what does "Revision: A1" mean? is it like "stepping" in CPUs?
>>56039901
For the most part, yes.
how can you tell before buying a graphics card which revision you are getting?
>>56039901
Yes, but graphics cards aren't often revised. Most stay at the initial revision throughout their lifespan. The GTX 780 is one exception, and the B1 revision models overclock far, far better than the A1 revision cards. Most B1s can get up to 1400MHz+, putting them above any 970.
>>56040013
A crystal ball.
>>56040013
By checking the serial numbers, you can determine the date of production from most AIB partners, which would let you determine revision.
>>56040134
Thanks senpai, I didn't know.
Time to start OC'ing.
>>56040134
This.
I think nVIDIA count from poly layer mask set A, metal layer mask set 1 (not 0).
I've also heard they use their own supercomputer (made by themselves, because that is what they do and where they make most of their cash) to validate by pure simulation: by the time they're getting samples back, they go from integration test straight to driver writing and shipping to board partners.
They do not hang about fixing errata. If it's fucked in some way they just monkey patch it in the video BIOS or the drivers: almost no hardware bug is too serious to work around. They do tool for this, but it's definitely flying by the seat of your pants, even compared to every other chip designer. Really bad drivers could absolutely brick your GPU.
Now you know why the Tegra was such an enormous trashfire. SoCs aren't quite so amenable to such a process.
AMD operate on a similarly accelerated schedule for their graphics cards. Competition is great for progress, not so hot for stability.
what about r9 series revisions ?
i can find jack shit regarding them .
is it a secret ?
my 390 has revision 80 .
no clue what to make of that
>>56040655
most R9's are the 285, 380, 290, 390 and all of them are AIB in most cases.
I have no idea about older cards so you have to ask others.
gainward """"""""""""golden"""""""""""""""""sample""""""""""""""
>>56041304
rest in peace your asic quality .
we hardly knew ye .
bagpipes
>>56041304
Daily reminder that ASIC quality has zero impact on overclocking ability and that you're a tech-illiterate who should fuck off back to /v/ if you think it does.
>>56041304
Most GTX 9xx have extremely low ASIC quality. Over 75% is very rare.
Not that it means anything. Since 100% quality chip might overclock like shit.
>>56041571
I know well what it does, i just find it funny that it's branded as golden sample and has shit asic quality