• 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.

How hyped are you?

  • A little hyped

    Votes: 98 15.7%
  • A lot of hyped

    Votes: 50 8.0%
  • WALNUUUUUTSSSSS

    Votes: 222 35.5%
  • Hyped enough to eat this whole bag of walnuts

    Votes: 63 10.1%
  • Hyped enough to bite this moose

    Votes: 37 5.9%
  • Hyped enough to scramble a dozen eggs

    Votes: 39 6.2%
  • Hyped enough to be even more hyped, like, cyclical or something

    Votes: 116 18.6%

  • Total voters
    625
Status
Not open for further replies.

Transistor

Hollowly Brittle
Administrator
Oct 25, 2017
37,193
Washington, D.C.
Oh, i don't think it'll happen, but if only a few thousand units could be available, consoles could be released next fall, but with a premium price, and in very limited quantities, and do massive launch with standard price next year
I mean. Of course it could, but that would be such bad PR that it wouldn't be worth what the price increase would get them.
 

Dimple

Member
Jan 10, 2018
8,569
Didn't see this posted but IGN made a video podcast where multiple developers have commented on what gamers can expect from both next gen graphics and new gameplay possibilities. I felt like there was quite a bit of interesting stuff in here. Surprised that I didn't see any discussion on any of this:

youtu.be

What Xbox’s 12 Teraflops ACTUALLY Do - Next-Gen Console Watch

Welcome back to Next-gen Console Watch 2020, our new show following all the news and rumors on the PlayStation 5 and Xbox Series X. Last week we got the full...

Couldn't help but laugh at Dornbush and Mcaffery trying to passively aggressively one up each other throughout the video.
 

Xeontech

Member
Oct 28, 2017
4,059
I dont get this.
1.8/1.3 = 1.38
6/4.2 = 1.42
12/9.2 = 1.3

Its the smallest difference out of those. There is no logic in comparing absolutes. That is like saying the difference between OG xbox and PS2 GPUs is nothing because its only 14gflop difference when its actually 3 times higher
No no no you're looking at it all wrong.

1.8 - 1.3 = .5 million game cubes duct taped together

6 - 4.2 = 1.8 million game cubes duct taped together

12 - 9.2 = 2.8 million game cubes duct taped together

The latter is clearly the largest wall of game cubes, which is all we care about.
 

Transistor

Hollowly Brittle
Administrator
Oct 25, 2017
37,193
Washington, D.C.
I see this more of a real world testing for launch units (reliability and such) than them making a few extra money. And well, having the chance for impatient to buy them.
Like I said though, the PR would be a disaster for them. Plus no guarantee the competitor would do it too, so even if it's a limited release, keeping the same price is key
 

Bazz Bazz

Member
Feb 5, 2020
140
I think E0 was a mistake. Oberon C0 is where we are at as of December 2019 (at least what Komachi report). I made the case for it early yesterday. See below:

Well, now I am of the opinion that we are at Oberon C0 and not E0. So maybe RGT has a good source.

Komachi lists CPU ID
[Family, Model, Stepping Identifiers]

ARL A0: 00840F00 (actual 0x840f00 - 10000100 0000 1111 0000 0000)
ARL E0: 00840F40 (actual 0x840f40 - 10000100 0000 1111 0100 0000)

The last two digits represent the revision (stepping + base model). We are at base model 4 (n+1) makes revision 5 hence E0.
We know that Ariel went through A0, B0, and also saw record of Oberon A0, B0.

However, the CPU iD does not explicitly give a name for an iGPU, but Ariel and Oberon do appear to share the same base family and extended family.

So when Komachi is listing the GPU name alongside the CPU-ID , he is simply guessing or going off known associations.



10000100 0000 1111 0100 0000
Stepping = Bits 3-0 = 0000 = 0
BaseModel = Bits 7-4 = 0100 = 4
BaseFamily = Bits 11-8 = 1111 = 15
Reserved = Bits 18-12 = 0000 = 0
ExtModel = Bits 19-16 = 0100 = 4
ExtFamily = Bits 27-20 = 1000100 = 4096/512 (reset 08h) = 8

Family = ExtFamily + BaseFamily = 15 + 8 = 23 = 0x17
Model = Concat BaseModel & ExtModel = 0x44
Revision = Convert BaseModel to letter (0=A) = E + Stepping (0) = E0


III-V, really interesting.

Same as this tweet:
twitter.com

Randy on Twitter

“@XcloudTimdog @LiquidTitan @RedGamingTech Arden data in Github has gfxclk. People missed it but Arden data in Github is 100% 3584 shader cores (/64 = 56CU) 1675MHz gfxclk 12TF 560GB/s BW (320bit bus 14Gbps)”

Is it your account ? haha

Edit: updated / confusion.
 

Expy

Member
Oct 26, 2017
9,866
III-V, really interesting.

Same as this tweet:
twitter.com

Randy on Twitter

“@XcloudTimdog @LiquidTitan @RedGamingTech Arden data in Github has gfxclk. People missed it but Arden data in Github is 100% 3584 shader cores (/64 = 56CU) 1675MHz gfxclk 12TF 560GB/s BW (320bit bus 14Gbps)”

Is it your account ? haha.=

As you have access to the data, what do you think about this one:
twitter.com

Randy on Twitter

“@XcloudTimdog @LiquidTitan @RedGamingTech Arden data in Github has gfxclk. People missed it but Arden data in Github is 100% 3584 shader cores (/64 = 56CU) 1675MHz gfxclk 12TF 560GB/s BW (320bit bus 14Gbps)”

For him, it looks like Arden has some references to 3584 shader cores.
Look at who this guy is @ing...
 

III-V

Member
Oct 25, 2017
18,827
III-V, really interesting.

Same as this tweet:
twitter.com

Randy on Twitter

“@XcloudTimdog @LiquidTitan @RedGamingTech Arden data in Github has gfxclk. People missed it but Arden data in Github is 100% 3584 shader cores (/64 = 56CU) 1675MHz gfxclk 12TF 560GB/s BW (320bit bus 14Gbps)”

Is it your account ? haha.=

As you have access to the data, what do you think about this one:
twitter.com

Randy on Twitter

“@XcloudTimdog @LiquidTitan @RedGamingTech Arden data in Github has gfxclk. People missed it but Arden data in Github is 100% 3584 shader cores (/64 = 56CU) 1675MHz gfxclk 12TF 560GB/s BW (320bit bus 14Gbps)”

For him, it looks like Arden has some references to 3584 shader cores.
Not me, but yeah, we deduced that long ago. Arden has a bunch of theoretical values. Nothing new there.

If you meant to post his comment to Komachi, then yes, I saw that the other day when BlacknGoldBlood asked me to look back at Oberon E0. I cross referenced his work with what sangreal had posted (showed me how to read CPU-ID) and it all checks out.

That tweet was in March. All that was derived from GitHub leaks back in December? November? Not sure.

DrKeo thinks that is the memory bandwidth.
 
Last edited:
Oct 27, 2017
4,645
Oh, i don't think it'll happen, but if only a few thousand units could be available, consoles could be released next fall, but with a premium price, and in very limited quantities, and do massive launch with standard price next year
Its not RRP/SRP you'd have to worry about in that scenario, its the (inevitable) fucking scalpers.
 

anexanhume

Member
Oct 25, 2017
12,914
Maryland
One thing worth pointing out is that if big Navi is indeed 505 mm^2 for 80 CU in RDNA 2, we need to go back and revisit what 60 CUs with Zen 2 should measure as (XSX). Jim noted this in his video, but it sounds as if RT and the other enhancements are definitely adding area.

Liabe Brave are you looking at this?
 

Nayeon

Member
Oct 29, 2017
329
Looks like we're getting back to talking about Github. Here we go again, let the pendulum swing back to 9.2 TF. And here I thought we moved past this after the AMD confirmation of RDNA 2 for PS5.
 

klik

Banned
Apr 4, 2018
873
Well there was almost 40% difference between PS4 and Xbone GPU so i guess everything is possible
 

III-V

Member
Oct 25, 2017
18,827
One thing worth pointing out is that if big Navi is indeed 505 mm^2 for 80 CU in RDNA 2, we need to go back and revisit what 60 CUs with Zen 2 should measure as (XSX). Jim noted this in his video, but it sounds as if RT and the other enhancements are definitely adding area.

Liabe Brave are you looking at this?
yeah it is clear we are not seeing the density improvements expected from N7+. RDNA2 is on 7NP.
 

Brees2Thomas

Member
Dec 27, 2019
1,525
Yeah, he has some conclusions I am in agreement with. But not all.
I just had a little back and forth with him on Twitter. He says there is nothing out there showing Oberon could be Navi 10 lite and he thinks Oberon is the final chip and was RDNA2 from the start. But I'm sure you already knew that. 😀
 

chris 1515

Member
Oct 27, 2017
7,074
Barcelona Spain
Not me, but yeah, we deduced that long ago. Arden has a bunch of theoretical values. Nothing new there.

If you meant to post his comment to Komachi, then yes, I saw that the other day when BlacknGoldBlood asked me to look back at Oberon E0. I cross referenced his work with what sangreal had posted (showed me how to read CPU-ID) and it all checks out.

That tweet was in March. All that was derived from GitHub leaks back in December? November? Not sure.

DrKeo thinks that is the memory bandwidth.

And DrKeo conclusion is the right one, this is data showing memory clock not GFCLK. There is GB/s and reading the feature this isnot the frequency of the GPU

ESG_AkqWoAEVOsg
 

Deleted member 12635

User requested account closure
Banned
Oct 27, 2017
6,198
Germany
Not me, but yeah, we deduced that long ago. Arden has a bunch of theoretical values. Nothing new there.

That tweet was in March. All that was derived from GitHub leaks back in December? November? Not sure.

DrKeo thinks that is the memory bandwidth.
You asked DrKeo but let me chime in, if it is allowed.

I have that option in my prediction since Dec as achieving the bandwidth mentioned in the Github leak would require to use 18Gbps GDDR6 modules or to heavily overclocked 16Gbps GDDR6 modules if the SOC would use a 256-bit memory bus.

DukeBlueBall also said more than once to me that he derived a 320-bit bus from the Project Scarlett video. DrKeo also asked me to make the change to a 320-bit bus which would allow for 14Gbps GDDR6 chips.

I rejected the idea several times because I thought 320-bit is a weird config but Fafalada stated into my direction that a 320-bit bus is not as uncommon as I described it in my argument that time.

I am more convinced about it now as the 14 Gbps solution is likely cheaper and the chips have a better availability than maybe 16Gbps GDDR6 and I do not even know if 18Gbps GDDR6 is even available in higher quantities.

Maybe anexanhume knows that about the 18Gbps GDDR6 memory modules.
 

anexanhume

Member
Oct 25, 2017
12,914
Maryland
The naming conventions for 7nm is so F'ing confusing.

N7+ is EUV and 7NP is supposed to be 7NM DUV+ right?
N7 - 7nm DUV
N7P - 7nm DUV with process enhancements (performance, not area)
N7+ - 7nm EUV, library incompatible with 7nm DUV (performance and area enhancements, ~18% over N7)
N6 - N7 compatible EUV (performance and area enhancements, ~15% over N7)

You asked DrKeo but let me chime in, if it is allowed.

I have that option in my prediction since Dec as achieving the bandwidth mentioned in the Github leak would require to use 18Gbps GDDR6 modules or to heavily overclocked 16Gbps GDDR6 modules if the SOC would use a 256-bit memory bus.

DukeBlueBall also said more than once to me that he derived a 320-bit bus from the Project Scarlett video. DrKeo also asked me to make the change to a 320-bit bus which would allow for 14Gbps GDDR6 chips.

I rejected the idea several times because I thought 320-bit is a weird config but Fafalada stated into my direction that a 320-bit bus is not as uncommon as I described it in my argument that time.

I am more convinced about it now as the 14 Gbps solution is likely cheaper and the chips have a better availability than maybe 16Gbps GDDR6 and I do not even know if 18Gbps GDDR6 is even available in higher quantities.

Maybe anexanhume knows that about the 18Gbps GDDR6 memory modules.

The problem with 18 Gbps modules is that they're going to be the cream of the crop. You don't want those for a box you're trying to sell tens of millions of. AMD recently made a mistake where they upclocked memory modules after designs were finalized and put AIB partners in a bind because they hadn't validated the cards at those memory speeds, meaning they could crash. It's not a trivial thing to hit these speeds.
 

MrDeveus

Member
Apr 26, 2019
833
Like I said though, the PR would be a disaster for them. Plus no guarantee the competitor would do it too, so even if it's a limited release, keeping the same price is key

Transistor, is it possible to do a broad style bet like for instance. I bet Series S gets announced sometime this coming week, if I am wrong I will leave new avatar up for 6 months regardless of reveal?
 

III-V

Member
Oct 25, 2017
18,827
I just had a little back and forth with him on Twitter. He says there is nothing out there showing Oberon could be Navi 10 lite and he thinks Oberon is the final chip and was RDNA2 from the start. But I'm sure you already knew that. 😀
No, I do not know or talk to this person. And, like I showed yesterday, Oberon had gfx10 libraries loaded onto it. (NAVI10 - RDNA1).
I think he is a console warrior if he is acting timdawg

The naming conventions for 7nm is so F'ing confusing.

N7+ is EUV and 7NP is supposed to be 7NM DUV+ right?
N7 - 7nm DUV
N7P - 7nm DUV with process enhancements (performance, not area)
N7+ - 7nm EUV, library incompatible with 7nm DUV (performance and area enhancements, ~18% over N7)
N6 - N7 compatible EUV (performance and area enhancements, ~15% over N7)


And DrKeo conclusion is the right one, this is data showing memory clock not GFCLK. There is GB/s and reading the feature this isnot the frequency of the GPU

ESG_AkqWoAEVOsg
For intra-GPU tests, such as with Oberon, the GPU clock is used for similar tests.

You asked DrKeo but let me chime in, if it is allowed.

I have that option in my prediction since Dec as achieving the bandwidth mentioned in the Github leak would require to use 18Gbps GDDR6 modules or to heavily overclocked 16Gbps GDDR6 modules if the SOC would use a 256-bit memory bus.

DukeBlueBall also said more than once to me that he derived a 320-bit bus from the Project Scarlett video. DrKeo also asked me to make the change to a 320-bit bus which would allow for 14Gbps GDDR6 chips.

I rejected the idea several times because I thought 320-bit is a weird config but Fafalada stated into my direction that a 320-bit bus is not as uncommon as I described it in my argument that time.

I am more convinced about it now as the 14 Gbps solution is likely cheaper and the chips have a better availability than maybe 16Gbps GDDR6 and I do not even know if 18Gbps GDDR6 is even available in higher quantities.

Maybe anexanhume knows that about the 18Gbps GDDR6 memory modules.
This is one area I really do not have a firm opinion on. The data seems to allow for different conclusions. Its probably a good time to revisit as a group.
 

ImaginaShawn

Banned
Oct 27, 2017
2,532
N7 - 7nm DUV
N7P - 7nm DUV with process enhancements (performance, not area)
N7+ - 7nm EUV, library incompatible with 7nm DUV (performance and area enhancements, ~18% over N7)
N6 - N7 compatible EUV (performance and area enhancements, ~15% over N7)
So this is an EUV process that is compatible with the N7 DUV process?
 

Judau

Member
Oct 28, 2017
4,797
So, WatchMojo posted a PS5 vs XSX video earlier. It states at the beginning that it's purely based on what they know so far about both consoles before going straight into comparing console design and giving the edge to PS5, solely because they can't imagine it being "uglier" than the Series X. I didn't actually watch all of it because the next comparison point was memory, which they also gave the edge to PS5 for reasons, I guess? Honestly, I think they jumped the gun, since there's not enough info on PS5 to compare it to the Series X.
 

Deleted member 12635

User requested account closure
Banned
Oct 27, 2017
6,198
Germany
This is one area I really do not have a firm opinion on. The data seems to allow for different conclusions. Its probably a good time to revisit as a group.

See the response from Anex below!

The problem with 18 Gbps modules is that they're going to be the cream of the crop. You don't want those for a box you're trying to sell tens of millions of. AMD recently made a mistake where they upclocked memory modules after designs were finalized and put AIB partners in a bind because they hadn't validated the cards at those memory speeds, meaning they could crash. It's not a trivial thing to hit these speeds.
Appreciated. I agree on the risks it would bring in.

cc DrKeo
 

chris 1515

Member
Oct 27, 2017
7,074
Barcelona Spain
So, WatchMojo posted a PS5 vs XSX video earlier. It states at the beginning that it's purely based on what they know so far about both consoles before going straight into comparing console design and giving the edge to PS5, solely because they can't imagine it being "uglier" than the Series X. I didn't actually watch all of it because the next comparison point was memory, which they also gave the edge to PS5 for reasons, I guess? Honestly, I think they jumped the gun, since there's not enough info on PS5 to compare it to the Series X.

lol compare a console we don't know the design and the specs.
 

score01

Member
Oct 27, 2017
3,701
If the quote above is the one people keep using for Jason, then he did not say they were close in power. Specifically gpu power. The exact wording was "similar specs".

That can mean a whooole lot. Both have SSDs, similar processor, RT, VRR, etc.

Jason also said that There won't be straight answers as to which console is most powerful for a while, if there ever are

This can only be true if the console are within the same ball park TF wise. No amount if RAM, bandwidth, SSD speed can make up for a ~3TF GPU difference.

In any case, we seem to be going round in circles here. Lines have been drawn and it seems that people have already formed their conclusions and set them in stone. Sony need to reveal their hand, whether it's 9TF or 12TF and end this madness... though one side should definitely be preparing to eat crow 😂
 

AegonSnake

Banned
Oct 25, 2017
9,566
Cant believe we still haven't got a word from Sony. Ppl Shitting bricks in here
Imagine having to come up with a marketing strategy to sell a 9 tflops console.

I bet they are shitting bricks too.

Either that or they got a 13 tflops monster like our prophet ruthetentic cookie said and they are so confident in their console they know it's gonna sell no matter what the competition does.
 

DrDeckard

Banned
Oct 25, 2017
8,109
UK
Sony are going to announce and not say shit about the teraflops, aren't they? Make everyone lose their minds.
 
Status
Not open for further replies.