Lucid Hydra gets official on graphics cards

[happy medium]

Hydra on your next gpu? I think this is better solution then the one they had on the motherboard. 40$ more money?, mabe not.

quote:
"However, there is one advantage here and that is that you can use cards from two different generations of GPUs which means that when you upgrade your graphics card, you could keep the old one to gain an extra performance boost"

http://www.semiaccurate.com/2010/08/30/lucid-hydra-gets-official-graphics-cards/

 

[Lonyo]

You might have forgotten a link.

 

[bryanW1995]

link?

ha, lonyo beat me to it!

 

[happy medium]

oooooooops, sorry guys, link added.

 

[Idontcare]

This got me thinking...with everyone claiming xfire scaling needs some improvement compared to SLI scaling, does hydra go any distance to helping close that delta?

And if it does...would AMD consider buying Hydra out and just using Hydra as their xfire glue-layer solution going forward?

 

[busydude]

This got me thinking...with everyone claiming xfire scaling needs some improvement compared to SLI scaling, does hydra go any distance to helping close that delta? And if it does...would AMD consider buying Hydra out and just using Hydra as their xfire glue-layer solution going forward?

Hydra's performance, as of today, does not offer compelling performance to be taken seriously. It does not seem to be a viable alternative for crossfire IMO.

Also, the recent Cat 10.8 application profiles have shown that corssfire does indeed have the capability to match SLI scaling efficiency. They just have to be consistent with their driver updates.

Its a similar case for Hydra, they have the hardware ready and its the drivers that needs to mature and support wide array of games just to be noticed.

 

[Idontcare]

Ouch...that settles that rather quickly.

 

[happy medium]

Hydra's performance, as of today, does not offer compelling performance to be taken seriously. It does not seem to be a viable alternative for crossfire IMO.

Those benchmarks are from last year and with a hydra chip on the motherboard.
This is a bit different, did you read the article?

Powercolor is puting a hydra chip on a 5770. With the chip on board the gpu , you can add a second gpu (4870) on any motherboard with 2 pci-e slots, to add performance.

Unless my eyes are getting worse?

 

[Sylvanas]

The benches Anand did a year ago were with early drivers and DX11 didn't even work back then. That has since changed and Lucid have been putting out new drivers at least every month now. It would be good to see a more up to date comparison.

 

[busydude]

Those benchmarks are from last year and with a hydra chip on the motherboard. This is a bit different, did you read the article? Powercolor is puting a hydra chip on a 5770. With the chip on board the gpu , you can add a second gpu (4870) on any motherboard with 2 pci-e slots, to add performance. Unless my eyes are getting worse?

Yes, I read the article. I know you can mix and match, did they release any performance figures? The testing that Anand did is all I have ever seen. If they have improved drivers and the performance, that's well and good. I have not seen any recent performance numbers so my opinion is based on the numbers that I have access to.

If their performance has increased, why haven't any hardware website done any testing or a review.

 

[happy medium]

Yes, I read the article. I know you can mix and match, did they release any performance figures? The testing that Anand did is all I have ever seen. If they have improved drivers and the performance, that's well and good. I have not seen any recent performance numbers so my opinion is based on the numbers that I have access to.

If their performance has increased, why haven't any hardware website done any testing or a review.

The Powercolor card is not due for a few more months, mabe it's not ready.

 

[bryanW1995]

why put the hydra chip on that gpu? seems to make more sense to use it on high end gpu's or mobos to more easily disguise the cost and/or offer better product differentiation. $40 on a 5770 is a lot of $$, but on a gtx 480 it would be negligible.

edit: uh oh, just read the article and charlie seems to agree with me. it's all downhill from here... either that, or I'm just as qualified as charlie to write about complicated technological gadgetry.

 

[GodisanAtheist]

The best thing about putting it on a graphics card, rather than on a board, is that you get to take the chip with you and it is not married to a certain platform.

Example:

You have a 4870. You use it for a year or two then you get the Hydra 5770. You use the Hydra 5770 + 4870 for a few years then you move the ole 4870 down into your lanbox and get an HD 7750 and you're still in business (assuming hydra stays on top of things with drivers).

In the middle of all this you're free to upgrade your motherboard/CPU or whatever else you want without sending the Hydra chip with it.

Tenuous at best but hey its something. Agreed on the "hide the cost in a high end part" although there is the possibility that high end part + hydra + new high end part tends to end up slower than just the new high end part by itself and totally defeats the purpose, while mid-range parts are actually additive.

 

[KingstonU]

I particularly like the notion that I could carry over my 4870 (which has worked fantastically for 2 years already) and just add it to a 7750 in my next upgrade and my $300 can benefit me for potentially ~5 years.

 

[Sylvanas]

I particularly like the notion that I could carry over my 4870 (which has worked fantastically for 2 years already) and just add it to a 7750 in my next upgrade and my $300 can benefit me for potentially ~5 years.

Yeah this is what I've been saying since release. A company which extends the value of my purchase will get my money, Lucid are doing this now and have had time to mature their drivers- It would be good to see how they're doing now. It doesn't look like they are stagnating either as this new product is testament to. Credit where it's due, Nvidia do this to a small extent with using an 'old' card as a PhysX card, better than nothing I guess.

Edit: here is a recent review with Unigine, 3Dmark and Crysis. The synthetics show pretty huge gains and crysis shows a small increase in minimums and a 10fps increase in avg when run maxed with 8xAA and enthusiast. Pretty good IMO.

 

[Dark Shroud]

Make a 6770 with one of those and you'll have a solid product for a 2+ years. Yeah an HD 77*0 with one of these would be an instant sale for me.

 

[Ben90]

Hydra's performance, as of today, does not offer compelling performance to be taken seriously. It does not seem to be a viable alternative for crossfire IMO.

Also, the recent Cat 10.8 application profiles have shown that corssfire does indeed have the capability to match SLI scaling efficiency. They just have to be consistent with their driver updates.

Its a similar case for Hydra, they have the hardware ready and its the drivers that needs to mature and support wide array of games just to be noticed.

While it is true that Hydra does not scale as well as SLI or Crossfire, it is a much more elegant solution. The only thing AFR rendering is good for is near perfect scaling. There is no doubt SFR is the future and I for one am glad we have companies like Lucid pushing SFR. Hopefully it gets the card companies looking at their own solution that can get scaling close to AFR.

While I do completely agree that Hydra in its current form is in no way ready for the mass market, I for one will not go multi-GPU without a SFR solution, and Hydra would be the only way to make that possible. Even with the lousy scaling.

 

[happy medium]

Edit: here is a recent review with Unigine, 3Dmark and Crysis. The synthetics show pretty huge gains and crysis shows a small increase in minimums and a 10fps increase in avg when run maxed with 8xAA and enthusiast. Pretty good IMO.

@ Ben90
Thats not bad scaling at all. No quite sli/crossfire but worth it.
Mabe it will get better with the on card solution?