, IC Diamond, Crucial Ballistix Elite 1866 9-9-9-27-2T 1.5V
All runs were using the XMP timings (fixed at 9-9-9-27-2T) - only the speed and voltage were changed. Verified to be error free in memtest86+ 5 RC1.
Highest core temperature during a 25-pass IBT run is reported, averaged across 2 runs.
1333 - 1.4V : 75C
1333 - 1.5V : 75C
1866 - 1.4V : 79C
1866 - 1.5V : 79C
Conclusion : Memory voltage doesn't seem to make a difference, but speed definitely does.
Edit : Update 2 [BCLK/multi]
Jack sh*t. Can't seem to get the Z87-GD65 to take the 1.25 BCLK ratio/strap.
Last edited:
I played with it a bit more, here's what I found.
4.6GHz CPU & cache required basically 1.3v to sort of work. My chip's cache doesn't like going past 4.5GHz, so I'll revisit 4.6GHz voltages later. I was far more interested in what 4.7GHz takes...
100x strap, 47x ratio on all cores, 45x ratio on cache, voltage offset +0.150v. DDR3 @ 1600MHz, 1.5v, and CL9. Speedstep and C states enabled. CPU-Z, as you can see, reports that the CPU gets 1.392v in a Prime95 Blend scenario.
I checked after 27 minutes of bench time and found that core temps reached 100C and throttled back the frequency. Welp, it's just a big air cooler, figured as much. Seems stable enough otherwise.
Tonight, I'm going to tune the cache ratio down to 39x or 40x and see if I can bring voltages down across all frequencies. I also want to try putting the 2 hotter cores at 45x or 46x and the 2 cooler cores at 47x or 48x. I wonder if Win7 is smart enough to put my games on the two faster cores?
4.6GHz CPU & cache required basically 1.3v to sort of work. My chip's cache doesn't like going past 4.5GHz, so I'll revisit 4.6GHz voltages later. I was far more interested in what 4.7GHz takes...
100x strap, 47x ratio on all cores, 45x ratio on cache, voltage offset +0.150v. DDR3 @ 1600MHz, 1.5v, and CL9. Speedstep and C states enabled. CPU-Z, as you can see, reports that the CPU gets 1.392v in a Prime95 Blend scenario.
I checked after 27 minutes of bench time and found that core temps reached 100C and throttled back the frequency. Welp, it's just a big air cooler, figured as much. Seems stable enough otherwise.
Tonight, I'm going to tune the cache ratio down to 39x or 40x and see if I can bring voltages down across all frequencies. I also want to try putting the 2 hotter cores at 45x or 46x and the 2 cooler cores at 47x or 48x. I wonder if Win7 is smart enough to put my games on the two faster cores?
WhoBeDaPlaya
2[H]4U
- Joined
- Dec 16, 2002
- Messages
- 2,642
Honestly, since I fold, I usually go for fixed Vcores and disable the power saving stuff (EIST, C1E, etc.)
Rattle
2[H]4U
- Joined
- Apr 10, 2007
- Messages
- 2,841
I have one incoming from sabrepc $310 deal hopefully 4.5 and I'll be happy, I am going to try and undervolt also as I always do.
I got a 4770k from Microcenter too so I'll chime in.
All tests were done at stock voltage, with turbo boost disabled and all other settings untouched. HSF is a zalman 9500 running at almost minimum speed (registers at ~1600 rpm). Prime95.
At 3.0ghz (underclocked but not undervolted) the hottest core reached a max of 55c
At 3.5ghz the hottest core reached a max of 61c
At 3.7ghz the hottest core reached a max of 64c
At 4.0ghz the hottest core reached a max of 83c
I didn't bother going above that, 83c is already above my comfort zone. So from 3.0->3.7 the temps went up by 9c, but to get another .3ghz took a 19c increase.
When I have time I might retest again with my fans on full blow just for fun, though I have trouble seeing myself going back to the leafblower days regardless of overclocks.
All tests were done at stock voltage, with turbo boost disabled and all other settings untouched. HSF is a zalman 9500 running at almost minimum speed (registers at ~1600 rpm). Prime95.
At 3.0ghz (underclocked but not undervolted) the hottest core reached a max of 55c
At 3.5ghz the hottest core reached a max of 61c
At 3.7ghz the hottest core reached a max of 64c
At 4.0ghz the hottest core reached a max of 83c
I didn't bother going above that, 83c is already above my comfort zone. So from 3.0->3.7 the temps went up by 9c, but to get another .3ghz took a 19c increase.
When I have time I might retest again with my fans on full blow just for fun, though I have trouble seeing myself going back to the leafblower days regardless of overclocks.
How were you able to buy 4 chips from MC? My MC checks and I can't buy more than one per month !
Well, yes, but because of the heat. I can ramp the voltage over 1.4v and 4.8GHz will boot just fine. Prime95 runs, but I hit 100C pretty quickly and throttle down.
I prefer offset right now. Manual mode isn't any better- you set it for 1.25v but the CPU-Z will still report >1.3v under load. Adaptive mode seems similar to offset, I'm just not 100% confident that it will give enough voltages when NOT turboing. I might play with it again, but offset is easier to understand while I'm gathering information.
Cool, I was wondering if it was the frequency bump or the tiny additional voltage causing most of the heat increase. Looks like it's mostly due to the frequency. When I run at stock with turbo, my results (with DDR @ 2133MHz) are:
1.216v CPU-z
78 max core after the first 10 Prime95 blends
Here's what I did yesterday. The offset voltage I used is indicated by +0.XX volts.
47x all cores DDR @ 1600
45x cache
Offset mode
+0.150v
Hits 1.392v in prime95, throttled due to peak core temps 100, 99, 89, 83
+0.135
Hits 1.376v in prime95, throttled due to temps again
+.100v and +0.125
Blue screen loading windows
45x cores and cache, DDR @ 2133
+0.0
Crash on boot
+.025
Blue screen loading windows
+.05 (1.280v cpu-z load)
Blender 20 passes, throttles
+.075
blend for 14 passes, throttled due to temps at times
45 core, 39 cache
Auto voltage setting
blue screen loading windows
+0.025v
blue screen loading windows
+.050v
Throttles due to temps at times
Sounds like Haswell is too hot right? Not necessarily. It totally depends on what you want to do with it. If you just use your computer for Prime95, then you're going to live in the 4.0-4.3GHz range due to heat. But if you just play games and browse the web, your temps are going to be fine:
Max core temp after 2 hours of Team Fortress 2 (32 man custom maps) @ 4.5GHz and DDR @ 2133: 62C.
Max core temp after 3 hours of Planetside 2 @ 4.7GHz and DDR @ 1600: 64 C.
I think this is great. Planetside 2 hasn't dipped below 50 fps yet, whereas my i7-930 @ 4GHz dipped to 31fps at times. And the lowest I saw in TF2 was 90 fps for a brief moment, otherwise staying around 120fps during battle (299fps otherwise) where my i7-930 used to dip into the 50's. And even more interesting, my i7-930 would hit peak temps of 79C. So as far as I'm concerned, haswell gives me ~1.5x performance improvement AND runs cooler than my old CPU on the games I play. Win-win. If I happen to run into a Prime95 scenario, then it'll just get very hot and throttle down to 4.4GHz. But it won't crash. Heck, I'm considering going for 4.8GHz or more now since my games have so much thermal headroom.
Those are some crazy high offsets you're inputting. On my P8Z77, with a .020 offset I get nearly 1.3V during turbo loads. I'd be really hesitant to use .1xxxx offsets.......but I guess this varies per board and BIOS. Also, like you hinted at - the thing about offset voltage is that the turbo voltage is unpredictable and involves a lot of guesswork/rebooting, and the non turbo voltages may not be high enough - causing BSODs while idle. It's really tough to deal with, although i'm not sure adaptive is any better in this respect.
And to piggyback on the unpredictable comment....there have been times where I've used offset voltage and the turbo voltage appears to be a certain value during prime (let's say, 1.25V) and then suddenly it jumps way up to 1.4V which causes instability. Ugh. I hate dialing offset voltage in, it's takes a lot of trial and error. Although, it's worth it in the end to prevent electromigration - obviously turbo voltage 24/7 isn't desirable since you're just outputting that much more heat from your system.
And to piggyback on the unpredictable comment....there have been times where I've used offset voltage and the turbo voltage appears to be a certain value during prime (let's say, 1.25V) and then suddenly it jumps way up to 1.4V which causes instability. Ugh. I hate dialing offset voltage in, it's takes a lot of trial and error. Although, it's worth it in the end to prevent electromigration - obviously turbo voltage 24/7 isn't desirable since you're just outputting that much more heat from your system.