How does binning for VF scaling work?  I'm not familiar with that concept.  Define "VF scaling".

talon951
How does binning for VF scaling work?  I'm not familiar with that concept.  Define "VF scaling".

VF scaling is short for Voltage Frequency scaling. I don't know all the specifics on the steps they take, but I do know 100% without a doubt that they do indeed bin chips for these cards, and I know that they use stock boost frequency binning as one part of the process. As evidenced by all the markings on different dies from cards in my own personal collection and has been well documented elsewhere. Those markings (EVGA current policy now mandates that they get erased) always directly translate Into the stock boost bin.

mylittlepwny2

talon951
How does binning for VF scaling work?  I'm not familiar with that concept.  Define "VF scaling".

VF scaling is short for Voltage Frequency scaling. I don't know all the specifics on the steps they take, but I do know 100% without a doubt that they do indeed bin chips for these cards, and I know that they use stock boost frequency binning as one part of the process. As evidenced by all the markings on different dies from cards in my own personal collection and has been well documented elsewhere. Those markings (EVGA current policy now mandates that they get erased) always directly translate Into the stock boost bin.

VF Scaling tests males most sense for a company like EVGA. True ATE IC/Chip/die level equipment that nVidia/AMT/Intel have costs millions of dollars to own and maintain. EVGA is at the mercy of nVidia and can only run a small hand full of tests on a custom socket fitting to one of their KPE PCB's just to wire up and test VF scale/curve without having to solder the chip down. Then binning them this way with their own process. 

UndesiredSanity

mylittlepwny2

talon951
How does binning for VF scaling work?  I'm not familiar with that concept.  Define "VF scaling".

VF scaling is short for Voltage Frequency scaling. I don't know all the specifics on the steps they take, but I do know 100% without a doubt that they do indeed bin chips for these cards, and I know that they use stock boost frequency binning as one part of the process. As evidenced by all the markings on different dies from cards in my own personal collection and has been well documented elsewhere. Those markings (EVGA current policy now mandates that they get erased) always directly translate Into the stock boost bin.

VF Scaling tests males most sense for a company like EVGA. True ATE IC/Chip/die level equipment that nVidia/AMT/Intel have costs millions of dollars to own and maintain. EVGA is at the mercy of nVidia and can only run a small hand full of tests on a custom socket fitting to one of their KPE PCB's just to wire up and test VF scale/curve without having to solder the chip down. Then binning them this way with their own process. 

Yep that's definitely part of it. But I'm not sure what other steps they might also take. But at the end of the day good scaling chips tend to have higher stock frequencies all else being equal. So they look for stock boost frequencies and use that as a large part of the determining factor when binning GPUs and deciding which ones are good enough to make the cut.

When comparing my 3090 kpe clocks to other 3090 kpe owners who used their cards on the stock coolers mine wasn’t that great.

Hasn't someone quoted Jacob if this Kingpin cards are binned? No one flat out asked prior?

AHowes
Hasn't someone quoted Jacob if this Kingpin cards are binned? No one flat out asked prior?

He's stated the FTW3 cards specifically are "binned" but thats not the same as what the guys/gals are talking about above.  They are "binning" cards to make sure they get the advertised clocks they are selling and that to me goes the same with the kingpins.  Now granted the kingpins are binned higher then a FTW3 card they aren't in the sure essence that EVGA is paying extra for "top binned" cards or spending time binning all dies and hand picking the top few.  Now maybe for some of those sponsored by EVGA for XOC, lol.

Yes,3090TI FTW3 ULTRA was binned,but it was only pretested to make sure it could run stably in stock bios.I'm thinking there is no difference between FTW3 BLACK and ULTRA,they are the same except their BIOS.Also,most 3090TI's PCB layouts are the same,except 3090TI KINGPIN and Galaxy HOF OC LAB EDITION.
EVGA did officially guarantee the ASIC quality for 980ti(four levels from 72% to 80%),clocks for 1080ti KINGPIN(2025Mhz),but 3090 and 3090TI KINGPIN were not.I thought EVGA had some issues when producing KINGPIN.
I owned all KINGPIN series except GTX 980 KINGPIN,and my 2080ti KINGPIN can both run in 2175mhz and higher with HYDRO COPPER Kit installed.I'm sure they are binned to run in higher clocks.
Maybe it is not wise to purchase the first batch of KINGPIN,as those really "top binned" cards would be sent to those overclockers for advertising.

My 3090 ti had 2040 wrote on it. That is where it runs mostly.

Is it normal that I can play FC5 @ 2205Mhz but Metro Exodus Enhanced Version I top out @ 2175Mhz. FC5 is pure rasterization and metro exodus is Ray traced and DLSS. Does using Ray Tracing lower the potential Max OC for a given game? I was getting pissed that Metro was CTD every now and then so I kept lowering my MSI Afterburner settings to +145C +900M and just gamed for nearly 40 minutes and all was perfect, where as before @ 2205 it would CTD within minutes. FC5 could care less, I can play @ 2230 and have no issues. Keep in mind, my temps never exceed 39c