confirmed: AMD's big Navi launch to disrupt 4K gaming

Zarathustra[H]

I Complain about Everything
Joined
Oct 29, 2000
Messages
29,979
You're absolutely correct.

AMD performance to value is amazing right now but for a straight-up gaming box it's not the fastest.
It's a tie in 90+% of configurations. Inteal only had an advantage in very low resolution very high refresh scenarios, which quite frankly are totally uninteresting.
 

ManofGod

[H]F Junkie
Joined
Oct 4, 2007
Messages
11,843
Similar performance does not mean it is the same chip. Jesus fucking Christ this is one of the dumbest arguments I have ever fucking seen.

Edit: And that's ignoring the fact that you're wrong. The 2080 ti and Titan XP do NOT have the same performance. On average, the 2080 TI will be noticeably faster than the XP.
Watch you mouth. Also, just because my point is correct does not make you wrong, did you ever consider that? Hey, if someone wants to spend $500 more for the same thing they could get the previous year for less...... :D
 

Snowdog

[H]F Junkie
Joined
Apr 22, 2006
Messages
10,424
LOL! What the heck? :D That does not make any sense.
I know, just using your nonsense "logic", back at you.

Trying to get though your thick skull, that performing the same, does not mean it actually is the same technology/chip.
 

Ready4Dis

[H]ard|Gawd
Joined
Nov 4, 2015
Messages
1,036
so tests and evidence mean nothing to you, only your personal feelings. I should not be surprised, but evidence trumps feelings, you lose, again.

1080 ti superseeded by 2080 ti, not the 2080. also i went from the Titan xp to the 2080 ti and had noticiable improvements in fps. went from janky 3440/1440 to buttery smooth.

lets see you move that goal post again...
As much is it pains me to say it, he is right in comparing 1080 ti and 2080... 1080 ti MSRP was $699, 2080 was $799.... It was Nvidia that moved the goal posts by moving everything up a category. The 2080ti @ $1200 was not the replacement for the 1080 @ $700. The 2080 was the direct replacement at $100 more.
 

Aireoth

2[H]4U
Joined
Oct 12, 2005
Messages
3,543
As much is it pains me to say it, he is right in comparing 1080 ti and 2080... 1080 ti MSRP was $699, 2080 was $799.... It was Nvidia that moved the goal posts by moving everything up a category. The 2080ti @ $1200 was not the replacement for the 1080 @ $700. The 2080 was the direct replacement at $100 more.
Price creep does not mean they are the same product, Ti is Ti.

I agree nVidia has a pricing problem, but its only a problem if AMD can deliver a high end video card that competes with the current gen, so far they only hit the best of the last gen.
 

Snowdog

[H]F Junkie
Joined
Apr 22, 2006
Messages
10,424
As much is it pains me to say it, he is right in comparing 1080 ti and 2080... 1080 ti MSRP was $699, 2080 was $799.... It was Nvidia that moved the goal posts by moving everything up a category. The 2080ti @ $1200 was not the replacement for the 1080 @ $700. The 2080 was the direct replacement at $100 more.
Are you blind? Or in part of the same club as him?

People are arguing against his nonsense claim, that Turing is just relabeled Pascal.

Pascal was essentially just Maxwell architecture again.

But Turing is a massively changed architecture, to not recognize that some massive if not willful ignorance.
 

Dayaks

Supreme [H]ardness
Joined
Feb 22, 2012
Messages
7,782
2080ti = Titan with RT Cores
2080 = 1080Ti with RT Cores
And so on. They was no real performance difference between the generations, overall, just added a at the time unused feature, relabeled the products and charged more for them.

However, the 5700 and 5700 XT is faster than the previous Vega 56 and Vega 64, uses significantly less power and costs less at MSRP then the Vega's, as well.
Yes, they are, performance is virtually identical. You forgot that the RT cores, which were not particularly useful at release, was added. Physical die sizes are not relevant because otherwise, by your definition, the 5700XT should be far slower than the Vega 64.
2080ti is around a 50% larger die and 25-45% more rasterized performance than a 1080ti.

nVidia’s net margin is ~20-25%. That’s a little high but any absurd price gouging like people are talking about would show up there.
 

ManofGod

[H]F Junkie
Joined
Oct 4, 2007
Messages
11,843
2080ti is around a 50% larger die and 25-45% more rasterized performance than a 1080ti.

nVidia’s net margin is ~20-25%. That’s a little high but any absurd price gouging like people are talking about would show up there.
Personally, I think they just moved the naming of the products up the line to make it look like they were significantly faster. After all, the 2080Ti is not all that much different from the previous Titan but hey, if they had labeled the 2080 as a 2080Ti, like they did in the past, no one would have bought one, since it was no better than the 1080Ti.
 

Ready4Dis

[H]ard|Gawd
Joined
Nov 4, 2015
Messages
1,036
Price creep does not mean they are the same product, Ti is Ti.

I agree nVidia has a pricing problem, but its only a problem if AMD can deliver a high end video card that competes with the current gen, so far they only hit the best of the last gen.
It's not price creep, they purposefully moved up their product stack... Price creep was the $100 increase from one generation to next (2080 being $100 more than 1080ti), not an almost doubling of price. That's not the same. If I was purchasing a card in a specific segment, the 2080 IS the same segment (price wise) as the 1080 to filled, with a little price creep. That was my only point.
 

Ready4Dis

[H]ard|Gawd
Joined
Nov 4, 2015
Messages
1,036
Are you blind? Or in part of the same club as him?

People are arguing against his nonsense claim, that Turing is just relabeled Pascal.

Pascal was essentially just Maxwell architecture again.

But Turing is a massively changed architecture, to not recognize that some massive if not willful ignorance.
Lol, says the man having a hard time taking the blinders off... No, my point had NOTHING to do with the architecture change. My point was simply that the 2080 filled the same price segment as the previous 1080ti, thus are more comparable. The 2080 was $100 more than 1080ti which is a normal price increase in a similar segment for a different architecture. There were many changes between these cards, and I would 100% percent take the 2080 over the 1080ti due to those changes, but if my market was $600-$700, then I'm not even considering the 2080ti, it's in a new segment all to itself. I'm not saying that's good or bad, you want the extra performance, then you pay for it. I'm not arguing about the platform differences, so no clue why your projecting your insecurities on me, but maybe you should take a look at your own bias before casting others into some 'club'.
 

Derangel

Fully [H]
Joined
Jan 31, 2008
Messages
18,955
Lol, says the man having a hard time taking the blinders off... No, my point had NOTHING to do with the architecture change. My point was simply that the 2080 filled the same price segment as the previous 1080ti, thus are more comparable. The 2080 was $100 more than 1080ti which is a normal price increase in a similar segment for a different architecture. There were many changes between these cards, and I would 100% percent take the 2080 over the 1080ti due to those changes, but if my market was $600-$700, then I'm not even considering the 2080ti, it's in a new segment all to itself. I'm not saying that's good or bad, you want the extra performance, then you pay for it. I'm not arguing about the platform differences, so no clue why your projecting your insecurities on me, but maybe you should take a look at your own bias before casting others into some 'club'.
You should actually take note of a conversation before jumping in blindly. What you said has fuck all to do with what was being talked about.
 

ManofGod

[H]F Junkie
Joined
Oct 4, 2007
Messages
11,843
You should actually take note of a conversation before jumping in blindly. What you said has fuck all to do with what was being talked about.
What he said has everything to do with what I said, despite what has been claimed to the contrary.
 

Ready4Dis

[H]ard|Gawd
Joined
Nov 4, 2015
Messages
1,036
You should actually take note of a conversation before jumping in blindly. What you said has fuck all to do with what was being talked about.
I did take note and read through the entire thing... "As much is it pains me to say it, he is right in comparing 1080 ti and 2080... 1080 ti MSRP was $699, 2080 was $799....", This is the correct GPU from both generation to be comparing. I NEVER said I agreed with his assessment, just that it should be the proper cards to compare. Please grow up if you can't have a simple conversation and overreact because someone does not 100% agree with your viewpoint. I feel there was significant changes (for the better) between the 10X0 and 20X0 and am looking forward to even more progress with Ampere (especially with faster RT cores). None of that changes my view that 2080 replaced the 1080ti and 2080ti was a new higher prices segment. It's like saying because AMD 3950x is their highest end CPU, it should be compared to a 2700x because they are both the top product from their generation... That's rediculous, there were new higher priced segments added with zen2, so I would compare similarly priced CPUs from one generation to the next. It's no different to me whether I'm talking about AMD, Intel, Nvidia, CPU, GPU, w/e. If you don't like it, well, I'd say I'm sorry, but I'm really not. That's my viewpoint. I still feel the merits of the 2080 over the 1080ti make it a better graphics card, and that's the ones I would compare to each other.
 

Snowdog

[H]F Junkie
Joined
Apr 22, 2006
Messages
10,424
I did take note and read through the entire thing... "As much is it pains me to say it, he is right in comparing 1080 ti and 2080... 1080 ti MSRP was $699, 2080 was $799....", This is the correct GPU from both generation to be comparing. I NEVER said I agreed with
Yes, it really pains you to leap to support someone of like mind, like manogod.

As Derangel pointed out you utterly ignored the point, one can only assume willfully(seems a shared trait with you and manogod), that no one is arguing against the point that Turing didn't improve perf/$$. That's been obvious since Turing launched in 2018. Everyone acknowledges that obvious fact.

The thing everyone was arguing against, was the manogod delusion that Turing was just Pascal with RT tacked on. A delusion I guess you share since you are so stridently defending your buddy over it..
 

Ready4Dis

[H]ard|Gawd
Joined
Nov 4, 2015
Messages
1,036
Yes, it really pains you to leap to support someone of like mind, like manogod.

As Derangel pointed out you utterly ignored the point, one can only assume willfully(seems a shared trait with you and manogod), that no one is arguing against the point that Turing didn't improve perf/$$. That's been obvious since Turing launched in 2018. Everyone acknowledges that obvious fact.

The thing everyone was arguing against, was the manogod delusion that Turing was just Pascal with RT tacked on. A delusion I guess you share since you are so stridently defending your buddy over it..
I'm ignoring what point? You are completely interpreting what I'm saying the way you want instead of taking it for what it is. Your ignoring what I'm saying and arguing someone else's points as if I made them. I get what you guys are arguing, I'm happy for you, really, keep it up your winning the internet today, lol. All I was saying was the proper comparison between generations was more in line with 1080ti to 2080, etc. If you managed to read into that that me and MangoGod are in some sort of synchronized thinking then thats on you, not me. I even said there were substantial generational updates that I think make it a much better card. Not sure that is in agreement with MwngoGod, it actually aligns with your argument... Oh no, that means you must be MangoGods bestest friend because I am obviously a shill like he is, and I agreed with your point in generational improvements, so...! Yeah, that makes as much sense as what your saying. OMG, someone isn't 100% agreeing that NVIDIA is perfect, they must be an AMD fan boi!! Yeah, no, some of use can be in the middle of you and MangoGod and see both viewpoints and use our own brains to have our own views that lay in the middle of the AMD camp and Nvidia camps.

Edit. PS this is getting very off topic, please PM if you feel the need to discuss anymore.
 

Aegir

n00b
Joined
Apr 15, 2020
Messages
58
People speaking of how performance in frames-per-second doesn't change between different GPU architectures makes me think of a recent thread I made about ATX vs ITX motherboards, where I began to ponder: Even if you put in the same RAM, the same GPU, and the same CPU, could there be distinct features that are missing from a smaller motherboard that have nothing to do with actual CPU/GPU performance?

If GPU-X runs Game-X at 120 frames per second, and GPU-Y runs Game-X at 120 frames per second, then are those cards the same? What if GPU-Y has an enhanced physics module, or processes light in a different way?

Even if the performance is the same, that is NOT the full story.

There is far more to technology than looking at one specific type of performance and then just making it smaller, faster, and more optimized.

The heart of technological innovation is inventing completely new technologies. These might seem more primitive in comparison with current technology, but all technology needs time to develop. The evolution of technology must be taken note of, or else important details and aspects regarding new features and their capabilities will be missed.
 

Marees

Limp Gawd
Joined
Sep 28, 2018
Messages
424
Rumor about Navi 23 aka the "Nvidia Killer" gpu

Beyond Navi21
We know from previous leaks that AMD is planning three RDNA2 dies as part of the Navi2x family. We also know that even the smaller Navi23 die is around Navi10 die size. Add to that the new RDNA2 improvements as well as the expected clock bump and you have a product that is far better than current Navi10 lineup

Having Navi23 this powerful means it’ll not be cheap
https://hardwareleaks.com/2020/05/23/exclusive-future-amd-gpu-stack-navi21-navi10-refresh/

I take it that he is implying that the RX 6700 XT will be $500+ & close to performance of RTX 2080 ti
 

gigatexal

Supreme [H]ardness
Joined
Jun 22, 2004
Messages
7,418
Rumor about Navi 23 aka the "Nvidia Killer" gpu



https://hardwareleaks.com/2020/05/23/exclusive-future-amd-gpu-stack-navi21-navi10-refresh/

I take it that he is implying that the RX 6700 XT will be $500+ & close to performance of RTX 2080 ti
open source gaming seems to favor the AMD line of cards -- the mesa drivers really are top notch -- so for me I think it'll be AMD still -- looking forward to what they bring out -- hopefully it kills everything so the 5700XTs drop in price a ton
 

Snowdog

[H]F Junkie
Joined
Apr 22, 2006
Messages
10,424
Rumor about Navi 23 aka the "Nvidia Killer" gpu



https://hardwareleaks.com/2020/05/23/exclusive-future-amd-gpu-stack-navi21-navi10-refresh/

I take it that he is implying that the RX 6700 XT will be $500+ & close to performance of RTX 2080 ti
You are guessing at what you think he is implying about a rumor? :D

There really has been nothing about any other chip yet, except the original Navi 21( ~500mm2, 80CU), rumor back in January (and probably earlier, as I couldn't be bothered to trace the original source).

Navi 21 (AKA Big Navi) would then be the NVidia killer, not something smaller. The smaller chips are just more guesses based on being smaller than the original rumor, which may also be a guess.
 

Snowdog

[H]F Junkie
Joined
Apr 22, 2006
Messages
10,424
Getting to the original sources to assess how solid these rumors seem:

The original Big Navi 80CU rumor that everyone seems to be riffing on since appears to be have started here (Jan 2020):
https://www.guru3d.com/news-story/amd-radeon-5800-xt-(big-navi)-to-get-80-compute-units,2.html

But this is just obvious extrapolation: When we were speculating late last year about Big Navi at CES I said this:
"A 20 Billion Transistor, 500 mm^2 Navi die, with 80 CUs (5120 SPs), 512 bit memory would leave the 2080 Ti behind. That's a true NVidia killer. "

It's just a rather obvious extrapolation. The next rumor that listed three different Navi 2 series parts (21,22,23) appears to have started with this tweet:
https://twitter.com/KOMACHI_ENSAKA/status/1255161800116416512?ref_src=twsrc^tfw

Which was gleaned from this Chinese forum posting:
https://www.ptt.cc/bbs/PC_Shopping/M.1588075782.A.C1E.html

To my eye, it looks like we have nothing. These rumors don't seem to be anything more than forum speculations (plus language barrier) turned into "leaks/rumors".

The only Die with actual numbers attached is the Biggest one, which is essentially identical to my simplistic double Navi 10, extrapolation from December.

The problem with the 80CU, ~500 mm2 speculation is that it's just Navi 10 X2. An easy guess that anyone could/would make. Doesn't mean it won't happen, but this is exactly the kind of speculation that gets turned into rumor.

Bottom line, I don't see that we have anything solid. Our own speculations are just as likely as the rumors so far.

All we really know is that Big Navi is coming with RDNA 2. It will almost certainly beat 2080Ti, but we don't know how it will fare vs the also unknown 3080 Ti.
 

Marees

Limp Gawd
Joined
Sep 28, 2018
Messages
424
Getting to the original sources to assess how solid these rumors seem:

The original Big Navi 80CU rumor that everyone seems to be riffing on since appears to be have started here (Jan 2020):
https://www.guru3d.com/news-story/amd-radeon-5800-xt-(big-navi)-to-get-80-compute-units,2.html

But this is just obvious extrapolation: When we were speculating late last year about Big Navi at CES I said this:
"A 20 Billion Transistor, 500 mm^2 Navi die, with 80 CUs (5120 SPs), 512 bit memory would leave the 2080 Ti behind. That's a true NVidia killer. "

It's just a rather obvious extrapolation. The next rumor that listed three different Navi 2 series parts (21,22,23) appears to have started with this tweet:
https://twitter.com/KOMACHI_ENSAKA/status/1255161800116416512?ref_src=twsrc^tfw

Which was gleaned from this Chinese forum posting:
https://www.ptt.cc/bbs/PC_Shopping/M.1588075782.A.C1E.html

To my eye, it looks like we have nothing. These rumors don't seem to be anything more than forum speculations (plus language barrier) turned into "leaks/rumors".

The only Die with actual numbers attached is the Biggest one, which is essentially identical to my simplistic double Navi 10, extrapolation from December.

The problem with the 80CU, ~500 mm2 speculation is that it's just Navi 10 X2. An easy guess that anyone could/would make. Doesn't mean it won't happen, but this is exactly the kind of speculation that gets turned into rumor.

Bottom line, I don't see that we have anything solid. Our own speculations are just as likely as the rumors so far.

All we really know is that Big Navi is coming with RDNA 2. It will almost certainly beat 2080Ti, but we don't know how it will fare vs the also unknown 3080 Ti.
Redgamingtech maintains that as per his sources "Nvidia Killer" refers to Navi 23. I will hunt around & get that link in a short while. It is from Aug-2019 I think.

For a long time I was thinking that Big Navi & "Nvidia Killer" refer to the same. (Which is why the title of this post has "big Navi" in it.)

After the recent rumor about the sizes of Navi 21, Navi 22, & Navi 23 chips, I came across the old rumor that Navi 23 is referred to as "Nvidia killer"

Coming back to the original post the AMD exec just says that they will disrupt 4k gaming. Doesn't say which of the 3 chips (or if all 3) will disrupt. Long road to go. Another 6 months before we can confirm if the claim is true.
 
Last edited:

Marees

Limp Gawd
Joined
Sep 28, 2018
Messages
424
Redgamingtech maintains that as per his sources "Nvidia Killer" refers to Navi 23. I will hunt around & get that link in a short while. It is from Aug-2019 I think.
Here is the post which links to Redgamingtech's rumour

https://hardforum.com/threads/big-navi-rumored-to-have-die-size-over-505mm2.1995923/post-1044572753

Just going thru the links in the OP & saw that Redgamingtech's source refers to Navi 23 (5700XT replacement) & not Navi 21 (big Navi) as the "Nvidia killer"

Does this imply a $400-$500 6700XT that performs better than a 2080 ti ?



http://www.redgamingtech.com/navi-20-series-is-known-internally-as-the-nvidia-killer-exclusive/

EDIT:
he acknowledges that possibility in the next post



http://www.redgamingtech.com/big-navi-released-date-naming-scheme-is-rx-6000-exclusive/
 

Snowdog

[H]F Junkie
Joined
Apr 22, 2006
Messages
10,424
Clear as mud, quoting your source:
A few things are possible here – either my source was mistaken and it was Navi 21 which was the ‘Nvidia Killer’, but he was very certain it was Navi 23. Another possibility is that Navi 23 is even a higher-end card than Navi 21, although this would be an interesting development – as normally the biggest SKU is created first then the lower end GPUs follow it.
Maybe Navi 23 is the big one, maybe Navi 21 is, maybe NVidia killer means performance, maybe it means value... :rolleyes: IOW nothing of consequence, but supposedly someone at AMD said they have a NVidia killer.

Doesn't matter what supposedly was said behind close doors to boost morale. The only NVidia killer people really care about in the real world is beating or at least matching NVidia top parts.

As far as value (relative pricing) goes, that would have been completely impossible for even someone in the AMD development team to know back in 2019. They could know, they were beating 2080Ti and call it NVidia killer based on that, but no one back then would have a clue on eventual pricing by shipping time.

That is something that will only be determined when the rubber hits the road and the pricing dust settles. But when the dust settles the cards will be essential be close in price/performance ratio. Because AMD doesn't want to gut their margins, and NVidia won't sacrifice large amount of market share, and they will have similar build cost structures, since they essentially share a fab and process now (some form of refined TSMC 7nm).

So there will be some price jockeying, and in the end, it will settle into the old pattern where AMD gives a small price/performance advantage over NVidia.
 

Marees

Limp Gawd
Joined
Sep 28, 2018
Messages
424
(plus language barrier)
Here is a translation by a twitter user, if it helps:

So, Twitter user Retired Engineer has graciously provided us with a translation of the other parts of the leak we weren’t comfortable speculating on when we had the poor Google translation. Most of this content is tangential and not directly related to the Navi die sizes but it’s worth looking into these extra parts.
above translation from below post:

One of the largest die sizes ever known to have been created!?

"The OP also talked about the console APUs, Arden (for Microsoft) and Oberon (for Sony). Arden has entered mass production in a volume similar to Navi 10 GPUs (that is the 5700 series) prior to the July 7 launch. Oberon also had been in mass production, but is now “nowhere to be seen.” This could mean that Sony is afraid that Microsoft’s hardware advantage could translate into better sales for the Xbox and production for the PS5 APU has ramped down in anticipation of poor sales. The Sparkman APU also got a mention since it is now in trial production, but the OP could not confirm whether or not this was another Microsoft SKU. There is still an expectation that Microsoft will launch a cheaper, less powerful Xbox, which is why OP suspects it could be ordered by Microsoft.

Finally, there was a fragment of info about CDNA, AMD’s upcoming compute oriented GPU architecture. Apparently it is codenamed Vega H. That’s about it for CDNA, which we know relatively little about."


https://adoredtv.com/big-navi-rumored-to-have-die-size-of-505-mm2/
 

Marees

Limp Gawd
Joined
Sep 28, 2018
Messages
424
Because AMD doesn't want to gut their margins,
If the speculated die sizes are correct, it makes the Navi 23 chip same size as PS5, Navi 10, or Polaris (the RX470/570 sold for $120 at some point)

When Navi 10 first leaked, based on chip size it was speculated that price would be in range of $250-$350, but it launched between $350-$500

Again based on chip size we can speculate that Navi 23 will be less than $500 but who knows if it is a huge boost in performance compared to 5700 XT maybe AMD might price it $500+
 

Marees

Limp Gawd
Joined
Sep 28, 2018
Messages
424
Still not seeing the supposed source of the die sizes to determine if they are claimed to come from AMD source, or if they are just forum speculation. If you start with double size Big Navi, you can just as well extrapolate to smaller dies under it.
The source for die sizes was supposed to be TSMC and not AMD

EDIT:

from AdoredTv

So, this forum user (who might be a TSMC employee) apparently knows the die sizes of three different GPUs within 5 mm2. Navi 21 is around 505 mm2 (consistent with a previous leak from the same user), Navi 22 is 340 mm2, and Navi 23 is 240 mm2. The die size of Navi 23 is an interesting point of data because we have already heard quite a bit about it from RedGamingTech, who I find has a decent track record. He was told that Navi 23 is referred to internally as the “Nvidia killer”, which is an odd name for such a small GPU, smaller than even the 5700XT. However, it might not be a killer in the sense of performance, but in value, as RedGamingTech has speculated recently.
https://adoredtv.com/big-navi-rumored-to-have-die-size-of-505-mm2/
 
Last edited:

cybereality

Supreme [H]ardness
Joined
Mar 22, 2008
Messages
5,325
Honestly, we know next to nothing for certain. And, judging from past performance, AMD may not have the goods.

But, if they do have a 2080 Ti killer, it won't be $500. Though if they did that, it would be amazing.
 

Snowdog

[H]F Junkie
Joined
Apr 22, 2006
Messages
10,424
If the speculated die sizes are correct, it makes the Navi 23 chip same size as PS5, Navi 10, or Polaris (the RX470/570 sold for $120 at some point)
PS5 dies appear closer in size to Navi 22.

Lets assume the speculated size claims are correct.
Navi 21: 505 mm2 - 80 CU claimed - Estimate significantly better than 2080 Ti performance.
Navi 22: 340 mm2 - 56 CU (Estimate from Size) - Estimate near, but under 2080 Ti performance.
Navi 23: 240 mm2 - 36 CU (Estimate from Size) - Estimate near, butt under 2080S performance.

Pricing estimates are too dependent on NV 3000 series. If these were out now with no new NVidia competetion. I would expect minimum $1000 for Navi 21, $700 for Navi 22, $500 for Navi 23.

Now if I extrapolate NVidia 3000 impact on top of these extrapolations, Navi 21 $800, Navi 22 $600, and Navi 23 $400.
 

ManofGod

[H]F Junkie
Joined
Oct 4, 2007
Messages
11,843
Nice but, it really does not say anything. Also, based upon a majority consensus, WCCF Tech is to be taken with a rather large grain of salt, anyway.

PS5 dies appear closer in size to Navi 22.

Lets assume the speculated size claims are correct.
Navi 21: 505 mm2 - 80 CU claimed - Estimate significantly better than 2080 Ti performance.
Navi 22: 340 mm2 - 56 CU (Estimate from Size) - Estimate near, but under 2080 Ti performance.
Navi 23: 240 mm2 - 36 CU (Estimate from Size) - Estimate near, butt under 2080S performance.

Pricing estimates are too dependent on NV 3000 series. If these were out now with no new NVidia competetion. I would expect minimum $1000 for Navi 21, $700 for Navi 22, $500 for Navi 23.

Now if I extrapolate NVidia 3000 impact on top of these extrapolations, Navi 21 $800, Navi 22 $600, and Navi 23 $400.
You can attempt to extrapolate all you please but, at best, it is simply a guess like everything else we have, information wise. You can expect a minimum of whatever you like but, I would not believe your opinion anyways.
 

Marees

Limp Gawd
Joined
Sep 28, 2018
Messages
424
PS5 dies appear closer in size to Navi 22.

Lets assume the speculated size claims are correct.
Navi 21: 505 mm2 - 80 CU claimed - Estimate significantly better than 2080 Ti performance.
Navi 22: 340 mm2 - 56 CU (Estimate from Size) - Estimate near, but under 2080 Ti performance.
Navi 23: 240 mm2 - 36 CU (Estimate from Size) - Estimate near, butt under 2080S performance.

Pricing estimates are too dependent on NV 3000 series. If these were out now with no new NVidia competetion. I would expect minimum $1000 for Navi 21, $700 for Navi 22, $500 for Navi 23.

Now if I extrapolate NVidia 3000 impact on top of these extrapolations, Navi 21 $800, Navi 22 $600, and Navi 23 $400.
For this performance, you have mentioned, the price expectation of $400 for Navi 23 (Rx 6700 XT ??) is fair

The rumor from the hardwareleaks guy was that there could be a vast performance improvement (despite being on similar nodes) due to RDNA 2 optimizations.

Assuming the 6700XT is better than a RTX 2080 super & worse than a RTX 2080 ti and also assuming that it has no Ray Tracing (since it is a small chip), what would be the price expectation if it released between Jan-2021 to Mar-2021 ??
Less than $500 or more ?
 

Snowdog

[H]F Junkie
Joined
Apr 22, 2006
Messages
10,424
For this performance, you have mentioned, the price expectation of $400 for Navi 23 (Rx 6700 XT ??) is fair

The rumor from the hardwareleaks guy was that there could be a vast performance improvement (despite being on similar nodes) due to RDNA 2 optimizations.

Assuming the 6700XT is better than a RTX 2080 super & worse than a RTX 2080 ti and also assuming that it has no Ray Tracing (since it is a small chip), what would be the price expectation if it released between Jan-2021 to Mar-2021 ??
Less than $500 or more ?
Well I have different assumptions in my extrapolation. I assume Navi 23 still has Ray Tracing, and with a smaller die(Than Navi 10), and some of that smaller die devoted to Ray Tracing, it won't quite reach 2080S performance level even with new gen performance bump. But close to 2080S for $400 seems pretty good to me.

Mind your these are extrapolation on top of extrapolations, based on sketchy rumors...
 
Top