Incredible: 4770 in XFire beats 4890 AND it's cheaper !!!

[ShadowFold]

Indeed,

No rebadge screams when its ATI.

But rebadge screams when its Nvidia (GTS 250 *cough cough*)

Because ATi doesn't rebadge? The HD 4770 is a new 40nm core. The 4830 isn't either. Neither is any of the HD 4000 series. So why would anyone "scream" rebadge?

 

[DrPepper]

Because ATi doesn't rebadge? The HD 4770 is a new 40nm core. The 4830 isn't either. Neither is any of the HD 4000 series. So why would anyone "scream" rebadge?

The GTS250 is a 55nm 9800GTX so its not a rebadge either ?

 

[silkstone]

Because ATi doesn't rebadge? The HD 4770 is a new 40nm core. The 4830 isn't either. Neither is any of the HD 4000 series. So why would anyone "scream" rebadge?

Because they're saying it's essentially a revamped 4830 core. and if it were to have 256-but it would be identical to the 4830 apart from having a smaller core die size. Hence rebadge, like the 250

 

[Darren]

Because ATi doesn't rebadge? The HD 4770 is a new 40nm core. The 4830 isn't either. Neither is any of the HD 4000 series. So why would anyone "scream" rebadge?

You've missed the point.

People are moaning that they want a 256-bit interface with DDR3 on the 4770 with a 40nm, that would almost be a rebadged 4830.

They want this

However,

They complain when Nvidia reduce the core to 55mn on the GTS 250 and kept its memory interface the same, and scream rebadge!


Edit:

They didn't reduce anything. The 9800GTX+ was 55nm long before the GTS 250 was. The GTS 250 is the same exact core as the 9800GTX+. I just ignore the people who say GDDR5 128bit is slow, they obviously don't know how this stuff works.

Some 9800 GTXs were 55nm!


But indeed, DDR5/128-bit seems perfect for me. I dont have money to blow on 4890s

 

[ShadowFold]

You've missed the point.

People are moaning that they want a 256-bit interface with DDR3 on the 4770 with a 40nm, that would almost be a rebadged 4830.

They want this

However,

They complain when Nvidia reduce the core to 55mn on the GTS 250 and kept its memory interface the same, and scream rebadge!

They didn't reduce anything. The 9800GTX+ was 55nm long before the GTS 250 was. The GTS 250 is the same exact core as the 9800GTX+. I just ignore the people who say GDDR5 128bit is slow, they obviously don't know how this stuff works.

 

[DrPepper]

They didn't reduce anything. The 9800GTX+ was 55nm long before the GTS 250 was. The GTS 250 is the same exact core as the 9800GTX+. I just ignore the people who say GDDR5 128bit is slow, they obviously don't know how this stuff works.

So if they gave the g92 a new name say gt201 it wouldn't be a rebadge.

 

[ShadowFold]

So if they gave the g92 a new name say gt201 it wouldn't be a rebadge.

That would be a core rebadge

 

[DrPepper]

That would be a core rebadge

But all they've done with the 4770 is renamed its core along with the die shrink. Nvidia kept it g92 even after it was shrunk. So technically it is a different core but same name. We can't tell what nvidia have changed on it apart from the node size.

 

[silkstone]

Both the 250 and 4770 seem like pointless cards. the 250 being the same performance as a 4830 for the same price and the 250 being identical to the 9800gtx+. But i guess that is beside the point. I'm sure the 4770 was only released to keep market share as was the re-badged 250.
Mind you the 4770 core might be very nice inside laptops.

 

[Darren]

Both the 250 and 4770 seem like pointless cards. the 250 being the same performance as a 4830 for the same price and the 250 being identical to the 9800gtx+. But i guess that is beside the point. I'm sure the 4770 was only released to keep market share as was the re-badged 250.
Mind you the 4770 core might be very nice inside laptops.

Obviously you haven't read any reviews lately. The GTS 250 much faster than the 4830. It is also a fair bit faster than the 4850!

Heck, even the 9800 GTX+ could own the 4850 half the time, the GTS 250 has no problems in that department.

Edit:

And I'm talking about overall performance from numerous of independent reviews. Not "hear say" from Chinese whispers.

 

[newtekie1]

Because ATi doesn't rebadge?

Sure they do, you just don't hear about it around here as much, and they haven't done it for a while, but they have done it.

For Example:
x300=x1050 Both shared the same RV370 core, both used DDR, both were available with a 128-bit memory bus and a 64-bit memory bus. The only difference was the x1050 had a core clock increase of between 25-75MHz depending on what model of x300 and x1050 you looked at.

They complain when Nvidia reduce the core to 55mn on the GTS 250 and kept its memory interface the same, and scream rebadge!

Some 9800 GTXs were 55nm!

They didn't reduce anything. The 9800GTX+ was 55nm long before the GTS 250 was. The GTS 250 is the same exact core as the 9800GTX+.

There were no 55nm 9800GTX cards, the 9800GTX+ was 55nm. And the GTX250 is not simply a rebadged 9800GTX+. The big difference between the 9800GTX+ and the GTX250 is in the PCB. It uses a PCB similar to the 8800GTS 512MB.

The process with the G92 went:

8800GTS
\/
Improved PCB
\/
9800GTX
\/
Die Shrink
\/
9800GTX+
\/
"Improve" PCB to reduce cost
\/
GTS250

 

[DrPepper]

Sure they do, you just don't hear about it around here as much, and they haven't done it for a while, but they have done it.

For Example:
x300=x1050 Both shared the same RV370 core, both used DDR, both were available with a 128-bit memory bus and a 64-bit memory bus. The only difference was the x1050 had a core clock increase of between 25-75MHz depending on what model of x300 and x1050 you looked at.

The 9550 I had was also renamed to x1050 which was quite confusing.

 

[silkstone]

Obviously you haven't read any reviews lately. The GTS 250 much faster than the 4830. It is also a fair bit faster than the 4850!

Heck, even the 9800 GTX+ could own the 4850 half the time, the GTS 250 has no problems in that department.

Edit:

And I'm talking about overall performance from numerous of independent reviews. Not "hear say" from Chinese whispers.

I don't like Nvidia because you have to pay extra to get SLI, as if buying a second graphics card wasn't putting enough money in their pockets.

 

[DrPepper]

I don't like Nvidia because you have to pay extra to get x-fire, as if buying a second graphics card wasn't putting enough money in their pockets.

How do you have to pay extra for SLI ?

 

[silkstone]

Extra cost for motherboard or being forced into using a Nvidia chipset. Blah

 

[silkstone]

silkstone,

So according to you, when you buy an ATI based graphics card e.g. 4850 ATI will give you a FREE motherboard with crossfire support so you can have two GPUs?

No, but the motherboard comes with Xfire support without having to use an extra chip or being forced to use a pure nVidia chipset as they aren't/weren't at the time i bought my mobo, very good for overclocking. I'm not getting into a nVidia vs. ATI debate, i just don't like them. I used nVidia all the time from my first voodoo 2 card (which could be SLI'd) This 4850 is the first ATI card i've had.

 

[DrPepper]

Extra cost for motherboard or being forced into using a Nvidia chipset. Blah

You could choose from nvidia chipsets or x58. I don't see a problem because if you want to use xfire you need to get an amd chipset or intel.

 

[silkstone]

You could choose from nvidia chipsets or x58. I don't see a problem because if you want to use xfire you need to get an amd chipset or intel.

Doesn't give you much of a choice tho does it? go high end Intel which costs $$$ or go nvidia, of which boards generally pale in comparison to similar priced boards with intel chipsets.
I don't know about in the states, but where i live there ain't many boards with nVidia chipsets and those there are aren't very good for the price.

Silk,

I'm only kidding

No worries, i'm just not wanting to attract the fanboys

 

[DrPepper]

Doesn't give you much of a choice tho does it? go high end Intel which costs $$$ or go nvidia, of which boards generally pale in comparison to similar priced boards with intel chipsets.
I don't know about in the states, but where i live there ain't many boards with nVidia chipsets and those there are aren't very good for the price.

I don't know about the states either but nvidia boards aren't that bad. I had a tri-sli one which sadly died but it was a great board and wasn't too expensive. I do wish nvidia come out with new chipsets though.

 

[KainXS]

Sure they do, you just don't hear about it around here as much, and they haven't done it for a while, but they have done it.

For Example:
x300=x1050 Both shared the same RV370 core, both used DDR, both were available with a 128-bit memory bus and a 64-bit memory bus. The only difference was the x1050 had a core clock increase of between 25-75MHz depending on what model of x300 and x1050 you looked at.







There were no 55nm 9800GTX cards, the 9800GTX+ was 55nm. And the GTX250 is not simply a rebadged 9800GTX+. The big difference between the 9800GTX+ and the GTX250 is in the PCB. It uses a PCB similar to the 8800GTS 512MB.

The process with the G92 went:

8800GTS
\/
Improved PCB
\/
9800GTX
\/
Die Shrink
\/
9800GTX+
\/
"Improve" PCB to reduce cost
\/
GTS250

The later version of the 9800GTX+ is an EXACT rebrand of the GTS250, same gpu, same PCB, same everything, the GTS250 was announced shortly after the newer PCB 9800GTX+ was released,

IT IS A REBRAND

ATI does do this but almost never use the EXACT same components like nvidia does.

 

[ShadowFold]

Ati did it in the X era, but I haven't seen any HD rebadges, except for the 740g board is a X1250.

 

[newtekie1]

The later version of the 9800GTX+ is an EXACT rebrand of the GTS250, same gpu, same PCB, same everything, the GTS250 was announced shortly after the newer PCB 9800GTX+ was released,

IT IS A REBRAND

ATI does do this but almost never use the EXACT same components like nvidia does.

You are correct, once the new cheaper PCB was put into production, most manufacturers started using it for the 9800GTX+ also. It is actually pretty stupid to use two PCBs for essentially the same card, especially when one is more expensive.

However, the new PCB was developed for the GTS250, in nVidia's eyes it should have been used for the GTS250 only, of course in their eyes the 9800GTX+ should have stopped production with the introduction of the GTS250 also...:shadedshu

Ati did it in the X era, but I haven't seen any HD rebadges, except for the 740g board is a X1250.

ATi might not have done any re-badges recently, but they have done some pretty bad stuff, stuff nVidia is just as quilty of also. Namely releasing two cards under the same name, with different specs. Making one version weaker than the other. A rebradge I can handle, but having two different cards under the same name really pisses me off.

 

[Binge]

Prove it and I'll happily hold my hands up and say I'm wrong.

However, looking it at from both the reviews currently out there, and from the raw specs. The 4770 is the better card.

I see nothing that points to the 4770 being a better card than the 4830... it sure as hell doesn't justify making a whole new card except to make early use the 40nm fab.

 

[a_ump]

well yea, that's always a smart move, to use a new fab on a cheap card to find and work out any kinks or bugs with the node. Anyone that says the HD 4830 is better than the HD 4770 or words it HD 4770 isn't better than HD 4830 is plain stupid. sorry to be blunt but there are numerous reviews out there that prove this in every single game(i thk ). as for the overclocking theory, yes you may find a few HD 4830's that can oc very nicely and match or outdo an oc'd HD 4770, however the majority of HD 4830's even when oc'ed will not match or beat the HD 4770 oc'd. Just wait for more users to purchase HD 4770's and see how far they can push the GDDR5 memory chips, as that's when it seems to get the most boost in performance. but the facts more most i believe would be, when your shopping for a card, do you look at how it compares to another card overclocked or at stock? and most would say performance at stock and just assume the same result only greater when both cards would be overclocked. at stock the HD 4770 wins no arguments to made against that statement, when overclocked vs an oc'd HD 4830 i don't know as there are no reviews, we can only speculate.

 

[newtekie1]

I see nothing that points to the 4770 being a better card than the 4830... it sure as hell doesn't justify making a whole new card except to make early use the 40nm fab.

If you have seen nothing to point you to believe that the HD4770 is better than the HD4830, then I suggest you go read some of the reviews again.

Even at stock speeds, with the rather poor memory bandwidth, it still manages to outperform the HD4830, even in resolutions all the way up to 1900x1200, and possible beyond(I haven't really paid attention beyond 1900x1200, but I would hope the people that can afford monitors with resolutions higher than that can afford something more than a $100 graphics card). When both are overclocked, the memory bandwidth of the HD4770 can easily matches the HD4830 and the core tends to at least match it, if not surpass it.

Add to that the fact that the HD4770 should be cheaper to manufacture, which should translate to cheaper prices to the consumer, and I don't see how you can say the HD4770 isn't better than the HD4830.

Right now the HD4830 only has two things going for it, better power consumption and it is cheaper. However, once the hype for the HD4770 dies down, the HD4770 should retail for less than the HD4830, leaving only power consumption to hold up the HD4830.