Overclock Help Silicon Lottery Binning Results Deidded i7-7700K

[lockheedmartinengineering]

I created a post back in November which is located here: http://www.tomshardware.com/answers/id-3559133/7700k-83c-spike-temps-deepcool-gammax-400.html

I now have my chip delidded and binned, upgraded to a Corshair H115i cooler, added more efficient fans, and I'm running a push pull config on my raidator via two 4 (four) 140mm corsair fans. Since I had my computer apart, I also added some WideBeam UV/RGB lights and UV reactive cable extensions. Also added some more RGB ram and a couple Hybrid Hardrives so I don't fill up my M.2 NVMe SDD. So I really like the way it came out.

OH, BY THE WAY, my temps are about 35-40 degrees while idle with an ambient temp of 24 degrees C. While gaming, my temps go up to around 50-55 degrees. In comparison, with the same game on my old air cooler, my temps would reach 75-78C. These are stock voltages with XMP enabled. I haven't applied any real overclocking voltages yet.

I wanted to get some input about the binning part. Silicon Valley sent me back a note saying:
"Your CPU should be stable at 5.1 GHz when using a 1.437 Vcore with a -2 AVX offset."

My question is, are those the only two settings that I should change? Set it and forget it?

BEFORE Delidding & Binning with Air Cooler:

After Binning & Delidding with AIO:

FULL SYSTEM SPECS:
INWIN 805: Front Panel Modified for Higher Air Flow Rate
Processor: Stock Clock Rates & Voltages. KABY LAKE INTEL 7TH GENERATION CORE I7-7700K
Cooler: Corsair H115i 280mm Radiator with 4 Corsair 140mm Fans Push Pull Configuration
MONITOR #1: SAMSUNG U28E590D 28 INCH UHD LED-LIT MONITOR WITH FREESYNC SUPPORT 3480X2160
MONITOR #2: ASUS VE278H 27" FULL HD 1920x1080 2MS HDMI VGA BACK-LIT LED MONITOR 169.00
FAN HUB: SILVERSTONE PWM FAN HUB SYSTEM CABLES (CPF04) 14.44
RAM: 32 GB 4 QUANTITY - XMP Enabled CORSAIR VENGENCE RGB 16GB DDR4 3466 (PC4-2700) C16-INTEL 100/200 SERIES PC MEMORY CMR16GX4M2C3466C16
Graphics: ASUS GEFORCE GTX 1080 8GB ROG STRIX GRAPHICS CARD (STRIX-GTX1080-A8G-GAMING)
Keyboard: Corsair Strafe RGB
Mouse: Logitech G502 Proteus Spectrum RGB
SSD: SAMSUNG 960 EVO SERIES - 500GB NWMe M.2 INTERNAL SSD (MZ-V6E500BW)
HDD #1: 2.5" Hybrid SSHD Segate 1TB (7200 RPM I think)
HDD #2: 2.5" Hybrid SSHD Segate Firecuda 2TB (7200 RPM I think)
FANS: 1 QTY ML 120mm as fan exhaust in back 75 CFM Max, 2 QTY Corsair HD120 in bottom as intake, 4 QTY LL140 has Radiator intake fans in the front
Power Supply: EGA SUPERNOVA 650 G2 80+ GOLD 650W FULLY MODULAAR EVGA ECO MODE 7 YEAR WARRANTY FREE POWER ON SELF TESTER 220-G2-0650-Y1 | 90% EFFICIENCY
Lights: CableMod WideBeam UV/RGB Strips, one 60cm strip and one 30cm strip.
Green Cable Extensions: Amazon UV Reactive
Radiator Hose & Wire Mod: PrimoChill Anti-Kink Coil

 

[lockheedmartinengineering]

Also guys, I need help to validate the need to overclock. I can overclock now, but I haven't experienced or noticed any lag or processor speed issues with any game, or any task. In what instances would I be able to notice a performance increase? I really want to overclock, but having an issue validating a reason. If it shortens the life of your CPU right?

BTW, I have all of my fans running at 100%. I don't mind the extra little fan noise, it really isn't that noticeable if you ask me. It makes sense that it would shorten the life of your fans. To that, I figure since an AIO lasts 3-5 years, maybe the fans will last at least that long when running them at 100% and I'll just change the fans with the radiator when the pump goes bad.

 

[1LiquidPC]

Hi there. First off, nice build The OC performance difference can vary greatly on what tasks you are performing. Your systerm is already very fast in games, at best in ones that are CPU bound, pick up some FPS. Others you wouldn't gain any. Encoding, rendering, ect. can improve a good bit, but again depends on software. Some things would again be about the same. Shortening the lifespan is up for debate depending on settings, usage, and lifespan of build. Most likely if OC settings are reasonable, temps good, ect., you would get a new PC before the CPU would have issues that could be related to OC.

You really don't need fans at 100%. I never really had a need to, unless airflow is very bad in a case to begin with. Again, up to you. Temps would be better, but more noise and lower fan life and higher power consumption, though not all that much.

Regarding Silicon Lottery, those really are the only 2 settings you need. They do actually add a bit extra than needed for a buffer of sorts regarding voltage. I sent my CPU to them for same reasons, in my own testing I got the same result for frequency for stability so I do trust their delidding and improved binning methods(They recently changed). I did drop the voltage a little and set 0 for AVX offset instead of exactly whats on the paper they send Be aware that the AVX offset doesn't work correctly on some vendor's boards, but that is another story. Hope this helps.

 

[lockheedmartinengineering]

Hi there. First off, nice build The OC performance difference can vary greatly on what tasks you are performing. Your systerm is already very fast in games, at best in ones that are CPU bound, pick up some FPS. Others you wouldn't gain any. Encoding, rendering, ect. can improve a good bit, but again depends on software. Some things would again be about the same. Shortening the lifespan is up for debate depending on settings, usage, and lifespan of build. Most likely if OC settings are reasonable, temps good, ect., you would get a new PC before the CPU would have issues that could be related to OC.

You really don't need fans at 100%. I never really had a need to, unless airflow is very bad in a case to begin with. Again, up to you. Temps would be better, but more noise and lower fan life and higher power consumption, though not all that much.

Regarding Silicon Lottery, those really are the only 2 settings you need. They do actually add a bit extra than needed for a buffer of sorts regarding voltage. I sent my CPU to them for same reasons, in my own testing I got the same result for frequency for stability so I do trust their delidding and improved binning methods(They recently changed). I did drop the voltage a little and set 0 for AVX offset instead of exactly whats on the paper they send Be aware that the AVX offset doesn't work correctly on some vendor's boards, but that is another story. Hope this helps.

Thanks so much. You have an awesome build as well!

I have another question. Since my chip is stable at "5.1 GHz when using a 1.437 Vcore with a -2 AVX offset," is it safe to assume it's stable at all settings that are lower than those two numbers? Or is it possible to encounter instability even at lower voltages and AVX offset? Thanks for such awesome help and input as it is greatly valued and appreciated. I'm assuming that this may be dependent on the board as you said. But, assuming it is stable at the settings stated by Silicon Lottery, will it more than likely be stable at lower settings?

 

[1LiquidPC]

Hi again. Sorry for late reply, but thanks for the compliments. Regarding your question, are you referring to lowering your speed to 5.0, 4.9, ect.? To my knowledge, Silicon Lottery only tests CPUs at a single voltage, most likely based on averages from their own testing based on several hundred CPUs. They basically set the voltage, and try to boot at a certain speed, if it passes, run tests for stability. If it fails, they lower the speed 1 notch, and try again. This is my theory, but based on some time in the forums it seems it would be close to this. The -2 AVX seems to be the default as mine had the same info on the printed page. I think it's again for a safety cushion for stability. To sum up, if you are stable at 5.1 with settings, you could lower the CPU a bit and take some voltage off, but in this territory, you are on your own really. If you use the boards they list on their QVL, you should be fine with the settings provided.

What you are asking here is what I did myself. I run my own tests, though I can't say for certain it's as stressful as Silicon Lottery's. I ended up keeping the CPU at tested speed, I got same result, but lowered the voltage a little. What's considered "stable" in OCing depends on who you ask. Some folks will say it's stable as long as what you work with runs fine with no issues. Others will only say it's truly "stable" if it runs full stress load 24-48 hrs on every test imaginable. Only you could decide which side to take on this one.