• 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.
Status
Not open for further replies.

JahIthBer

Member
Jan 27, 2018
10,382
Oh I immediately watched the entire video sir. I love how in depth and nuanced the guys at DF approach tech discussions. Best in the business no debate, in my view.

The raw data just lines up two perfectly when you look at actual benchmarks, like the ones I posted. In that video the author tested 53 games with both the RX 580 and 5700 XT paired with a 9900K, and it was hitting 2x performance on the nose way too often.

Why would Microsoft explicitly state 2x the power if they were actually at 2.25x or slightly more? The 2.25~2.3x figure is where 12TF would put them, when a 5700 XT is 2x at 9.75 TF.

The TLDR is that 12TF RDNA is way beyond 2x the GPU power of the Xbox One X. Microsoft would tout that loudly.
Im not sure how literal we should take the 2x, it might just be "yeah my engineers told me it's like 2x as powerful, pretty good huh?"" people don'tt really casually say "oh it's 2.25x faster"
But if it's truly a 5700 XT, 2x does seem to match up almost too well, since even if you count the larger bus width & likely faster GDDR6 that the SX provides over the 5700 XT, you have to couple that with lower clocks due to console thermal limits (say 1.9 down to 1.8) either way your ending up with 2x over the XBX.
 

bcatwilly

Member
Oct 27, 2017
2,483
Digital Foundry is just doing their job and remaining skeptical of these consoles in terms of thermals and everything as they should until we have final specs reveal for both. However, the one thing that I think they are not paying enough attention to here based on the interview information from Phil Spencer and Jason Ronald is that the form factor of the Xbox Series X was absolutely developed per direction to the engineering team to target a high performance console without regard for typical thoughts on design constraints or aesthetics. So their design relative to cooling and such may permit them to achieve things that currently don't make sense based on console history.
 

Deleted member 12635

User requested account closure
Banned
Oct 27, 2017
6,198
Germany
Digital Foundry is just doing their job and remaining skeptical of these consoles in terms of thermals and everything as they should until we have final specs reveal for both. However, the one thing that I think they are not paying enough attention to here based on the interview information from Phil Spencer and Jason Ronald is that the form factor of the Xbox Series X was absolutely developed per direction to the engineering team to target a high performance console without regard for typical thoughts on design constraints or aesthetics. So their design relative to cooling and such may permit them to achieve things that currently don't make sense based on console history.
Hey, I have that job too to be sceptical ;)
 

Ra

Rap Genius
Moderator
Oct 27, 2017
12,207
Dark Space
Im not sure how literal we should take the 2x, it might just be "yeah my engineers told me it's like 2x as powerful, pretty good huh?"" people don'tt really casually say "oh it's 2.25x faster"
But if it's truly a 5700 XT, 2x does seem to match up almost too well, since even if you count the larger bus width & likely faster GDDR6 that the SX provides over the 5700 XT, you have to couple that with lower clocks due to console thermal limits (say 1.9 down to 1.8) either way your ending up with 2x over the XBX.
Yeah see Microsoft is now saying twice the GPU power, while people have been stuck talking in twice the Teraflops. It's an entirely different discussion.

What is providing twice the GPU power throughput at a 4K resolution? Literally the 5700 XT.

Now, it is going to be a custom part and not just an off the shelf 5700 XT, don't misunderstand me. It's going to be a more mature 7nm process that provides benefits on which I can't speak. As has been the case in the last couple generations, the GPUs will have features of the next-generation GPU, like the ray-traycing.

It makes sense.
 

MrKlaw

Member
Oct 25, 2017
33,059
Yeah see Microsoft is now saying twice the GPU power, while people have been stuck talking in twice the Teraflops. It's an entirely different discussion now.

What is providing twice the GPU power at a 4K resolution? Literally the 5700 XT.

Now, it is going to be a custom part and not just an off the shelf 5700 XT, don't misunderstand me. It's going to be a more mature 7nm process that provides benefits on which I can't speak. As has been the case in the last couple generations, the GPUs will have features of the next-generation GPU, like the ray-traycing.

It makes sense.

it does.

but also makes sense to say twice the power to mean twice the TF because 6->12 despite the architectural changes

depends who MS is communicating to and what they're trying to get across

your view and measurements are very logical and I think fits well with what we might get - but too many unknowns to be completely sure. And of course people will always push to believe the biggest number
 

AegonSnake

Banned
Oct 25, 2017
9,566
DF has been underestimating this gen since the wired reveal. Go back and watch their comments on RT even on videos after Sony said its hardware accelerated six months later.

I guess they are being skeptical of 12 Navi tflops but whatever. This just tells me they don't know any more than we do. I expected df to be more knowledgeable than the gcn tflops brigade.

Disappointed to say the least.
 

MrKlaw

Member
Oct 25, 2017
33,059
5700XT effective performance (more CUs and lower clock speed), plus RDNA2 features like RT and VRS I think is a sensible, pragmatic option

will likely punch above a 5700XT if VRS is utilised we'll, and RT will mean it's more feature rich too
 

anexanhume

Member
Oct 25, 2017
12,913
Maryland
3D printed mockup

rgpqrme5nm441.jpg
 

modiz

Member
Oct 8, 2018
17,844
Oh I immediately watched the entire video sir. I love how in depth and nuanced the guys at DF approach tech discussions. Best in the business no debate, in my view.

The raw data just lines up two perfectly when you look at actual benchmarks, like the ones I posted. In that video the author tested 53 games with both the RX 580 and 5700 XT paired with a 9900K, and it was hitting 2x performance on the nose way too often.

Why would Microsoft explicitly state 2x the power if they were actually at 2.25x or slightly more? The 2.25~2.3x figure is where 12TF would put them, when a 5700 XT is 2x at 9.75 TF.

The TLDR is that 12TF RDNA is way beyond 2x the GPU power of the Xbox One X. Microsoft would tout that loudly.
12TF RDNA is closer to 2.6x higher than X1X actually.
 

Deleted member 1589

User requested account closure
Banned
Oct 25, 2017
8,576
DF has been underestimating this gen since the wired reveal. Go back and watch their comments on RT even on videos after Sony said its hardware accelerated six months later.

I guess they are being skeptical of 12 Navi tflops but whatever. This just tells me they don't know any more than we do. I expected df to be more knowledgeable than the gcn tflops brigade.

Disappointed to say the least.
Can't blame them.

They work with what they know. Right now we are going with good faith with nothing factual that can support our speculations.

That will hopefully change once more information about RDNA2 comes around (if it's good news). That will show if 12 teraflops for the Xbox Series X is viable or not.
 

sncvsrtoip

Banned
Apr 18, 2019
2,773
One thing to clarify. 5700xt has almost 2x more fps in 4k than rx580 but we have take into account that numbers of frames also depands on cpu calculation. So if we have 2x more fps and the same time for cpu calculation that means 5700xt is more than 2x faster than rx580. In conclusion even slightly downlocked 5700xt is good for Spencer "twice gpu performance" ;)
 

Chamber

Member
Oct 25, 2017
5,279
Y'all are looking wayyyyyyy too deep into a statement lmao. Learn to just take things at face value.
Agreed.

Allowing the press to run with the 12TF number just to be proven wrong later on would be the most unnecessary PR blunder I've seen in gaming in a long time. I'm amazed to see DF running with that speculation.
 

Deleted member 12635

User requested account closure
Banned
Oct 27, 2017
6,198
Germany
Agreed.

Allowing the press to run with the 12TF number just to be proven wrong later on would be the most unnecessary PR blunder I've seen in gaming in a long time. I'm amazed to see DF running with that speculation.
Have you actually watched the video? Because they actually don't run with that number ... they question it. Or did I just misunderstood what you wanted to say?
 

III-V

Member
Oct 25, 2017
18,827
DF has been underestimating this gen since the wired reveal. Go back and watch their comments on RT even on videos after Sony said its hardware accelerated six months later.

I guess they are being skeptical of 12 Navi tflops but whatever. This just tells me they don't know any more than we do. I expected df to be more knowledgeable than the gcn tflops brigade.

Disappointed to say the least.

giphy.gif
 

sleepr

Banned for misusing pronouns feature
Banned
Oct 30, 2017
2,965
Guys it's 12 TF RDNA, it would make no sense otherwise. The box design even hints at this thing running hot as F, it screams "high-tech".
 

Deleted member 1589

User requested account closure
Banned
Oct 25, 2017
8,576
Yeah I've been wondering if there's actually split chips inside that console than one single monolith.

duolith?
 

Pharaoh

Unshakable Resolve
Member
Oct 27, 2017
2,675
DF has been underestimating this gen since the wired reveal. Go back and watch their comments on RT even on videos after Sony said its hardware accelerated six months later.

I guess they are being skeptical of 12 Navi tflops but whatever. This just tells me they don't know any more than we do. I expected df to be more knowledgeable than the gcn tflops brigade.

Disappointed to say the least.

Yeah, I've been avoiding DF more and more lately. But I love John Linneman and still watch his content.
 

Deleted member 10747

User requested account closure
Banned
Oct 27, 2017
1,259
Agreed.

Allowing the press to run with the 12TF number just to be proven wrong later on would be the most unnecessary PR blunder I've seen in gaming in a long time. I'm amazed to see DF running with that speculation.
This is also why i believe the 12tf. Would they really risk it again to explain every little thing?
 
Jul 6, 2018
174
Man 2020 is gonna be one of the best year in gaming
What? Because of Gears Tactics, FFVII Remake Prologue, Last of Us 2, and Ghost of Tsushima?

New consoles never launch with a great lineup. You might get one amazing game per new console if you are lucky: e.g. Halo: CE, Zelda: BoTW, or Altered Beast. But there's no way you're getting awesome lineups from Series X or PS5 at launch. That's just crazy talk.
 

JahIthBer

Member
Jan 27, 2018
10,382
DF has been underestimating this gen since the wired reveal. Go back and watch their comments on RT even on videos after Sony said its hardware accelerated six months later.

I guess they are being skeptical of 12 Navi tflops but whatever. This just tells me they don't know any more than we do. I expected df to be more knowledgeable than the gcn tflops brigade.

Disappointed to say the least.
A lot of things have changed, AMD have confirmed HW RT in RDNA 2 & $499 for Consoles seems set in stone, im pretty sure both PS5 & SX are using 40+ CU GPU's since they are coming out in late 2020 after all, but not 100%.
I think it's fine to be a bit conservative.
 

Fafalada

Member
Oct 27, 2017
3,066
To put it another way, if we're talking pr. What sounds better? "Twice the power" or "nearly three times the power"?
Inflating the power in non-numerical ways could take you WAY higher than 3x - just add VRS and RT and even 12 'GCN' TF would run you a much larger multiple of X1X regardless. The problem is there's pretty much no upper-bound how far you can take this(it is after all, actually moving the goal-posts), and competition can do it too, ending up in PR contest no customer can keep up with pretty quickly.

Spec inflation didn't really do 360 or PS3 any favours, and it didn't hurt PS4 or X1 to come out with actual "lower" official GFlops announcements than their predecessors.
 
Oct 25, 2017
17,904
Agreed.

Allowing the press to run with the 12TF number just to be proven wrong later on would be the most unnecessary PR blunder I've seen in gaming in a long time. I'm amazed to see DF running with that speculation.
This is also why i believe the 12tf. Would they really risk it again to explain every little thing?
It wouldn't be much of a blunder at all. MS only gave multipliers and those are true. The conclusions drawn from that are on the people who decided to do that. Besides, the vast majority of people have no clue what a TF is anyway. This would have no bearing on anything.

The 'strongest console' thing is protected by how they've worded that as well.
 
Jul 6, 2018
174
Yeah, I've been avoiding DF more and more lately. But I love John Linneman and still watch his content.
If Microsoft wasn't factoring in IPC increases from architectural changes then they couldn't claim that the 8 core Zen 2 in the Series X CPU is 4 or 5 times more powerful than the One X.

Why would they be focused on a metric like tflops for their GPU comparison when they ignore it for their CPU comparison?
 

Expy

Member
Oct 26, 2017
9,865
With all these people apparently having sources about next-gen console specifications and performance targets, I'm surprised DF doesn't seem to have any clue about these. Kind of puts into question what kind of sources they have, they must not be very close to the developers and/or manufacturers.

Of course, this is assuming they don't already know and just are pretending not to know until their NDA expires and they can talk about it.
 

anexanhume

Member
Oct 25, 2017
12,913
Maryland
Yeah I've been wondering if there's actually split chips inside that console than one single monolith.

duolith?

Split discrete packages is somewhat antithetical to unified memory. Perhaps chiplets on same package, but 0.01% chance they're discrete packages.

Inflating the power in non-numerical ways could take you WAY higher than 3x - just add VRS and RT and even 12 'GCN' TF would run you a much larger multiple of X1X regardless. The problem is there's pretty much no upper-bound how far you can take this(it is after all, actually moving the goal-posts), and competition can do it too, ending up in PR contest no customer can keep up with pretty quickly.

Spec inflation didn't really do 360 or PS3 any favours, and it didn't hurt PS4 or X1 to come out with actual "lower" official GFlops announcements than their predecessors.

Preach.
 

Chamber

Member
Oct 25, 2017
5,279
It wouldn't be much of a blunder at all. MS only gave multipliers and those are true. The conclusions drawn from that are on the people who decided to do that. Besides, the vast majority of people have no clue what a TF is anyway. This would have no bearing on anything.
That's why I think it's implausible though. You have multiple press outlets running with 12TF, and I'm assuming they didn't arrive at the number with napkin math, it'd be easy to just issue a correction with no harm done. It looks worse to wait 3 months and come out and say it's actually 9-10TF and they gain nothing from letting the lie sit so long.
 

JahIthBer

Member
Jan 27, 2018
10,382
That's why I think it's implausible though. You have multiple press outlets running with 12TF, and I'm assuming they didn't arrive at the number with napkin math, it'd be easy to just issue a correction with no harm done. It looks worse to wait 3 months and come out and say it's actually 9-10TF and they gain nothing from letting the lie sit so long.
Well the 5700 XT is Vega 64 level, which is around 12 TFLOPS. There isn't really a lie here. No one has officially said RDNA TFLOPS, just insiders.
 
Status
Not open for further replies.