GTX 480

[Sunin]

Curious if anyone is going to do a build with a Big ADV and a pair of 480's or more?

If estimates are accurate a quad 480 + Big Adv 920 sys

could pull 100k by itself and some estimates put it at 150k.

480 is expected to be about 25-50% faster than a GTX 295 which my GTX 295 out put around 19k,

19k + 4.75k = 23.75k
19k + 9.25k = 28.25k x 4 = 110k PPD
+ Big Adv of what 25k + puts you anywhere from 100k to about 135k depending on OC's you could see 150k from a single box.

Wattage pulls are estimated at 300w per 480, but actualls will probably be in the 250 range, plus the 250 or so from the 920/mob = 1250 watts needed to run, which if you got 125k PPD out of the system and need 1250 watts to run = 100PpDpW which I believe sets a new system high. The only thing that does better is a system that is pure Big Adv and no gpu.

I'm already looking at a build of a EVGA Classified + 2 x 480 + a raid 5 of 60gb SSD harddrives. This would not only be a kick butt gaming system but would output around 75k PPD

What are you all thinking of doing when the GTX 480's hit?

 

[lehmann]

I think everyone would like a duel xeon 6 core on evga's duel overclockable board and 4 of these 480.s but thats a huge chunk of cash. im guessing 5 grand to do something like that. Im sure that atlas folder guy will buy a hand full of both.

 

[sirmonkey1985]

the real question is how close are nvidia's numbers.. no one can really answer that since even the demo's were done with tesla cards.. and well you cant believe anything a company claims until there are legit numbers from a second party.. this goes for amd, intel, ATI, and Nvidia.. i also have my doubts this thing will run at 250w full load.. still going with the 300w estimate seeing as how much larger the gpu is vs ATI's gpu.. but we will have to see when they actually get a working consumer level GPU.. they have had long enough to perfect it so i wont bother saying it will be slower then the HD5k.. only an idiot would claim that.. but 20% gain in gaming performance doesnt always mean its a gain for folding since vram speed has no effect on folding but a huge effect on gaming.. also we have to wonder what the clocks will be on the shaders.. to many questions not enough answers to give a definitive answer on what the folding performance will be.. though i wish nvidia would hurry the heck up so i can buy a 5870..

 

[Elledan]

I'll be waiting on benchmarks before I decide anything.

Fermi's design should be much more suited to CUDA and OpenCL tasks like the F@H GPU cores due to its more generic design. That alone would lead me to believe that it'll wipe the floor with GT200(b) cards in GPGPU performance. Matters like power usage and such could still make me want to decide to wait for the next revision, though.

 

[capreppy]

My Sig Rig is due for a refresh.

I am actually looking at eVGA's Dual 1366 Board with a pair of 480's.

 

[tjmagneto]

Looking = cheap. Buying = $$$$. I think I'll pay off a car or two before putting any more big bucks into my rig even though the new card sounds great.

 

[Kendrak]

Looking = cheap. Buying = $$$$. I think I'll pay off a car or two before putting any more big bucks into my rig even though the new card sounds great.

I'll upgrade Nexus's mobo to the OC-able EVGA one before I drop crazy coin on a brand new GPU.

I like buying used anyways

 

[DooKey]

I'm going to be selling my 5870s soon so I can buy a couple of Fermi's.

 

[zero2dash]

The points are appealing but I'm currently in the "money talks" camp. How much will they cost?
I won't be buying 2 cards and taking a huge loss on my GTX's, but if the loss would be minimal (or the cost of the Fermi cards) would be low I might consider tossing 2 in my i7 and migrating the GTX's back to the UD3P they came from. Again though money talks. If we're talking $250+ per card (which is probably likely) it'd almost be cheaper ($ to PPD) to buy another open box X58 Extreme, a 920 from MC and a 6gb ram kit and make a boxless -bigadv runner.

 

[sirmonkey1985]

The points are appealing but I'm currently in the "money talks" camp. How much will they cost?
I won't be buying 2 cards and taking a huge loss on my GTX's, but if the loss would be minimal (or the cost of the Fermi cards) would be low I might consider tossing 2 in my i7 and migrating the GTX's back to the UD3P they came from. Again though money talks. If we're talking $250+ per card (which is probably likely) it'd almost be cheaper ($ to PPD) to buy another open box X58 Extreme, a 920 from MC and a 6gb ram kit and make a boxless -bigadv runner.

lol were talking $450+ here most likely some where in the mid 500 dollar range if the cards score around what nvidia said they would.. if they actually fall below the 20% margin that nvidia stated between the HD5k series and the GT100 then they will be in the 450ish range..

 

[zero2dash]

[spits out diet mountain dew]
*Per card*?
They're insane.

nVidia is so far behind the current 'gpu warZ' but if they want to price like they're in the lead, I won't try to stop them...I'll just point and laugh from the sidelines. What a bunch of idiots. Maybe Charlie [Demerjian] is onto something. Huang is bat$#!t crazy.

If ATi cards Folded worth a damn I would have nothing but ATi cards right now; Folding is the only reason I have nV cards.

 

[Kowan]

[spits out diet mountain dew]Folding is the only reason I have nV cards.

Folding is the only reason I don't have ATi cards.

 

[Sunin]

To me it will be appealing to manage less boxes overall. Right now I have 5 boxes that fold 100% and 1 that is 50/50. I won't stop the GPU folding that I have. I would probably convert my 2x 295 over to the ATI/AMD platform, can't beat 45w processor. Then have one of these either single or dual 920 setups. What mobo is EVGA coming out with that will have twin 1366s? How many 16x PCIE?

Will that mobo support on board raid?

Because you could probably put a good build together for about 2k for the 75k PPD range by June/July

 

[Kendrak]

To me it will be appealing to manage less boxes overall. Right now I have 5 boxes that fold 100% and 1 that is 50/50. I won't stop the GPU folding that I have. I would probably convert my 2x 295 over to the ATI/AMD platform, can't beat 45w processor. Then have one of these either single or dual 920 setups. What mobo is EVGA coming out with that will have twin 1366s? How many 16x PCIE?

Will that mobo support on board raid?

Because you could probably put a good build together for about 2k for the 75k PPD range by June/July

My understanding is the board will have a bunch of extras including raid. It will however require xeons not normal i7s. So the chips will be aprox 50 to 100% more expencive than the similar i7.

 

[Tobit]

So the chips will be aprox 50 to 100% more expencive than the similar i7.

E5520's are only $100 more than a Core i7 920 (based on NewEgg pricing, not MC insane deals). If, on the dual EVGA board, we can OC these to 3.5GHz, the dent in the wallet wouldn't be that bad to consolidate maximum cores in a single box.

 

[Mr. Pedantic]

My understanding is the board will have a bunch of extras including raid. It will however require xeons not normal i7s. So the chips will be aprox 50 to 100% more expencive than the similar i7.

But because the chips are picked specifically for server and corporate functions, they will also typically use less voltage to attain any given clock and last longer as well.

I'm already looking at a build of a EVGA Classified + 2 x 480 + a raid 5 of 60gb SSD harddrives. This would not only be a kick butt gaming system but would output around 75k PPD

RAID5 is no good with SSDs, since it pretty much negates their biggest selling point - i.e. high rates of random reads/writes. It would be much better to go with RAID1+0; you'd only lose 60GB of space, but performance would be much better in gaming as well as in general OS performance.

I'm going to be selling my 5870s soon so I can buy a couple of Fermi's.

I would wait. For one thing, there's no real telling how Fermi performs in comparison to RV800. For another, by the time the GTX480 actually comes out the GPU3 client could very well have come out. I would wait for independent reviews of the GTX480 and at least an estimate of AMD folding performance increases with GPU3 before committing to anything.

 

[Kendrak]

E5520's are only $100 more than a Core i7 920 (based on NewEgg pricing, not MC insane deals). If, on the dual EVGA board, we can OC these to 3.5GHz, the dent in the wallet wouldn't be that bad to consolidate maximum cores in a single box.

I agree, but it is 52% more than MC price for a 920. And you start your oc 400mhz lower.

 

[10e]

I think it would be a frickin' travesty if the GTX 480s used 300 watts each. I'd wait for the smaller/slower variants and see how much less they consume.

Just powering them makes my "circuit breaker" finger itchy. Alternatively with the GPU3 OpenMM client coming soon, I'd have to see what the ATI cards do.

Vijay has said on his blog that there might actually be a slight PPD reduction with it on nVidia cards, and that coupled with theoretical performance enhancements for ATI might swing the Folding@home tide back to red team. It's not so bad considering ATI has been better at utilizing power for performance and not heat.

There's also the other issue in that folding@home performance does not always match gaming performance, as they are seemingly two different beasts, so all bets are off until the cards are reviewed for gaming and the GPU afficionados start reporting on folding speed.

 

[Kendrak]

Don't get me wrong, I think dual proc is the way to go..... If your serious about -bigadv. It's just not going to be cheap.

 

[APOLLO]

Vijay has said on his blog that there might actually be a slight PPD reduction with it on nVidia cards, and that coupled with theoretical performance enhancements for ATI might swing the Folding@home tide back to red team. It's not so bad considering ATI has been better at utilizing power for performance and not heat.

I can't comment on the power consumption part, but Ati never had the advantage unless we're taking into account that they were in F@H when no one else had a GPU presence. Since nVidia's debut, there has been no real comparison between the two companies in regards to general performance, CPU utilization and I would say for certain cards even in the heat department.

Don't get me wrong, I think dual proc is the way to go..... If your serious about -bigadv. It's just not going to be cheap.

Multiprocessor setups were never an easy route economically speaking. In F@H, the best option right now is an i7 rig. I'm saying this even though I'm a decades-old diehard multi-socket guy. Coming from someone who almost never builds single-socket systems and has been folding practically forever, that means something.

Why? Because $ for $ there has never been a platform that is so demonstrably superior in almost every aspect of folding and relatively affordable to boot. You can find an inexpensive single-socket S1366 board, slap a couple of GPUs and a 920, OC to around 4GHz and run -bigadv. The combined PPD for the money is almost staggering and hitherto unimagined.

The problem with Gainestown and the soon to be released Gulftown systems aside from the motherboard and processor cost is OC or the lack thereof. Big difference compared to the previous S771 where there were workarounds both with CPU mods and software OC. That's one of the big reasons I haven't touched this platform yet.

-----------------------------------------------------------------------------------------------------------------------------------------------

Regarding EVGA dual S1366 board: the EVGA board might be a technological marvel and an enthusiast's panacea, but at what cost? Unless one is also contemplating building a rendering machine, photo/video editor or HPC, for folding it's way down in the PPD/$ ratio. The motherboard alone will be in the $500-600 range as a conservative estimate. One will also need a killer (or two) PSU ($300-400) from everything I've been gathering around several forums lately. Coupled with the 8-core VM limitation and the drawbacks start to mount. I guess one can always run two or three VMs. Unless of course native Linux is the goal, but then whither the seven PCI-E slots? You will also need a special case to fit it in. Strictly for F@H, the EVGA board is just not an economical way to achieve very high production when everything is taken into consideration, IMHO.

 

[Jathanis]

What are you all thinking of doing when the GTX 480's hit?

Drooling a lot over all of the other builds on here, and cursing uncle sam for taking any and all of my folding cash for the year...

 

[gkanakis]

Well put Apollo!

 

[Kendrak]

Looks like they announced the announcement that Fermi will be out in about a month.
http://hardforum.com/showthread.php?t=1496795

*yawn*

The bit's I've seen is it will cost $600 a card.

Here is some lol'z for you all... look at the comments...
https://www.sabrepc.com/p-174-xfx-geforce-gtx-480-2gb-gddr5-pci-express-x16-retail.aspx

 

[lehmann]

Looks like they announced the announcement that Fermi will be out in about a month.
http://hardforum.com/showthread.php?t=1496795

*yawn*

The bit's I've seen is it will cost $600 a card.

Here is some lol'z for you all... look at the comments...
https://www.sabrepc.com/p-174-xfx-geforce-gtx-480-2gb-gddr5-pci-express-x16-retail.aspx

where still a good 2 months away if where lucky. and boy are they getting there asses cursed on there facebook page

 

[zero2dash]

and boy are they getting there asses cursed on there facebook page

Good! they should be!
enough is enough with this BS. Release the card already; until you have a physical card with a street date and an MSRP, STFU about everything else. I don't care about hearing that there's going to be an announcement where they announce an announcement or a "we're almost there" or posting some lame stupid 5 minutes in MS Paint crap.

Show us the cards or piss off already; man this is irritating. I really *really* wish the GPU3 client was close, or at least some new updated GPU2 client with ATi optimizations. It really is getting under my skin that I have nV cards at this point because I'm sick of all the nonsense with the new cards and the old nonsense of the die shrinks and product rebadges. It's a 9800GTX even though it's really just a souped up G92 GTS, oh wait a minute it's a GTX+, oh snap no now it's a GTS 250 something or other, no way now it's a GTX 250 Super Hyper Rainbow Edition. STFU nVidia.

 

[Elledan]

So much anger about a piece of hardware

 

[Mr. Pedantic]

I think it would be a frickin' travesty if the GTX 480s used 300 watts each. I'd wait for the smaller/slower variants and see how much less they consume.

If GT200 is anything to go by, there won't be any smaller variants. The chips just cost too much to make.

Good! they should be!
enough is enough with this BS. Release the card already; until you have a physical card with a street date and an MSRP, STFU about everything else. I don't care about hearing that there's going to be an announcement where they announce an announcement or a "we're almost there" or posting some lame stupid 5 minutes in MS Paint crap.

That's the thing. Apart from a few ray-tracing and tessellation demos, NVidia hasn't released much about Fermi. And there are three possible reasons for this: 1) it's what they've done before, 2) they don't have enough cards to waste one on tech demos/previews, or 3) they are embarrassed about GF100's performance compared to what they claimed. And I think 2 and 3 are most likely.

Show us the cards or piss off already; man this is irritating. I really *really* wish the GPU3 client was close, or at least some new updated GPU2 client with ATi optimizations. It really is getting under my skin that I have nV cards at this point because I'm sick of all the nonsense with the new cards and the old nonsense of the die shrinks and product rebadges. It's a 9800GTX even though it's really just a souped up G92 GTS, oh wait a minute it's a GTX+, oh snap no now it's a GTS 250 something or other, no way now it's a GTX 250 Super Hyper Rainbow Edition. STFU nVidia.

I would bet quite a bit that there is still quite a lot of rebranding yet to come. Mostly with G92 cards, NVidia really seems to have EOL'd GT200. So I think we should be seeing a 40nm version of the GTS 250/9800GTX+/9800GTX/9800GT/8800GTS/8800GT. If NVidia and TSMC ever get their respective shit together.

Regarding EVGA dual S1366 board: the EVGA board might be a technological marvel and an enthusiast's panacea, but at what cost? Unless one is also contemplating building a rendering machine, photo/video editor or HPC, for folding it's way down in the PPD/$ ratio. The motherboard alone will be in the $500-600 range as a conservative estimate. One will also need a killer (or two) PSU ($300-400) from everything I've been gathering around several forums lately. Coupled with the 8-core VM limitation and the drawbacks start to mount. I guess one can always run two or three VMs. Unless of course native Linux is the goal, but then whither the seven PCI-E slots? You will also need a special case to fit it in. Strictly for F@H, the EVGA board is just not an economical way to achieve very high production when everything is taken into consideration, IMHO.

Because of the bonus, unless it is at least $700, it may still be quite a viable investment for folding in terms of ppd/$.

 

[zero2dash]

So much anger about a piece of hardware

Yeah but seriously, how long has Fermi been talked about? It's been too long and I grew sick of the smoke and mirrors a long time ago (as have a lot of other people). To quote the great Judge Judy "don't pee on my leg and tell me it's raining".

In a perfect world they'd slash prices on the 260/275/280/285s and save us all the trouble of waiting for some cards that may or may not arrive before we (as a people) have flying cars and teleportation chambers, but of course their argument would be "no, we're not going to slash prices since people still buying those old outdated and outperformed cards are the only reason we're making any money whatsoever these days". It's just retarded, utterly retarded.

I'm not saying I was going to rush out and buy 10 Fermi cards; however I want the damn things to pop up in the pipeline because it will drive down prices and drive up competition. (Not like ATi really has any at this point in time though. )

 

[Vaulter98c]

OK, lets not get too pissed off here, yáll are over looking a few things

1) Alot of the rage about Fermi has been started by ATI fanboys, and I can't blame them, the last few years have been really hard on them, only just turning around now. It's not the end of the world that he held a fake at a conference, they had real units in the Demo's, who cares if the plastic he is holding is fake or not? Seriously, it is nothing more than a prop in a speech, who cares?

2) This is partly Nv to blame, but no one has any official numbers on anything involved with Fermi. People are getting tired of Nv not releasing info so they flock to any where else that will throw out some numbers and claim "inside sources". They only do this for attention/traffic, they know that because Nv isn't putting out number that people will look elsewhere, and that leads to revenue in the web. I mean honestly, if everyone has an inside source, why the hell does no one have the same numbers?

3. I could go on and on, but don't base the price off of what one website puts out for a preorder. That will not be the MSRP, the MSRP hasn't even been anounced. All they are doing is jacking the prices to sucker a few people out of money. I mean shit, look at the damn page, it says
Price 429.99
Savings: -70 $
Total 499.99

Seriously? If you don't smell price gouging here I think your sniffer is broken

How about we just sit it out and wait for Fermi to actually show it's head before we get all flame wars' y

Every company has their lemons, every company has their pushed back dates, its how the world works people

 

[APOLLO]

Because of the bonus, unless it is at least $700, it may still be quite a viable investment for folding in terms of ppd/$.

It still doesn't seem to come up as being a worthwhile folding investment over something else like an i7 rig or two for reasons I already outlined in detail further up, but made all the more relevant by recent threads of users with ordinary non-OCable dual S1366 boards, and their problems seeing decent performance over an equivalent single processor.

Things may change when the board is released but I'm not extremely optimistic. I just don't see that it's worthwhile for dedicated folding because of the total system cost, JMO. Calculate the total cost and determine if it's worth it for you, I did. Now, if you'll be doing a ton of other things with it and use the board in a primary system, then that's fantastic. I may do the same come time to replace my main system, but not right away, probably not until 2011.

So far, where it concerns strictly folding, it looks like it would be best implemented in native Linux to see the fullest use of all those available cores. But why splurge top dollar on a halo motherboard for that reason with SEVEN PCI-E slots? To run WINE? Seriously folks.

BTW, off topic but when you quote someone, you never include their username. I didn't realize you had quoted me until I reread your post, LOL. Not a biggie, but never seen this before.

 

[nitrobass24]

If the EVGA Board was E-ATX and didnt need a special PSU or more than one just to turn on, then i would buy it, for my gaming rig. But as it stands it comes with too many caveats.

 

[zero2dash]

OK, lets not get too pissed off here, yáll are over looking a few things

Speaking only for myself - my nV rage is because their PR department is crap, they know it, but they keep doing the same crap.
Announcing an announcement for 9am today and riling up everyone who thought maybe it would be SOMETHING other than what Charlie Demerjian jokes, err, says on SA and then announcing a further data/announcement instead of something concrete everyone cares about like street date or MSRP is what ticks me off.

My thing is - save your announcements. When you want to say something important that 99.9% of people care about, then do it. But don't pull some nonsense and drum up the hype of an announcement, but then turn around and not saying anything on that day at that time other than "oh well we're going to show something at this date down the road". It's the little boy who cried wolf.

 

[Mr. Pedantic]

It still doesn't seem to come up as being a worthwhile folding investment over something else like an i7 rig or two for reasons I already outlined in detail further up, but made all the more relevant by recent threads of users with ordinary non-OCable dual S1366 boards, and their problems seeing decent performance over an equivalent single processor.

Most dual-socket boards are no good because of their insane price and because of non-OC-ability. Since it's an eVGA board, I would be disappointed if it didn't overclock in spades, and since it's not really aimed at an enterprise market I would suspect it would have to be a lot cheaper as well.

Things may change when the board is released but I'm not extremely optimistic. I just don't see that it's worthwhile for dedicated folding because of the total system cost, JMO. Calculate the total cost and determine if it's worth it for you, I did. Now, if you'll be doing a ton of other things with it and use the board in a primary system, then that's fantastic. I may do the same come time to replace my main system, but not right away, probably not until 2011.

We will have to see. I wouldn't buy it in any case, I don't have the money, but my i7 is fine enough as a CPU for now. Top of my upgrade list would probably be to get an SSD, new GPU, or change to WC.

So far, where it concerns strictly folding, it looks like it would be best implemented in native Linux to see the fullest use of all those available cores. But why splurge top dollar on a halo motherboard for that reason with SEVEN PCI-E slots? To run WINE? Seriously folks.

Most cases have 7 expansion slots. It's called flexibility.

BTW, off topic but when you quote someone, you never include their username. I didn't realize you had quoted me until I reread your post, LOL. Not a biggie, but never seen this before.

Yeah, I know. I type out the quote tags. Using the buttons seem really awkward.

1) Alot of the rage about Fermi has been started by ATI fanboys, and I can't blame them, the last few years have been really hard on them, only just turning around now. It's not the end of the world that he held a fake at a conference, they had real units in the Demo's, who cares if the plastic he is holding is fake or not? Seriously, it is nothing more than a prop in a speech, who cares?

I think folding is probably the last thing on people's minds when it comes to GF100. The fact of the matter is that NVidia fanboys across the world (or at least the internet) have been spreading rumours and hearsay generated by NVidia itself, saying that GF100 is a totally new architecture, and it will completely blow away Cypress as a gaming processor. Which is fine in itself. Except GF100 is over a quarter late, yields are still horribly low, it is far too hot, too power-hungry, and too big, all the signs point to NVidia having to slash shader processors and clocks just to get enough samples off their wafers to test the chips, nobody has proper, trusted and proven benchmark results about any part of GF100 performance, and all the while NVidia and their fanboys are still saying that it'll be the cat's pyjamas, bees' knees, or whatever other cliche you want, and it'll blow the socks of Cypress, despite a leak saying that it'll be priced at over $100 more, with sub-par performance in almost every single relevant aspect of hardware operation.

2) This is partly Nv to blame, but no one has any official numbers on anything involved with Fermi. People are getting tired of Nv not releasing info so they flock to any where else that will throw out some numbers and claim "inside sources". They only do this for attention/traffic, they know that because Nv isn't putting out number that people will look elsewhere, and that leads to revenue in the web. I mean honestly, if everyone has an inside source, why the hell does no one have the same numbers?

Apparently this is how they conducted the GT200 launch as well. But also apparently (I wasn't into tech then) GT200 didn't really meet many of the expectations that had been generated - it was too slow, too hot, and too big. And by all indications, GF100 is a much bigger failure.

How about we just sit it out and wait for Fermi to actually show it's head before we get all flame wars' y

Because there are all these people whose 9800GX2s, 8800 Ultras, 9800GTX's etc have started failing, and want an upgrade/replacement for their card(s). The problem with this is that we've been telling them for the last quarter that they have 2 options; they could go with Cypress/Hemlock, or they could wait "a few weeks" for Fermi - because that's how long NVidia keeps telling us it will take for them to come out. I don't know how long other people have been saying this, but for me that's been the situation for about a quarter. It's not on. Aside from the fact that it's just plain bad PR, it's also dishonest and insulting to customers.

Every company has their lemons, every company has their pushed back dates, its how the world works people

That's not how the semiconductor industry works. Delays don't just pop out of the bushes on the path to success. Delays come about because of lack of foresight and/or lack of organization. Cypress came out on time and was a success because AMD planned and organized and realized that that was the way that they had to go with TSMC's botch-ups if they wanted their cards out in time to coincide with Black Friday, Christmas, DX11, and Windows 7. NVidia did not plan and did not organize, and they are suffering for it. Besides which, they are doing something they said they'd never do again, introduce a new architecture on a completely new manufacturing process (the largest chip ever made, on what is arguably the crappiest manufacturing process ever, to boot). NVidia is having delays from what they put out in roadmaps because they were sloppy. They didn't do the homework, they didn't cover all the angles, they didn't learn from their mistakes. And there are mistakes aplenty.

 

[Vaulter98c]

Not to argue, but you are trying to tell me that no computer company has had to delay a product because unforeseen events? Ya...

And its funny, the logic you use to say Fermi is such crap is the same logic you say fanboys use to say it will be awesome, you do realize that. You say Nv kids are using fake numbers to justify it, well you are using fake numbers to disprove it. So, how about you drop the act and we just wait for REAL evidence

Until then, everyone's numbers are fake, no point to fight over them

EDIT

This will be the day we can start complaining

Quote from NVIDIA

Quote:
Apologies for the confusion around our most recent GF100 update. To clarify, launch date for GeForce GTX 480 and GTX 470 is March 26, 2010

 

[EvilAlchemist]

-One will also need a killer (or two) PSU ($300-400) from everything I've been gathering around several forums lately..

All my Dual E5530 System are using SeaSonic 400w 90+ Bronze PSU - about $40 from Newegg. All you really need if you are running headless.

I agree, those board will be expensive but if they allow overclocking, maybe worth the additional costs.

 

[Vaulter98c]

Ive got a 750 running my Quad Socket (ok, its not really stable but thats a memory issue/bios flash needed)

 

[Mr. Pedantic]

Not to argue, but you are trying to tell me that no computer company has had to delay a product because unforeseen events? Ya...

I never said anything like that. It's just that product-related delays do not come about randomly.

And its funny, the logic you use to say Fermi is such crap is the same logic you say fanboys use to say it will be awesome, you do realize that. You say Nv kids are using fake numbers to justify it, well you are using fake numbers to disprove it. So, how about you drop the act and we just wait for REAL evidence

These are undisputed facts:

1) Fermi is late. Very late.
2) Even when it wasn't late, NVidia admitted it was very, very hot, and very, very power-hungry. Those are not disputed.
3) NVidia originally planned for 512 SPs for the GTX480, which apparently gave it a 60% advantage over Cypress. Now it has been cut to 448, a 12.5% drop. This alone reduces its lead over Cypress to 40%, not 60.
4) TSMC's 40nm process sucks. The best way to minimize errors and increase yields on a sucky process is to use a small die to work out the kinks, since a smaller die means less probability an error will occur on any single die. AMD did this with the 4770. Their attempt worked. NVidia did this with the G210, GT220, GT240, and tried with the GT200 architecture. Neither companies did perfectly. However, AMD did a damn sight better than NVidia. The 4770 was quite big, whereas the GPUs NVidia tried were too small, and then they tried for too big a jump to GT200 (and later Fermi when GT200 was scrapped for 40nm).

With just those facts, the prospects aren't rosy for NVidia. Add on rumours of clock cuts as well (say bye-bye to 40% faster than Cypress) but with the same power and heat numbers, and it starts looking even worse.

 

[Salavat23]

Good! they should be!
enough is enough with this BS. Release the card already; until you have a physical card with a street date and an MSRP, STFU about everything else. I don't care about hearing that there's going to be an announcement where they announce an announcement or a "we're almost there" or posting some lame stupid 5 minutes in MS Paint crap.

Why would they stfu about everything else? Their main profit generator isn't the high end market segment, and never will be.

Plus, they shouldn't be rushing anyways. They compete well already in the current market. The last thing they want to be is in a situation where the product is not at the level they want it to be at, and they haven't ironed out their yield issues yet.

Show us the cards or piss off already; man this is irritating. I really *really* wish the GPU3 client was close, or at least some new updated GPU2 client with ATi optimizations. It really is getting under my skin that I have nV cards at this point because I'm sick of all the nonsense with the new cards and the old nonsense of the die shrinks and product rebadges. It's a 9800GTX even though it's really just a souped up G92 GTS, oh wait a minute it's a GTX+, oh snap no now it's a GTS 250 something or other, no way now it's a GTX 250 Super Hyper Rainbow Edition. STFU nVidia.

Who cares if it's re-branded. Is the re-brand at a high price? No. Is it slower? No. If you don't care enough to put in 2 minutes of research time into a product, then you get what is coming at you.

The auto market re-brands products all the time, and you never hear anyone complaining about that. When was the last time you saw some raging that the fact that the Chevy Uplander was basically a re-branded Pontiac SV6? SSDs are also often rebranded from manufacturer to manufacturer. Yet no one has any issues with that.

Plus, who says that Nvidia is in an urgent position to release any high end product right now? First of all, no matter what they release for their high end, they will lose money. Next, they are already fairly competitive at the very high end. And lastly, the last top range card released by Nvidia, the GTX 295, came out just over a year ago. They are still well within the average time it takes to release a new line-up of cards.

 

[APOLLO]

Most dual-socket boards are no good because of their insane price and because of non-OC-ability. Since it's an eVGA board, I would be disappointed if it didn't overclock in spades, and since it's not really aimed at an enterprise market I would suspect it would have to be a lot cheaper as well.

Overclocking isn't everything and I have a Skulltrail to prove it. It doesn't OC that much better than my Supermicro boards unless you run very expensive processors that are 3GHz or higher stock, which can take high voltages. Beyond the ~$400 mark, a motherboard quickly succumbs to the Law of Diminishing Returns, regardless of feature set. That has been my experience.

We will have to see. I wouldn't buy it in any case, I don't have the money, but my i7 is fine enough as a CPU for now. Top of my upgrade list would probably be to get an SSD, new GPU, or change to WC.

I wouldn't buy it even if I did have the money unless it proves to be a revolutionary or watershed product that Skulltrail failed to be.

Most cases have 7 expansion slots. It's called flexibility.

True, but what I meant was with all those PCI-E slots, most folders will not be buying this board to run native Linux and WINE to take advantage of it. Don't get me wrong, I love Linux and wish I was more versed in it, but it just doesn't make sense to me from a folding standpoint. Those PCI-E slots are begging for heavy GPU use and I cannot justify dumping ~$700 on a dual-socket board to run Linux in order to fully exploit its CPU potential. This will completely change once -bigadv will be released in Windows A3, but until then this board is not the best buy for a folder's budget.

Yeah, I know. I type out the quote tags. Using the buttons seem really awkward.

You lost me there. You type out the quote tags...??

Seriously, not intending on being overly critical or disputative but reading your last few posts makes it all the more evident. It's like sailing an ocean without any navigational tool, for the lack of a better analogy. I could see the possible reasons for doing it this way, but it makes it hard for someone reading to know exactly who you are replying to, unless we continue glancing back through the thread to find the relevant posts.

 

[zero2dash]

they haven't ironed out their yield issues yet

Really?

Who cares if it's re-branded. Is the re-brand at a high price? No. Is it slower? No. If you don't care enough to put in 2 minutes of research time into a product, then you get what is coming at you.

Nice attitude.

The auto market

....has nothing to do with this.

SSDs are also often rebranded from manufacturer to manufacturer. Yet no one has any issues with that.

Intel builds crippled Kingstons. Name others.
Intel, Indilinx, Samsung, now SandForce. 4 MLC makers. Indilinx provides chips to several companies, as does Samsung. Intel is Intel only except for a few Kingstons (V series of which has no TRIM support). SandForce is brand new and OCZ only ATM.

Plus, who says that Nvidia is in an urgent position to release any high end product right now?

...apparently they do otherwise they wouldn't be saying shit about it.