AMD to Give RV770 a Refresh, G200b Counterattack Planned

[LordEC911]

I'm not claiming the article is stating that the G200 scales better than the RV770.

All I was saying, is that it contradicts a widely held view by ATI fans that the RV770 scales better than the G200 in terms of performance gleaned from clockspeed increases, which simply isn't true when you look at the data.

Both the G200 and the RV770 scale similarly in performance when you increase the clockspeed, but because the G200 has more headroom, you can get more out of it.

Ummm and where are these facts?
And G200 has more overclocking headroom? Please elaborate...

You say your not claiming that G200 scales better than RV770, you are claiming that it contradicts what ATi fans say about RV770 scaling better than G200? That is exactly what I said...

What evidence is there, besides Charlie's article (not exactly prime evidence) that the Gt200b is facing bad yields?

No release date yet. Unless they sneak it in, which is unlikely, they only have a few weeks left to get it in before the end of the year.

[waver_01]

wallets am cry

very good news, I was going to wait for GTX 270...but who knows now

It better have 1GB vram btw

[munim]

Sounds like a bunch of armchair speculation to me.

[AMDDeathstar]

No, nvidia was rename and resell this sounds like they'd actually do some tweaks and adjustments to it and not sell it as a "new" generation but a revision within the current generation. Maybe like a 4890 or something.

*cough* Wrong *cough*

9800GTX -> 9800GTX+

My EVGA 9800GT Superclocked Edtion is just a rebrand 8800
Anywhere there was HDMI on the box was covered with a sticker

[Warboy]

My EVGA 9800GT Superclocked Edtion is just a rebrand 8800
Anywhere there was HDMI on the box was covered with a sticker

That wasn't the discussion........

[Carfax]

It's called common sense, even I can see it. The GT200 faced yield problems in the beginning. So add a Die shrink. The Chance of a yield problem will increase. Of Course Nvidia is starting to do good with the GT200 as of lately. But it's just common sense case. So there is some facts hidden (and I mean hidden since I don't like charlie) in Charlie's article. ATi seems to have a easier time doing a die shrink then Nvidia. That's simple to example too. The GT200b vs the R870 was a joke if you couldn't tell tho.

Common sense while useful, isn't exactly evidence. Also, I would imagine that the die shrink would alleviate yield problems, and not exacerbate them.

The GT200 was a very large chip in terms of transistor density, too large for the 65nm process in truth. Cramming that many transistors in such a small space is no easy thing. The die shrink to 55nm should help though, and reduce thermal and power constraints which will improve yields and ramp up the clockspeeds.

Anyway, I'm hoping that Nvidia releases the die shrink before Christmas, so I can use the step up program to upgrade to the GTX 290.

[Warboy]

Common sense while useful, isn't exactly evidence. Also, I would imagine that the die shrink would alleviate yield problems, and not exacerbate them.

The GT200 was a very large chip in terms of transistor density, too large for the 65nm process in truth. Cramming that many transistors in such a small space is no easy thing. The die shrink to 55nm should help though, and reduce thermal and power constraints which will improve yields and ramp up the clockspeeds.

Anyway, I'm hoping that Nvidia releases the die shrink before Christmas, so I can use the step up program to upgrade to the GTX 290.

Die Shrinks can do alot of things, Some postivies are Reduce Thermal output, and Improve clock speeds. Yes sometimes they can improve yields. But in Nvidia's case. No It won't. Just because of a die shrink doesn't make it better. The Typical thought when the term Die Shrink comes to mind. Most people align it with "Better Clocks, Cooler, and more successful" and that's a very stereotypical idea. Sometimes it's true, Sometimes it's not.

[AliG]

Like nVidia? Monkey see monkey do?

this would be different though, as it would have a new chip, while what nvidia did, the 9800gtx and 8800gts 512mb used the same processor, one just had better VRMs (so essentially the 9800gtx was like a non-reference 8800gts 512mb)

[AMDDeathstar]

That wasn't the discussion........

So it's not about doing minor tweaks and releasing as new product
My bad, sorry

[YukonTrooper]

[SNiiPE_DoGG]

*snip*

hahaha nice one

[Carfax]

Ummm and where are these facts?

I didn't say anything about "facts." I just read the reviews and it didn't correlate with what ATI fans were saying about the RV770 scaling, at least from where I was standing.

And G200 has more overclocking headroom? Please elaborate...

From what I've seen, you can usually get a 20% overclock (give or take a few) from the GT200 (in particular the 260) fairly easily using stock cooling.

Try that on the RV770, and you'll probably need a voltage mod and/or improved cooling.

You say your not claiming that G200 scales better than RV770, you are claiming that it contradicts what ATi fans say about RV770 scaling better than G200? That is exactly what I said...

Dude, I think you need to re-read my statements.....or I need a drink

[villa1n]

I didn't say anything about "facts." I just read the reviews and it didn't correlate with what ATI fans were saying about the RV770 scaling, at least from where I was standing.


rom what I've seen, you can usually get a 20% overclock (give or take a few) from the GT200 (in particular the 260) fairly easily using stock cooling.

Try that on the RV770, and you'll probably need a voltage mod and/or improved cooling.


Dude, I think you need to re-read my statements.....or I need a drink

I ve been following this thread, and I think i need to point out something that seems to be missing in this discussion.
You allude to the fact you can get a 20% overclock on the gt200, which is true, and on r700, lets say you can only get on average, a 10-15%. None of that really matters in the slightest, unless you have some obsession with a particular architecture. The only thing that matters is how much performance the card puts out, given the overclock, and since they are priced equally, scaling doesnt really matter that much either. What matters is the final numbers these cards drop, and how reliable the cards remain while oc'd. So if a normal 260 oc'd cant match a 4870 oc'd in performance, the scaling and percentage of overclock are all just nice talking points, but pragmatically are meaningless. Now if you want to post up numbers showing a normal gt260 oc'd 20% beating a 4870 clocked at 830mhz... by all means set the record straight, but if your going to talk about relative percentages that don't equate to performance, please save us the posts.

[jaredpace]

Common sense while useful, isn't exactly evidence.

GT200b has had 3 respins because of poor yields

[Macadamia]

This is probably RV770, A14.

[Eastcoasthandle]

I ve been following this thread, and I think i need to point out something that seems to be missing in this discussion.
You allude to the fact you can get a 20% overclock on the gt200, which is true, and on r700, lets say you can only get on average, a 10-15%. None of that really matters in the slightest, unless you have some obsession with a particular architecture. The only thing that matters is how much performance the card puts out, given the overclock, and since they are priced equally, scaling doesnt really matter that much either. What matters is the final numbers these cards drop, and how reliable the cards remain while oc'd. So if a normal 260 oc'd cant match a 4870 oc'd in performance, the scaling and percentage of overclock are all just nice talking points, but pragmatically are meaningless. Now if you want to post up numbers showing a normal gt260 oc'd 20% beating a 4870 clocked at 830mhz... by all means set the record straight, but if your going to talk about relative percentages that don't equate to performance, please save us the posts.

Finally,
He has alluded to this very thing before in another news thread. Yet he still keeps pointing out that OC potential is more important then results.

[Carfax]

the scaling and percentage of overclock are all just nice talking points, but pragmatically are meaningless.

Meaningless to you perhaps, but many other people use them as a reference.

Now if you want to post up numbers showing a normal gt260 oc'd 20% beating a 4870 clocked at 830mhz... by all means set the record straight, but if your going to talk about relative percentages that don't equate to performance, please save us the posts.

The percentages are for reference only, to extrapolate the gain one would get from overclocking. We're not dealing with rocket science here

There aren't any reviews that I know of that directly compare an overclocked 4870 and an overclocked GTX 260, which is why people resort to these measures.

Most people simply look at separate reviews of both cards where they were overclocked (most review sites do this nowadays) and then at the scaling to see how they perform with higher clockspeeds.

From what I've seen, both the GTX 260 and the 4870 scale similarly when you increase the clockspeed. In some games, this results in a near linear increase in performance while in others, there is barely an increase.

Regardless, because you can get a higher percentage overclock out of the GTX 260 than you can on the 4870, the former ends up outperforming the latter; except in certain circumstances where the HD 4870 is very strong.

[003]

Hmm... well it's always good to stay competitive, and ATI has been on the offensive lately. But I sure will be pissed if my 2x 4870X2s are outdated so quickly. I was hoping there would be nothing between the 4870X2 and RV870...

[Carfax]

Finally,
He has alluded to this very thing before in another news thread. Yet he still keeps pointing out that OC potential is more important then results.

You are vastly over simplifying what I said.

I never stated that OC potential is more important than results. Thats ridiculous!

The OC potential does have an impact on the results however, and only a fool would say otherwise.

[Omastar]

this is very interesting, I am in the market for a beast right now.....

Might as well pick up a 4870X2. Didn't you have 4870 Crossfire? How is that not beast-like?

[Eastcoasthandle]

You are vastly over simplifying what I said.

I never stated that OC potential is more important than results. Thats ridiculous!

The OC potential does have an impact on the results however, and only a fool would say otherwise.

It's simple, here let me quote you:

Regardless, because you can get a higher percentage overclock out of the GTX 260 than you can on the 4870, the former ends up outperforming the latter; except in certain circumstances where the HD 4870 is very strong.

Because this is the foundation of your argument it simply falls flat on it's face. This portion of your post alone alludes to your claims that a higher OC potential is more important. Because you believe it will perform better. But you also say that in certain circumstances the 4870 is still stronger. That alone contradicts your very own claims.

[Sly Fox]

Let's all just agree that we need some faster cards already!!!

The 4870X2 is pretty beastly, but not yet enough

[jaredpace]

All i gotta say is give me gddr5 or nothing. And that a 5870X4 sounds sweet.

[SNiiPE_DoGG]

Might as well pick up a 4870X2. Didn't you have 4870 Crossfire? How is that not beast-like?

ha yeah well the current plan is to pick up a 4870x2 but if this is available before christmas then I may jump for it instead

[Klarko]

These will come before 09?