bitbydeath
Member
Well they obviously had to wait for the hardware to become available before it could be used.
-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
- Status
Well they obviously had to wait for the hardware to become available before it could be used.
CrustyBritches
Gold Member
1. So what? DF has has already confirmed this.
2. I claimed that under NEO Overview Dev Kit 2 was "Final Spec", and the time from Dev Kit 2 to Test Kit 2 was under 3 months. Then you confirmed that with your own links that you didn't read. I never made any claim regarding time from Dev Kit to launch. That's your own hallucination.
---
I'm still waiting for you to post even a single data point that confirms your claims about consoles. P.S.- It won't happen.
bitbydeath
Member
1.Prove it. It must be them speaking with Sony and not a wink wink take our word for it confirmation.
2. You seem to be re-writing history here. I’ll let you just have the last word because I can’t be bothered shifting back to where you thought the hardware was final and wouldn’t change from 36-40CU.
The links I provided a few posts up.
Gonzalo was a qualification sample - final design. So they will now redesign a chip for the magic RDNA2 RT 1 year until console release? If there's any hardware for RT acceleration it's already in and there's no need to wait for some big RDNA2.
Last edited:
bitbydeath
Member
As linked above the Pro didn’t receive its final dev kit until 3 months out from launch so there is plenty of time in comparison.
Final dev kit doesn't mean the APU will be suddenly changed. It means final clocks, final memory config, storage size etc.
Last edited:
bitbydeath
Member
Yeah, they’re probably already running RDNA2 with Prospero given the noise from that.
What noise? Tales from someone's ass? As I have repeated many times there's no proof that RDNA can't be adjusted for RT hardware with the chip ending something between RDNA and RDNA2. We don't even know if the AMD's solution in RDNA2 will be used for ray tracing in playstation chip. Maybe they designed something a bit different.
If anything, Prospero only confirms the connection with Gonzalo, Flute, Oberon and not some RDNA2 big Navi chip.
Last edited:
SlimySnake
Flashless at the Golden Globes
While i agree that Gonzalo leaks are tied to data i wouldnt read too much into them. These are just tests that got through the cracks. we went from 1 Ghz to 1.8 ghz in three months and then 2.0 ghz three months later. So at the very least, we know the chip is still being worked on and tweaked. I distinctly remember DF and Apisak saying that the second gonzalo chip was in qualifying which means it was closed to being ready for production, and yet three months later we got oberon which had an even higher clock.
Another problem with following data leaks like these is that you keep switching expectations every few months. When the 1 GHz data leak came out, everyone assumed a 64 CU part. That got completely disregarded after the 1.8 ghz leak. So the first data point was essentially BS. useless data point that set up wrong expectations.
Meanwhile, WCCFTECH had predicted that the PS5 will have a big navi two months prior to the first Gonzalo leak. He specifically said that the 40 CU Navi will NOT be the one that gets featured in next gen consoles. That was in November. In June of 2018, literally a year and a half ago, he said Navi 10 would be a small GPU, not a high end one and would launch in 2019. He also said the PS5 would release in holiday 2020. not to mention the fact that he was the first person to say that Navi would be a new architecture when everyone assumed it would be GCN. He said this a year before the actual announcement. In fact, he mentioned big Navi in June of 2018 so you are wrong about komachi beating him to it. the only thing he was wrong about was that vega 7 would not get released but things change so he might have been right about it 8 months before it actually launched.
So yeah, they were wrong about 1 thing out of the 7 things they posted. That's not 75%.
So we have a source who has been consistent about big Navi for over a year now vs some data miners who find something new every few months and then refuse to answer questions about it. I asked Apisak what the actual Gonzalo score was multiple times and he didnt reply. he ignores 90% of the questions about gonzalo. he gets like two comments per tweet and didnt bother replying to any comments asking for more detail. Komachi actually threw a full on fit when people asked him to clarify the Oberon leak to see if he could see CU counts. He then posted some bizarre chat that shows that the ps5 apu is 300mm2. we dont know that guy in the chat. we dont know his sources either. but you trust him. fine. I trust their numbers too but they are very fluid and i wouldnt put too much stock in them, especially when we have websites with a proven track record saying the complete opposite.
WCCFTECH is not the only one saying the PS5 and Scarlett will use a bigger GPU. kleegamefan said that the next gen consoles are using gpus that have not yet been released by AMD and they will have double digit tflops. Jason Schrier said they are both aiming to beat Stadia's 10.7 tflops number. Reiner and Colin Moriarty both said PS5 is more powerful than MS's most expensive SKU Anaconda. Why would MS go with an 8 tflops GPU as its benchmark console?
You are also ignoring the fact that the 5700 series does not support RT. It simply cannot be the GPU that goes in there. It's entirely possible that we end up getting 36 CU Navi but it will not be the 5700 series Navi 10 GPU as we know it. I agree that thermals dont add up, but you are also working backwards to prove the Gonazlo and Oberon clocks. You are ignoring AMD's own roadmaps that show 7nm+ RDNA 2.0 GPUs that feature ray tracing. You are ignoring how the 2.0 36-40CU 5700 GPU would be a 200-235w on 7nm which isnt particularly realistic either. You are ignoring 90% of leaks from journalists with track records in favor of two data miners who hide and throw tantrums when asked to provide more details than the cryptic shit they provide. wtf does 20000+ mean anyway? why not give an actual number.
Sosokrates
Report me if I continue to console war
It was a bit unclear his tweets were, Albert penello tweeted to him that Dev kits can change a lot, then reiner replied saying its based of target specs, but they can change to, so penello point still stands.
Sosokrates
Report me if I continue to console war
What will be interesting to see where the consoles land in resolution and framerate differences.
This gen quite a few devs aimed for higher then ps4 specs to do 1080p.
What if 3rd parties are aiming for 4k 30/60fps on stadia specs?
If that is the case, then the differences between scarlett and ps5 will be less important if they are both more powerful then the 4k "base" stadia spec.
This gen quite a few devs aimed for higher then ps4 specs to do 1080p.
What if 3rd parties are aiming for 4k 30/60fps on stadia specs?
If that is the case, then the differences between scarlett and ps5 will be less important if they are both more powerful then the 4k "base" stadia spec.
CrustyBritches
Gold Member
DF already confirmed Gonzalo is PS5's SoC. I'm sure you'd like not to read much into that.
Strawman, you're not talking about Apisak or me. I've been consistent with estimates based on power draw and historical precedence.
Gave their source credit for the 40CU already.
Wccftech's source for PS5 "Big GPU" = their own speculation until stealth updated again. No doubt he has a occasionally reliable desktop GPU source.
Irrelevant.
All rumors with no specs/data tied. Just like wccftech did not cite their source, but stated personal speculation for PS5. Benji's "13TF GPU", later to be changed to support DF's Gonzalo build is a prime example of that pitfall. I don't discount possibility of double-digit consoles either.
Where are you reading this? Only known data for Scarlett is the 350mm²-380mm² die size and use of mixed 14Gbps memory in reveal render.
This is my argument. I'm only concerned with 2 things: 1. Overall performance relative to best 5700XT-based PC 2. Number of CUs being 40 or under. Not arguing it's series. This is literally my argument, but add 40CU possibility and RT.
What you say here doesn't make sense. That number is Fire Strike Overall score when paired with Flute-level CPU. This was made clear he could not see Graphics score, only Overall. 36-40CU Navi GPU+Ryzen 5 1600/2600 = 20K+ Overall score across hundreds of results. 60-64CU+Flute-level 1700x/Ryzen 2600 CPU doesn't = 20K+ overall score.*edit*To clarify 60-64CU score will be more like 23-24K overall with Flute-level CPU, so 23K+.
---
1. Do you believe Gonzalo/Oberon/Prospero is PS5 like DF says?
2. I gave wccftech credit where credit is due.
3. What about a 20K overall score confuses you?
4. We agree that future 36-40CU GPU with RT is possible.
5. We agree on console power consumption limits and thermals are an issue, why go up to 60/64CU if 40CU is an issue?
*edit* 6. Do you believe Codemasters have had the Dev Kits since August, as stated by their dev?
Last edited:
Sosokrates
Report me if I continue to console war
I think they will both be more then 36cus,
I think scarlett with have 48 (44 active) and ps5 will have a similar amount.
I think scarlett with have 48 (44 active) and ps5 will have a similar amount.
SlimySnake
Flashless at the Golden Globes
Not sure why you are putting so much faith in DF. They know nothing more than we do. Richard literally had to talk to Apisak to get his info. He knows just as much about Gonazlo as we do. Richard has talked to devs who said they were happy with the PS5 and yet he couldnt even get a tflops number out of them. DF is as useless as you or I. lets not use them as this word of god that confirms everything. they couldnt even confirm a tflops number or who was more powerful after talking to devs like Reiner or Colin did who arent even tech focused like Richard is. Like i said, utterly useless.
I do believe that Gonzalo, Flute and Oberon are all PS5 chips. But what we do not know from these data mining is the CU count. You are looking at ONE Gonzalo benchmark and using it to base off your entire prediction. And worse, dismissing almost half a dozen reporters with their own sources in the dev industry.
How is pointing out the fact that the first gonzalo leak was 1 ghz strawman? It literally establishes the fact that these specs are fluid. You are basing your entire prediction on gonzalo's test and yet you have no idea which gonzalo version was used for the test. Was it the 1ghz one? Was it 1.8 ghz? Was it something in between that wasnt caught by Apisak? It definitely wasnt the 2.0 ghz oberon. So the test is already out of date.
The 20k+ number is confusing because it could be 21K or 22K or 20.9k. I asked him for the real number. I am not asking for the graphics score, just the actual number. Why give us 20k+ and not the actual number? BTW, my rtx 2080 and i7 8700 at 3.2 ghz gave me an overall score of 21.5k. That doesnt really tell us much.
Your data might be reliable but its incomplete and out of date. For instance, lets look at this tweet.
if the 5700 series is Navi 10 and the PS5 IS NOT based on the 5700 series like you admitted then why is Gonzalo here Navi 10 LITE? Navi 10 doesnt even have RT.
Yes, we agree on several different things. Gonzalo is PS5. 36-40 CU is a possibility. Thermals are an issue. What we dont agree with is how you are dismissing and mocking ALL speculation that does not follow these incomplete data points. Data that keeps changing every few months. Data that simply doesnt jive with reality. 2.0 ghz clocks on a 7nm process in a console? How? Navi 10 LITE with RT? So Navi 10 LITE can have RT but not more CUs?
Again, if you want to go by data only, thats fine by me. But this whole shtick of being a data nazi and shutting down posters like bitbydeath is bizarre just because he chooses to believe actual journalists over some twitter drama queens who post random stuff without context. You talk about historical precedence but when we try to use it, you ask for data knowing full well we have none and neither do you when it comes to thermals or CU counts and yet it hasnt stopped you from using historical precedence to speculate.
Here are some historical precedents.
- PS4, PS4 Pro and X1X all used the x800 lineup AMD had available at the time. Shouldnt historical precedent suggest they would go wit the 5800 fat die?
- PS4 Pro and X1X had MORE CUs than the 480 and 580 by 4 and 8 respectively. They went wide and slow because of thermals. Why would they suddenly shift their strategy now?
- Prior to PS4, sony had no problems going with dies 400-500mm2. They took a big loss on consoles because they could.
- MS went with a 7 billion transistor die on a 16nm . If silicon cost is so important, number of transistors on a die is important. 5700xt is 10 billion transistors on 7nm. Zen 2 is 2 billion. Cut down Zen 2 should be under 1 billion. 16nm to 7nm should allow them to increase transistors by a much bigger amount. Ignoring the X1 APU which had a massive ESRAM die inflating the transistor count, the PS4's GPU transistor count was 2.8 billion transistors, 36 CU polaris was 5.7 billion transistors. So if they can make a massive 2x jump when going from 28nm to 16nm, why would they settle for a 1.5x jump now that they have 2.3x more space on the die? nearly 3x if you count 7nm+ improvements. Again, historical precedent suggests a much bigger jump.
As for how they can get 56 active CUs in a console form factor. At the risk of getting dismissed because of my lack of data, allow me to speculate:
- 7nm+ gives us a 20% size reduction plus 10% in power savings. So if a 60 CU part is around 360mm2, it will be 300mm2 on 7nm+ with much better thermals. This is taken straight from AMD's own roadmaps. RDNA 2 GPUs = 7nm+.
- HBM2 - Consumes much less power than GDDR6, takes up much less place on the die. Smaller die, less power usage, better overall thermals.
- Sony's Heatsink Patent - sits under the apu and cools both sides.
- Cut down CPU as the flute leak suggests (hey im using data!) reducing size even further by around 30 mm2 since each 8MB L3 cache seems to take up 10mm2. Again, smaller die, less power usage, better overall thermals.
- RDNA 2.0 efficiencies. RDNA 2.0 gpus should be more power efficient compared to RDNA 1.0 gpus.
CrustyBritches
Gold Member
"I do believe that Gonzalo, Flute and Oberon are all PS5 chips." You are believing them, and bizarrely inserting yourself into this, while downplaying your own sources...Apisak, DF. DF confirmed independently with their own source that Prospero is the PS5 codename.
Highly unlikely. This is why he said "Gonzalo - score 20000 up"...many scores. He is not a primary English speaker, don't expect perfect phrasing. I only care to establish the lowest possible combos for 20k+ Overall = either 5700+2600 or 5700XT+1600 as baseline.
I use a mix of Fire Strike score, Apisak saying "5700XT+3700X = Next-gen console simulator", Reddit OQA PCB die size, Digital Foundry's test system, 5700XT power consumption, Benji's almost 13TF statement for a likely pre-Navi dev kit delete and switchover, and even Jason Schreier's own comments which basically confirmed they weren't based on Gonzalo Dev Kits.
The 1GHz speed was gone by April 10, 2019 and replace by 1.8GHz speed 13F8 chip. Test scores coming in June 10, 2019. 20K+ Scores reveal June 25, 2019. Flute tested with 1.8GHz core on July 21, 2019. It's possilbe Oberon is the other chip mentioned as ES2G, but 200MHz is neither here nor there for FS Overall score.
It's a batch of scores, starting at baseline 20K, hence "20K up". 20000 is an actual number.
Irrelevent and a lie. I go hard at bitbydeath because that's how he came at me(earlier this year). We have a bet and get along fine. Who is "we", when you mean you? He can speak for himself.
Nice slander. My interaction with bitbydeath is always like this, and none of your fucking business
He's based everything on insider leaks, which aren't as airtight as the 60/64CU crowd would suggest:
Jason Schreier didn't really say what he's been characterized as saying, and he was pissed:
"I don't now specs, don't know teraflops. A dev told me that people in studios are basing info on limited info, one of the assumptions is that everything's going to be twice as powerful". | DF's newest testbed was 3x PS4 Pro and 2x Xbox One X.
---
Benji was basing info off pre-navi dev kits, and later flipped to DF's stance, deleted his tweet, deleted his Reee- account said something like, "welp, learn my lesson won't do that again":
^Newest DF video from October 19th, 2019 based on 5700 and 5700 XT.
---
Matt from Resetera flip-flopped and "guessed" that Xbox will be more powerful:
---
Klee stated people won't be impressed by specs as games. Simply stated double-digit TF, which I don't disagree with.
---
These were the basis of the 60CU, ~13TF rumors. Seems highly probable by the Schreier and Benji comments that Gonzalo-based kits weren't the source for these statements. Concerning my 36/40CU and 44CU estimates, they fit everyone of these claims quite nicely.
If wccftech has a source for their PS5 claim and it's solid, then I will believe it.
P.S.- I've repeated myself ad nauseum. I'm done with this thread until the specs leak or are officially announced.
Last edited:
I already wrote last week what AMDs nomenclature means.
Navi 10 means first Navi chip hitting the market.
This was ALWAYS high spec cards, but this time it was mid range card. What that means is, back in Jan when Gonzalo leaked as "Navi10_Lite" there likely was plan to have high end Navi as first one on market. This changed in next few months and Navi 10 became 5700.
What this practically means is PS5s chip was likely cut down version of full Navi 10 before plans changed. Once plans changed, Navi 10 (10 meaning first car to hit market) became 5700. This makes sense now because Navi 10 "lite" in consoles would be below 5700 level (that means 32CUs which makes 0 sense).
Navi 10 means first Navi chip hitting the market.
This was ALWAYS high spec cards, but this time it was mid range card. What that means is, back in Jan when Gonzalo leaked as "Navi10_Lite" there likely was plan to have high end Navi as first one on market. This changed in next few months and Navi 10 became 5700.
What this practically means is PS5s chip was likely cut down version of full Navi 10 before plans changed. Once plans changed, Navi 10 (10 meaning first car to hit market) became 5700. This makes sense now because Navi 10 "lite" in consoles would be below 5700 level (that means 32CUs which makes 0 sense).
Lunatic_Gamer
Gold Member
In the meantime at Sony’s headquarters things seem to be going pretty “smooth”. 
twinfinite.net
PS5 Games' Development "Proceeding Smoothly;" Sony Expects "Highly-Engaging Gameplay Experiences"
During Sony's quarterly financial conference call for investors and analysts, chief financial officer Hiroki Totoki talked about the upcoming PS5 and more.
twinfinite.net
bitbydeath
Member
The very first information we received about Prospero before it was ‘confirmed’.
Confirms the GPU is late (5800).
Just like codemasters.
And of course there is this.
A Story About Two Tiny, Fuzzy Photos Of A 'PlayStation 5 Prototype'
Short of the annual iPhone, few future devices get fans chatting like the next generation of consoles from Sony and Microsoft. Both companies have teased what’s next recently — with Sony revealing some details about the PlayStation 5 in April, and Microsoft showing off a flashy video teasing...
www.gizmodo.com.au
Now you can choose to believe Prospero or not but this was our first introduction to it being called that and a lot of the data lines up.
SlimySnake
Flashless at the Golden Globes
False. You said the same to bhunachicken in this post below.
You also said this to the entire thread when dismissing the WCCFTECH article.
And then finally this to me.
This is hilarious. 1 score = many scores? You are literally using a reddit leak yourself and chastising the entire thread for not using data points?
Your timeline doesnt take into account when the gonzalo tests were done. only when they were found by Apisak. The only test we have the date for is the flute test.
Lastly, i never mentioned benji or matt. Schrier was pissed at getting a thread made about his post, but in his original post he simply stated that both makers were aiming to be higher than the Stadia number.
You are getting way too caught up in this. This is all just speculation. Either you are right or I am. doesnt really matter. The problem is your attitude and dismissal of anyone who dares to look at anything other than Gonzalo leaks which is bizarre because you yourself dont abide by your own rules since you have no issues using the Reddit OQA leak, jason's clarification but not the original post, and a benji retraction of all things.
Lort
Banned
Ah yup .. its all speculative.
Some people are treating my post like im saying i know ... i dont but i am suggesting thy MS has more capability and incentive to release the faster console. That was also their intention .. if they succeed or not is another matter but technically they should be able to.
bitbydeath
Member
Since it’s the same tech it’ll likely only be separated by secret sauce. Which given we’re in the 4K+ range won’t amount to much no matter who is ahead.
SlimySnake
Flashless at the Golden Globes
If you look at this just a bit differently you would arrive at 56 CUs, not 32 CUs.
You said first chip is always a high spec card. So back in Jan, Navi 10 Lite was actually a 60 CU card (56 active).
This actually makes sense because if you look at the november 12th article on WCCFTECH, Navi 12 is described as 40 CUs. That gives us:
Navi 14 - 24 CUs
Navi 12 - 40 CUs
Navi 10 - 60 CUs (or 56 or 64 or whatever they can manage on a bigger die)
With Navi 10 Lite being the cut down version. 52-56 CUs.
The current naming convention makes absolutely no sense. navi 14 is 24 CUs and Navi 10 is 40? What the hell is navi 12 then? a 28-32 CU part? There is no such product.
You are right, their plan to have a high end navi as the first to market changed and navi 12 became navi 10 and navi 10 became 5700. While the original navi 10 is now the big Navi which will go in the next gen consoles; previously known as Navi 10 LITE as per the Komachi leak in Jan.
Sosokrates
Report me if I continue to console war
From what we know, yes I agree.
TeamGhobad
Banned
we're not getting navi10, it'll be downsized navi 12 cause navi 12 is the only one with RT. 40 WORKING CU's is enough. anything more than that is gravy.
Sosokrates
Report me if I continue to console war
We already know that next gens are based on RDNA 1.5 or RDNA 2 because they have both been confirmed to have hardware ray tracing.
The amount of cu's they have will be dictated by the cost and the power goal they want to hit.
I
The amount of cu's they have will be dictated by the cost and the power goal they want to hit.
I
Negotiator
Member
Some people think we're getting 6-core Ryzen... *SMH*
Mass Shift
Member
In the meantime at Sony’s headquarters things seem to be going pretty “smooth”.
PS5 Games' Development "Proceeding Smoothly;" Sony Expects "Highly-Engaging Gameplay Experiences"
During Sony's quarterly financial conference call for investors and analysts, chief financial officer Hiroki Totoki talked about the upcoming PS5 and more.
Well I would hope so. But even if it wasn't I sure wouldn't expect Sony HQ to come out and say things are going poorly or even slowly. It's a financial conference call, you'd want to project nothing but confidence toward your shareholders.
Fake
Member
At least match with some devs saying they're 'happy' with PS5 devkit.
Yikes.
Last edited:
Barakov
Gold Member
Weird. I thought it was generally accepted that PS5 and Scarlett would have 8 cores. We KNOW for a fact that PS5 is going to have 8 cores and there's no way Scarlett is going to have less than that. I guess people will downplay the specs right until the end.
Thedtrain
Member
I’m convinced it’s the same person a few times over
Imtjnotu
Member
My favorite is the 12 core ryzen next box rumor
Insane Metal
Member
I believe it's possible. For the next next xbox... or the next xbox X
Last edited:
Negotiator
Member
What bothers me with PCMR folks is that they have accepted the price gouging status quo in the market. This status quo is ruining PC gaming and it's driving former PC gamers to consoles. How hard it is to understand that? This mentality doesn't help their hobby to flourish.
I fully understand that hardware manufacturers have to make a profit, but $500 CPUs (like 9900k) and $1199 GPUs (2080 Ti) is straight into scalping territory. It's against our consumer interests to justify that.
Just because Ryzen 5 3600 costs $200, it doesn't mean consoles cannot have an 8-core Zen 2. I even consider 3600 overpriced, considering the fact it only costs $30. AMD could sell it at $150 and still make a decent profit.
It's pointless to try to explain technicalities like the I/O die 12nm + 7nm chiplet and how consoles are going to fuse those 2 chips into a single 7nm monolithic die. People don't wanna listen, all they want is to downplay consoles until the very end. Same story every gen.
Last edited:
pawel86ck
Banned
The force is strong with this one
. SonGoku has left us for good and I thought the fight is over, but maybe you can be our leader and lead Team 12-14TF to a new battle? You give us hope 
Mass Shift
Member
Indeed. I'll never understand the need for such pathetic positions. The technology in the consoles only get to dazzle us once in their entire existence. And the PS4 and XB1 didn't even do that the only opportunity they were presented.
But as you say, it happens at the start of EVERY gen like clockwork.
Fake
Member
Was 5th gen ryzen right? Lol.
Imtjnotu
Member
Some people actually believed it was zen 3
Fake
Member
Good times lmao.Some people actually believed it was zen 3
StreetsofBeige
Gold Member
We need some SonGoku posts. Where is he?
Negotiator
Member
He must be busy with real-life stuff. I miss his enthusiasm and number crunching posts.
Imtjnotu
Member
Probably arrested during the Hong Kong protests
Sosokrates
Report me if I continue to console war
Or if one of them is willing to take a bigger loss.
Hardware sauce could be a thing to, like sony's daul heatsink patent could allow insane clocks that even conventional vapour chamber can't match
bitbydeath
Member
Will it still amount to much of a visual difference though? The higher the resolution the harder it gets to spot differences.
magnumpy
Member
1080p can't be the resolution forever. granted it gets to be less of an improvement each time it is improved (256x224 games say hi) but 1080p is seriously long in the tooth.
4k does sound enticing! xD
DragoonKain
Neighbours from Hell
Since the new Xbox is backwards compatible, is there any reason to buy an Xbox One X now? I still have a launch one and barely use that, but I thought about buying an XBX so I can play 4K.
Or should I just wait til the new one launches?
Or should I just wait til the new one launches?
Last edited:
FranXico
Member
That argument can always be made for any console. If you want to play enhanced games now, there's no reason not to buy it now. It's still around a year before next gen starts.
DragoonKain
Neighbours from Hell
Well that’s the thing I barely use my Xbox. I haven’t used it since 2017. So waiting a year would be totally easy for me. sounds like an easy decision to wait then I guess unless the backwards compatibility isn’t universal.
- Status