Are you going for reference Big Navi/RTX 3xxx or waiting for AIB offers?

Nebell

2[H]4U
Joined
Jul 20, 2015
Messages
2,379
Pretty sure I'm going to wait on MSI Gaming X version of 3080Ti as I have good experience with Gaming X cards.
 
Going aib for Nvidia or evga.... I need a card under 11" to fit in my case.
 
I'm guessing that AMD will finally do away with the reference blower also. I'd guess something like either the Radeon VII or the 5600XT depending on the power draw.
 
I'm going reference 3080/3080 Ti as soon as they're released. Looking to replace my 1080 Ti, so whatever I can get my hands on.
 
If I bite I'll probably just get the reference model. Video card OC'ing has been temperamental and noisy for me, so I'm fine just rolling with the stock one.
 
Maybe I'm old but AIB still means 'add in board' and what we're talking about here are dedicated external GPUs plugging in to PCIe slots, so by default they're AIB's, so why do people keep using this phrase when referring to the differences between AMD/NVIDIA default/reference designs over custom partner boards? It used to be about reference designs over custom designs, especially since some vendors will sell reference designs and custom designs so the current (wrong) 'AIB'' tag doesn't make much sense here either. Please help me out in understanding this craziness.
 
I'm going AIB as the reference cards don't have a 0 db mode. But since I have a mini itx case im limited to ref dimensions. 10.5" dual slot.
 
I'll probably wait and see. I water cool so a day 0 buy isn't going to do me any good; plenty of time to wait and see which if any aftermarket cards have significantly better power hardware and higher power limits as a result and then get a full cover block to match. If they all turn out to be duds, or not worth the markup anyway, I can fall back on a reference board instead.
 
Maybe I'm old but AIB still means 'add in board' and what we're talking about here are dedicated external GPUs plugging in to PCIe slots, so by default they're AIB's, so why do people keep using this phrase when referring to the differences between AMD/NVIDIA default/reference designs over custom partner boards? It used to be about reference designs over custom designs, especially since some vendors will sell reference designs and custom designs so the current (wrong) 'AIB'' tag doesn't make much sense here either. Please help me out in understanding this craziness.
And this is the first time you've seen people use a term incorrectly? This happens all the time, and it either dies off or becomes the norm. If you want to try to go around and correct everyone with the hopes it'll die, good on you. I just read it, know they mean partner boards and/or not reference design (because even board partners can still in release reference designed boards) and move on with my life. It's too short to worry about silly things like this. All that said, your are correct about the term not really meaning what people use it for.
 
Maybe I'm old but AIB still means 'add in board' and what we're talking about here are dedicated external GPUs plugging in to PCIe slots, so by default they're AIB's, so why do people keep using this phrase when referring to the differences between AMD/NVIDIA default/reference designs over custom partner boards? It used to be about reference designs over custom designs, especially since some vendors will sell reference designs and custom designs so the current (wrong) 'AIB'' tag doesn't make much sense here either. Please help me out in understanding this craziness.

The third parties are called "AIB Partners" but out of laziness people just abbreviate to "AIB." This is just netspeak. There are tons of abbreviations in everyday life that make no sense when you think about them, and really only exist for expediency.
 
I like AIB designs better, but I don't like waiting more, so probably get a reference of whatever AMD has.
 
Maybe I'm old but AIB still means 'add in board' and what we're talking about here are dedicated external GPUs plugging in to PCIe slots, so by default they're AIB's, so why do people keep using this phrase when referring to the differences between AMD/NVIDIA default/reference designs over custom partner boards? It used to be about reference designs over custom designs, especially since some vendors will sell reference designs and custom designs so the current (wrong) 'AIB'' tag doesn't make much sense here either. Please help me out in understanding this craziness.
I omitted the word "partners" after aib. Add that word and your concern is moot.
 
Yes, that is what the term stands for, but I believe it is now used to refer to 3rd-party partner designs. Language evolves.
 
Didn't AMD say they're next reference design was not going to be blower style or am I misremembering something I heard/read? If so, might not make to big of a difference. Nvidia reference design are normally pretty good, but obviously partner boards can can come with a free extras or slight differences. Reference boards are typically easiest to find water blocks for if that's your thing.
 
Didn't AMD say they're next reference design was not going to be blower style or am I misremembering something I heard/read? If so, might not make to big of a difference. Nvidia reference design are normally pretty good, but obviously partner boards can can come with a free extras or slight differences. Reference boards are typically easiest to find water blocks for if that's your thing.

The new 5600XT's "reference" card isn't a blower. I don't think we'll see another reference blower from them. You might see some reference PCB's with blowers depending on the manufacturer as there are always a few of them.
 
I’m going to just buy reference. Slap on a Heatkiller and add to the custom loop. My oc’d 2080ti doesn’t get hotter than 39c under max stress/)
 
Waiting for benchmarks. Not sure what the differences are for reference cards but every GPU I have owned was an AIB.
 
I think I'm skipping this generation. Gains won't be enough to even worth the hassle. I'd be surprised if it will be over 20% faster in anything but RTX.
 
Nvidia has a much better track record with reference cooling in the last few generations, and the leaked pics of the reference 3080 (non-ti) looks very innovative/cool (pun intended).
 
I've been so impressed with the ASUS Strix 2080Ti that I'll probably wait for the 3080Ti variant.

One thing I won't do is buy anything made by ZOTAC.
 
I believe you are going to be wrong. I guess we'll find out this winter.
I don't think he will be honestly. What ever AMD releases probably be around 2080ti speed at best. Nvidia been doing nothing but milking people. I expect HDMI 2.1 and a big RTX push with the 3xxx series. We going to be lucky if there isn't a price increase.
 
I've been so impressed with the ASUS Strix 2080Ti that I'll probably wait for the 3080Ti variant.

One thing I won't do is buy anything made by ZOTAC.

Well they're owned by the same company that owns Sapphire, curious why Zotac = bad?
 
I've bought Zotac before, and it worked fine, but the resale value was much less.
 
AIB for sure.. getting another ASUS Strix oc. Wondering how the resale value of the 2080ti will hold up after the release of ampre.
 
I think the 2080 Ti would retain some value, even after the 3000 series shipped.

I managed to get around $1,000 for my Titan X Pascal. I believe that was like 2 years ago.

So I got to enjoy the Titan X for like 2 years (I bought it when it released) and only lost $200 bucks. Fair deal.
 
I think a lot of people would see 2080ti performance at upper midrange price (vs $1000-1200+ even recently) as a nice upgrade especially if it comes with other features like updated hdmi etc. Where the 2080 ti falls in the new pecking order will also probably determine resale price.
 
I believe you are going to be wrong. I guess we'll find out this winter.

I'm cautiously optimistic. 12nm to 7nm is ~3x as many transistors for a given die size. Assuming pascal is 50/50 shading and raytracing, the rumor mill claiming +50% shading and 4x faster ray tracing is plausible. Shading would go from 0.5 pascals of transistors to 0.75, ray tracing from 0.5 to 2.0 for a total of 2.75x as many transistors as pascal has. That fits comfortably within the 3x larger transistor budget, and leaves room for either adding other things (eg more or bigger and faster memory controllers, or more video en/decode options), or a slight step back in die size from the monsters that powered the 2000 series.

Even if the rumor mill is overstating it by a factor of two, it'll finally be enough to justify replacing my 1080, so I'm almost certainly going to be buying a 3080/ti later this year.
 
Back
Top