Overclock.net › Forums › Intel › Intel CPUs › *Official* Intel DDR4 24/7 Memory Stability Thread
New Posts  All Forums:Forum Nav:

*Official* Intel DDR4 24/7 Memory Stability Thread - Page 265

post #2641 of 5714
Lower tRFC will scale up performance, with diminishing returns, as far as you care to take it because less time is spent refreshing. However, stability will get flaky very rapidly as you near the minimum values a particular set of ICs can tolerate at a given clock speed. The difference between completely stable (or as close as one can get) and having to reinstall your OS can be razor thin.

tREFI works similarly to tRFC (former is time between refreshes, later is how long a refresh lasts), though higher is better and gains tend to be more linear. The line between stable and unstable is usually much more subtle though as well as more temperature dependent.

With regard to tWCL, you can test and see if the value is actually sticking pretty easy with memory write benchmarks. If it's offering benefit, then there is no reason not to use it, as long as you can be convinced of stability.
Primary
(15 items)
 
Secondary
(13 items)
 
In progress
(10 items)
 
CPUMotherboardGraphicsRAM
5820K @ 4.2/3.5GHz core/uncore, 1.175/1.15v Gigabyte X99 SOC Champion (F22n) Gigabyte AORUS GTX 1080 Ti (F3P) @ 2025/1485, 1... 4x4GiB Crucial @ 2667, 12-12-12-28-T1, 1.34v 
Hard DriveHard DriveHard DriveCooling
Plextor M6e 128GB (fw 1.06) M.2 (PCI-E 2.0 2x) 2x Crucial M4 256GB 4x WD Scorpio Black 500GB Noctua NH-D15 
OSMonitorKeyboardPower
Windows 7 Professional x64 SP1 BenQ BL3200PT Filco Majestouch Tenkeyless (MX Brown) Corsair RM1000x 
CaseMouseAudio
Fractal Design Define R4 Logitech G402 Realtek ALC1150 + M-Audio AV40 
CPUMotherboardGraphicsRAM
X5670 @ 4.4/3.2GHz core/uncore, 1.36 vcore, 1.2... Gigabyte X58A-UD5 r2.0 w/FF3mod10 BIOS Sapphire Fury Nitro OC+ @ 1053/500, 1.225vGPU/1... 2x Samsung MV-3V4G3D/US @ 2000, 10-11-11-30-T1,... 
RAMHard DriveHard DriveHard Drive
1x Crucial BLT4G3D1608ET3LX0 @ 2000, 10-11-11-3... OCZ (Toshiba) Trion 150 120GB Hyundai Sapphire 120GB 3x Hitachi Deskstar 7k1000.C 1TB 
CoolingOSPowerCase
Noctua NH-D14 Windows 7 Pro x64 SP1 Antec TP-750 Fractal Design R5 
Audio
ASUS Xonar DS 
CPUMotherboardGraphicsRAM
i7-6800K @ 4.3/3.5GHz core/uncore, 1.36/1.2v ASRock X99 OC Formula (P3.10) GTX 780 (temporary) 4x4GiB Crucial DDR4-2400 @ 11-13-12-28-T2, 1.33v 
Hard DriveHard DriveCoolingOS
Intel 600p 256GB NVMe 2x HGST Travelstar 7k1000 1TB Corsair H55 (temporary) Windows Server 2016 Datacenter 
PowerCase
Seasonic SS-860XP2 Corsair Carbide Air 540 
  hide details  
Reply
Primary
(15 items)
 
Secondary
(13 items)
 
In progress
(10 items)
 
CPUMotherboardGraphicsRAM
5820K @ 4.2/3.5GHz core/uncore, 1.175/1.15v Gigabyte X99 SOC Champion (F22n) Gigabyte AORUS GTX 1080 Ti (F3P) @ 2025/1485, 1... 4x4GiB Crucial @ 2667, 12-12-12-28-T1, 1.34v 
Hard DriveHard DriveHard DriveCooling
Plextor M6e 128GB (fw 1.06) M.2 (PCI-E 2.0 2x) 2x Crucial M4 256GB 4x WD Scorpio Black 500GB Noctua NH-D15 
OSMonitorKeyboardPower
Windows 7 Professional x64 SP1 BenQ BL3200PT Filco Majestouch Tenkeyless (MX Brown) Corsair RM1000x 
CaseMouseAudio
Fractal Design Define R4 Logitech G402 Realtek ALC1150 + M-Audio AV40 
CPUMotherboardGraphicsRAM
X5670 @ 4.4/3.2GHz core/uncore, 1.36 vcore, 1.2... Gigabyte X58A-UD5 r2.0 w/FF3mod10 BIOS Sapphire Fury Nitro OC+ @ 1053/500, 1.225vGPU/1... 2x Samsung MV-3V4G3D/US @ 2000, 10-11-11-30-T1,... 
RAMHard DriveHard DriveHard Drive
1x Crucial BLT4G3D1608ET3LX0 @ 2000, 10-11-11-3... OCZ (Toshiba) Trion 150 120GB Hyundai Sapphire 120GB 3x Hitachi Deskstar 7k1000.C 1TB 
CoolingOSPowerCase
Noctua NH-D14 Windows 7 Pro x64 SP1 Antec TP-750 Fractal Design R5 
Audio
ASUS Xonar DS 
CPUMotherboardGraphicsRAM
i7-6800K @ 4.3/3.5GHz core/uncore, 1.36/1.2v ASRock X99 OC Formula (P3.10) GTX 780 (temporary) 4x4GiB Crucial DDR4-2400 @ 11-13-12-28-T2, 1.33v 
Hard DriveHard DriveCoolingOS
Intel 600p 256GB NVMe 2x HGST Travelstar 7k1000 1TB Corsair H55 (temporary) Windows Server 2016 Datacenter 
PowerCase
Seasonic SS-860XP2 Corsair Carbide Air 540 
  hide details  
Reply
post #2642 of 5714
Quote:
Originally Posted by Praz View Post

Hello

Anyone running memory higher than 3200MHz on either Haswell-E or Broadwell-E need to do so with their eyes open. At these speeds OC Socket settings are being manipulated behind the scenes to levels that I personally am not comfortable with for 24/7. I understand how this may not be evident when the available user settings do not give a direct indication of what and to how high of a level voltages are being applied.

Quote:
Originally Posted by Jpmboy View Post


rut-oh. my cpu is being juiced in a clandestine manner - really?

Praz, I've had some situations where I experienced significantly higher voltages than I specified in the BIOS (much as you warn) when I read the Aida measurements, but I'm assuming the reported voltages are correct, hopefully? I ended up putting lower than the expected in the BIOS and by several tries got a reading around what I was looking for.
post #2643 of 5714
So I been having issues with my Ram stability at XMP settings at 3200mhz no matter what I do with SA and VCCIO. I lower it down to 3000mhz and it is stable with HCI memtest with auto on SA and VCCIO. Guess I will settle for 3000 and just try to tighten timings or something. Disappointing though since its binned for 3200.
My latest Rig
(16 items)
 
  
CPUMotherboardGraphicsRAM
AMD Ryzen 7 1700 AsRock X370 TaiChi EVGA GTX 1080 ACX 3.0 G.Skill 32MB 3200mhz DDR4 
Hard DriveHard DriveCoolingCooling
Samsung EVO 500GB SSD Samsung EVO Pro 512MB M2 Drive XSPC Raystorm Pro AlphaCool ST30 360 
CoolingOSMonitorKeyboard
Corsair SP120 x6 push/pull Windows 10 Acer Predator X34 Corsair K70 RGB 
PowerCaseMouseMouse Pad
Corsair AX1200 Inwin 909 Logitech G502 Proteus Spectrum Corsair MM300 
  hide details  
Reply
My latest Rig
(16 items)
 
  
CPUMotherboardGraphicsRAM
AMD Ryzen 7 1700 AsRock X370 TaiChi EVGA GTX 1080 ACX 3.0 G.Skill 32MB 3200mhz DDR4 
Hard DriveHard DriveCoolingCooling
Samsung EVO 500GB SSD Samsung EVO Pro 512MB M2 Drive XSPC Raystorm Pro AlphaCool ST30 360 
CoolingOSMonitorKeyboard
Corsair SP120 x6 push/pull Windows 10 Acer Predator X34 Corsair K70 RGB 
PowerCaseMouseMouse Pad
Corsair AX1200 Inwin 909 Logitech G502 Proteus Spectrum Corsair MM300 
  hide details  
Reply
post #2644 of 5714
Quote:
Originally Posted by Praz View Post

Hello

Both tWTR and tRTP are limited by the burst length. A setting less than 4 will be valid only if the board is applying the value as an offset instead of an absolute.

With this idea, it would really help if you or @Raja@ASUS can give us an insight (given your association with Asus) whether Asus applies these values as Offsets or Absolute.

Has this been mentioned elsewhere?
post #2645 of 5714
Thread Starter 
Quote:
Originally Posted by greg1184 View Post

So I been having issues with my Ram stability at XMP settings at 3200mhz no matter what I do with SA and VCCIO. I lower it down to 3000mhz and it is stable with HCI memtest with auto on SA and VCCIO. Guess I will settle for 3000 and just try to tighten timings or something. Disappointing though since its binned for 3200.


Can you confirm what kit you are using?
post #2646 of 5714
Quote:
Originally Posted by djgar View Post


Praz, I've had some situations where I experienced significantly higher voltages than I specified in the BIOS (much as you warn) when I read the Aida measurements, but I'm assuming the reported voltages are correct, hopefully? I ended up putting lower than the expected in the BIOS and by several tries got a reading around what I was looking for.


Aida doesn't read the "voltages" Praz is referring to.
post #2647 of 5714
Quote:
Originally Posted by Raja@ASUS View Post

Aida doesn't read the "voltages" Praz is referring to.

I am curious as to what voltages are being referred to?
PC
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7-6700K eVGA Classified K eVGA GTX970SSC G.Skill F4-3600C16-16GTZ 
Hard DriveOptical DriveCoolingOS
2x Samsung 960 EVO M.2 RAID0 Samsung SH-S223L Custom Loop D5 Vario, Thermochill PA140.3, Heat... W7 U SP1 x64 
MonitorKeyboardPowerCase
Viewsonic VX2770 Logitech Corsair AX1200, APC RS1500 LCD Thermaltake VG4000SNA  
Mouse
Logitech 
  hide details  
Reply
PC
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7-6700K eVGA Classified K eVGA GTX970SSC G.Skill F4-3600C16-16GTZ 
Hard DriveOptical DriveCoolingOS
2x Samsung 960 EVO M.2 RAID0 Samsung SH-S223L Custom Loop D5 Vario, Thermochill PA140.3, Heat... W7 U SP1 x64 
MonitorKeyboardPowerCase
Viewsonic VX2770 Logitech Corsair AX1200, APC RS1500 LCD Thermaltake VG4000SNA  
Mouse
Logitech 
  hide details  
Reply
post #2648 of 5714
Quote:
Originally Posted by greg1184 View Post

So I been having issues with my Ram stability at XMP settings at 3200mhz no matter what I do with SA and VCCIO. I lower it down to 3000mhz and it is stable with HCI memtest with auto on SA and VCCIO. Guess I will settle for 3000 and just try to tighten timings or something. Disappointing though since its binned for 3200.

I'm guessing those are double-sided 8GB sticks, which put quite a strain on the IMC. I happen to have the same components as you (CPU, mem, mobo) so I can do a test for comparison, though of course the silicon lottery disclaimer applies as always.
post #2649 of 5714
Quote:
Originally Posted by Jpmboy View Post

oh well, I'm waaay off the Reservation at this point, no sense in turning back... guess I have an on-going robustness test of sorts
The song Suicide Blonde comes to mind. redface.gif
64GB

Thank you very much for taking a look! Yes these settings have been used for a little over a week and stressed in all sorts of manners including gsat. No errors from that or memtest86+.
1) I have heard mixed stories about tREFI. On one stance if you have no issues with stability you should max it out. One the other I have heard the maximus hero viii actually sets these values very well and should always be left on auto. Taking what you said into account I think I will move that value back to auto and see what I can do as far as further adjusting terciary timings.
2) If I am not having an issue using the current timings should this value still be adjusted?
3) When I was asking questions on the ram addict club forum the guys there told me that I should always set to a value of 9 despite CAS. Them knowing much more than me I listened haha. Is this a problem ? I only asked bc it was more than ones opinion in the thread to adjust this to 9 despite CAS
4) I know the value of 16 is well below 4*(tRRD)=24. Im guessing 24 is being applied there.. Triyingto reduce tRRD further resulted in code 3E.

1) Try 2x what the board sets on auto.. so far this has been fine with a continuous ram disk (and days of up time) and overnight sleep (suspend to ram) cycles. But there's always an inherent risk.
2) does setting it that low actually improve the performance? If yes, then stand pat.
3) remember, the guys in the ram addict club are fairly extreme in their uses and voltages (which is fine). I'm running 9 with cas 13 also, but got there differently. wink.gif
4) No idea what value the bios/board/mc is applying to the timing error. By setting it to 4x tRRD you know what value is being used. If you can lower it from there and gain performance and stability/reliability - the board may be running an offset as Praz and Raja have suggested.
These two guys know this sheet like you (we) know your day job... I tend to listen to them, and then wander off the margins at my own risk. wink.gif[/quote]


Quote:
Originally Posted by Blameless View Post

Lower tRFC will scale up performance, with diminishing returns, as far as you care to take it because less time is spent refreshing. However, stability will get flaky very rapidly as you near the minimum values a particular set of ICs can tolerate at a given clock speed. The difference between completely stable (or as close as one can get) and having to reinstall your OS can be razor thin.

tREFI works similarly to tRFC (former is time between refreshes, later is how long a refresh lasts), though higher is better and gains tend to be more linear. The line between stable and unstable is usually much more subtle though as well as more temperature dependent.

With regard to tWCL, you can test and see if the value is actually sticking pretty easy with memory write benchmarks. If it's offering benefit, then there is no reason not to use it, as long as you can be convinced of stability.

Hey guys thank you very much for the detailed explanation! I spent some time testing and this is what I found. Note setting were proven stable so the rest is under the assumption all testing was completed at stable operating conditions.
1) changing tREFI to auto then changing it to 2*Auto and finally adjusting it to MAX value shows zero changes in perceived stability and AIDA throughput. I also tested with superPI and 32M receives same time. With that should I just leave the values set to auto or is there something about the longer refresh that will help other things not being seen by this particular test method?
2,3 & 4) Changing both values higher (tRFC to 320, auto then back to 250), ( tFAW from 16 to 24)and( tWCL from 9 to 13) all show very slight but quantifiable improvements in throughput on AIDA with a four run average. So I guess it seems to actually be applying the values so im assuming I should leave these alone since there are no issues?

Again thank you for the help it goes a long way in understanding the relationship between all the timings. maybe @Raja@ASUS can chime in and fill me/us in on what is really going on here. thumb.gif
post #2650 of 5714
Quote:
Originally Posted by Blameless View Post

Lower tRFC will scale up performance, with diminishing returns, as far as you care to take it because less time is spent refreshing. However, stability will get flaky very rapidly as you near the minimum values a particular set of ICs can tolerate at a given clock speed. The difference between completely stable (or as close as one can get) and having to reinstall your OS can be razor thin. Warning: Spoiler! (Click to show)
tREFI works similarly to tRFC (former is time between refreshes, later is how long a refresh lasts), though higher is better and gains tend to be more linear. The line between stable and unstable is usually much more subtle though as well as more temperature dependent.

With regard to tWCL, you can test and see if the value is actually sticking pretty easy with memory write benchmarks. If it's offering benefit, then there is no reason not to use it, as long as you can be convinced of stability.

This is SOOOO important... and the corruption happens without any overt signs or warnings. ohno-smiley.gif
x99
(22 items)
 
Z370
(8 items)
 
x299
(18 items)
 
CPUCPUMotherboardGraphics
i7 5960X i7 6950X Asus Rampage V Extreme 10 2 GTX Titan X Pascal 
RAMHard DriveHard DriveHard Drive
GSkill 3200c14 TZs 8x8GB @ 3400c13 Intel 750 NVMe 400GB 2x Plextor SSD 256 Raid 0 (Win 7) Samsung NVMe 950 Pro M.2 
Optical DriveCoolingCoolingCooling
Plextor 810 2x XSPC RX360 Koolance 380i D5 
CoolingOSOSOS
Aquaero 6 Windows 10 x64  Windows 7 Pro Linux Mint 
MonitorKeyboardPowerCase
Seiki 50" 4K monitor (720P-2160i) 240-30Hz Das Keyboard Model S Pro Corsair AX1200 Case Labs SM8 
MouseAudio
Steelseries Rival Ultimate Ears TF10 
CPUMotherboardGraphicsRAM
8700K ASUS Maximus X Apex GTX 1080 2x8 GB G.Skill 4400c19 
Hard DriveCoolingOSCase
Samsung 960 Pro 360Rad + Chiller Win 10 HWBOT OBT 
CPUCPUMotherboardGraphics
7740X 7980XE ASUS Rampage Vi Apex 2x Nvidia Titan Xp SLi 
RAMHard DriveHard DriveHard Drive
G.Skill 3600c15 2x8GB kits Intel 900p WD Raptor 1T Plextor SSDs Raid 0 
Optical DriveCoolingOSOS
Plextor DVD/BR R/W (very) Custom Water Windows 10 Pro Windows 7 Pro 
MonitorKeyboardPowerCase
ASUS 144Hz Ducky Corsair 1500i Microcool Benchetto 101 
MouseAudio
Steelseries Rival Ultimate ears TF10s (IEMs) 
  hide details  
Reply
x99
(22 items)
 
Z370
(8 items)
 
x299
(18 items)
 
CPUCPUMotherboardGraphics
i7 5960X i7 6950X Asus Rampage V Extreme 10 2 GTX Titan X Pascal 
RAMHard DriveHard DriveHard Drive
GSkill 3200c14 TZs 8x8GB @ 3400c13 Intel 750 NVMe 400GB 2x Plextor SSD 256 Raid 0 (Win 7) Samsung NVMe 950 Pro M.2 
Optical DriveCoolingCoolingCooling
Plextor 810 2x XSPC RX360 Koolance 380i D5 
CoolingOSOSOS
Aquaero 6 Windows 10 x64  Windows 7 Pro Linux Mint 
MonitorKeyboardPowerCase
Seiki 50" 4K monitor (720P-2160i) 240-30Hz Das Keyboard Model S Pro Corsair AX1200 Case Labs SM8 
MouseAudio
Steelseries Rival Ultimate Ears TF10 
CPUMotherboardGraphicsRAM
8700K ASUS Maximus X Apex GTX 1080 2x8 GB G.Skill 4400c19 
Hard DriveCoolingOSCase
Samsung 960 Pro 360Rad + Chiller Win 10 HWBOT OBT 
CPUCPUMotherboardGraphics
7740X 7980XE ASUS Rampage Vi Apex 2x Nvidia Titan Xp SLi 
RAMHard DriveHard DriveHard Drive
G.Skill 3600c15 2x8GB kits Intel 900p WD Raptor 1T Plextor SSDs Raid 0 
Optical DriveCoolingOSOS
Plextor DVD/BR R/W (very) Custom Water Windows 10 Pro Windows 7 Pro 
MonitorKeyboardPowerCase
ASUS 144Hz Ducky Corsair 1500i Microcool Benchetto 101 
MouseAudio
Steelseries Rival Ultimate ears TF10s (IEMs) 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel CPUs
Overclock.net › Forums › Intel › Intel CPUs › *Official* Intel DDR4 24/7 Memory Stability Thread