It's a lane multiplexer, just like the NF200 chip.Quote:
Originally Posted by villa1n
It's a lane multiplexer, just like the NF200 chip.Quote:
Originally Posted by villa1n
Ahhh ok, so if the bandwidth exceeds the normal, they will still recieve the x16, there will just be added latency then.. like the nf200, because the interconnect from the hydra to the pci-e controller is still only 16 lanes. Gotcha. Thanks.Quote:
Originally Posted by aka1nas
yeah unless its a total flop imma get a board with one in a few monthsQuote:
Originally Posted by Wiggy McShades
need a new cpu/board/gpu anyway, it just should be nice to have the 8800gt working alongside the 5870.
so if i were to disable it when using linux, could i get the 8800gt to be the [sole] provider of graphics? how does this work?
Lucid guys at IDF told me today that what is coming in a month only allows two cards form the same vendor. Their software is not yet ready for mixed vendor operation. It works but will be released in the next quarter. Since it's a software limitation, moherboards launching in a month should be able to run Nvidia+ATI later.
i was waiting for thisQuote:
Originally Posted by xoqolatl
but kinda ridiculous to release it if its not working properly
later can mean a year later :rofl:
I'm seriously considering the msi bigbang and an 860 if the chip adds any performanceQuote:
Originally Posted by AndrewZorn
Lucid guy said that you must have a regular video driver (he had catalysts 9.9 running) along with a Hydra driver. Hydra can be enabled or disabled in flight by Hydra control panel residing in system notification area. He also said it's compatible with all PCI-E compliant devices including chipsets other than P55. It's all up to motherboards manufacturers now.
is it up to the manufacturers, or are there limitation, like Intel not wanting this to make it to AMD boards?Quote:
Originally Posted by xoqolatl
I think this is great, but I want it on an AMD setup. I want a 5870, but I don't really have a good use for my 4890 then. Way too loud and hot and powerful for my HTPC, and my wife's would not be able to utilize it either. So, I could sell it, but would rather make use of it, since its a great card.
no problem blocking it .. plenty of ways .. easiest would probably be:Quote:
Originally Posted by SKYMTL
if (upstream_pcie_bridge_vendor==LUCID)
disable_driver();
or even dont load the driver if any lucid pcie device is detected in the whole system.
QFT, I was thinking this would be fairly easy for Nvidia to implement with their drivers.Quote:
Originally Posted by W1zzard
Seeing that they're prepared to block PhysX if an ATI card is in the system, who knows what stance they'll take?
I can just picture it now, they'll claim unreliability issues and suggest using good ol' SLi as the necessary alternative.
doesn't nvidia have their "not nvision" convention coming up?... people should just yell out asking if they support lucid, and just generally make it apparent that if they do block the lucid hydra chip, that we the consumer will crucify them :)
... there's nothing more convincing than an angry mob, given the publicity it would give them
I don't know if this has been posted before, but upon searching 'MSI Big Bang' I came across this website.
http://event.msi.com/mb/bigbang/
30 days and ~5 hours remaining til release?
I'm really curious about how this thing behaves at some regards in the end.
If Hydra distributes the rendering work amongst the cards connected to it, and then re-assemble the image, I wonder how it will behave with differences in final output from different cards (for example, different cards implement AA and AF filters in a different way, different cards show -or may show- different color calibrations, different cards are subject to different optimizations). Won't this generate any kind of visual artifacts?
And what about full screen shaders and things like that? This problem is intrinsic with the way Hydra works, and is not present for example with AFR. How will it work around that?
I am curious too about the different graphic APIs support (how it works with a DX10 and a DX11 card on the system? can you use the DX10 card to make some work when rendering in DX11, or only when you are using DX10 or less?).
And I'm also curious about other non-graphical tasks such as GPGPU accelerated aplications (OpenCL, CUDA, and all). Will Hydra be able to distribute the work amongst all the videocards, or will it only bypass it to one of them?
------------
As a sidenote, I'd like to make a comment about the thing of disabling Hydra from the videocard drivers, the chip being transparent to drivers, and so. The chip itself works over the drivers of the videocards, so is not dependent on them, but the videocards themselves need their drivers to render. Don't be mistaken on that. So NVIDIA/ATI drivers are going to be needed anyway if the cards are meant to work... so if some of them would want to avoid working with this chip, or with cards from another IHV, for whatever reason, they could do it easily. So easily as NVIDIA do to not allow PhysX working if an ATI card is installed on the system: check on the drivers if the Hydra chip / other vendors GPUs are installed on the system, and refusing to work in that case.
Any new Info yet?
Most curious about scaling, latency and driver functionality. If scaling is better than linear, as originally advertised...then that would be something. Many have speculated that a chip like this will cause latency, but will it be noticeable? Let's hope not!
Also, since its not affected by drivers, I wonder how features not natively supported by a game can be forced using the hydra software vs. the likely no longer useful Nvidia control panel/nHancer. I like the ability to force AA, AF and certain quality settings that are not always supported by a give game.
Hm, one Radeon5870X2380GTXHalfbreedFTW rig coming up.Quote:
Originally Posted by Chumbucket843
Yeah something like 8800GT for HD movies/Bluray and fire up the 4870x2 if I wanna game on my HTPC? That would be sweet and energy efficient.Quote:
Originally Posted by AndrewZorn
Just thinking this could allow more than just 4 GPUS in theory
I don't see how the performance improvement can be more than linear... so using 2 cards and performance being more than double, using three and being even more than triple, and so?Quote:
Originally Posted by Hu1kamania
About videocard drivers, think that even if Lucid Hydra itself isn't affected by them, the cards use them the same way as always. Hydra intercept the calls that should arrive to the videocards (to its drivers to be more accurate) and first distribute it and then send them to the videocards (obviously, to their drivers --the cards can't work without drivers, Hydra works without the videocards drivers, so isn't affected by them, but the videocards don't...).
What I don't know is how it would be handled when you have drivers from different vendors installed because you're using cards from different vendors (HD5870 now and then if I like it GTX380 too? :rolleyes:). You could set them on different configurations. Well, you for sure will do, since they have not the same options for everything. I'm curious too about how it will be solved... when multivendor support is really implemented and it's not a promise.
you can quote me:
Hydra will be able to run every vga with anyother but because of the problems people will use it just for runnig 2 identical vgas.
abomination!!Quote:
Originally Posted by Chumbucket843
i can only hope sli/xf get terminated
why should we be bound to certain hardware because of sli/xf ??
lucid hydra should get more support.. i hope it doesnt end up only on certain mobos
I cant wait to see the results. I have a 4870 and GTX285 just waiting to be paired up.
:banana::banana::banana::banana:, if I do that then I need make another water loop.... DAMN IT! Oh wait, SWEET!
Well it's been 30 days :)
3 days left...Quote:
Originally Posted by tdream
http://event.msi.com/mb/bigbang/