Gtx 580m?

As some of you may know the GTX 580 ( desktop ) has just been released recently.
OcUK GeForce GTX 580 1536MB GDDR5 PCI-Express Graphics Card []

however all of the others are currently on pre-order.

how long till we can expect a mobile version?

 

I doubt we will see the mobile version any time soon, the 400M fermi cards haven't been fully implemented to the market yet, eg: 470M and we just recently see some new 400M cards appearing in laptops, eg: 435M/445M in dell XPS, and they are just on pre-oder now.

I am more excited about the new ATI 6000 mobile cards, the prices of new Nvidia cards(460M/470M/480M) are just exorbitant compared to ATI equivalent, I liked Nvidia in the past but it's losing its appeal to me by each passing day.

 

Got me excited for a moment to see what I was expecting but

 

I'm willing to bet the 580m is a revision of the 470m.

 

I just found some screenshots of this new GPU, it´s supposed to be smoking fast:



If the 480m is hot I can´t imagine having a 480m on steroids in a laptop...

 

I thought I read somewhere that the 580 actually runs a bit cooler than the 480 ._.

 

Desktop GTX 580 only runs cooler because it has a limiter chip that throttles the card down when stress tests are run.

The GPU-Z guys found a way to get past the limiter and discovered the total power on the GTX 580 around 350W.

 

HAHAHA!! Laugh of the day .. but jokes aside , the 480M runs cool.. someone in my college's gaming club has a sager and woohh its cool.. way cooler than my G73 but louder , with 3 fans and not to mention a copper heatsink... you can't have everything

Wow.. worse than GTX480... anyways , how would you get past the limiter lol? But 350W... lol.. that absurd.

 

I only pay attention to mobile GPU's that cost under $1k, so therefor I am not even remotely interested in the release of the 580M.

 

But does that really matter if it still performs better in games even with the limiter?

Either way, if the GTX 580M is based on the GTX 460 (or maybe 560?), it'll probably run cooler regardless.

 

well the limiter limit only in furmark and such software so it likely consume more power in games where the limiter don't kick in thus thier actual consumsion number are simply lies

 

As I understand it, the issue is that the limiter only serves to hide the the card's power draw by throttling when it detects specific applications like OCCT or Futuremark that are designed to stress test the GPU. Other apps and games that aren't on the limiter chip's "list" can still push the card to it's max power draw.

For overclockers it can be especially problematic when the settings that look safe after a stress test still can't be trusted to not end up frying your card when running something not on the list.

 

Is this limiter on all the GTX 580 range? sounds like my next GPU is an ATI..

 

Just fire up Stalker CS with god rays and everything else maxed out and see if the card starts smoking. I tried maxing out Stalker CS last year @ 1680x1050 with 8x AA using a PII x4 @ 3.5ghz and HD4850 xfire and I still got a slide show but it was a great looking slide show

 

Ah. But it still runs cooler while gaming than the GTX 480, doesn't it?

 

Only because the cooler is better. The same cooler that can render the 5870 silent makes the 580GTX as loud as a GTX280.

I don't think you can put that cooler into a laptop.

Would be interesting to see vapor chambers though.

The GF106 is still better at lower power envelopes, we will need to wait for the GTX560 for new mobile chips.

 

i could keep all my computer under 20c with that coler alone

 

With the limiter on it you can probally get a program like throttlestop to over power the limiter, but it will decrease battery time but if your on a cord then it doesn't matter.

http://forum.notebookreview.com/alienware-m11x/524257-how-supercharge-m11x-core-i5-i7-um-cpus.html

This is just for the m11xR2 (I think) but he will update it to work with other things to.
For the m11xR2 it made the time in wPrime go from 30 secs down to 23secs (Estimated)
Use at your own risk as always

 

That weak joke is getting really old.

Clevo has already proven, that the 480M can be easily cooled. The GPU barely hits 80C under load.

 

yea...i always wondered where this really bad joke originated from. at least it makes it easier for us to filter out the nonsense posters from real infomation.

@ alstrike: i used to have a m1530 as well so i know all about how hot that thing can get. please quit listening to, and spreading, misinformation. the highest i've seen my 480M hit was 64c - 65c under hours of gaming. the same games would make my 8600M GT hit 80's-90's.

 

wait, so you're telling me that the 480 runs COOLER than the 5850/5870/460? Because I remember reading temps under load (furmark) for those cards and they were all usually in the 80s.

 

Again, the problem with FERMI is not temps themselves, they can be easily controlled and current mobile solutions show that. The thing with fermi is the power consumption, which is mostly prominent on the high end GPUs.

 

sorry, let me specify. i was comparing my 480M's temps to my 8600M GT's temps with my cryo LX. my NP8850 gets around low-mid 70's with the back propped up. i rarely use it off my cooler, so i don't remember the exact number.

what he said.

don't 5870's run hotter than the 480M's in the sager? i know my buddy's g73 runs a bit hotter than my NP8850.

 

But power consumption correlates with the temperatures... You are right about the original GF100 having higher power consumption but all the designs based off the GF104 match AMD's offerings.

 

Actually a 6870 uses 10% less power and outperforms a GTX460.

 

I'm lucky, since my company pays for the electricity and water bills!

Fermi all the way.

 

First IMO comparing a Clevo with 480M to a G73jh to make a statement of whether 480M runs cooler or not is stupid. The cooling systems are different with different results, one being the ultimate cool and the other being the ultimate in decibels. If you want to make such a statement, put a HD5870M in a NP8850 then compare. That's the only way to do it, compare the GPUs in the exact same laptop.

But... A MSI with a single fan cooling both an i7 and HD5870M runs around 70C while gaming and under 70C for vantage benchmarking. I've seen some insane temps from Alienware after the user redid the thermal paste and thermal pads. And my G73jh even with Furmark runs at 78C without direct air intake.

Tell your friend to redo his thermals. Bad assembly of the GPU on the G73jh does not mean the HD5870M runs hot, all it means is there was some serious issues with Asus when assembling the G73jh. It seems they have resolved that with the G73jw, but too late for your friend.

Just for benchmarking fun, 850/1200 (same clock setting as desktop HD5770). No notebook cooler or having the back removed etc. No modifications to the heatsink. And you'll find there are plenty of other G73jh owners having similar results and the only thing they did, was just use a different thermal paste, which IMO is by no means a modification, that's just G73jh being assembled the way it should have been stock.

 

And that is true to some extent, but you have to remember that cooling methods are not the same for all gpus. Put a crappy fan with a small heatsink and watch a 480m or any high end GPU melt away. Also, powerconsumption to temperature ratio is not linear, and it heavily depends on the physical die, size etc besides the cooling solution.

All nvidia offerings consume more power than AMDs offering on a similar price range. Their best offerings are indeed the GF104, but even those still consume a bit more than comparable AMD offerings.

 

And they require massive cooling. Even the GTX 460M on a G73jw compared to G73jh is nearly twice the size.

 

Uhmm.. wasn't the GTX 480m released about 3 months ago? Why such a quick update?

In my experience all manufacturers have serious issues when assembling their computers, especially the thermal paste part. Which is why I go barebones when I have the chance; doesn't Asus void your warranty if you even touch your GPU? or is that just the CPU? I'm not sure.

 

No update yet. There isn't even news about it. The GTX 580 is 512 core GF110, it's so massive there is no way Nvidia would use it for a basis for any mobile GPU. So it's just wishful thinking I guess by the OP? Nvidia is just playing name game sine AMD have released 2 of their new series and the others are expected soon. IMO the GTX 580 does not deserve a 5xx name. It should be GTX 485, but then IMO Nvidia is nearly 50% marketing

The GTX 470M was just released and that should hold Nvidia over for a while IMO. And as others have noted, even for desktop the GTX 4xx series is not complete yet, so I'd expect to see some more 4xx GPU released for mobile in the future. 470M should be faster than the HD5870M so IMO that probably is Nvidia's current flagship that has reasonable efficiency.

 

The GF100 GPU it's based off of is supposedly already going EOL.

 

Yup, the GF100 GTX 465 (480M) was essentially murdered a while back, by the launch of the GF104 GTX 460. Nvidia itself basically said, "oops, this is what the 465 was supposed to be."

Back on topic:

Expect disappointment from the high-end of 500M.

The best we'll see out of this fabled '580M', will be a chip with the full desktop GTX 460's specs unleashed. But then, the required downclock will leave it within 5-10% of the GTX 470M anyway. One can only hope, that Nvidia at least prices 500M competitively, with whatever Mobility 6000 turns up as. Again, expect disappointment in that regard also.

Nvidia should just give up on the mobile market, until 28nm is available. They're tapped out. Spent. The current Fermi architecture just doesn't have any headroom left for the mobile market, even GF104. Well, wait, the 470M would be impressive, if only it was $300 cheaper.

Because you have to flip the entire motherboard, plus remove the keyboard, to access the CPU/GPU, Asus will void your warranty for attempting that, period.

The greatest downside of the G73, is that it's internals aren't simply behind the back panel.

 

The joke comes from Nvidia and the community.

Both the community and Nvidia mocked their Geforce FX 5800 chip for being hot, noisy and impotent when it came to DX9. The jokes showed leaf blowers, grills and hair dryers.

It has been applied to any particularly power hungry chips since.

 

I don't understand how these latest 4xxm and 5xx chips run soooo hot? Especially because my 320m is one of the coolest running and fastest integrated GPU's.

Why can't Nvidia put some of the power saving features from the 320m(and 9400m) in the new 4xx and 5xx chips?

 

They have, the idle power numbers are very impressive. They are however over 3 billion transistors compared to your chips few hundred million.

 

i know, and understand, exactly where the lame joke stems from. all i'm saying is that it's not even really true besides that the card does use a bit more power.

it's more of a bad/exaggerated fanboy joke to me. :shrugs:

@ ruckus: i wasn't making my temp comparsions between a 5870-equipped g73 and a 480m-equipped NP8850. the sager 8760's come with 5870's, and they are essentially the same shell as my NP8850.

 

ATI 6X series would be nice anyway with the new M17X R3.

 

We've been over this. They don't run hot.

 

The 580M is very likely going to be a 470M with the chip fully enabled. It will likely be faster than the 6xxx series as the 470 is already faster than a 5870. Wait and see of course, but in Jan you'll probably see the R3 with Sandy Bridge and 6xxx series and 470's, with a 58x either as an option in January or shortly thereafter.

I remember when the 1730 sold with the 8700 SLI and folks felt stupid when the 8800 came out (the "real deal" in performance back then) a short time later. 470's will be nice in the R3, but the 5xxM will be a worthy chip that will last quite a while.

28nm is coming along slowly and both vendors aren't likely to get mobile chips based on it out for another year or bit longer, probably at the next Intel refresh after this or a bit later than that even. That gives the 6xxx series and 5xxx series a clear year of running room before they are outdone.

Meanwhile AMD has been steadily improving drivers, proving that Nvidia can't just sit around idle. If either side does that, it will go badly in the market for them. Competition, however is good for us as consumers.

 

I can dream....:wub:

Maybe they'll release the 570M, I think thats more plausible than 580m.

 

That's very doubtful, because Fermi has hit the ceiling, in mobile terms.

Unlocking any more of the GF104 will have to be accommodated by lowering the clocks below the 470M's frequencies. It will be a 480M situation all over again, meaning more cores but the performance will not match. Nvidia has no headroom left here. Maybe 5%.

The mobile Barts PRO will be untouchable, until 28nm GPUs hit.

 

I think most of whatever we see new its just going to be rename gpus, as kevin said, until we 28nm from both we wont see much reason to upgrade, even the upcoming 6970m/6870m probably wont be much better.

 

You will not see this for many months likely ATI will again dominate mobile GPU'S with there new line of 6k Cards NO DOUGHT.

 

I don't think the 5xxx is "dominating" by any stretch of the word. We'll have to see if the 6xxx series is anything more than an evolutionary upgrade, which I think is the most likely scenario.

That said, I don't think the 460/470's are "dominating" either. It's more of a bit of this bit of that scenario, which is what you'd expect from solid competition.

Since it looks to me like the 6xxx series will be "on time" when Sandy Bridge comes, to me the real failure would be Nvidia being late "again"...


Oh and one tiny note on the comment above regarding a fully enabled GF104. Realize that the 470 has more cores enabled, yet goes slower than the 460, but still outperforms the 460, so don't think it's guaranteed that a fully enabled GF104 wouldn't be both possible and a crowning performer to the mobile line.

 

By fully enabled do you mean GF104(GF114) with 336 cores/192-bit, 336 cores/256-bit or 384 cores/256-bit?

For desktops Barts already beats the GF104 in performance, perf/watt, and perf/mm (AKA cost) and the only reason why that wouldn't carry over directly to mobile cards is because AMD lowballed Blackcomb's clocks/TDP.

As for GTX580M's debut my guess is it shows up the first week of March when Sager announces their new machines at CeBit 2011. Lowest spec I'd speculate for is 288 cores/256-bit (desktop GTX 460 SE) and highest is 336 core/256-bit....though Nvidia might be able to save some TDP going 192-bit and using 6 VRAM modules instead of 8. All the full GF114 GPUs will be needed for desktop GTX 560.

28nm in Jan 2012, brings Cayman's new Southern Islands arch to mobiles...same to greater perf out of less SPU and 10% savings on die area. Nvidia will need to do more than die shrink Fermi.

 

80% of DX11 cards are HD5xxx.

How would you describe dominating? I suppose you're right, that's not really dominating. That's annihilating.

 

You know we're talking about performance. Please take your crusade elsewhere.

 

Does anyone know how many watts the 580 gtx uses. I just got a decent sized check and I'm going to use it to build a new desktop. I want to eventually run two of these in SLI and need to know how big of a PSU to buy.

 

Peak total power with a mild oced i7 quad core will be around 770-800W (depending on the chips you get, each has its own individual voltage).

You will be looking at around 290W peak per card.

Yeah, you wont be seeing that in a laptop lol.