Mobile Kepler review

[Gikaseixas]

I think the average performance increase over the GTX580 should be something around 60%.

You're being way too generous to the green guys IMO %wise

 

[Quantos]

I for one am not ready to extrapolate the results shown here to the discrete card(s). Let's just take this for what it is, a mobile GPU, and not get ahead of ourselves with predictions based on this. :hmm:

 

[Gikaseixas]

I for one am not ready to extrapolate the results shown here to the discrete card(s). Let's just take this for what it is, a mobile GPU, and not get ahead of ourselves with predictions based on this. :hmm:

I agree

 

[sontin]

You're being way too generous to the green guys IMO %wise

GTX580 is ~1.4x faster than GTX560TI.

GTX680@GTX560TI Clock will be ~2x faster than GTX560TI.

That would be a performance increase of 1.42x over the GTX580.

I'm more optimistic because in the worst case i don't need to buy the GTX680. :awe:

 

[PingviN]

GTX580 is ~140% faster than GTX560TI.

Not even close dude.

http://www.techpowerup.com/reviews/MSI/GTX_560_Ti_448_Cores_Twin_Frozr_III/27.html

You're about as off as you can get.

 

[Imouto]

Brau brau.

 

[sontin]

Thx guys, changed it from % to x.

 

[tviceman]

That would be awesome but it probably doesn't scale linearly like that, due to various bottlenecks. Maybe even memory bottlenecks.

I think he was joking around.

 

[Lonbjerg]

Not even close dude.

http://www.techpowerup.com/reviews/MSI/GTX_560_Ti_448_Cores_Twin_Frozr_III/27.html

You're about as off as you can get.

A chart that mixes 1024x768 noAA settings into the mix is useless.
IMHO...

 

[railven]

Thx guys, changed it from % to x.

1.4x == 140%

Am I the only one who still doesn't see what he is trying to get across?

 

[Will Robinson]

No you aren't the only one.
I'm afraid he may have unrealistic expectations of Kepler's performance.

 

[swilli89]

Not even close dude.

http://www.techpowerup.com/reviews/MSI/GTX_560_Ti_448_Cores_Twin_Frozr_III/27.html

You're about as off as you can get.

Did you really just set the 100% baseline to the 448 Core version? I believe he said 560TI not "560TI 448 Cores Edition"

There is cherry picking then there is this, cherry picking + straight up using a different product to prove a point.

 

[SirPauly]

Not even close dude.

http://www.techpowerup.com/reviews/MSI/GTX_560_Ti_448_Cores_Twin_Frozr_III/27.html

You're about as off as you can get.

Not really! Actually using watered down results that offer 1024 x 768 with no AA -- really isn't helpful.

http://www.computerbase.de/artikel/...60-ti-448-core/5/#abschnitt_leistung_mit_aaaf

The GTX 560 TI and GTX 580 are listed here, with 40 percent , actually really close.

 

[PingviN]

Did you really just set the 100% baseline to the 448 Core version? I believe he said 560TI not "560TI 448 Cores Edition"

There is cherry picking then there is this, cherry picking + straight up using a different product to prove a point.

The GTX560Ti is in the charts too.

I even link to the entire performance summary, are you too freaking lazy to click the link and check the resolutions you find relevant?

 

[SirPauly]

The GTX560Ti is in the charts too.

I even link to the entire performance summary, are you too freaking lazy to click the link and check the resolutions you find relevant?

Sure, but it's your tone of "You're about as off as you can get," to a 40 percent difference -- when it actually is -- was puzzling

 

[Arachnotronic]

i can very accurately calculate gk104 gtx 680 by this information on mobile part.

1536 virtual cores act more like 768 real cores divided by 512 gk110=150%

the 1004mhz vs 700mhz balance each other out due to the bandwidth constraints.

150% gf110
+6% give
-6% take
--------------

=gfk104 is anywhere from 111% to +132.65% of a 7970 give or take 28%!!!!!


i would put money on it

Buy NVIDIA stock, then.

 

[SirPauly]

1.4x == 140%

Am I the only one who still doesn't see what he is trying to get across?

With a base-line comparison some use 1.4x or for example 1.6x:


http://www.google.com/imgres?hl=en&...3&tbnw=176&start=0&ndsp=49&ved=1t:429,r:2,s:0

 

[taltamir]

I'm glad most people are impressed with the performance... but if i read the above statement... I'm more scared than impressed..

I noticed that, this is too hot and I deem it unacceptable.
I hope for their sake that this is due to a poorly ventilated machine rather then their GPU.

 

[railven]

With a base-line comparison some use 1.4x or for example 1.6x:


http://www.google.com/imgres?hl=en&...3&tbnw=176&start=0&ndsp=49&ved=1t:429,r:2,s:0

It's essentially the same meaning, 1.4, 1.6, 140%, and 160%. He edited his post but I didn't understand why if he said the same thing.

EDIT: OHHHH I saw, he said 140% faster, where he meant 40% faster. Reading comprehension fail on my part haha.

 

[Vesku]

I noticed that, this is too hot and I deem it unacceptable.
I hope for their sake that this is due to a poorly ventilated machine rather then their GPU.

It's a thin form factor, so I can't imagine it being very well ventilated. Still, is this going to be a user tweakable setting? I don't particular like the bottom of my notebook becoming a furnace.

Can't wait to see some bitcoin numbers, looks like both GPU makers are using lots of small compute units now.

 

[Meaker10]

^^ There are two variations that are shipped with the same moniker.

From AT:



555M 'B' spec is the one that Alienware uses in its M14x.

There is also:

C: 144 CUDA Cores, 16 TMUs, 16 ROPs, Core Clocks: 590MHz, Shader Clocks: 1180MHz, 128bit DDR3

(MSI only)
D: 144 CUDA Cores, 16 TMUs, 16 ROPs, Core Clocks: 753MHz, Shader Clocks: 1506MHz, 128bit GDDR5

So in summary a GT555M can be based on two entirely different chips, can be equiped with different widths, DDR3 of GDDR5 at completely different clocks covering different performance segments.


It's the biggest cluster **** in GPU naming history.

Oh and it's been renamed the GT635M now.

 

[Meaker10]

That is intentional of Nvidia, as they did that with their Geforce 540m line as well. Extrapolating that, that should give us a rough idea of what kind of models the 625m and 655m are as well:

625m - 550mhz core, 1100mhz shader, 384 CUDA cores, DDR3 @ 900mhz
640m - 625mhz core, 1250mhz shader, 384 CUDA cores, DDR3 @ 900mhz
655m - 700mhz core, 1400mhz shader, 384 CUDA cores, DDR5 @ ~800-900mhz

That should also put the 620m at at 700-750mhz core with 192 CUDA cores, albeit on a 64bit bus.

The Geforce 540m isn't the most potent of GPUs, but I will be the first to admit that it is surprisingly capable of delivering a very adequate and comfortable gaming experience for most games I've thrown at it (WoW, SC2, Deus Ex: Human Revolution, BF3 etc).

Looks like the 640m will perform a little bit better.

This is wrong because the lower chips are just renamed little and mid range fermis.

GT620->630 are using the same core as the 540m
GT635 is a rebranding of the 555m and ALL that entails.
GT640M -> GT660M are new little kepler chips.
GTX670M -> GTX675M are renamed GTX570M and GTX580M respectivly.
GTX680M is the new mid kepler.

Yeah, have fun with that.

 

[Lonbjerg]

This is wrong because the lower chips are just renamed little and mid range fermis.

GT620->630 are using the same core as the 540m
GT635 is a rebranding of the 555m and ALL that entails.
GT640M -> GT660M are new little kepler chips.
GTX670M -> GTX675M are renamed GTX570M and GTX580M respectivly.
GTX680M is the new mid kepler.

Yeah, have fun with that.

Then it would be a good time to look at the performance...and not "specs"...

 

[ElFenix]

i don't know how a notebook with 1/3 the resolution of a tablet can be called 'ultra'

 

[Meaker10]

Then it would be a good time to look at the performance...and not "specs"...

Except each has drastically different power consumption, a different feature set and very large gaps in performance.