• Ever wanted an RSS feed of all your favorite gaming news sites? Go check out our new Gaming Headlines feed! Read more about it here.
  • We have made minor adjustments to how the search bar works on ResetEra. You can read about the changes here.

Overall maximum teraflops for next-gen launch consoles?

  • 8 teraflops

    Votes: 43 1.9%
  • 9 teraflops

    Votes: 56 2.4%
  • 12 teraflops

    Votes: 978 42.5%
  • 14 teraflops

    Votes: 525 22.8%
  • Team ALL THE WAY UP +14 teraflops

    Votes: 491 21.3%
  • 10 teraflops (because for some reason I put 9 instead of 10)

    Votes: 208 9.0%

  • Total voters
    2,301
Status
Not open for further replies.

th1nk

Member
Nov 6, 2017
6,266
I was actually slightly disappointed that the zoom through Spider-Man city on PS5 was not completely smooth but had hiccups. I saw it in two different perspectives, so it is not the codec. I thought PS5 would get rid of the stuttering. But i guess it was a stress test and games can be optimized better.
 

Gohlad

Avenger
Oct 28, 2017
1,072
New leak

PS5 is made with infinity ram and cerny is actually Dr strange which helps.

Xbox next is made from the soul stone Phil had to sacrifice a lot of ms staff to get the soul stone

At this point, I only want the console that either has the power stone, the reality stone, or both!

I was actually slightly disappointed that the zoom through Spider-Man city on PS5 was not completely smooth but had hiccups. I saw it in two different perspectives, so it is not the codec. I thought PS5 would get rid of the stuttering. But i guess it was a stress test and games can be optimized better.

No, actually those where intentional stops, so that the audience could see that there is no pop-in happening. It was also mentioned in the Wired article that Sony did a few intentional stops during the PS5 demo, to show that the textures and clutter etc. were already loaded and would not pop-in.
 
Oct 27, 2017
699
I love this thread so much.

Leak that Navi is awesome:

- Sony fanboys: Navi was developed by Sony. MS won't get it! Moneyhatted architecture!
- XBOX fanboys: Bullshit - XBOX is Navi too plus some next gen shit. 3D chess!

Leak that Navi is shit:

- Sony fanboys: Everyone is getting Navi!
- XBOX fanboys: Navi sucks, we're on the mega custom Vega next level shit!

Both consoles are the same spec:

- Sony fanboys: All along I've said it's the games that matter. Who gives a shit about flops. 20% is nothing.
- XBOX fanboys: We're buying all the studios. Game pass is life.

You forgot the astroturfers or don't we talk about those?
 

big_z

Member
Nov 2, 2017
7,797
I was actually slightly disappointed that the zoom through Spider-Man city on PS5 was not completely smooth but had hiccups. I saw it in two different perspectives, so it is not the codec. I thought PS5 would get rid of the stuttering. But i guess it was a stress test and games can be optimized better.

It was kind of clear they were pushing it to the max and while SSD is fast it still has limitations. It'll be interesting to see how much Sony markets the improved speed as it's not really anything new outside of the console space and 99% of games will never need data so fast it's constantly pulling data like in that demo. Still it's going to be a huge improvement over this gen.
 

III-V

Member
Oct 25, 2017
18,827
I love this thread so much.

Leak that Navi is awesome:

- Sony fanboys: Navi was developed by Sony. MS won't get it! Moneyhatted architecture!
- XBOX fanboys: Bullshit - XBOX is Navi too plus some next gen shit. 3D chess!

Leak that Navi is shit:

- Sony fanboys: Everyone is getting Navi!
- XBOX fanboys: Navi sucks, we're on the mega custom Vega next level shit!

Both consoles are the same spec:

- Sony fanboys: All along I've said it's the games that matter. Who gives a shit about flops. 20% is nothing.
- XBOX fanboys: We're buying all the studios. Game pass is life.
Almost as if there is a set of predetermined responses
 

amstradcpc

Member
Oct 27, 2017
1,768
Size is important, if the chip is 312mm2 that means they probably don't target highest possible console apu on 7nm but rather go path similar to ps4
Aiming son


And translation :

I don't know more about the specific details of AMD fans' Chengdu station. Because Chengdu didn't bring a camera when it rained, the photos used in this article are all from Chengdu Station activities AMD China and Sapphire Technology Officials.

The point is that I personally can make a small interview with Leo as an AMD fan. It is also the biggest achievement of this event.

I always like sapphire.jpg

Q: I am the author

L: Sapphire Product Manager leo

J: Jenny Public Relations Director Jenny

Q: Hello Leo, sapphire is a well-received graphics card manufacturer in the mainland, but the sapphire warranty and after-sales system has become a short board on the barrel. Is there any solution or plan for this sapphire?

L: Thank you, Sapphire does have a plan to extend the warranty time in the future, but because you know the reason (laugh), it is temporarily impossible to implement, and can only be implemented after the end of the warranty period of certain batches of graphics cards.

Q: I just saw the atomic (atomic) as the flagship series of AMD's highest public and the possibility of restarting at the booth?

J: You have to know that there are only 20 cards in the world and 5 cards in the mainland.

L: Sapphire as an AIB can not only consider technical, commercial decision is also very important, such as the profit range did not meet expectations or the target group is smaller than our estimate, then this series may not exist anymore.

Q: Even if AMD completely surpasses its friends in the business of independence in the future, will Sapphire not restart the atomic series?

L: It depends on the market (laugh)

Q: Will Radeon7 be out of the public?

L: No, at the moment we don't have Radeon7's non-public plan. AMD's public version of Radeon7 has already reached many of the flagship grades of the company.

Q: In the public interaction section just now, there is a series of A powder in the poison series. As far as I personally observe, many of the powders in the mainland are hoping to restart the poison series. What is the plan for this sapphire?

L: You will see poison on the next generation of graphics cards.

Q: Really (excited)

L: We have completed the work of navi poison, and it is a water-cooled version.

Q: Before the release of Vega in the previous two years, some netizens were asking if sapphire had Vega poison. The reply of sapphire would be there, but it still did not come out.

L: Actually, the relevant design was completed at the time, but as I said earlier, our evaluation results show that Vega poison is not profitable, and even it is difficult to return. Moreover, AMD's Vega's revenue on the super memory is higher than that of the overclocking core , leaving us with a very limited income, so we did not put Vega poison in the end.

Q: It is understandable that a company must ensure that it can make profits by legal means. So, can you disclose some information about Navi (expected)?

L: I can't tell you too much because of some reason you know (laugh) Navi currently has two versions, one version is priced at $499 and the other is $399.

Q: So the price of the mainland, a lot of powder is expected to be poison, they think that super platinum is a younger brother

L: At present, we have not yet determined the price of the mainland, but you know it yourself, isn't it (laugh)

Q: Then, if you are a competitor, you will use the 20 series as your main competitor.

L: Stronger than 2070

Q: What is the rumor that the Navi core of 5120sp+rt core will appear?

L: You have to know that Navi does not exist by the size of the core, there is no related product at present.

Q: So AMD will consider hardware investment in the light chasing acceleration module?

L: You can look forward to next year's new architecture.

Q: What can you tell me about the new architecture? Is it a complete replacement for the current GCN?

L: The new architecture AMD is not completely finalized at present. You need to know that some products are not suitable for the next-generation architecture, whether it is for us or for friends (this is why both ANs have vest cards). It is said that products above the thousand yuan level have been determined to fully adopt the next generation architecture (referring to the successor of Navi)

Q: Is Navi going to be released on July 7 this year?

L Yes, you know 7nm July 7th girl (laugh)

Q: Then on May 27th, you can see AMD's next-generation processor and Navi graphics card.

L: I don't know much about CPU (after all, sapphire only does graphics cards) but the graphics card will definitely show on Computex on May 27.

Q: Will you participate in this year's Computex?

L: Of course

Q: Will the next-generation host adopt Navi's solution (nonsense)?

L: Yes, but still because you know why, I can't tell you too much, I hope to see you on Computex.

=========================================================== ===

In general, this AMD Chengdu station I am still very happy to play, originally wanted to interview the AMD Greater China Vice President Ji Zong, but unfortunately when interviewing LEO, the season has always left, probably the only event of this event Sorry
If the 2070 like version is water cooled Navi must be a very hot chip, 250 watts at least. It doesnt bode well with an efficient design.
 
Oct 25, 2017
3,595
I don't see how your math works for a 44 CU maximum on this die size. It's 4% smaller than PS4 Pro, but on 7nm rather than 16nm node. That means there's room for at least 100% more transistors. Yet you say that a 20% increase in CUs is the highest possible. (Going from 40 with 4 disabled for yield to 48 with 4 disabled.)

Yes, there's a new CPU design and new I/O, but it makes no sense they'll be 80% bigger even with double the trace density. (And "only" double is actually conservative for the node switch.) There will almost certainly be plenty of room for more than 44 CUs. Indeed, all the way up to GCN limit is probably possible, even with this chip smaller than PS4 Pro.

There are other reasons regarding power and cooling why the APU may not be as dense, which you mentioned in passing. But I really doubt the physical size of the chip (if we even trust this rumor) is a constraint.

Like i said, is just a speculation.
7nm Vega VII 64CU is 331mm2.
We don't know about Navi.
If Navi CUs size are about the same as Vega VII, them we have a problem.
Also, more CUs, lower Clock Speeds, even worse with I/O and 8C/16T Zen 2 CPU in the same package.
If the 1.8GHz is true, makes more sense going with less CUs for thermal and consumption.

I really want a 64/56 CUs @ 1.8GHz with Zen 2 8C/16T @ 3.2GHz... but i doubt.
That thing will be hungry and hot af.
 

Darth Smurf X

Member
Oct 25, 2017
2,037
Hoth, WI
Depends. Do you like Navi or Vega?

Alt reply: Pastebin

vega-hdstance.gif
vs.
avatar-sequels-china.jpg


Hmmm...
 

sncvsrtoip

Banned
Apr 18, 2019
2,773
Aiming son
If the 2070 like version is water cooled Navi must be a very hot chip, 250 watts at least. It doesnt bode well with an efficient design.
It don't have to be very hot on basic but if medium segment chip want to be over 2070 (so basically radeon vii) levels the clocks are probably overclocked significantly and watercooling is mandatory. I don't think we should wish rtx2070/radeon vii performance in next consoles unless navi is huge architecture improvement over vega
 
Oct 27, 2017
4,643
I wanted to make a similar point but I'll quote Shifty from B3D since he made it first
Average lifetime device spend exceeds $700•Average launch year lifetime device spend exceeds$1,600"

$1600 for launch consoles. I'm guessing these spends included hardware costs? Makes the notion of loss-leading hardware all the more viable if it means securing the lion's share of sales.

That's 70B in revenue once they hit 100m install base. And $1600 for launch units definitely legitimizes the idea of loss-leading IMO.

I'm reading through the PDF and I'd actually like to see the numbers that contribute to these — I kinda wanna study just how disproportionate it is so I can see how many "casuals" a "hardcore" is worth lol.
 
Last edited:

Pheonix

Banned
Dec 14, 2018
5,990
St Kitts
We don't know details about Navi.
We can speculate based on the Reddit leak, calling the SOC being 312mm2 @ 7nm.
If is true, based on the size and the Gonzalo leak, we can estimate around 40~44CU @ 1.8GHz.
No way in hell a SOC this size will have 56CU, plus Zen 2 and IO and 1.8GHz. Not gonna happen.
For comparasion PS4 Pro GPU is 36CU @ 911mhz = 4.2TF.
While I don't dispute that we may see a 48U GPU in that chip.... just want to point out that the size of the APU isn't why it will only be 48CU. Think about it.

PS4pro is 321mm2. Fits a 8ore CPU + 265bit mem controller + I/O + 40CU all at a 16nm proces.

The rumored PS5 chip is 312mm2. That's less than 4% smaller. And is on a 7nm process. Which at the very least offers an at least 2x density over 16nm. And keep in mind that everything in that chip scales downwards with the fabrication. And its also going to have a 256bit mem controller.

And somehow, you expect the CU count in the GPU to only be able to increase by 8CU? that's like 20% increase.To put this into context, going from 28nm to 16nm (similar to going from 16nm to 7nm) there was a larger size reduction in chip size from the og PS4 to the PS4pro, and yet there was a 100% increase in GPU CU count.
 
Nov 12, 2017
2,877
I love this thread so much.

Leak that Navi is awesome:

- Sony fanboys: Navi was developed by Sony. MS won't get it! Moneyhatted architecture!
- XBOX fanboys: Bullshit - XBOX is Navi too plus some next gen shit. 3D chess!

Leak that Navi is shit:

- Sony fanboys: Everyone is getting Navi!
- XBOX fanboys: Navi sucks, we're on the mega custom Vega next level shit!

Both consoles are the same spec:

- Sony fanboys: All along I've said it's the games that matter. Who gives a shit about flops. 20% is nothing.
- XBOX fanboys: We're buying all the studios. Game pass is life.
Ahahah i love It

.perfect
 

AegonSnake

Banned
Oct 25, 2017
9,566
We don't know details about Navi.
We can speculate based on the Reddit leak, calling the SOC being 312mm2 @ 7nm.
If is true, based on the size and the Gonzalo leak, we can estimate around 40~44CU @ 1.8GHz.
No way in hell a SOC this size will have 56CU, plus Zen 2 and IO and 1.8GHz. Not gonna happen.
For comparasion PS4 Pro GPU is 36CU @ 911mhz = 4.2TF.
The Pro was 325mm2, smaller than the 350mm2 PS4 despite having double the number of CUs. they were able to double the CUs while reducing the size of the die going from 28nm to 16nm.

why wouldnt they be able to double the CUs when going from 16nm to 7nm? the zen 2 7nm cpu shouldnt take more space than the jaguar 16nm cores and even if it does, there is definitely more than enough space on the die to accomodate a 60 CU GPU and a Zen 2 along with some rtx cores.

again, 16nm to 7nm is a 2.3x increase in chip density. they should be able to put in 2.3x the Pro CUs which gives us 82 CUs on a 325mm2 chip. Or 92CUs if we also count the 4 disabled CUs on the pro. but let's ignore that and go by the actual CU count.

  • we know the PS5 chip is even smaller. around 5% smaller.
  • we know 36CUs take up 325mm2 on a 16nm die
  • we know going to 7nm is 2.3x reduction.
  • this gives us roughly a 141mm2 chip for 36CUs
  • or 155mm2 chip for 40 CUs or 170 mm2 chip for 44 CUs.
so tell me, where is the rest of the 150 mm2 worth of space on the die going? are they really going to have that much room for RT cores? highly doubt that. you can look at RT cores on rtx gpus and they are around 10-20% the number of shader units. that gives us a 200mm2 chip.

so what else is in there thats taking up 111mm2 worth of space? the answer is very simple. 60-64 CUs. with 4-8 disabled you get a 56-60CU GPU at 12.9 to 13.8 tflops.

there is absolutely no reason to not use all the CUs available to you when going to 7nm. now if they wanted to settle for a cheap $399 console then they would release a 64 CU GPU at 1.0 ghz which wouldnt require cooling and give us 8 tflops. thats where all the 8 tflops rumors originated from. but now we know they are going with 1.8 ghz. potentially 1.85 ghz. there is zero reason to increase the clockspeeds so high unless they are adding in fancy vapor chamber cooling. they can get 10 tflops with 56 CUs at 1.2 ghz. so why bother going to 1.8 ghz only to settle for 40? makes zero sense.

While I don't dispute that we may see a 48U GPU in that chip.... just want to point out that the size of the APU isn't why it will only be 48CU. Think about it.

PS4pro is 321mm2. Fits a 8ore CPU + 265bit mem controller + I/O + 40CU all at a 16nm proces.

The rumored PS5 chip is 312mm2. That's less than 4% smaller. And is on a 7nm process. Which at the very least offers an at least 2x density over 16nm. And keep in mind that everything in that chip scales downwards with the fabrication. And its also going to have a 256bit mem controller.

And somehow, you expect the CU count in the GPU to only be able to increase by 8CU? that's like 20% increase.To put this into context, going from 28nm to 16nm (similar to going from 16nm to 7nm) there was a larger size reduction in chip size from the og PS4 to the PS4pro, and yet there was a 100% increase in GPU CU count.

i swear i didnt read your post before i wrote mine. i did take like half an hour to write it because my wife decided to dump both my toddlers in my room who as usual wrecked havoc all around me. it was like trying to write a forum post during the battle of the bastards.

i do love how we think alike. haha
 
Last edited:

AegonSnake

Banned
Oct 25, 2017
9,566
I love this thread so much.

Leak that Navi is awesome:

- Sony fanboys: Navi was developed by Sony. MS won't get it! Moneyhatted architecture!
- XBOX fanboys: Bullshit - XBOX is Navi too plus some next gen shit. 3D chess!

Leak that Navi is shit:

- Sony fanboys: Everyone is getting Navi!
- XBOX fanboys: Navi sucks, we're on the mega custom Vega next level shit!

Both consoles are the same spec:

- Sony fanboys: All along I've said it's the games that matter. Who gives a shit about flops. 20% is nothing.
- XBOX fanboys: We're buying all the studios. Game pass is life.
the funniest thing about this post is how not a single point is true. i get it you are kidding, but all jokes are rooted in some kind of truth. this has none of it.

the navi was made for sony rumors originated in the last thread or the thread before the last thread and everyone ignored it as bs. the 20% is nothing comments have been made by both camps. both when ms had the lead and now when rumors suggest sony has the tflops advantage.

there is very little talk about gamepass or sony first party studios here even on slow news days. you must be thinking of the first party threads.
 

sncvsrtoip

Banned
Apr 18, 2019
2,773
The Pro was 325mm2, smaller than the 350mm2 PS4 despite having double the number of CUs. they were able to double the CUs while reducing the size of the die going from 28nm to 16nm.

why wouldnt they be able to double the CUs when going from 16nm to 7nm? the zen 2 7nm cpu shouldnt take more space than the jaguar 16nm cores and even if it does, there is definitely more than enough space on the die to accomodate a 60 CU GPU and a Zen 2 along with some rtx cores.

again, 16nm to 7nm is a 2.3x increase in chip density. they should be able to put in 2.3x the Pro CUs which gives us 82 CUs on a 325mm2 chip. Or 92CUs if we also count the 4 disabled CUs on the pro. but let's ignore that and go by the actual CU count.

  • we know the PS5 chip is even smaller. around 5% smaller.
  • we know 36CUs take up 325mm2 on a 16nm die
  • we know going to 7nm is 2.3x reduction.
  • this gives us roughly a 141mm2 chip for 36CUs
  • or 155mm2 chip for 40 CUs or 170 mm2 chip for 44 CUs.
so tell me, where is the rest of the 150 mm2 worth of space on the die going? are they really going to have that much room for RT cores? highly doubt that. you can look at RT cores on rtx gpus and they are around 10-20% the number of shader units. that gives us a 200mm2 chip.

so what else is in there thats taking up 111mm2 worth of space? the answer is very simple. 60-64 CUs. with 4-8 disabled you get a 56-60CU GPU at 12.9 to 13.8 tflops.

there is absolutely no reason to not use all the CUs available to you when going to 7nm. now if they wanted to settle for a cheap $399 console then they would release a 64 CU GPU at 1.0 ghz which wouldnt require cooling and give us 8 tflops. thats where all the 8 tflops rumors originated from. but now we know they are going with 1.8 ghz. potentially 1.85 ghz. there is zero reason to increase the clockspeeds so high unless they are adding in fancy vapor chamber cooling. they can get 10 tflops with 56 CUs at 1.2 ghz. so why bother going to 1.8 ghz only to settle for 40? makes zero sense.
Another flexible math, 28nm to 16nm is simillar as 16nm to 7nm. So if sony takes the same path (256bit, not very big chip) that 4.2tf/1.84tf = 2.28x So 2.28x 4.2tf = 9.58tf
Edit: fixed wrong calculation
 
Last edited:

ppn7

Member
May 4, 2019
740
While I don't dispute that we may see a 48U GPU in that chip.... just want to point out that the size of the APU isn't why it will only be 48CU. Think about it.

PS4pro is 321mm2. Fits a 8ore CPU + 265bit mem controller + I/O + 40CU all at a 16nm proces.

The rumored PS5 chip is 312mm2. That's less than 4% smaller. And is on a 7nm process. Which at the very least offers an at least 2x density over 16nm. And keep in mind that everything in that chip scales downwards with the fabrication. And its also going to have a 256bit mem controller.

And somehow, you expect the CU count in the GPU to only be able to increase by 8CU? that's like 20% increase.To put this into context, going from 28nm to 16nm (similar to going from 16nm to 7nm) there was a larger size reduction in chip size from the og PS4 to the PS4pro, and yet there was a 100% increase in GPU CU count.

Again, 16nm to 7nm maybe is 2x more density, but how about the heat on 7nm, the clock speed of PS4 is under 1000mhz, Zen 2 could be probably bigger than Jaguar ? and according to TSMC, from 16nm to 7nm is +35/40% faster or more powerful at same TDP. Then what about the price, 7nm is a new node, still high priced but who care ok, maybe it will be cheaper in 2020 when PS5 will arrive.

And so the Sapphire leaks talking about 399$ and 499$ SKU for Navi, stronger than 2070, which one, the 499$ only or maybe both of them are stronger then RTX 2070, and without RT cores, so we have to wait
 

RoboPlato

Member
Oct 25, 2017
6,808
Isn't that similar to what happened with Xbone? In the late stages before release, MS increased both CPU and GPU speeds. Not saying they did it as a response to ps4, but it was a response to something.
Those were tiny clock speed increases. A lot of these rumors imply changes to architecture or huge increases to clock speed.
 

Pheonix

Banned
Dec 14, 2018
5,990
St Kitts
The Pro was 325mm2, smaller than the 350mm2 PS4 despite having double the number of CUs. they were able to double the CUs while reducing the size of the die going from 28nm to 16nm.

why wouldnt they be able to double the CUs when going from 16nm to 7nm? the zen 2 7nm cpu shouldnt take more space than the jaguar 16nm cores and even if it does, there is definitely more than enough space on the die to accomodate a 60 CU GPU and a Zen 2 along with some rtx cores.

again, 16nm to 7nm is a 2.3x increase in chip density. they should be able to put in 2.3x the Pro CUs which gives us 82 CUs on a 325mm2 chip. Or 92CUs if we also count the 4 disabled CUs on the pro. but let's ignore that and go by the actual CU count.

  • we know the PS5 chip is even smaller. around 5% smaller.
  • we know 36CUs take up 325mm2 on a 16nm die
  • we know going to 7nm is 2.3x reduction.
  • this gives us roughly a 141mm2 chip for 36CUs
  • or 155mm2 chip for 40 CUs or 170 mm2 chip for 44 CUs.
so tell me, where is the rest of the 150 mm2 worth of space on the die going? are they really going to have that much room for RT cores? highly doubt that. you can look at RT cores on rtx gpus and they are around 10-20% the number of shader units. that gives us a 200mm2 chip.

so what else is in there thats taking up 111mm2 worth of space? the answer is very simple. 60-64 CUs. with 4-8 disabled you get a 56-60CU GPU at 12.9 to 13.8 tflops.

there is absolutely no reason to not use all the CUs available to you when going to 7nm. now if they wanted to settle for a cheap $399 console then they would release a 64 CU GPU at 1.0 ghz which wouldnt require cooling and give us 8 tflops. thats where all the 8 tflops rumors originated from. but now we know they are going with 1.8 ghz. potentially 1.85 ghz. there is zero reason to increase the clockspeeds so high unless they are adding in fancy vapor chamber cooling. they can get 10 tflops with 56 CUs at 1.2 ghz. so why bother going to 1.8 ghz only to settle for 40? makes zero sense.



i swear i didnt read your post before i wrote mine. i did take like half an hour to write it because my wife decided to dump both my toddlers in my room who as usual wrecked havoc all around me. it was like trying to write a forum post during the battle of the bastards.

i do love how we think alike. haha
Hehe... maybe you should try giving them a detailed explanation of TFs and the like, if nothing else it puts them to sleep.
Again, 16nm to 7nm maybe is 2x more density, but how about the heat on 7nm, the clock speed of PS4 is under 1000mhz, Zen 2 could be probably bigger than Jaguar ? and according to TSMC, from 16nm to 7nm is +35/40% faster or more powerful at same TDP. Then what about the price, 7nm is a new node, still high priced but who care ok, maybe it will be cheaper in 2020 when PS5 will arrive.

And so the Sapphire leaks talking about 399$ and 499$ SKU for Navi, stronger than 2070, which one, the 499$ only or maybe both of them are stronger then RTX 2070, and without RT cores, so we have to wait
As I said I am not disputing that there could be only 48CU in that APu. I am just saying that it won't be because that's all they could fit in there, Especially if rumors of Navi being a more gaming-focused GPU (so no compute bloat) are true.. There may be other things that they are adding into the APU that may take up space, there may be multiple mem controllers, some RT stuff......a leprechaun. My point though is that if looking at the previous GCN based chips.... there is no reason why they shouldn't be able to fit more than 48CU.

Hell, has it occurred to you that they could put in 56CU, disable 4CU to leave only 52CU active and clock that at 1600Mhz to get around 10.6TF. Or go with 60CU and clock it at 1400Mh for 10.7TF.
 
Status
Not open for further replies.