I thought that there were limits to how far the K-series can go and X-series was unlimited?Originally Posted by THUMPer
I thought that there were limits to how far the K-series can go and X-series was unlimited?
flow man:
du/dt + u dot del u = - del P / rho + v vector_Laplacian u
{\partial\mathbf{u}\over\partial t}+\mathbf{u}\cdot\nabla\mathbf{u} = -{\nabla P\over\rho} + \nu\nabla^2\mathbf{u}
I think you are looking too much into it. The K is fine for the everyday user/overclocker. X is for e-peen guys who run benchmarks all day. One won't necessarily overclock better than the other.
Well, it helps with my understanding of it though...(i.e. what are the limitations? etc...)
flow man:
du/dt + u dot del u = - del P / rho + v vector_Laplacian u
{\partial\mathbf{u}\over\partial t}+\mathbf{u}\cdot\nabla\mathbf{u} = -{\nabla P\over\rho} + \nu\nabla^2\mathbf{u}
Is it better to up ONLY the ratio or some combination of ratio + base clock?
flow man:
du/dt + u dot del u = - del P / rho + v vector_Laplacian u
{\partial\mathbf{u}\over\partial t}+\mathbf{u}\cdot\nabla\mathbf{u} = -{\nabla P\over\rho} + \nu\nabla^2\mathbf{u}
Base clock adjustment is rather limited, the reason why you bought a K processor with unlocked multiplier.
Multiplier is best for easy overclocks in 100 MHz increments. If you want to fine tune that, mess with BCLK.
Smile
I didn't know if there was like some kind of "sweet" spot where a combination means that you wouldn't have to up the ratio quite nearly as high, but you'd still be able to get to the same OC, with the slight added side effect of increasing the memory speed as well.
*edit*
Also, would it be better to let it turbo UP to the desired OC speed or would it be better just to set it in the CPU ratio and disable TurboBoost altogether?
*edit #2*
Also I'm at 4.5 GHz now @ 1.35 Vcore. No changes to any of the other voltages or BCLK speed. Topping out at 311 W at full load at the wall and according to Core Temp 153 W on the CPU itself, with a max temp of 68 C while running the "very high" setting of Intel Burn Test.
(JUST shy of 100 GFLOPs. *snaps fingers* dam.)
*edit 3#*
haha...it just BSOD'd. Bumping up the Vcore voltage again...
Call me Mr. Cosell.
*edit #4*
4.5 GHz @ 1.36 Vcore. Max 69 C @ 153.2 W. ~320 W at the wall.
4.5 GHz @ 1.38 Vcore with HTT enabled. Max. 72 C @ 168 W. ~352 W at the wall.
*edit #5*
Another quick update - it looks like that the crash sim program either runs differently or is a little more intensive than some of the benchmarks (possibly because it uses the CPU in a different way as compared to the benchmarks). I ended up having to up Vcore to 1.42 V in order to get the crash sims to be stable at 4.5 GHz. But it doesn't quite use the CPU as much compared to IBT. *shrug* Interrresting. Running my second run right now - estimated time: 2.8 hours.
Last edited by alpha754293; 04-21-2012 at 10:15 AM.
flow man:
du/dt + u dot del u = - del P / rho + v vector_Laplacian u
{\partial\mathbf{u}\over\partial t}+\mathbf{u}\cdot\nabla\mathbf{u} = -{\nabla P\over\rho} + \nu\nabla^2\mathbf{u}
Why is this? I was able to get 4.8 stable on a $209.00 Intel DX79TO. Are the Giga X79's that bad?
What makes ya say that?
Because of the heating up mosfets or because of degradation?
AMD 7800X3D@watercool-Heatkiller IV Pro| ASUS Crosshair X670E Hero | NVidia Founders Edition 4090 @watercool_Waterblock H²O@Mo-Ra3|Seasonic Fokus-GX 1000W| 2*16 GB G.Skill@F5-6000J3038F16GX2-TZ5N | Samsung 990 Pro 2 TB |Gigabyte Aorus NVMe Gen4, Samsung 850 PRO 1 TB| Cooler Master CoSmoS II |LG OLED42C21LA 106 cm (42 Zoll)|Windows 11 Prof. 64 bit
Sound: AVR-X3600H@5.1 Nubert-NuVero70
Because there were still issues with the bios coding as you need to pump ridicilous voltages going from 4.7 to 4.8. Under cold the boards work fine, not on air or water... haven't tested the latest biosses though, focus is somewhere else now :p
Same CPU works fine on ASUS, ASRock and MSI boards, with lower voltages then on the GB's...
Question : Why do some overclockers switch into d*ckmode when money is involved
Remark : They call me Pro AsusSaaya yupp, I agree
Yeah, I hear ya.
It is easy to get 4.6MHz@1.308V Vcore stable. But for same damned reason I do not succeed to hit 4.7 MHz straight.
AMD 7800X3D@watercool-Heatkiller IV Pro| ASUS Crosshair X670E Hero | NVidia Founders Edition 4090 @watercool_Waterblock H²O@Mo-Ra3|Seasonic Fokus-GX 1000W| 2*16 GB G.Skill@F5-6000J3038F16GX2-TZ5N | Samsung 990 Pro 2 TB |Gigabyte Aorus NVMe Gen4, Samsung 850 PRO 1 TB| Cooler Master CoSmoS II |LG OLED42C21LA 106 cm (42 Zoll)|Windows 11 Prof. 64 bit
Sound: AVR-X3600H@5.1 Nubert-NuVero70
Wow...that's substantially less voltage than what I'm running at right now. Damnnn....
flow man:
du/dt + u dot del u = - del P / rho + v vector_Laplacian u
{\partial\mathbf{u}\over\partial t}+\mathbf{u}\cdot\nabla\mathbf{u} = -{\nabla P\over\rho} + \nu\nabla^2\mathbf{u}
Here is Intel Burn Test WITH HTT (at 4.5 GHz @ 1.42 V)
ibt 4.5 GHz w HTT.jpg
And here it is without.
ibt 4.5 GHz wo HTT.jpg
And this is why I would want to disable HTT.
flow man:
du/dt + u dot del u = - del P / rho + v vector_Laplacian u
{\partial\mathbf{u}\over\partial t}+\mathbf{u}\cdot\nabla\mathbf{u} = -{\nabla P\over\rho} + \nu\nabla^2\mathbf{u}
ATTO Express Pro Tools for the Intel 520 180 GB on SATA 6 Gbps (at 3.2 GHz stock)
atto ept intel 520.jpg
ATTO Express Pro Tools for the OCZ Vertex 3 240 GB on SATA 6 Gbps (at 3.2 GHz stock)
atto ept vertex 3.jpg
HDTach for the Intel 520 180 GB on SATA 6 Gbps (at 3.2 GHz stock)
hd tach intel 520.jpg
HDTach for the OCZ Vertex 3 240 GB on SATA 6 Gbps (at 3.2 GHz stock)
hd tach vertex 3.jpg
I forgot to save the h2benchw results, but I think that swapping on the Intel 520 was somewhere between like 60-70 MB/s while on the Vertex 3, it's more like 70-80 MB/s.
Here is the HDTach for the Intel 520 180 GB on SATA 6 Gbps (at 4.5 GHz at 1.42 V)
hd tach intel 520 2.jpg
Here is the HDTach for the OCZ Vertex 3 240 GB on SATA 6 Gbps (at 4.5 GHz at 1.42 V)
hd tach vertex 3 2.jpg
H2bench -- by Harald B”geholz & Lars Bremer / c't Magazin fr Computertechnik
Version 3.13/Win32, Copyright (C) 2009 Heise Zeitschriften Verlag GmbH & Co. KG
Dutch translation by F&L Technical Publications B.V.
ATA disk: INTEL SSDSC2CW180A3
serial #: CVCV202103MP180EGN
firmware: 400i
Supported UDMA modes: 0 1 2 3 4 5 6
UDMA mode 5 active.
acoustic management not supported.
Capacity: 351646785 sectors=171703 MByte, CHS=(21889/255/63)
Checking timer for 10 seconds (Win32) ............. Ok.
timer resolution: 0.321 æs, 3.118 MHz
timer statistics: 236298447 calls, min 0.00 æs, average 0.02 æs, max 1.92 æs
Reading some sectors to warm up... done.
interface speed test with block size 128 sectors (64.0 KByte):
sequential read rate medium (w/out delay): ...332.1 MByte/s
sequential transfer rate w/ read-ahead (delay: 0.21 ms): ...332.1 MByte/s
Repetitive sequential read ("core test"): ...237.6 MByte/s
5 seconds breather........Ok.
sequential write rate medium (w/out delay): ...410.5 MByte/s
sequential transfer rate write cache (delay: 0.17 ms): ...412.7 MByte/s
Repetitive sequential write: ...419.4 MByte/s
data integrity check (first 20480 sectors fully checked)
writing test patterns ...done.
short breather ... OK.
reading some sectors to flush cache ... OK.
reading test patterns ...done.
Zone measurement read: calibrating... 333.3 MByte/s at 50% of total capacity.
reading 999 sample points (2748 blocks of 128 sectors = 171.75 MByte)
estimated runtime: 9 minutes...done.
sustained data rate read: average 340074.0, min 320064.0, max 347739.1 [KByte/s]
Zone measurement write: calibrating... 409.7 MByte/s at 50% of total capacity.
5 seconds breather........Ok.
estimated runtime: 7 minutes...done.
sustained data rate write: average 420202.4, min 388104.7, max 430186.7 [KByte/s]
Measuring random access time (whole disk):
reading... 0.16 ms (min. 0.08 ms, max. 0.73 ms)
writing... 0.17 ms (min. 0.08 ms, max. 2.28 ms)
random access time in lower 504 MByte
reading... 0.15 ms (min. 0.05 ms, max. 0.36 ms)
writing... 0.15 ms (min. 0.03 ms, max. 2.34 ms)
Running application profile `swapping' ...124642.1 KByte/s
Running application profile `installing' ...365773.5 KByte/s
Running application profile `Word' ...278040.8 KByte/s
Running application profile `Photoshop' ...257440.5 KByte/s
Running application profile `copying' ...418296.4 KByte/s
Running application profile `F-Prot' ...169483.3 KByte/s
Result: application index = 242.0The SSDs are VERY sensitive to the state of performance that they're in when you start testing them. The differences in the HDTach graphs probably is the most visually telling tale, but even the discrepancies with h2benchw and the swapping performance indicate the same thing.H2bench -- by Harald B”geholz & Lars Bremer / c't Magazin fr Computertechnik
Version 3.13/Win32, Copyright (C) 2009 Heise Zeitschriften Verlag GmbH & Co. KG
Dutch translation by F&L Technical Publications B.V.
ATA disk: OCZ-VERTEX3
serial #: OCZ-9D7P653025TDKA66
firmware: 2.15
Supported UDMA modes: 0 1 2 3 4 5 6
UDMA mode 5 active.
acoustic management not supported.
Capacity: 468857025 sectors=228934 MByte, CHS=(29185/255/63)
Checking timer for 10 seconds (Win32) ............. Ok.
timer resolution: 0.321 æs, 3.118 MHz
timer statistics: 236314042 calls, min 0.00 æs, average 0.02 æs, max 1.92 æs
Reading some sectors to warm up... done.
interface speed test with block size 128 sectors (64.0 KByte):
sequential read rate medium (w/out delay): ...325.5 MByte/s
sequential transfer rate w/ read-ahead (delay: 0.21 ms): ...321.1 MByte/s
Repetitive sequential read ("core test"): ...345.2 MByte/s
5 seconds breather........Ok.
sequential write rate medium (w/out delay): ...410.8 MByte/s
sequential transfer rate write cache (delay: 0.17 ms): ...411.5 MByte/s
Repetitive sequential write: ...417.0 MByte/s
data integrity check (first 20480 sectors fully checked)
writing test patterns ...done.
short breather ... OK.
reading some sectors to flush cache ... OK.
reading test patterns ...done.
Zone measurement read: calibrating... 329.1 MByte/s at 50% of total capacity.
reading 999 sample points (3663 blocks of 128 sectors = 228.94 MByte)
estimated runtime: 12 minutes...done.
sustained data rate read: average 345140.2, min 316167.9, max 359401.5 [KByte/s]
Zone measurement write: calibrating... 416.8 MByte/s at 50% of total capacity.
5 seconds breather........Ok.
estimated runtime: 9 minutes...done.
sustained data rate write: average 420258.8, min 384829.6, max 428152.2 [KByte/s]
Measuring random access time (whole disk):
reading... 0.24 ms (min. 0.09 ms, max. 0.57 ms)
writing... 0.19 ms (min. 0.09 ms, max. 2.50 ms)
random access time in lower 504 MByte
reading... 0.16 ms (min. 0.05 ms, max. 1.43 ms)
writing... 0.16 ms (min. 0.03 ms, max. 2.46 ms)
Running application profile `swapping' ...123020.8 KByte/s
Running application profile `installing' ...346335.0 KByte/s
Running application profile `Word' ...284403.3 KByte/s
Running application profile `Photoshop' ...256911.7 KByte/s
Running application profile `copying' ...417226.6 KByte/s
Running application profile `F-Prot' ...162826.6 KByte/s
Result: application index = 238.4
So the way that I've got it set up right now - I have Win 7 on the Intel 520 and swap on the Vertex 3 only because the Vertex 3 uses the Sandforce 2281 controller which is known to have the BSOD issue.
I think that the two runs that I ran last night didn't really work because they're reporting times that's even LONGER than when it was at stock, so I'm re-running those now.
Otherwise, for the Neon collision, the results are as follows:
(Time, run state/settings)Code:962 stock at 3.2 GHz, with HTT, EIST, C1E, C3, C6, Turbo; swap file on Hitachi 500 GB 7.2krpm SATA 1012 stock at 3.2 GHz, with HTT; swap file on Hitachi 500 GB 7.2krpm SATA 1174 stock at 3.2 GHz, without HTT; swap file on Hitachi 500 GB 7.2krpm SATA 800 4.5 GHz@1.42 Vcore, with HTT; swap file on Hitachi 500 7.2krpm SATA 939 4.5 GHz@1.42 Vcore, without HTT; swap file on Hitachi 500 7.2krpm SATA 926 4.5 GHz@1.42 Vcore, without HTT; swap file on OCZ Vertex 3 240 GB SATA 6 Gbps 841 4.5 GHz@1.42 Vcore, without HTT; swap file on OCZ Vertex 3 240 GB SATA 6 Gbps; with pfile numproc 6
And for those that are REALLY interested, this is what the Neon crash sim looks like:
http://www.youtube.com/watch?v=3ppuDedRZDM
I'm only running the first 30 milliseconds of it.
And here's what the 3 car collision simulation looks like:
http://www.youtube.com/watch?v=HhOJmGVhklY
And here's the EDM CH4 Combustion sim:
http://www.youtube.com/watch?v=xDAgloQQAh0
(which I don't think it's nearly that exciting as the others, but it's a piece of the SI ICE sim puzzle)
Last edited by alpha754293; 04-22-2012 at 07:15 AM.
flow man:
du/dt + u dot del u = - del P / rho + v vector_Laplacian u
{\partial\mathbf{u}\over\partial t}+\mathbf{u}\cdot\nabla\mathbf{u} = -{\nabla P\over\rho} + \nu\nabla^2\mathbf{u}
How much vtt voltage do ya fire on the chip?
Last edited by Fritz_the_germ; 04-22-2012 at 09:59 AM.
AMD 7800X3D@watercool-Heatkiller IV Pro| ASUS Crosshair X670E Hero | NVidia Founders Edition 4090 @watercool_Waterblock H²O@Mo-Ra3|Seasonic Fokus-GX 1000W| 2*16 GB G.Skill@F5-6000J3038F16GX2-TZ5N | Samsung 990 Pro 2 TB |Gigabyte Aorus NVMe Gen4, Samsung 850 PRO 1 TB| Cooler Master CoSmoS II |LG OLED42C21LA 106 cm (42 Zoll)|Windows 11 Prof. 64 bit
Sound: AVR-X3600H@5.1 Nubert-NuVero70
There's something that isn't right, it's downright impossible to need 1.45 Vcore at only 4.5.
Forget about using the offset mode for now, dial in ~1.35 manual and 1.15 VTT & VCCSA. Go with medium LLC and if needed increase CPU current capability to 130% if auto doesn't cut it. Even in the worst possible case I couldn't imagine a 3930K needing 1.4V for 4.5.
What's LLC? and what's VCCSA?
flow man:
du/dt + u dot del u = - del P / rho + v vector_Laplacian u
{\partial\mathbf{u}\over\partial t}+\mathbf{u}\cdot\nabla\mathbf{u} = -{\nabla P\over\rho} + \nu\nabla^2\mathbf{u}
Uh...I'm not sure. I left it at auto.
flow man:
du/dt + u dot del u = - del P / rho + v vector_Laplacian u
{\partial\mathbf{u}\over\partial t}+\mathbf{u}\cdot\nabla\mathbf{u} = -{\nabla P\over\rho} + \nu\nabla^2\mathbf{u}
Here's a screenshot of the desktop with EasyTune 6 Voltage I page shown.
desktop.jpg
Maybe you guys would be able to decipher and make more sense of it than I will. But it's 1.42 V and that's apparently what's needed to make the crash sim program stable (at least based on these current settings anyways).
PCH is 1.54 V.
flow man:
du/dt + u dot del u = - del P / rho + v vector_Laplacian u
{\partial\mathbf{u}\over\partial t}+\mathbf{u}\cdot\nabla\mathbf{u} = -{\nabla P\over\rho} + \nu\nabla^2\mathbf{u}
For 4.5ghz, auto settings are well within a safe range. If you wanted to drop idle vcore a little you could try raising LLC/Load Line Calibration.
Your load temps are pretty good. Some chips that clock very high on low vcore run hot, and some that require more vcore, run very cool.
I have a couple 3930k's. The first one runs 4.8ghz stable at 1.358vcore load, but runs very hot. The other needs a lot more vcore for the same 4.8ghz but the chip never gets hot. All chips are different, but as long as your temps are ok, which yours seem fine, I wouldnt worry about it. You reached your goal, temps are good, vcore is a little high but nothing to be overly concerned about, and if you wanted to spend more time tweaking, you can probably lower it bit.
If I had my choice, for a 24/7 system, Id take the higher vcore, cooler running chip any day. Low vcore chips are great for screen shot purposes, but as log as you are in a safe range, it doesnt matter. Heat is just as dangerous as high voltage.
Just my opinion.
Last edited by OC Nub; 04-22-2012 at 11:02 AM.
Yea....with people saying stuff about like offset voltages and stuff - I don't even see that option in the BIOS unless I'm looking at the wrong thing/wrong spot.
I would think that it would be a bit of an optimization. I'm no ECE, but could it be that the chips that run with a lower Vcore is because it has to draw a higher current, which makes it hotter? Of course, the alternative would be to run it with a higher Vcore, which then in turn drops the current requirement (with power being constant), but you can up the Vcore say to like 10 V lest you'd fry that thing.
Like I said, out of all the testing that I've done, the highest I've been able to hit is 73 C. Course that's also with the H80 running at full blast. Right now, the CFD run that's going through doesn't even need that.
*edit*
The other interesting side note (or at least it's interesting to me) is the combustion CFD that I'm running which is a VERY simple model generates about 20 GB of data each time I run it.And it runs in under an hour on this system.
flow man:
du/dt + u dot del u = - del P / rho + v vector_Laplacian u
{\partial\mathbf{u}\over\partial t}+\mathbf{u}\cdot\nabla\mathbf{u} = -{\nabla P\over\rho} + \nu\nabla^2\mathbf{u}
PCH is waaaaaaay high, try 1.05-1.10v.
Your screen hot shows PCH at 1.10v, which is perfectly fine.
*shrug* I left everything at default except for Vcore.
flow man:
du/dt + u dot del u = - del P / rho + v vector_Laplacian u
{\partial\mathbf{u}\over\partial t}+\mathbf{u}\cdot\nabla\mathbf{u} = -{\nabla P\over\rho} + \nu\nabla^2\mathbf{u}
LLC = Load Line Calibration. You should set it to medium due to vdroop if it isn't set to that already.
Stop looking at voltages shown in the Gigabyte app, keep an eye on the voltage in CPU-Z and find the highest voltage reached, it will fluctuate slightly under load. According to your screenshot that appears to be 1.368, which seems pretty standard and the temps are perfectly fine too.
I'm not. Actually, I just installed it because someone was asking me what my Vtt is set at and I can't go peek into the BIOS setting while a run is going.
The Vcore that I'm reporting is out of the BIOS setting, NOT the app.
And those measurement programs kinda suck because it does NOT reflect what's actually being set in the BIOS. (I don't think that it EVER does or EVER has - at least not in this case. And not for my Q9550 daily driver either.)
Don't you think that a program that reads the BIOS settings would be more accurate than those that try to "measure it" ex post facto?
flow man:
du/dt + u dot del u = - del P / rho + v vector_Laplacian u
{\partial\mathbf{u}\over\partial t}+\mathbf{u}\cdot\nabla\mathbf{u} = -{\nabla P\over\rho} + \nu\nabla^2\mathbf{u}
You want to re-enable Hyper-Threading now?1012 stock at 3.2 GHz, with HTT; swap file on Hitachi 500 GB 7.2krpm SATA
1174 stock at 3.2 GHz, without HTT; swap file on Hitachi 500 GB 7.2krpm SATA
800 4.5 GHz@1.42 Vcore, with HTT; swap file on Hitachi 500 7.2krpm SATA
939 4.5 GHz@1.42 Vcore, without HTT; swap file on Hitachi 500 7.2krpm SATA
What do you mean the "BIOS setting"...what you set in the BIOS? That is what is reported by EasyTune, and also something you can adjust in Windows through EasyTune.
Uhh...why do you think they are called measurement programs? They read results from the thermal diodes and voltage sensors on the motherboard that take real-time measurements of actual temperature and voltage. Those same readings are in your BIOS too, but they aren't "settings".
A lot of people here use HWMonitor, and no, a program that reads BIOS settings would certainly be the farthest thing from accurate. Software/on-board sensor readings are not the most accurate either, but are obviously much closer than guessing. Buy yourself a digital multimeter, find good measurements and test for yourself if you don't trust the board readouts.
EDIT: It looks like your load Core voltage is about 1.37v when you have 1.42v set, no?
Last edited by BeepBeep2; 04-22-2012 at 06:29 PM.
Smile
Posting Permissions
Bookmarks