nVidia wins this round - Charlie D.

[Troll Trolling]

http://semiaccurate.com/2012/01/19/nvidia-kepler-vs-amd-gcn-has-a-clear-winner/

Charlie hates nVidias. Charlie says nVidia wins.

Three possibilities: Charlie is wrong, Charlie is right or Charlie is trolling. EDIT: or Charlie was hacked.

Charlie.

 

[Arkadrel]

For the doubters, both of the current cards we saw have two DVI plugs on the bottom, one HDMI, and one DP with a half slot cooling grate on one of the two slot end plates. The chip is quite small, and has 8 GDDR5 chips meaning a 256-bit bus/2GB standard, and several other features we can’t talk about yet due to differences between the cards seen. These items won’t change the outcome though, Nvidia wins, handily.S|A

cant tell if charlie is serious or sarcastic.

How does a 256bit bus chip get more memory bandwidth than a 580?
Nvidia's 580 already has 384bit bus.

lmao look at the forum inputs:

MrGiggleNutz Jan 19, 2012 at 1:34 pm #
After I read this I had to check who the author is….. I can’t believe my eyes!

JPB Jan 19, 2012 at 1:39 pm #
Wow, someone hacked Charlies account?

Lapdog Jan 19, 2012 at 1:40 pm #
Charlie got DDoS’d.

Robert Varga Jan 19, 2012 at 1:42 pm #
LOL

flyck Jan 19, 2012 at 1:45 pm #
That would need a huge bandtwith efficiency increase?
given they would need 1600MHz GDDR5 ram to best the current 580 in that area.

Skiddywinks Jan 19, 2012 at 1:51 pm #
Wait, seriously? Or is this a joke?
Did hell just freeze over?


Sky Jan 19, 2012 at 1:57 pm #
Seriously? Even nVidia fanboys won’t believe this let alone normal people….

DS Jan 19, 2012 at 1:59 pm #
Is Charlie trying to out-OBR OBR?


Hardware guy Jan 19, 2012 at 2:03 pm #
My browser doesn’t handle the sarcasm tag.

turtlefly Jan 19, 2012 at 2:05 pm #
The only way that would be possible is if they use XDR2.


Stig Jan 19, 2012 at 2:06 pm #
this is definite confirmation that CD has lost his moles in AMD during the last swept-out with in the Sunnyvale company, and has buried axes with green goblin! well show must go on ;-)

^ had me laughing

 

[Troll Trolling]

Yeah, that is what I'm wondering right now.
The article's objectiveness is confusing.

Also, they would have to clock the memory it beyond 6GHz to get to the GTX's 580 bandwidth.
Unless Charlie is referring only to perf/mm² and perf/W, or nVidia made some crazy magic, I find it hard to believe that GK104 is faster than GNC.

 

[badb0y]

It's OK I can always sell the 7970 and buy Kepler.

 

[Olikan]

Yeah, that is what I'm wondering right now.
The article's objectiveness is confusing.

troll trolling may have been trolled XD

well, kepler seems be very bandwidth eficient to beat a 7970, using 256 bit rate

 

[mamisano]

He could be setting a trap to see if someone copies his information without giving credit. Especially since it is so contrary to what he has been saying about Kepler of late.

 

[TakeNoPrisoners]

A little early to say who wins at this point. No real benchmarks are even out.

 

[OILFIELDTRASH]

It's OK I can always sell the 7970 and buy Kepler.

That gets to be a costly cycle. Believe me.

When I can get a significant upgrade from my gtx 480 for $350 I'm in. Probably next gen though.

 

[Skurge]

lol, I haven't read any of his articles in ages cause they mostly nonsense, but I just had to click on the link and see for myself.

Ofcourse he just says nvidia wins, he doesn't even say why.

 

[notty22]

The GK 104 has the 256mb bus, those they saw..

The short story is that Nvidia (NASDAQ:NVDA) will win this round on just about every metric, some more than others. Look for late March or early April availability, with volumes being the major concern at first. GK104 cards seen by SemiAccurate all look very polished and complete, far from rough prototypes or “Puppies“. A2 silicon is now back from TSMC, and that will likely be the production stepping barring any last second hitches. Currently, there aren’t any.

 

[Troll Trolling]

Some numbers:

Let's shoot really high and say nVidia somehow manages to clock the memory to 7GHz (highest, according to Wikipedia), that puts them with about 85% of the bandwidth of the HD 7970, or HD 7970 got 17% more datarate.

Seems odd to me. Manly when considering that:

- At stock, GTX 580 got 4GHz GDDR5
- Overclocked, they can't reach 5Ghz: 1, 2, 3 , 4

 

[Crap Daddy]

Because something is happening here
But you don’t know what it is
Do you, Mister Jones?

 

[Troll Trolling]

Because something is happening here
But you don’t know what it is
Do you, Mister Jones?

Maybe this is what is going to happen:

Come gather 'round people
Wherever you roam
And admit that the waters
Around you have grown
[...]
For the times they are a-changin'

 

[RavenSEAL]

 

[Arkadrel]

Nvidia must have had a breakthrough in memory controller design, because that was one of their issues right? haveing trouble matching AMDs design, and thats why AMD went with smaller bus size + faster ram speeds, and nvidia did bigger bus size + slower ram speeds.

That and maybe even changed to XDR2? ... otherwise... 256bit bus beating a 384bit bus card?
without memory bandwidth bottleneck? how?



Also if this GK104, is a ~560-570 legue card, but is rated at 225watts....
how much will the 780 one use?

 

[RavenSEAL]

Nvidia must have had a breakthrough in memory controller design, because that was one of their issues right? haveing trouble matching AMDs design, and thats why AMD went with smaller bus size + faster ram speeds, and nvidia did bigger bus size + slower ram speeds.

That and maybe even changed to XDR2? ... otherwise... 256bit bus beating a 384bit bus card?
without memory bandwidth bottleneck? how?



Also if this GK104, is a ~560-570 legue card, but is rated at 225watts....
how much will the 780 one use?

On a serious note, it just sounds to me like another BS blog. This guy seems worse than Fudzilla.

 

[blastingcap]

I had to triple check that today was not April 1, 2012.

I wonder if those who continually trash Charlie are now going to take this as, ahem, an article of faith.

Edit: I wonder if Charlie got bored and had fun with this one, or just wanted a spike in site traffic or something. Or if it IS true, maybe he wanted to have a placeholder now, rather than later, so that he can say "told ya so" despite not having given much firm data.

 

[Lonyo]

Nvidia must have had a breakthrough in memory controller design, because that was one of their issues right? haveing trouble matching AMDs design, and thats why AMD went with smaller bus size + faster ram speeds, and nvidia did bigger bus size + slower ram speeds.

That and maybe even changed to XDR2? ... otherwise... 256bit bus beating a 384bit bus card?
without memory bandwidth bottleneck? how?



Also if this GK104, is a ~560-570 legue card, but is rated at 225watts....
how much will the 780 one use?

It wasn't one of their issues, it's basically (according to Anand) just AN issue which is common to all and basically impossible to overcome.
Anand indicated that even 6.5Gbps would be impossible, and AMD are already running 5.5 with a 384-bit bus.

Yes, NV have struggled more than AMD to get their RAM speeds higher, but AMD and NV both almost certainly won't be able to get that high using GDDR5, assuming NV didn't make some actually magical breakthrough which was probably seen as impossible from an engineering POV.

 

[RavenSEAL]

I find it hard to believe that nVidia would (if this "Charlie" person is indeed right) use XDR2 on their cards when the market doesn't demand it.

If so, we are looking at slightly higher performance with a price difference of maybe up to $100+ between the GTX 680 and the 7970.

 

[blackened23]

Is someone at NV waging an internet war or something?

Wait, so Kepler is coming in Feb according to rumors.

But SA says April or May.

Another source says october.

2 days ago rumors suggested that GK104 would equal 7970.

Now charlie says otherwise.

Screw it, i'm not believing anything until real hardware is given to reviewers. If nvidia is leaking "rumors" to websites, it wouldn't be very surprising seeing as they did this a year before Fermi, proclaiming it was ready when it wasn't.

 

[bryanW1995]

It wasn't one of their issues, it's basically (according to Anand) just AN issue which is common to all and basically impossible to overcome.
Anand indicated that even 6.5Gbps would be impossible, and AMD are already running 5.5 with a 384-bit bus.

Yes, NV have struggled more than AMD to get their RAM speeds higher, but AMD and NV both almost certainly won't be able to get that high using GDDR5, assuming NV didn't make some actually magical breakthrough which was probably seen as impossible from an engineering POV.

Charlie is so anti-nvidia, I see only 3 possibilities:
1. He is off the persona-non-grata nvidia list finally
2. He was writing tongue-in-cheek
3. NV is using xdr for gk104

 

[Arkadrel]

Charlie:

.....The chip is quite small, and has 8 GDDR5 chips meaning a 256-bit bus/2GB standard,.......

Hes saying its still GDDR5....

so is techpowerup:

http://www.techpowerup.com/159009/GeForce-Kepler-104-%28GK104%29-Packs-256-bit-GDDR5-Memory-Bus-225W-TDP.html

GeForce Kepler 104 (GK104) Packs 256-bit GDDR5 Memory Bus, 225W TDP

I cant see how a 256bit bus, wont be memory bandwidth bottlenecked, at those speeds though..

 

[blastingcap]

Charlie is so anti-nvidia, I see only 3 possibilities:
1. He is off the persona-non-grata nvidia list finally
2. He was writing tongue-in-cheek
3. NV is using xdr for gk104

4. Site traffic needed a boost today
5. Site got hacked
6. He was fed misinformation

Btw people are saying it is GDDR5 and realistically that is all it can possibly be for a gaming card. Note that NVDA had major memory controller problems for Fermi. If they cleared that problem up, and then some, that may account for the discrepancy.

 

[Skurge]

What about the rest of the card? Memory bandwidth alone doesn't make a card.

 

[Troll Trolling]

4. Site traffic needed a boost today
5. Site got hacked
6. He was fed misinformation

Btw people are saying it is GDDR5 and realistically that is all it can possibly be for a gaming card. Note that NVDA had major memory controller problems for Fermi. If they cleared that problem up, and then some, that may account for the discrepancy.

Although Charlie has been spot on on nVidia stuff lately.