Nvidia Geforce 'Maxwell' Thread

[CAT-THE-FIFTH]

TBH, a GTX660,HD7850 or R9 270 don't consume that much power.

My Xeon E3 1220,H67 mini-ITX motherboard and a GTX660 is lucky to break 200W at the wall during gaming,and is usually well below that.

Plus my card is also pre-overclocked too.

If I had an IB or Haswell CPU,it would be even lower.

A decent 300W to 400W PSU would run any of the three cards.

That means a £30 to £40 PSU.

Anything under £30 is usually a pile of crap anyway with poor construction,unless there is a clearout sale on a decent budget model.

 

[Suarez7]

I wouldn't trust a cheap PSU again even on a low end card, so many of them break even when not pushed and when the cheap ones give way they usually take your whole system.

The lowest I'd go would be the Corsair CX series for a low end build.

 

[CAT-THE-FIFTH]

The XFX PRO 450W is a Seasonic S12II design so is pretty decent,and even the CX430W V3 is Okish.

TBH,as a result I am not sure whether the GTX750TI makes much sense over the GTX660,HD7850 or R9 270.

Any systems we spec,build or suggest will have decent quality PSUs.

Personally,I think the GM107 is a mobile GPU,Nvidia has hyped up for desktop. It sounds more of use there,although I expect boost will be absent or curtailed.

 

[humbug]

I wouldn't trust a cheap PSU again even on a low end card, so many of them break even when not pushed and when the cheap ones give way they usually take your whole system.

The lowest I'd go would be the Corsair CX series for a low end build.

Ain't that the truth, Cheap PSU's can be a death sentence to the rest of your rig.

 

[CAT-THE-FIFTH]

I also think it is a response to Iris Pro as Intel was targeting the GTX650M.

Apple ended up replacing Nvidia GPUs with Intel Iris Pro IGPs in some of the iMacs and MBPs too.

Iris Pro was created especially for Apple.

Hence they are a massive threat to Nvidia especially for a high value customer like Apple.

Since the latest refresh of the Apple laptops and desktops will be in a few months time, as they do updates every six months,I can see the GM107 being their answer to Iris Pro,especially since Apple will probably be the first to receive Broadwell chips.

If Nvidia can get them in many of the Apple products,it would mean Broadwell Iris Pro will be fighting a faster GPU now,even in low end iMacs and various Apple laptops.

 

[Rroff]

Yep all cards once running past their normal speed seem to use a great deal more power, I'm sure I seen a post here of a guy with one 780 pulling 800w from the wall with an overclock.

Seems wrong to me - my whole system pulls 370watt in BF4 with CPU at 4.4ghz and GPU at 1188 and under 400 with max GPU clock. No problems with boost dropping out - can run full clocks all day long.

 

[Suarez7]

Seems wrong to me - my whole system pulls 370watt in BF4 with CPU at 4.4ghz and GPU at 1188 and under 400 with max GPU clock. No problems with boost dropping out - can run full clocks all day long.

This guy was putting like 1.35v through his card I think, maybe more.

 

[vsg28]

270X performs better than a GTX 580 and 6970. If we didn't have generational performance increases something would be very wrong.

Sure, but that doesn't make good propaganda for Nvidia marketing

 

[CAT-THE-FIFTH]

BTW the hardware.fr analysis of the GM107 hardware design is quite detailed:

http://translate.googleusercontent....l.html&usg=ALkJrhg1ZZurn5AVLp9tTmnnxEcoqJPsbg

n its global communication, Nvidia Kepler present as architected around SMX monolithic face Maxwell whose SMM were partitioned for greater efficiency:


This presentation of things however, is not quite correct. More than a technical reality, it is actually easy to sell and prepared by the marketing to decorate various items, which may mislead many journalists simplified technique history.

The reality is different and more complex. SMX Kepler GPUs are already partitioned into four GPUs as Maxwell, the main difference at this level is to be sought in the resources that are shared by pairs these partitions. To represent this, we changed the architecture diagrams Nvidia to get closer to reality, to the best of our knowledge of different architectures:

Nvidia has really simplified the SMM 2 main points:
- Pooling of texturing units per pair of scores as the GK208 and GK20A (Tegra K1), which is to increase the ratio of calculation units per unit texturing, a natural evolution
- Removal of the block units further calculation shared between 2 sheets of Kepler, it boostait the theoretical power of 50%, but had a rather weak performance in practice

So when Nvidia announces a performance by "heart" up 35%, this does not mean that the main units of computation have been improved, but this block was removed misused. This results in a significantly smaller when the SMX during heavy work in SMM calculation, its performance is very close, on the order of 90% by NVIDIA. Of course, in the game, when the texturing becomes more important, the new SMM may have to settle for 50% performance SMX.

Other small optimizations have been implemented: latency reduction of 32-bit units, fusion of L1 cache and texture cache, shared memory extended, some faster ... what instructions to give a boost to GPU computing.

They basically took the GK100 series design,and shared parts between each pair of units in a SMX to save space and power. It also means a reduction in the processing power of certain workloads.

So basically they took Kepler and gimped it even more to save power and transistor count. OTH,it also means the higher end chips might not be the same design,as they might be more focused on processing power than power consumption.

 

[humbug]

BTW the hardware.fr analysis of the GM107 hardware design is quite detailed:

http://translate.googleusercontent....l.html&usg=ALkJrhg1ZZurn5AVLp9tTmnnxEcoqJPsbg

Spoiler

They basically took the GK100 series design,and shared parts between each pair of units in a SMX to save space and power. It also means a reduction in the processing power of certain workloads.

So basically they took Kepler and gimped it even more to save power and transistor count. OTH,it also means the higher end chips might not be the same design,as they might be more focused on processing power than power consumption.

Oh for goodness sake...

This explains why it falls flat on its face in things like Crysis 3 which uses a lot of texture rendering.

Sorry, but its not good to see GPU's being interfered with like that, take something away from it to make the 'paper' numbers look a bit better here and there....

We will end up loosing what could be starting to get known as a proper GPU just for the sake of some watts shaved off, which looks good in the marketing gumph and saves money on components. i mean look at this thing, there is nothing to it, it looks like it was knocked together for a fiver.

 

[Davedree]

Oh for goodness sake...

This explains why it falls flat on its face in things like Crysis 3 which uses a lot of texture rendering.

Sorry, but its not good to see GPU's being interfered with like that, take something away from it to make the 'paper' numbers look a bit better here and there....

We will end up loosing what could be starting to get known as a proper GPU just for the sake of some watts shaved off, which looks good in the marketing gumph and saves money on components. i mean look at this thing, there is nothing to it, it looks like it was knocked together for a fiver.

you can't say things like that Humbug, it would be the same as saying the 680 was just a mid range imposter.

The problem with it is the price, and the fact we are still on 28nm

 

[Boomstick777]

No idea what those guys ^^ are talking about.

The compute performance is back with a vengeance, along with mining performance / gaming performance and extreme low power use. Maxwell is looking like a beast of an architecture. All that performance and only 640 cores. AMD need to pray Nvidia don't release a 3000 shader core part on 20nm. God only knows how much power AMD would need to use to match the performance.

 

[Suarez7]

lol Boom, too early to start the engines on the hype train yet.

 

[Boomstick777]

lol Boom, too early to start the engines on the hype train yet.

Nvidia are about to release the Kraken I tells ya

I'll say it now, the GTX 880 is going to wipe the floor with anything AMD have coming. Quote me on this in Q4 when the cards are out

 

[humbug]

No idea what those guys ^^ are talking about.

The compute performance is back with a vengeance, along with mining performance / gaming performance and extreme low power use. Maxwell is looking like a beast of an architecture. All that performance and only 640 cores. AMD need to pray Nvidia don't release a 3000 shader core part on 20nm. God only knows how much power AMD would need to use to match the performance.

Oh come on Broomstick. here we go again with this hyperbolic nonsense "AMD need to pray"

The same sort of thing keeps getting repeated over and over again by Nvidia enthusiasts.

Look, the last one was with the GTX Titan, it was insisted upon that AMD could not match the GTX Titan on 28nm, its impossible, its a fact and all the rest of it.
It would be far to big, it would consume far to much power and so it went on by self proclaimed experts. Nvidia FTW.

And yet here we are now with an AMD Hawaii 28nm GPU thats Much smaller than the GTX Titan and Faster with good power levels.

Your face is already dripping with Egg Yoke and now you have stuck your face back out into the firing line yet again.

Sucker for punishment?

 

[Gregster]

Well to be fair, AMD had almost a year to beat the Titan and then they only just managed it (debateable in most games) and a week later, nVidia release the full fat Titan (780Ti) and retake the performance crown, so AMD are playing 'catch up'.

The 750Ti mining well and at such a low power draw will surely see the miners flocking to nVidia and at least putting the prices of AMD cards back to normal, so not a bad thing.

 

[Boomstick777]

Oh come on Broomstick. here we go again with this hyperbolic nonsense "AMD need to pray"

The same sort of thing keeps getting repeated over and over again by Nvidia enthusiasts.

Look, the last one was with the GTX Titan, it was insisted upon that AMD could not match the GTX Titan on 28nm, its impossible, its a fact and all the rest of it.
It would be far to big, it would consume far to much power and so it went on by self proclaimed experts. Nvidia FTW.

And yet here we are now with an AMD Hawaii 28nm GPU thats Much smaller than the GTX Titan and Faster with good power levels.

Your face is already dripping with Egg Yoke and now you have stuck your face back out into the firing line yet again.

Sucker for punishment?

No need to get serious bumhug, just having a bit of fun . Dripping with Egg Yoke, I have no idea what you mean you weirdo, anyway Maxwell is poised is to be a great arch. Nvidia went light last time with the first round of Kepler cards. If they release a full fat Maxwell part from the get go on 20nm. AMD could be in trouble given they have to stick with GCN, hot/power hungry.. While Nvidia Maxwell architecture's performance per watt is phenomenal.

My prediction is the GTX 880 will be the return of Nvidia having a genuine big lead over AMD. You heard it here first

Can quote me in Q4 , Maxwell destroying whatever GCN cards come next.. Even a washed out Mantle API won't help em

 

[humbug]

Well to be fair, AMD had almost a year to beat the Titan and then they only just managed it (debateable in most games) and a week later, nVidia release the full fat Titan (780Ti) and retake the performance crown, so AMD are playing 'catch up'.

The 750Ti mining well and at such a low power draw will surely see the miners flocking to nVidia and at least putting the prices of AMD cards back to normal, so not a bad thing.

I'm all for Nvidia taking up the Mining crown, Retailers can stop price gouging AMD GPU's and do it to Nvidia for a change

No need to get serious bumhug, just having a bit of fun . Dripping with Egg Yoke, I have no idea what you mean you weirdo, anyway Maxwell is poised is to be a great arch. Nvidia went light last time with the first round of Kepler cards. If they release a full fat Maxwell part from the get go on 20nm. AMD could be in trouble given they have to stick with GCN, hot/power hungry.. While Nvidia Maxwell architecture's performance per watt is phenomenal.

My prediction is the GTX 880 will be the return of Nvidia having a genuine big lead over AMD. You heard it here first

Can quote me in Q4 , Maxwell destroying whatever GCN cards come next.. Even a washed out Mantle API won't help em

Nvidia did well to get the power consumption down, but its not what you think it is, while i'm sure there is some genuine fine tuning going on they also achieved it by striping out some of the GPU's features or the performance of those features.
A bit like pulling one of the HT leads off your Cars engine to save fuel.

I hope AMD don't do the same, tho my fear is they will have no choice now.

 

[CAT-THE-FIFTH]

lol Boom, too early to start the engines on the hype train yet.

He already fired the rockets and got the fusion drive ready.

A bit like the infamous Rollo.

But he was funny.Unintentionally.

 

[Gregster]

Ohhh and it's "Egg Yolk"