Guest Editorial – NVIDIA SFF-Ready Program is Misguided and Not Guiding Informed Purchase Decisions
- Thread starter confusis
- Start date
Marketing came up with this SFF-Ready stuff. Clearly they didn't listen to any engineers.
Looking forward to the next article. And I would still love to see 200w and below cards available with single slot, or at least 'ITX'. Triple slot, triple fan 100-150w cards are ridiculous.
Also on my shortlist are cases like the Phanteks "ITX" case that are larger than needed to fit a MicroATX motherboard (and after accounting for ridiculous "feet" are only an inch from being able to take full ATX)
Looking forward to the next article. And I would still love to see 200w and below cards available with single slot, or at least 'ITX'. Triple slot, triple fan 100-150w cards are ridiculous.
Also on my shortlist are cases like the Phanteks "ITX" case that are larger than needed to fit a MicroATX motherboard (and after accounting for ridiculous "feet" are only an inch from being able to take full ATX)
Got something nice to add to the door frame analogy: In CSS, there is padding and margin.
Padding is the inner additional space of an HTML element,while margin is the space surrounding the element.
So for the analogy of the door frame, what we need to know is the margin (the whole outside dimensions), NOT the padding, which could be eg. include the door angles.
cu, w0lf.
Padding is the inner additional space of an HTML element,while margin is the space surrounding the element.
So for the analogy of the door frame, what we need to know is the margin (the whole outside dimensions), NOT the padding, which could be eg. include the door angles.
cu, w0lf.
At the end of the day, the SFF-ready program feels to me like the usual kind of "decision-maker" thing: Its the fifth iteration of anothers persons work, which was done by at least half a dozens persons before in at least four different ways, that had been declined or ignored by the big heads on top for years onwards; and then suddenly somebody finally brought this topic up as a problem ..
.. so here we are, the regurgitated, over and over and over reworked idea, half-way good, rest total garbage, but "we finally need to ship it", D.O.A. final edition.
cu, w0lf.
.. so here we are, the regurgitated, over and over and over reworked idea, half-way good, rest total garbage, but "we finally need to ship it", D.O.A. final edition.
cu, w0lf.
It's a mess for everyone designing SFF cases without having the budget or contacts to have a hands-on experience with every card from every vendor in current generation. If you don't have a list with actual measurements for each card measured the same way, it's hard to set a cut-off dimensions at some specific arbitrary spot because some of the cards that you think are inside may be outside because the dimensions were measured excluding bracket.
The worst case scenario is that you assume specific set of the cards to be exactly fitting and then most of them do not fit because of dimensions being not what you expected them to be.
At some point I thought that maybe the solution for this problem would be to force everyone to have the cable connectors at the reference PCB height line, so for any oversized/taller cards at least the oversize wouldn't affect the space for cable moving as well. Some vendors do it this way, but not everyone and a lot cards keep being oversized with connector placement shifted.
There's this saying going like "we sent people to moon 50 years ago and yet here we are struggling with centering an image on the website" which makes this CSS analogy quite good here. We could go forward with the SFF form factors a lot faster if the approach was to solve problems instead of ignoring them, but here we are struggling to explain someone how big is the object he wants to buy.
I think it's just the business/marketing approach to doing things: They focused first on those who want to have a simple choice while also limiting this to a close group of partner companies, so it's just a bean counter approach to try and gain whenever you can and gains here are not letting others in.
Great article - I agree that it feels like they're trying to reinvent this stuff as a marketing stunt.
Realistically they just need to consider from an engineering and product design perspective - we need to know the accurate dimensions and shape of a product, so give us accurately measured and documented illustrations.
Simples.
Realistically they just need to consider from an engineering and product design perspective - we need to know the accurate dimensions and shape of a product, so give us accurately measured and documented illustrations.
Simples.
Ha-ha, sad but true.
Just give us the exact W*H*D values for the GPU box, measured to the same standard for all GPUs.
21s century: we have internet, powerful GPUs but don't have exact dimensions of them(
It's not 100% correct but I found Technpowerup's GPU data base is a great place to start when looking into compatibility and gpu dimensions. The length is accurate 99% of the time but the thickness was hit or miss last time I was seriously digging through gpus.
Techpowerup's gpu data base link
Techpowerup's gpu data base link
Curiosity
Too busy figuring out if I can to think if I shoul
Platinum Supporter
Bronze Supporter
M...M...M...M...Multi-Tier...Subscriber...
- Apr 30, 2016
- 724
- 838
That not 100% is kind of risky. How do you estimate 99% of the time to be correct? This depends on how many cards you can physically check.
I remember checking two almost identical reference AMD cards between generations (6900XT / 7900XT or something like that) having different thickness but looking at the photos said otherwise.
Anyway the problem with techpowerup is that it has a database for GPU and the reference design, which usually has a dimensions that are properly measured while there's no way to check board partner models here, so it doesn't really help that much since those have their dimensions all over the place.
Luckily I have found that in most cases gpu dimensions and pictures on gpu manufacturer's site are correct, there are of course some unfortunate bugs as saper showed but I can rely on them (more transparent way of showing them would be preferred of course) .
What I would like to see the most are some size limits, for eg official size classes of gpus like we had before (267*111*42 for reference gpus that was valid for a long time, 305 for "high power" ones etc etc). Now we have almost 4slot, 160mm tall and 360mm long gpus without any real benefit since you can cool 450w with much smaller cooler, things are just getting out of hand. I want to be able to design a case that will always work for a certain class of gpu, forever, and not be obsolete on the next gen release.
What I would like to see the most are some size limits, for eg official size classes of gpus like we had before (267*111*42 for reference gpus that was valid for a long time, 305 for "high power" ones etc etc). Now we have almost 4slot, 160mm tall and 360mm long gpus without any real benefit since you can cool 450w with much smaller cooler, things are just getting out of hand. I want to be able to design a case that will always work for a certain class of gpu, forever, and not be obsolete on the next gen release.
Even if they are correct, if it's not stated where it's measured, it's still not clear to someone who looks for a card to fit in specific size. Us (case vendors) stating acceptable dimensions in multiple ways doesn't make it clearer if you don't know how it was measured for the card.
Yeah, ITX sized/170mm long, 10.5"/267mm and 12"/305mm would make a lot of sense and making just the 12" to be the form factor that allows 2.5 slot width.
But it's not happening unless nvidia does something about it. SFF is still a niche and vendors having a lot more beefy cards over those small ones show it. Actually if you take a look at the fin direction in open air cooled cards, most of the cards have them lengthwise so it's better for a case with airflow, but potentially terrible for a SFF case that doesn't have front-to-back airflow, but just vertical airflow or even the card is handling it's airflow on it's own and such radiator is blocked off from the front of the case.
Another problem for me is that FE cards are just scalped right away and are not accessible until the end of generation when people sell them off. Those were the go-to cards for SFF builders before and now if you're late for your order, there's no FE card anymore.
Similar threads
