New video: the most ridiculous Lucidlogix (Hydra chip) interview... ever!

[grimREEFER]

btw, near 100% scaling doesnt really mean anything. you can say sli and crossfire have near 100% scaling.

you really have to see benchmarks before any of this means anything.

[Chruschef]

i'm surprised that anandtech's article hasn't been posted on this yet..

link to the article



ati/nvidia running at the same time


ati/nvidia cards running bioshock


... i think that should stop any disbelievers. anandtech does comment, that they can't comment on the performance until they get the hardware themselves.

anyone who doesn't believe that this will give some legitimate performance gains is in for an awakening; i don't think they could justify a $1.50 per lane otherwise (16x PCI-E lane = 24.00$). we just can't be to sure how much of a performance gain yet, and there could be other issues. even if this chips flop entirely, if it's possible ATi and nvidia cards will work together in unison ATi and nvidia will simply have to accept it and take it on, i think.

[Manicdan]

OK... I understand now.

You simply don't believe your eyes. Everything is FAKE until someone else benches the Hydra 200 within a game, right? That video is fake (nothing to discuss there), press release is fake (nothing to discuss there, either) and MSI announcements and the highly anticipated BIG BANG is fake too (because you don't have performance figures).

I think many are just unwilling to admit that this technology is actually here, period!

if those are your points, you missed everything i said

i know its here, i like the idea, i would buy a mid range video card more often if i knew i could keep my old ones. instead i only upgrade about once every other year and basically trash the previous one.

you sound like everyone on here is throwing tomatoes at you, when all we want is to know how it will be in real life. will it offer atleast 80-90% scaling in all situations? will it cost 100$ premium to the motherboard? will intel let these go on AMD platforms? will you be able to use dx11 if one card is just dx10?

design and concept, this is great, but when it comes to cost and performance, will it only be a good buy for the people who spend hundreds of dollars on video cards every year?

until these things are known, i have no idea if i should ever get one. so until i know them, i honestly dont care about "OMG ITS HERE" statements. just like the new Camaro that came out, its sick looking, but i dont plan on buying one due to real life blocking my dream world. and when i see one go buy, do you think i stick my head out the window drooling, no, im a little more mature than that.

[Computurd]

I was in a chat session that had shamino, and he indicated that EVGA has tested the hydra chip, and that it does in fact work, however he would not give specifics, but did state that once the drivers mature EVGA will be looking at putting it on a board. It was a public chat for the p55 motherboard fellas.

[tdream]

Everything will be based off the results.

100% fps scaling = holy grail
80% fps scaling = great product
CF/SLI levl scaling = interesting quirkly product/mix n match gpus
less than that = total failure

[kadozer]

Surprised this isn't getting more excitement. MSI Big Bang board is scheduled to be released October.

[RVWinkle]

Do you guys remember in the 90s when that board vendor made a multi gpu driver? I think it was Wicked3d or something.

[keithlm]

This is "kind of" similar to going from software (i.e. onboard) raid to a dedicated hardware raid card. It's not hard to see the benefits.

But then there are still some skeptical people that honestly believe that software/onboard raid is "just fine" because often benchmarks don't show many of the advantages of using dedicated hardware raid cards.

Further speculation: I wonder if they can make a PCIE card with the Hydra chip in it that plugs in and allows you to have an external enclosure with like 4 or 8 video cards in it. Of course then we have to worry about the bandwidth. But it could be nice.

[Lestat]

he not acting so passively about the technology, you have to understand its their technology so when they do talk about it, they talk in an every day normal way because it is normal for them.

i wouldnt worry about the dual video cable, i think that is simply the driver installation process. you need to view the video from the card your installing drivers for maybe? plus to recognize the monitor aspect ratio, resolution, etc. plus showing PCPer and others how they switch between the active and secondary video cards, probably on the fly, or with a simple reboot.

this is preetty freakin scarey stuff in terms of the real world implications... especially when they are talking about real world use of a GPU as a processor...

[Chruschef]

Further speculation: I wonder if they can make a PCIE card with the Hydra chip in it that plugs in and allows you to have an external enclosure with like 4 or 8 video cards in it. Of course then we have to worry about the bandwidth. But it could be nice.

answered yourself; as well as it's a PCI lane switch

[Xoulz]

Not interested 'till performance numbers come out

Yep... Big Bang

[Xoulz]

if those are your points, you missed everything i said

i know its here, i like the idea, i would buy a mid range video card more often if i knew i could keep my old ones. instead i only upgrade about once every other year and basically trash the previous one.

you sound like everyone on here is throwing tomatoes at you, when all we want is to know how it will be in real life. will it offer atleast 80-90% scaling in all situations? will it cost 100$ premium to the motherboard? will intel let these go on AMD platforms? will you be able to use dx11 if one card is just dx10?

design and concept, this is great, but when it comes to cost and performance, will it only be a good buy for the people who spend hundreds of dollars on video cards every year?

until these things are known, i have no idea if i should ever get one. so until i know them, i honestly dont care about "OMG ITS HERE" statements. just like the new Camaro that came out, its sick looking, but i dont plan on buying one due to real life blocking my dream world. and when i see one go buy, do you think i stick my head out the window drooling, no, im a little more mature than that.

Oh!

So, you do believe Lucid's claims..! ? It's just that your really don't care to get excited about it or discuss it, because you "dont plan on buying one due to real life blocking my dream world". (ie: cynical)



Sorry bro, I started this thread because that video (in full) makes some pretty outrageous claims, (if true) that is going to shake the foundation of gaming, MOBO chip priority and mainstream acceptance.. based on nothing other than at least being as quick as SLI or Xfire. HYDRA doesn't need a performance gap to completely usurp SLI & Xfire.

So, with the incredible claim that it's near 100% scaling... makes the rest of this mind boggling. As I know NOBODY associated with Intel would be allowed to go on utter such claims, if it wasn't true..


Nothing I have seen indicated this company cannot deliver!

[annihilat0r]

Xoulz: Did Lucid indeed claim 100% scaling?

[Your_Boss]

I think Lucid made that near 100% claim regarding their first Hydra 100 chip. So this is 2nd gen..so hopefully it will be even better..if only slightly.

As for the running two different chips (ATI/NV) in one computer to enhance it's performance I think is HUGE!. Not sure if people realize this. This means you don't have to sell your older cards(well, some, if they're too old).

Most people have a 2nd, or 3rd PCIE slot(right now) in their motherboards that have nothing in them. If this thing is real and it can do what it claims (even if the scaling between different chips is only 50%) then I think most people would love to have this on their motherboard.

For ATI and NV, however, this could be a problem and a blessing.

Problem - They sell less video cards simply because people will just use all their old cards in their unused PCIE slots until they need more performance.

Blessing - NV and ATI no longer need to work on xfire and SLI, when almost every MB has the Hydra chip on it then there's no need. Saves them time, money and effort.

So, who knows? I hope it's true only to give the consumer more options.

Remember that little Lucid icon in the system tray? OFF / ON haha! You choose!!

[Chickenfeed]

It is still hard to say how gpu mixing will work with differing amounts of VRAM. I'm still not convinced they managed to work around that limitation but we shall see soon enough I suppose.

[zalbard]

Everything will be based off the results.

100% fps scaling = holy grail
80% fps scaling = great product
CF/SLI levl scaling = interesting quirkly product/mix n match gpus
less than that = total failure

CF scaling is nearly 100% in some games...

[jmke]

CF scaling is nearly 100% in some games...

what games/resolution?

[Farinorco]

AFR is already "near 100% scaling" from a theoretical point of view, given that AFR works giving a 2nd frame to be rendered by the 2nd GPU at the same time that the 1st one is rendered in the 1st GPU, thus doubling the number of frames you process per time unit. Then, interactions with other parts of the system like the CPU, and the fact that the games are not only graphics rendering but there are lots of other things that has to be computed before the actual frame can be rendered, is what sets the real world scenario. So a claim of near 100% scaling is actually meaningless IMHO. Let's see how it turns to be in the real world.

Anyway, what I find more interesting from this technology is not the performance aspect versus AFR and other common multi-GPU techinques but the posibility to use a heterogeneous selection of videocards, with different chips and even different companies... so you can add a card to your system without changing the old one.

I've some doubts about real world practicality of this that I would like to know how they turn out to be...

What happens when your different cards are capable to render different versions of APIs and standards? (for example, one card with DX11 and other with DX10.1)

Is NVIDIA in the end preventing their users to enable PhysX support if any ATi card installed on the same system is detected?

Even if it's appearently similar, the output image from different cards it's not exactly the same (different filters, for example, but not the only thing). Won't these image rendering differences cause any kind of flickerings, artifacts or glitches when running?

Is this Hydra solution managing every GPU using scenario, as it could be GPGPU software processing, or only 3D rendering? It's probable that the GPGPU thing starts to become more important from now on. How is it solved when you have different cards with this Hydra chip?

There are lots of questions that will only be answered when this is on the hands of different people and start to comment what are their experiences with it. Right now, this chip could be either a revolution or an epic fail with no (or little) real world use.

[*GameOver*]

Slow down for just a sec guys...

Intel is a big hand in investing in the Lucid Hydra Chip... IT just so happens that only a month before the tech hits the open market that Intel combines the NB and SB into one chip and moves the PCIe Lanes to the CPU... Creating lots of new room on its board. What benefit does this have? The hydra chip only requires 1 PCIe 16 lane to communicate with the CPU then the Hydra chip gives out 2 PCIe x16 lanes, OR 4 8x lanes. This looks to me (and this is just speculation) That this new platform was designed around making this chip work... Think about it. Who would benefit more from this working over anyone els? Intel. It allows them to kick NVidia to the curb and say F your NV200 licensing fees we have our own solution. ATI shouldn't care because their crossfire is already multy-platform anyways...

It all adds up and it adds up really quick for Intel. IF this works and we see anywhere from 80 to 90% scaling even just across Matched GPU's it will totally reshape the way we spend our money. I just find it funny that 1 month before the Hydra 200 launch Intel brings out the perfect platform to support it.

(as a side note) Has there been anything about a multy Larabi set up in the off die version? hurmmmm a 80 to 100% scalable solution sounds fairly juicy does it not?

[Manicdan]

Oh!

So, you do believe Lucid's claims..! ? It's just that your really don't care to get excited about it or discuss it, because you "dont plan on buying one due to real life blocking my dream world". (ie: cynical)



Sorry bro, I started this thread because that video (in full) makes some pretty outrageous claims, (if true) that is going to shake the foundation of gaming, MOBO chip priority and mainstream acceptance.. based on nothing other than at least being as quick as SLI or Xfire. HYDRA doesn't need a performance gap to completely usurp SLI & Xfire.

So, with the incredible claim that it's near 100% scaling... makes the rest of this mind boggling. As I know NOBODY associated with Intel would be allowed to go on utter such claims, if it wasn't true..


Nothing I have seen indicated this company cannot deliver!

i never said i believe their claims, all i said is i know its here (meaning they are making it the production stage)

and btw if someone asks is the performance closer to 0% or 100% and the answer is 51% then you can say its near 100%, its called marketing. unless they show you tests, they would rather use these undefined terms to promote a view that is only suppose to manipulate people into thinking they understand whats going on. ever see how toothpaste commercials like to use the word "better" but they could simply be talking about the box is made from recyled paper and is better for the environment. until these open words have been defined with exact details (like through tests), they are marketing tools, not facts.

CF and SLI are already near 100%, on select games in select settings with select hardware configurations on select operating systems, and cost you nothing more. (now was that a fact or a marketing tool?)

[Decami]

btw, near 100% scaling doesnt really mean anything. you can say sli and crossfire have near 100% scaling.

you really have to see benchmarks before any of this means anything.

uh...no, no you cant. While scaling means the performance of the card as a whole, then what it actually does as a second card. It doesnt matter what numbers you make up, you cant say that.
I'm guessing Lucid has a reason to be saying near 100% scaling. The reason Nvidia nor ATI never said this, is because, obviously, they didn't have a reason.

CF scaling is nearly 100% in some games...

hmm, must be for mere milliseconds when turning face into a wall or something.
and lets not forget micro stutter, which is non existent in this process.

[Xoulz]

i never said i believe their claims, all i said is i know its here (meaning they are making it the production stage)

and btw if someone asks is the performance closer to 0% or 100% and the answer is 51% then you can say its near 100%, its called marketing. unless they show you tests, they would rather use these undefined terms to promote a view that is only suppose to manipulate people into thinking they understand whats going on. ever see how toothpaste commercials like to use the word "better" but they could simply be talking about the box is made from recyled paper and is better for the environment. until these open words have been defined with exact details (like through tests), they are marketing tools, not facts.

CF and SLI are already near 100%, on select games in select settings with select hardware configurations on select operating systems, and cost you nothing more. (now was that a fact or a marketing tool?)

My friend.. this is their second chip, the HYDRA 100 works and has been proven. You make it sound as if these claims are trickery!

When they claim near 100% scaling... they aren't hiding behind some verbal technicality..! Show some respect to this community, your cynicism is remarkable. Tri-SLI has never equaled 3x teh performance..



Toothpaste comercial...? I'm done arguing with you. sorry!

[Manicdan]

My friend.. this is their second chip, the HYDRA 100 works and has been proven. You make it sound as if these claims are trickery!

When they claim near 100% scaling... they aren't hiding behind some verbal technicality..! Show some respect to this community, your cynicism is remarkable. Tri-SLI has never equaled 3x teh performance..



Toothpaste comercial...? I'm done arguing with you. sorry!

im trying to teach you how to read into what is marketing, and what is proven facts. becoming too excited over something is a fast way to getting hurt (kinda the same as relationship problems). in every post you read my words as if im badmouthing hydra, ive never said anything bad. ive only asked questions that ive yet to see answered, and depending on the answer affects how well it will be accepted by the masses, not just enthusiasts.

[Tao~]

1) if all is working why didnt hydra 1 make it to the market ?

2) it seems very obvious to me that the capabilities of GPU_1+GPU_2+Hydra will be limited to that of the the older GPU (or basically the intersection)
eg. I have a GTX260 which is DX10, and then I add a HD5870 which is DX11. Now i play a DX11 game...how will GTX260 handle it?
The product (Hydra) giving scaling close to 100% is definitely insane. Imagine the barriers in architecture,drivers,OS,API it will have to accommodate. I mean, when I see scaling with the same cards , its difficult to believe.
(note that insane != impossible )

PS: I have not much idea how API's work, plz correct me if i have misinterpreted something.

[annihilat0r]

uh...no, no you cant. While scaling means the performance of the card as a whole, then what it actually does as a second card. It doesnt matter what numbers you make up, you cant say that.
I'm guessing Lucid has a reason to be saying near 100% scaling. The reason Nvidia nor ATI never said this, is because, obviously, they didn't have a reason.



hmm, must be for mere milliseconds when turning face into a wall or something.
and lets not forget micro stutter, which is non existent in this process.

Maybe not %100 but in super high resolutions where the case is extremely GPU bound, raw FPS performance added by a second card can near %100... of course these are too few and far between.

Microstuttering is wholly another issue though and another thing which I'm curious of in the Hydra chip.