What should we ask ATI?

A

AtomicMoose

[H]F Junkie
Joined
Sep 26, 2001
Messages
11,803
Post questions here...Kyle is going to put them together and try to get an interview from ATi.

See this thread if you have no idea what I am talking about. ;)
 
If the PR chick is hot: What are you doing this evening? :D (incl. the big smile)

Seriously though:
Are you looking into other applications to utilize the massive calculation power that graphics cards offer? Have you been working on creating some form of standardized development environment, compilers and or libraries? Does ATi think that there is a market for GPUs to be treated more as coprocessors for daily tasks? Does the change in the Vista driver model positively affect the versatility of GPUs?
 
Will R600 (next gen ATI chips) be supported.

Expected performance (in %) increase over existing 1900 series.

Road map for next 2-4 generations if that's a viable question.

Will this shorten video card life. Warranty still apply (not sure if right person for those).
 
what sort of improvements can we expect to see now that they have access to AMD's fabs/tech. ect. How will this affect both companies?

will we see something along the lines of an ATI to AMD specific buss for data movement ( if they have access to each others tech, having a dedicated buss where the chips can openly 'talk' would be a great idea... i think... ) im thinking long the lines of, if the processor see's the vid card on the pci-e buss it can engage a link to it without software overhead (firmware maybe)

will AMD use any of ATI's memory management tecniques (ala that nifty ring buss...) (think about the bandwith possibilitys here... 2gb of 1000mhz gddr3 on a ring buss... with dual/quad core... yeah... ) vise versa, will ati use any of amds cache tech?

last but not least... heat... you can only slap so much cooler on a card, what is there plan to combat there worst enemy?
 
Are we going to see support from ATI's professional line or just their consumer GPU line?

How will a crossfire/multi gpu setup perform, the same, double or somewhere in between? Does the amount of memory and speed of memory affect performance or is it mainly the gpu that is the limit.
 
will this "Stream" technology be open source? I.e. if I wanted to write software for it, would I be given the option to?


I'm not sure if these have to be F@H related, but I was just curious - I'll post back later with a F@H related question.
 
Suprized this hasn't been said yet...

With next gen video cards looking to comsume just about 2x-3x the power the entire rest of the system does, what are they doing to combat that? How does the merger with AMD benefit ATi's power comsumption design team.

I may think of a few more
 
because there is now a dedicated thread for this, I'll repost my questions here:

Thanks a ton Kyle!

I have 3 questions. ;)

How did this whole partnership between ATi and Stanford come about? <- or generally, a background story, "how the concept of GPU folding was born" would be nice, at least in this case from the perspective of ATi.

Have their been specific ATi driver optimizations planned to assist the performance of FAH? <- and furthermore, (this may be beyond the scope of her knowledge as it's a technical question) what types of optimizations have been made.

Does ATi see General Purpose computing on a GPU changing any fundamentals of how GPUs are designed on an architectural level, and have ATi's current lineup of products been designed with this emerging market in mind to any extent?

Hope an interview can be arranged, cant wait to read it, thanks again Kyle, you totally rock!
 
I'm not sure if this should be directed at ATi or Stanford, but how exactly will we know if our cards are overclocked too high and will be throwing out bad units? Not exactly any Prime95's for GPUs. Not sure if it applys, but I figure this is as good as a place as any to ask.

 
That reminds me of another question:
What kind of operating temperatures should we expect from running a GPU at full load 24/7, and in ATi's oppinion, are these cards up for that type of thermal abuse?

PS to the above posted.. ATi Tool is the closest thing to Prime that exists for these GPUs, I'm sure there will be
 
My concerns center around heat/cooling/warranty for 24/7 folding on the GPU.

i.e., Will the card be able to be cooled sufficiently to survive 24/7 folding without voiding the warranty ? And what will be their attitude toward overclocking the GPU when added to these issues ?

Another question which most likely should be directed to Stanford is: Will concurrent folding on both CPU and GPU be possible or desirable/undesirable ?

8607
 
What kind of interface bandwidth is required? - specifically, how would this work with an x16 card in an x4 or x1 slot? AGP x1600?

 
Will we see effects of running F@H in the background while running other graphic intensive task (games).

What memory will the F@H GPU client use. I assume the system ram. But then again when I assume things it never works out

The proteins thaat F@H GPU client folds will these be the same as the normal F@H client?

Will there be any way for us to use say 50% of the GPU like what we do for procs.

Will we have something like task manager to show what the GPU is doing?



 
Support for X1600!!!!!

Seriously, adding X1600 support would open the door to a lot more people. The X1600 isn't nearly as cost effective, but has a lower absolute cost of entry in terms of PSU, cooling, and card acquisition.

Many who would be looking at having to buy new PSUs to get an X1900 going would have no problem at all getting an X1600 folding. My 2 cents.
 
What is going to be the upgrade path on this, i.e. are future generations of cards and architecture differences going to continue to be used, or are we looking at a one pop on the 1900/1800 series and then going to be waiting a few more years due to tech changes?
 
KodiakStar said:
While the Open Beta is set to make use of the high end ATI Cards, are there any future plans to include graphics cards by other brands such as NVIDIA?

Will there be similar options for the GPU client to run the client in an "idle" mode on the video cards, and throttle down when needed. Additionally, will the client be able to monitor the GPU Core Heat Temps and shut F@H GPU client down if needed?
Click to expand...

Didn't realize there was a new thread for this.
Not sure if its an ATI only interview or ati and F@H...

so first q might have to be nixed.
 
I'm buying 4 of the SECOND fastest card at the time of a stable client, or Jan 07, whichever is later. THANK YOU for doing something to help more science get done. My only question therefore must be: Why are Canadians so freakin' cool? :D

Cheers,

 
In the future will it be possible to use a lower spec ati card (eg. x1600xt) as a physics processing unit?



...eh I read about something being planned along those lines
 
Will ATI sell the [H] a group buy on cards.... ;)
 
marty9876 said:
Will ATI sell the [H] a group buy on cards.... ;)
Click to expand...
Hell, I'd be onboard for that...

This comp's going to be on Folding anyway, might as well kick it up yet another notch.

For the question: Will there be any guidance on power requirement, since the GPU will be operating at a much higher utilization percentage than a more typical installation?



 
marty9876 said:
Will ATI sell the [H] a group buy on cards.... ;)
Click to expand...

Now THAT's a hell of an idea -- no doubt there would be some minimum total required... so put me down for 2 ! I'm with you on this Marty ---




 
You must log in or register to reply here.
Back
Top