GF110 is actually GF100B

[Janooo]

... as shown here: http://www.gpu-tech.org/content.php/144-…and-GF110s-real-name-is-GF100B?

 

[SlowSpyder]

It seems to be more or less a tweaked GF100 with the bugs worked out. It is what Nvidia announced back around when the 5870 was launched.

The GTX480 not being able to launch with all 512 cores and more than 700MHz clock speed actually worked out for Nvidia though. It was still fast enough to claim the single GPU crown, so they really didn't need to bring out a 512SP part until they ironed out the kinks. I think naming it a GTX485 or 490 would have made more sense, but whatever.

 

[tviceman]

I think I'd need Idontcare's expertise to back me up on this, but a B-layer spin implies only changes to the metal layer. GF110 has a different transistor count, and as Anand already showed in clock-for-clock comparisons, GF110 does better than GF100's 512 core theoretical performance.

 

[ChrisAttebery]

There isn't a standard in the industry, but usually letters are full mask sets, while numbers are metal revisions. A1 is a metal only fix to A0. B0 is a full mask set.

 

[Daedalus685]

I think we are over thinking the code name a touch.

The B is just part of the name, there is no reason they couldn't refer to it internally as GF100B A1 and so forth. I don't think the B necessarily needs to even relate to the mask/metal revisions. All of the review shots simply show the GPU as labeled "GTX580" so I'm not sure mind you.

Whether they call it the GFbibitybopityboo, GF100B, or GF110 we know it is the GTX580, and is a corrected/improved version of what we saw as the GTX480.

 

[edplayer]

I think we are over thinking the code name a touch.

This is nothing.

You should have seen how crazy the girls got over the new AMD 6 series.

 

[AdamK47]

I must have missed seeing that part when I used nvflash to update the BIOS on my three 580s.

 

[Keysplayr]

Does this have any significance to anyone except the OP? And whatever that is, could they share it with the rest of us?

 

[Axon]

Padma Lakshimi is hot.

...

I posted that because I'm not sure what this thread is about, so my post is ironic, like the Charlie Brown Christmas tree or something

 

[Keysplayr]

Padma Lakshimi is hot.

...

I posted that because I'm not sure what this thread is about, so my post is ironic, like the Charlie Brown Christmas tree or something

I sort of agree where you're coming from, unless the OP can shed a little light on what he believes is significant about this?

 

[Ben90]

This is the most disturbing news I have heard in years

 

[Teizo]

I sort of agree where you're coming from, unless the OP can shed a little light on what he believes is significant about this?

Maybe he was/is trying to make the case that the 580 is just a rebranded 480 or something.

 

[tannat]

Does this have any significance to anyone except the OP? And whatever that is, could they share it with the rest of us?

It's definitely interesting for us that follow graphic cards and technology. And it is at least not a repost of old news. Thread crapping is not very interesting though.

Back to topic:

G200 to G200B was a die shrink from 65 to 55 nm but mainly the same chip if i recall. (GTX280 to GTX285).

There is definitely no node jump involved from gf100 to gf100B/GF110

But the B is not necessarily related to a B spin as in A0, A1... B1 batches. It just indicates a new version of the same chip.

That GF110 is only an improved respin of GF100 is well known anyway. Well, it may only be binned GF100 chips for all that we know. But anyway no one has probably seen it as anything than a fix and what GTX480 may have been from the beginning.

Claimed are minimal changes in transistor count and 2% change in die size but nvidias batting average on truth has not peaked the last seasons so I guess noone really knows how big the changes are

 

[ginfest]

I sort of agree where you're coming from, unless the OP can shed a little light on what he believes is significant about this?

Look it has no relevance. This is another in a long line (see the post above) of anti-NV crap! I wish someone would put an end to the foolishness. Note to the "pushers"-not everyone who reads the video forums wants to see this stuff . This is not "news"

 

[Grooveriding]

NV fanboys unite! Form Voltron!

OP this is no surprise to anyone, it's already known GTX 580 is a GTX 480 with 512 shaders instead of 480 and a higher core clock.

 

[notty22]

A link to a random forum post with someone exclaiming "looks like Charlie was right" caused me to throw up in my mouth a little.
Remember the gtx 580 couldn't be made ,no one can buy one, paper launch etc.

 

[Keysplayr]

Look it has no relevance. This is another in a long line (see the post above) of anti-NV crap! I wish someone would put an end to the foolishness. Note to the "pushers"-not everyone who reads the video forums wants to see this stuff . This is not "news"

Oh. Ok.

 

[OCGuy]

Well in that case 6970 better beat the crap out of it if it is only a 480 re-brand!

That would be embarassing if a next-gen part couldnt beat a re-touched 480. That would be like 5870 not beating GT200b.

 

[tannat]

Look it has no relevance. This is another in a long line (see the post above) of anti-NV crap! I wish someone would put an end to the foolishness. Note to the "pushers"-not everyone who reads the video forums wants to see this stuff . This is not "news"

If you refer to my post, I only call it as I see it. I have no horse in the race for any team.

Is there anything in my post that is not correct?

I (and I'm sure other people) are usually interested in what the modifications and changes between new and old GPUs are. GTX580 gives an improvement of about 17% over GTX480. That's true disregarding what real changes have been done, and if there are no major changes so be it.

Rest is only technological curiosity, whether Barts or cayman versus cypress or GF100B/GF110 versus GF100.

These questions are what keeps such forums as this alive, if my joke about nvidias truthfulness offended you, i'm sincerely sorry but it wouldn't been fun if there was not any truth in it.

 

[SlowSpyder]

A link to a random forum post with someone exclaiming "looks like Charlie was right" caused me to throw up in my mouth a little.
Remember the gtx 580 couldn't be made ,no one can buy one, paper launch etc.

I think it is fair to say Charlie was dead wrong on the number of Fermi chips Nvidia would make (didn't he say they would only make 10k or one run of wafers or something like that?).

But, he did say that the 512SP part was not able to be built, I read that in his forum post somewhere. So depending on how you want to inerpret his statement, you could certainly say he was correct about the GF100 not being manufacturable if we are talking about a fully functional chip. It's like AMD claiming the RV770 was going to be an 850MHz part that can hit 1GHz for overclockers. They had to make the 4890 for that, which has much of the same DNA as the 4870, but it is a different chip with a few tweaks. The GF110 is a pretty similar scenario. So if Fud reported that an 850MHz RV770 is not manufacturable, we would have to agree with him I guess.

As far as this thread is concerned, I think we all know that the GTX580 is more or less like the 4890. A tweaked and improved part built very closely from it's predecessor.

*edit - I do have to wonder, if a 512SP part is what AMD expected for the original Fermi launch what will the 6970 do performance-wise? Or will the 32nm cancellation hold it back? I am looking forward to the next few weeks.

 

[Skurge]

You guys always have to go down this path? There is nothing anti-nV about this thread. You all are too paranoid. Its nice to discus something other than framerates. I learn quite a bit about how GPUs are designed when things like this are discussed, but when some members come here and start shouting "anti-nV", "pointless thread", etc. It just degenerates into more fanboy nonsense.

 

[tannat]

Well in that case 6970 better beat the crap out of it if it is only a 480 re-brand!

That would be embarassing if a next-gen part couldnt beat a re-touched 480. That would be like 5870 not beating GT200b.

This is ignoring technology and just going by names.

If HD6970 is less than 400mm2 and beats the 520mm2 GTX580 at the same node it's quite astonishing actually and cred should be given where cred is due.

If HD6970 is larger than GTX580 and has been aimed to beat it but fail. That would be a loss and actually embarrassing as that would mean that AMD would loose their advantage in technology efficiency in one step.

However, even if GTX480 was comparably inefficient in many aspects they seemed to be mainly due to leakage and unability to reach aimed frequencies at decent powerdraw. The architecture may have quite much more room for performance with minor adjustments than GTX480 (Which anyway is a very fast card) indicated at first try.

A GTX480 done right can be and is a really good card. Noone expects magic at the same node and HD4870 is only one example of a huge success that never beat or aimed to beat the competitors best chip.

 

[OCGuy]

This is ignoring technology and just going by names.

If HD6970 is less than 400mm2 and beats the 520mm2 GTX580 at the same node it's quite astonishing actually and cred should be given where cred is due.

If HD6970 is larger than GTX580 and has been aimed to beat it but fail. That would be a loss and actually embarrassing as that would mean that AMD would loose their advantage in technology efficiency in one step.

However, even if GTX480 was comparably inefficient in many aspects they seemed to be mainly due to leakage and unability to reach aimed frequencies at decent powerdraw. The architecture may have quite much more room for performance with minor adjustments than GTX480 (Which anyway is a very fast card) indicated at first try.

A GTX480 done right can be and is a really good card. Noone expects magic at the same node and HD4870 is only one example of a huge success that never beat or aimed to beat the competitors best chip.

Oh no..the die-size argument.....sorry, not going there. Although you know it is bad when that is what you grasp on to.

Then again, we have no idea what the 6970 is going to bench at, so this is all just speculation anyway.

 

[Mr. Pedantic]

*edit - I do have to wonder, if a 512SP part is what AMD expected for the original Fermi launch what will the 6970 do performance-wise? Or will the 32nm cancellation hold it back? I am looking forward to the next few weeks.

By the same token, if a 512SP Fermi is what AMD expected for the original Fermi release, what the hell were they thinking with the 5870?

 

[3DVagabond]

By the same token, if a 512SP Fermi is what AMD expected for the original Fermi release, what the hell were they thinking with the 5870?

That's what the 5970 was for