Nvidia GeForce GTX 1000 Series (Pascal) MegaThread: FAQ and Resources

Page 124 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
not sure if there will be reference design but it seems nvidia do want this particular gpu to fit in mini ITX size. it kind of reminds me of GTX980 for notebook. back then full fledged "high-end" GPU inside a notebook is something that never be done before. maybe this time nvidia want to prove that you can really fit GTX1080 performance into mini ITX form?
 
no idea if this is a new model or simply a mini 1070/80 or what? i sure hope it's not another new model like we need a 1060ti or 1070ti or even a misleading 1070 4GB or whatever else they think up from time to time.
 
The last I knew, NVidia manufactures all the cards. They are then shipped to the board partners for cooler mounting and other finishing. I don't know what the split is, design-wise, between NVidia and the board partners for special or unique products. One thought about that is that NVidia's manufacturing engineers and technicians must think they can execute a design, regardless of who makes the design decisions. This is simple common sense. I see all these various cards with power delivery differences and think that the partners have been in this loop for quite a while.

There must be a drop dead date for the board partners to receive their boards and another for the partners to finish their first release products in order to serve NVidia's timeline for releasing a new product line.

I know I'm rambling so I'll stop now. What do y'all think?
 
Nvidia doesn't even manufacture the chips, let alone the cards. All done by suppliers. Nvidia is a design firm.

AIB partners are supplied with the reference design and they do make those. They also design and manufacture their own custom circuit boards and coolers.

Chips and memory come from the fabs like TSMC and Global Foundries (I think Samsung is also making some this time around, haven't confirmed that anywhere as far as I know) and are sent to the AIB for assembly into finished products. They may also use suppliers for construction.
 
It is interesting though they would do that. I kind of thought the next step was going to be readily available MXM cards for the new STX and uSTX form factors.

Wouldn't mind a desktop CPU and a pair of mobile GTX1080 or what have you with a single giant heatsink or waterblock on them.
 
the AIO pc's you can get are almost like that. very different parts than normal desktop more akin to laptops than desktops. is why they are hard to upgrade cause nothing is stock off the shelf parts.

but yah nividia does not make anything really, they design it and farm out the work. the various brands are free to use the reference design or their won, which of course they use both. some are simply custom coolers on reference board but many are totally unique boards and coolers.

they are so unique that the rumor sites are able to see a leaked pic and know what brand it is by the board design. they match them up easily to known cards and can identify a lot about a pic just from the board design.
 
i'd love to see vertical slots for them like ram slots. so many ram slots on a mobo anyway, cut it down and give me some nice vertical slots for this fancy drives. would let it get more air anyway since it is such a hot item.

just my opinion anyway. that and $2 may get you a Dr Pepper depending on where you go.
 
so....1080ti event?

https://videocardz.com/66013/nvidia-invites-you-to-geforce-gtx-gaming-event

same day as AMD upcoming event. WhyCry said there is no reason for nvidia to reveal 1080ti or pascal refresh on the event when AMD most likely not going to talk much about Vega on AMD upcoming event (it is more for Ryzen). but i don't really think nvidia have to wait for AMD each time they want to release something new. 1080 still have no competition but it still not stopping nvidia to release titan XP back in august last year.
 
lol, yah that would be a noteworthy card. but really even a few 1080 cards are small enough for some rather small builds. not micro anything but pretty small anyway. 1070's can go mini already just waiting on 1080 models to show.....
 
looking at the new MSI itx card, i realized they actually already had multiple mini cards. this one is a touch shorter and has lower clock speeds but it's not the first itx card tyey have released.

so not sure at all why it is even a thing??? that and i wish they'd separate the 3gb and 6gb 1060 models on the site to make it easier for me to figure out which models i have on the page 1 list and which are missing.......
 
It is interesting that they have single slot quadros, i'm just not too sure how strong a p4000 actually is, Wikipedia say it's near a 1070, There are just NO single slot cards anywhere near gaming level, but professional GPUs are supposed to be for rendering pretty heavy workloads, which go beyond what games do to a card, so... why can't we have them?

Maybe the pro cards actually have "higher quality" chips that can withstand the higher temps, i doubt profresionals closely monitor their temps like gamers do, or they put them in big server rooms that are well climate controlled.
 
that's an interesting question James. never thought of it that way. don't know the specifics on the workloads of pro cards but i do know it is a different type of calculation. does this make it "harder" on the gpu over the types done by a gaming gpu?

does seem odd that they can do single slot pro cards but not gaming ones worth anything. may just be that there is only a small market for such a card. shorter cards are coming into their own and we see lots of them now. perhaps there is just not much call for single slot gaming cards so they don't bother with them. just guessing but that's been my thoughts on the topic
 


or GTX2060. GTX660 did use GK106. but GTX760 actually did use cut down version of GK104
 
They haven't released the clocks, and they've obviously been binning 14 SM unit chips for a long while. My guess would be a very low voltage/frequency to keep it under 105W, or no boost. And I am certain it will be loud. Not too much different from the TDP of an MXM Type B and gaming laptops can handle those.

You can take a Quadro and game with it, usually works pretty well as compared to its direct counterpart.

I don't think they do too much binning on the high end, if anything they use the chips that won't go fast. Basically just take the biggest pieces of silicon they've got and slap more memory on them.
 
that has always been the case for professional card. and they also usually clocked lower. take tesla P4 for example. it is GP104 based with power rated at 50w-75w. but the interesting part is it's performance. it's single precision (FP32) was rated at 5.5Tflops. that is around the same performance at 980ti stock performance (5.6tflops FP32).

http://images.nvidia.com/content/pdf/tesla/184457-Tesla-P4-Datasheet-NV-Final-Letter-Web.pdf
https://www.pcper.com/reviews/Graphics-Cards/NVIDIA-GeForce-GTX-980-Ti-6GB-Review-Matching-TITAN-X-650

granted there is no video output on tesla based card. but if the thing can really running games isn't that mean nvidia can have 980ti performance with only sub 75w of power?
 
if they could get 980ti performance at >75w, i'm sure the card would be out there already. don't think nvidia/amd would have left that great feature on the table for future gens.

i often wonder if there is some power to be saved as we saw with things like undervolting the amd cards and so on. but i don't think it is 200w or more of savings to be had. the 980ti is a 250w+ gpu, so i doubt there is that much savings to be had but nvidia just decided to ignore it
 
in case of tesla P4 it might have to do more with it's specialize task. but if they really do have something that can be as fast as 980ti for only 75w of power they don't really need to outright show that kind of leap right away. instead they can still give us that incremental leap each year and continue to milk us haha.

nvidia might do a lot better than AMD in terms of financial wise but still their primary source of income still coming from gaming GPU. in Q4 alone gaming GPU contribute 1.3bil of nvidia total revenue. their second largest segment by revenue right now is their data center division which in Q4 give nvidia almost 300mill. that is still less than 25% from what their gaming segment can generate. because of that nvidia have to play smart with how they release their gpu to maximize their income.
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom