Overclock.net › Forums › Intel › Intel CPUs › Haswell Overclocking Guide [With Statistics]
New Posts  All Forums:Forum Nav:

Haswell Overclocking Guide [With Statistics] - Page 1164

post #11631 of 19637
Quote:
Originally Posted by treome View Post

Tried up to 1.450v for 4.8 but BSOD IBT High 10 Loop

What are the temperature you're getting with such high voltage? I have 3 radiator loop and highest core hits 94C with 1.395vcor :/
post #11632 of 19637
Quote:
Originally Posted by marrawi View Post

What are the temperature you're getting with such high voltage? I have 3 radiator loop and highest core hits 94C with 1.395vcor :/

82C -- 2 x 240 mm push/pull in series with 7970 GPU
Swiftech Apogee HD
Koolance AR979
post #11633 of 19637
Quote:
Originally Posted by schlonzo View Post

Ok i found this thread in the HWInfo forum, stating that on ASROCK Z87 Motherboards the CPU VCORE value read out by HWInfo, is actually the half value of the VCCIN. Makes perfect sense to me, as my VCCIN is 1,9V. (2*0,95V)
Maybe its the same problem with HWMonitor?

If i use HWInfo v.4.36.2150 I can't find a Vcore, but under Sensors->CPU->all 4 Core VIDs show 1.3V.
Thats the Vcore, right? There are no other voltages related to the cpu.


So I don't have a crappy board.
Just an average crappy chip. arrowheadsmiley.png


I'm sorry to burst your bubble, but you do have a crappy board.
I have the same board, and there are a few things that show it's "quality" (or rather the lack of it).
First, after two months the first sata port is gone. It "died" on it's own. I checked with 4 drives (of which 2 ssd's) and half of dozen cables.

Second, the onboard audio is crap!, much worse than 'bog standard' xonar dg. They advertised it to be superior to standard onboard audio through it's ALC1150 and two amps but the reality of it is, that the sound is muddy at low volume and at high volume it's mostly noise (especially the treble).

Third, and this is more pertinent to this thread, the OC unfriendliness of the board (probably due to the cheapness of the components).
- Even at 4.1GHz @ 1.220V i need 1.9V VCCIN!, if i drop it to 1.850V it will bsod after few minutes.
- The lack of 'one to five' settings for the LLC that are present on the "non M" asrock boards, replaced with 'enable/disable'. If it had the same physical components as the "non M" boards, than it should have had the same settings for LLC.
- Any change to the multiplier or voltage or cache or ram etc.., will automatically set the LLC to disable (equivalent to 'one') and also, automatically change the input to manual @ 1.9V.
- It doesn't have an Vcore sensor implemented (i think for the entire 1150 line).

Another thing that might be related to this board (since i heave not read about this anywhere else):
I was trying the "cache on stock at 1.15V" that was recommended by the OP (only that i had it set to 1.2V) for my 4.2GHz OC, and could not reach stability even at 1.310 VID. So i thought of ramping the cache voltage to 1.250V (still at stock '3.4') for 4.2 multi, only this time i went back to 1.250 VID and it took more then an hour to crash.
Tested some more, and i ended up at 42@1.265V adaptive (on manual i need 1.280V) and the cache at 40@1.255V (adaptive).

The conclusion is: for my 42@1.265 i needed cache at 1.25x regardless whether at 34 or 40 multiplier.
Edited by angelotti - 4/5/14 at 3:07am
post #11634 of 19637
The best I've managed was 4.5 core and 4.4 uncore @ 1.350v and 1.225v respectively. Temps up to 70 whilst gaming.

But I've noticed that after saving OC profile and reverting to defaults both cpuz and HWinfo give higher-than-stock results. (4670k stock is 3.4Ghz)

post #11635 of 19637
Quote:
Originally Posted by Nyrii View Post

The best I've managed was 4.5 core and 4.4 uncore @ 1.350v and 1.225v respectively. Temps up to 70 whilst gaming.

But I've noticed that after saving OC profile and reverting to defaults both cpuz and HWinfo give higher-than-stock results. (4670k stock is 3.4Ghz)

the stock settings allow a single core to boost up to 3.9.
Cable box
(23 items)
 
Phat Bits
(16 items)
 
Lament Box
(16 items)
 
CPUMotherboardGraphicsRAM
i7-4790K 4.8ghz @ 1.245v Asus Maximus Hero VI 4GB Asus Strix GTX 980 Gskil Trident 2400 CL10 
Hard DriveHard DriveHard DriveCooling
128GB sandisk extreme SSD 2 tb Western digital media intel 530 128 gb SSD EK Coolstream 360mm PE Rad  
CoolingCoolingCoolingCooling
EK FC980 Strix Water Block swiftech MCP-35X EK XRES 140 DDC Top Gentle typhoon AP-15 X 3 
CoolingCoolingCoolingCooling
Fractal Design 140mm 1000RPM X3 Fractal Design Venturi High Flow Series 140mm F... Bitspower DDC Heatsink Black EK Coolstream XTC 140mm Rad 
CoolingOSMonitorKeyboard
EK Supremecy MX  WIN 10 AOC 2460 144hz 1080p Hcman Mechanical  
PowerCaseMouse
Seasonic G-750  Fractal Design R5 G200 logitech  
CPUMotherboardGraphicsRAM
4770k 4.5ghz @1.29v asus z87-A EVGA 4gb GTX 970 FTW 1510mhz 8000mem 1.23v Adata XPG 8GB @ 2400mhz 11-13-13-35-1 
Hard DriveHard DriveCoolingCooling
OCZ VERTEX 3 90GB 1tb 7200 Seagate hybrid  D5 vario Primo chill 120mm D5 Compression Res 
CoolingCoolingCoolingCooling
Swiftech MCR-320 XP 360mm rad Swiftech MCR-320 XP 120mm rad EK Supreme LTX Cpu block Biitspower ACX GTX 970 Water block 
OSKeyboardPowerCase
win7 64bit ultimate G13 Seasonic M211 620 watt Nzxt Guardian 
  hide details  
Reply
Cable box
(23 items)
 
Phat Bits
(16 items)
 
Lament Box
(16 items)
 
CPUMotherboardGraphicsRAM
i7-4790K 4.8ghz @ 1.245v Asus Maximus Hero VI 4GB Asus Strix GTX 980 Gskil Trident 2400 CL10 
Hard DriveHard DriveHard DriveCooling
128GB sandisk extreme SSD 2 tb Western digital media intel 530 128 gb SSD EK Coolstream 360mm PE Rad  
CoolingCoolingCoolingCooling
EK FC980 Strix Water Block swiftech MCP-35X EK XRES 140 DDC Top Gentle typhoon AP-15 X 3 
CoolingCoolingCoolingCooling
Fractal Design 140mm 1000RPM X3 Fractal Design Venturi High Flow Series 140mm F... Bitspower DDC Heatsink Black EK Coolstream XTC 140mm Rad 
CoolingOSMonitorKeyboard
EK Supremecy MX  WIN 10 AOC 2460 144hz 1080p Hcman Mechanical  
PowerCaseMouse
Seasonic G-750  Fractal Design R5 G200 logitech  
CPUMotherboardGraphicsRAM
4770k 4.5ghz @1.29v asus z87-A EVGA 4gb GTX 970 FTW 1510mhz 8000mem 1.23v Adata XPG 8GB @ 2400mhz 11-13-13-35-1 
Hard DriveHard DriveCoolingCooling
OCZ VERTEX 3 90GB 1tb 7200 Seagate hybrid  D5 vario Primo chill 120mm D5 Compression Res 
CoolingCoolingCoolingCooling
Swiftech MCR-320 XP 360mm rad Swiftech MCR-320 XP 120mm rad EK Supreme LTX Cpu block Biitspower ACX GTX 970 Water block 
OSKeyboardPowerCase
win7 64bit ultimate G13 Seasonic M211 620 watt Nzxt Guardian 
  hide details  
Reply
post #11636 of 19637
Thread Starter 

Does anybody here with a 4770k want to help me test something with chess? You must have a hyperthreaded CPU, preferably a Haswell one. It is said by the developer of Houdini chess engine that HT is bad for Houdini's engine strength, but just as importantly, that the decrease in strength is not as easily measured. Often people like to measure engine speed and relative strength by looking at kilonodes per second, but in this case it is deceiving. HT is said to increase kilonodes per second but decrease the efficiency of the search, decreasing depth and driving down the strength of the engine in the end. Robert Houdart said that chess engine software is in itself poorly optimized for HT, and this applies to basically all chess engines.

 

I want to see if this is true with Stockfish engine. So if you have free time, I can figure things out. One major reason why I got 4670k is because HT I believe is bad for chess for engines across the board. My idea is, you use Arena 3.5 GUI, latest beta Stockfish engine, load it up. Hash size to uhm, 512mb for example. So we'll have the engine analyze a position for a fixed amount of time and look at the ending search depth and how long it took to get to that depth.  And we might have to run this a few times, possibly with different positions to hit the nail in the common. 

 

If HT is as detrimental to engine strength for Houdini as it is for Stockfish, we should see a higher total nodes searched on HT on, but lower search depth. This means the engine searched for more stuff, but the search ended up with less results. Kind of like having a higher frequency on a CPU but a lower benchmark score due to having crappier IPC.

 

Unfortunately I do not have a HT enabled CPU or know a friend who has one.

Quote:
Originally Posted by Wirerat View Post


the stock settings allow a single core to boost up to 3.9.
I was looking at turbo and for me, it never turbos to 3.9 because the CPU tries to split a single core load among all four cores. So a 2003 single-thread bottlenecked videogame goes to 4.6 or whatnot because all four cores are fired up, but at like 8% per core. If I set core affinity to 1 core only though, then the higher speed kicks in. But it's not that important of a discovery.

Edited by Darkwizzie - 4/5/14 at 5:51am
Undelwalt (2017)
(29 items)
 
 
Celapaleis (2013)
(16 items)
 
CPUMotherboardGraphicsRAM
7600k @ 5.197/1.392 1.392v z170 Asus Hero EVGA 1080ti SC Black (2100/6250) x2 8gb GSkill Trident Z 3600 @ 3804 15-15-15-32-2T 
Hard DriveHard DriveHard DriveCooling
Samsung 950 Pro 512gb (Undelwalt) WD Red 2tb (Pack Yak II) Seagate Expansion Drive 5tb (Phoenix Down II) 3x560mm HardwareLabs Nemesis GTX 
CoolingCoolingCoolingCooling
XPSC Raystorm Pro Watercool Heatkiller 1080ti Full Cover Waterblock EK XTOP Revo Dual D5 (Serial, PWM, v4) EK ZMT (1/2 - 3/4 ID OD) 
CoolingCoolingCoolingCooling
14xEK ACF (Compression Fittings) EK x4 250 (v2) Distilled Water + PT Nuke (Copper Sulphate) 24x140mm Silent Wings 3 
OSMonitorKeyboardPower
Windows 10 Pro 64bit Catleap 27 Inch 2560x1440 IPS 60hz Display Coolermaster Storm Trigger (Brown Switches) EVGA P2 1000w 
CaseMouseMouse PadAudio
Corsair Air 540 Logitech G Pro Tek Syndicate "Raze the World" Desk Mat O2 + Odac by Mayflower Electronics (Rev A) 
AudioAudioAudioAudio
Sennheiser HD 800 2xKrk Rokit 6 G2 Blue Yeti (Grey) Rode PSA-1 Arm 
Other
Aquaero 6 LT + Aluminum Heatsink 
CPUMotherboardGraphicsRAM
i5 6600k Delid @ 4.848/4.848 Asus z170 Hero MSI GTX 980ti @ 1499/4082 1.25v/134% Power 2 x 8gb Gskills Ripjaws V 3131 16-16-16-32 
Hard DriveHard DriveHard DriveHard Drive
Samsung Pro Series 840 256gb (Celapaleis) 1 TB Samsung Drive (Pack Yak) Samsung Pro Series 850 256gb (Celapaleis Reprise) Phoenix Down (External Backup) 
CoolingCoolingOSMonitor
Noctua D14 Kraken x61 Push-Pull for GPU Windows 10 Pro Catleap 2560x1440 @ 60hz 
KeyboardPowerCaseMouse
CM Storm Trigger Brown Switches EVGA Supernova 1000w P2 Corsair 540 Air Razer Abyssus 
Mouse PadAudioAudioAudio
Steelseries QcK Mini Blue Yeti Microphone Objective Dac + Amp Sescom AB Switch 
AudioAudioAudioAudio
Krk Rokit 6 G2 x2 Radioshack Ground Loop Isolator (For Rokits) Wooaudio Aluminum Headphone Stand Sennheiser HD 800 
AudioOther
Logitech c920 Webcam NZXT Sentry Mix 2 Fan Controller 
  hide details  
Reply
Undelwalt (2017)
(29 items)
 
 
Celapaleis (2013)
(16 items)
 
CPUMotherboardGraphicsRAM
7600k @ 5.197/1.392 1.392v z170 Asus Hero EVGA 1080ti SC Black (2100/6250) x2 8gb GSkill Trident Z 3600 @ 3804 15-15-15-32-2T 
Hard DriveHard DriveHard DriveCooling
Samsung 950 Pro 512gb (Undelwalt) WD Red 2tb (Pack Yak II) Seagate Expansion Drive 5tb (Phoenix Down II) 3x560mm HardwareLabs Nemesis GTX 
CoolingCoolingCoolingCooling
XPSC Raystorm Pro Watercool Heatkiller 1080ti Full Cover Waterblock EK XTOP Revo Dual D5 (Serial, PWM, v4) EK ZMT (1/2 - 3/4 ID OD) 
CoolingCoolingCoolingCooling
14xEK ACF (Compression Fittings) EK x4 250 (v2) Distilled Water + PT Nuke (Copper Sulphate) 24x140mm Silent Wings 3 
OSMonitorKeyboardPower
Windows 10 Pro 64bit Catleap 27 Inch 2560x1440 IPS 60hz Display Coolermaster Storm Trigger (Brown Switches) EVGA P2 1000w 
CaseMouseMouse PadAudio
Corsair Air 540 Logitech G Pro Tek Syndicate "Raze the World" Desk Mat O2 + Odac by Mayflower Electronics (Rev A) 
AudioAudioAudioAudio
Sennheiser HD 800 2xKrk Rokit 6 G2 Blue Yeti (Grey) Rode PSA-1 Arm 
Other
Aquaero 6 LT + Aluminum Heatsink 
CPUMotherboardGraphicsRAM
i5 6600k Delid @ 4.848/4.848 Asus z170 Hero MSI GTX 980ti @ 1499/4082 1.25v/134% Power 2 x 8gb Gskills Ripjaws V 3131 16-16-16-32 
Hard DriveHard DriveHard DriveHard Drive
Samsung Pro Series 840 256gb (Celapaleis) 1 TB Samsung Drive (Pack Yak) Samsung Pro Series 850 256gb (Celapaleis Reprise) Phoenix Down (External Backup) 
CoolingCoolingOSMonitor
Noctua D14 Kraken x61 Push-Pull for GPU Windows 10 Pro Catleap 2560x1440 @ 60hz 
KeyboardPowerCaseMouse
CM Storm Trigger Brown Switches EVGA Supernova 1000w P2 Corsair 540 Air Razer Abyssus 
Mouse PadAudioAudioAudio
Steelseries QcK Mini Blue Yeti Microphone Objective Dac + Amp Sescom AB Switch 
AudioAudioAudioAudio
Krk Rokit 6 G2 x2 Radioshack Ground Loop Isolator (For Rokits) Wooaudio Aluminum Headphone Stand Sennheiser HD 800 
AudioOther
Logitech c920 Webcam NZXT Sentry Mix 2 Fan Controller 
  hide details  
Reply
post #11637 of 19637
What happen if I set 46x (stable already) with all Cstates ON and EIST OFF, Turbo OFF? is it ok?
post #11638 of 19637
Quote:
Originally Posted by Darkwizzie View Post

Does anybody here with a 4770k want to help me test something with chess? You must have a hyperthreaded CPU, preferably a Haswell one. It is said by the developer of Houdini chess engine that HT is bad for Houdini's engine strength, but just as importantly, that the decrease in strength is not as easily measured. Often people like to measure engine speed and relative strength by looking at kilonodes per second, but in this case it is deceiving. HT is said to increase kilonodes per second but decrease the efficiency of the search, decreasing depth and driving down the strength of the engine in the end. Robert Houdart said that chess engine software is in itself poorly optimized for HT, and this applies to basically all chess engines.

I want to see if this is true with Stockfish engine. So if you have free time, I can figure things out. One major reason why I got 4670k is because HT I believe is bad for chess for engines across the board. My idea is, you use Arena 3.5 GUI, latest beta Stockfish engine, load it up. Hash size to uhm, 512mb for example. So we'll have the engine analyze a position for a fixed amount of time and look at the ending search depth and how long it took to get to that depth.  And we might have to run this a few times, possibly with different positions to hit the nail in the common. 

If HT is as detrimental to engine strength for Houdini as it is for Stockfish, we should see a higher total nodes searched on HT on, but lower search depth. This means the engine searched for more stuff, but the search ended up with less results. Kind of like having a higher frequency on a CPU but a lower benchmark score due to having crappier IPC.

Unfortunately I do not have a HT enabled CPU or know a friend who has one.

Funny, 39x came up for me a lot even on sc2, because it's only loaded like max one core, 30-40% another, 5% on the last two - i expected maybe 38x or lower, but it seemed to run at the full turbo unless presented with high load across many cores

x264 perma-37x's for me
Insert Name Here
(14 items)
 
  
CPUMotherboardGraphicsRAM
6700k Asus Maximus VIII Hero Gigabyte Aorus Xtreme 1080ti Corsair LPX 2x8GB 3200c16 
Hard DriveHard DriveCoolingOS
Old Seagate HDD Samsung 850 EVO Thermalright Silver Arrow SB-E SE Windows 7 Home Premium 64 bit 
MonitorKeyboardPowerCase
Asus PG258Q (240hz + Gsync) WASDKeyboards.com v1 semi custom w/ mx browns, ... Superflower Golden Green HX550 Air540 
MouseMouse Pad
Logitech G Pro Qck+ 
  hide details  
Reply
Insert Name Here
(14 items)
 
  
CPUMotherboardGraphicsRAM
6700k Asus Maximus VIII Hero Gigabyte Aorus Xtreme 1080ti Corsair LPX 2x8GB 3200c16 
Hard DriveHard DriveCoolingOS
Old Seagate HDD Samsung 850 EVO Thermalright Silver Arrow SB-E SE Windows 7 Home Premium 64 bit 
MonitorKeyboardPowerCase
Asus PG258Q (240hz + Gsync) WASDKeyboards.com v1 semi custom w/ mx browns, ... Superflower Golden Green HX550 Air540 
MouseMouse Pad
Logitech G Pro Qck+ 
  hide details  
Reply
post #11639 of 19637
subscribe
post #11640 of 19637
why does my 1.64 cpuz show vid, not vcore anymore?
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel CPUs
Overclock.net › Forums › Intel › Intel CPUs › Haswell Overclocking Guide [With Statistics]