Kugoson (Son goku) sounds like a believable codename for a GPU.
-
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.
Next-gen PS5 and next Xbox speculation launch thread |OT5| - It's in RDNA
- Thread starter Mecha Meister
- Start date
- OT
- Status
Kugoson (Son goku) sounds like a believable codename for a GPU.
Shades of the terrible messaging from 2013 coming back.....
Matt booty also made another mistake.he came out and said 4x performance refers to combination of things like gpu CPU ram ssd etc... But the next day Phil said it was just for cpu
First, RX 570 has a TDP rated to 120W and RX 590 of 175W. As for power consumption, I rather compare the RX 570 to 580. RX 590 is pushing far beyond its optimal clocks in terms of thermals. Or even the RX 480 to the RX 470. You're taking two extreme exemples to explain a difference in power consumption.
Compare the RX 5700 to XT.
The difference in thermals isn't that extreme for comparable clocks.
Some people here are talking about adding 35% more CU at the same clocks as the RX 5700 XT for the same power consumption.
This is going to be a really stupid question, but it just popped up in my head, given the talk about PS5 having this 8GB pool of fast RAM and the rest slower.. but how much RAM is used or can be used to actually render a frame? Obviously the answer will be varying but I mean, is there a limit? Can the GPU only draw from a limited amount of RAM to render a frame. Can it use all of it or is there a power or bandwidth limit it will hit first. I'm guessing RAM is only utilised in fairly small parts to render a frame, with the rest being used to store all sorts of other information that needs to be ready at a moments notice, but isn't all utilised simultaneously.
I know NOTHING besides what I gleam from forums here like this, clearly.
Just trying to work out if Sony's solution of a shared pool of varying memory, combined with super fast streaming off SSD will be at a disadvantage or not compared to just a huge lump of GDDR6 when it comes to real world rendering.
I know NOTHING besides what I gleam from forums here like this, clearly.
Just trying to work out if Sony's solution of a shared pool of varying memory, combined with super fast streaming off SSD will be at a disadvantage or not compared to just a huge lump of GDDR6 when it comes to real world rendering.
dont you know that the most reliable method of comparing power levels is through dragon ball characters ;p
Didn't you took Klob's provided document as fact ( alongside with hmqgg's posts ) that Anaconda being more powerful is set in stone? Yeah, theirs statements makes sense, but Reiner's tweets doesn't makes sense at all. Right
And yet some people are reading books from Klob's HOLY provided document and hmqgg's posts. That's OK, i presume. I smell a hypocrisy here from a 100 miles.
Last edited:
HBM2 hype
PS4 got leaked out from a desktop linux commit ? are you sure you are following the thread closely ?
Sorry, I was referencing TBP. I specifically reference RX 590 because its TBP matches RX 590, suggesting it's GPU is pushed similarly hard.
I would really like to see a 5600 with 36 CUs and lower clocks to compare.
Previous codenames were found via commits and/or 3DMark results, as stated in DF article. Yup, pretty sure I can read. Thanks for checking.
Furthermore, if you read the DF article, you can see he expresses the exact same skepticism and provides a potential answer, too.
Last edited:
The RX 5700 has 36CUs and lower clocks. Rated at 180W of TBP.
The RX 5700 XT had 40CUs and higher clocks. Rated at 225W of TBP.
I'm aware. There is no comparable to RX 570 in terms of TBP, which is what I'm talking about. If you get another ~30W back by clocking 125MHz lower and disabling 4 CU like 570 does to 580, then those clocks make a hell of a lot more sense for a console. The Xbox One X clocks are very close to RX 570 and it has more CUs than any Polaris GPU.
In summary:
RX 590: 36 CU, 1545/1468 MHz core, 7000 MHz memory, 225W TBP.
RX 570: 32 CU, 1244/1168 MHz core, 6000 MHz memory, 150W TBP.
5700 XT: 40 CU, 1800 MHz core (gameclock), 14000 MHz memory, 225W TBP.
5600???: 36 CU, 1500 MHz core, 14000 MHz memory, ~160W TBP???
Last edited:
Your assertion that Klob did something like is laughable, but the fact that you used the term xbot shows me that you won't be around long.
After cernys wired interview, some ms guys produced a document (mid april) of specs featuring anaconda and lockhart but blacked out the Teraflop number.
Maybe they knew it wasnt higher than PS5s?
We already got a confirmation about this document being real. Klob just made his own version with dream specs
Right, but then again, it's about a modulation to get the same power. Not more.
People here are arguing that they'd add 14 (!!!) more CUs (for the record, Xbox One X has 2 more CUs than RX 580) and the same 1800mhz clockspeed as RX 5700 XT to reach 12.44 Tflops. Even at a lower clockspeed, like 1500mhz for 10.3Tflops. You're still adding 14 more CUs, and that doesn't make up for the loss of 300mhz.
It's interesting that none of the marketing buzzwords seem to match what Microsoft shown at E3 with them using words like "immersive". It's almost like this could have been made up.I can only assume this one.
Apologize if there was any hate speech used. Im not sure of the colloqial term for fanboys.
The whole "company X saw Y and then made a change to an an integrated semi-custom APU without any impact to its release date" is complete nonsense. This isn't a company just picking up a random GPU for sale and slotting it into a PCIE slot of a tower PC, this is a multi-year highly customised piece of electronics that has to be rigorously tested, ordered in bulks, and finalised in specifications well in advance.
You save about 33% TBP in that scenario. They're talking about adding 35% more CUs. It will all come down to clocks.
A few corrections here:
RX 590: 36 CU, 1545 MHz core, 225W TBP.
RX 580: 36 CU, 1340 MHz core, 185W TBP.
RX 570: 32 CU, 1244 MHz core, 150W TBP.
5700 XT: 40 CU, 1900 MHz core, 225W TBP.
5700 : 36 CU, 1900 MHz core, 180W TBP.
The 590 is the exception here. It's a model that is pushing far beyond it's expected optimal clock to reach higher performances.
I have troubles to see how one can add 14CU to the 5700XT while maintaining HIGHER performances... at the same or lower TDP.
And in that scenario, you're at 10.3 Tflops. Not 12.44 like some expects.
There's a possibility to have more CUs. But it'll be at lower clocks. To the point it'll nearly negate the performance gain.
Why would they start marketing when the console is 18 months out and the they have not even announced a name?
And/or ? That's your answer to my specific question ?
I am sorry but you can't approximate "linux commits" with "3DMarks results". Both things are very different kind of 'leaks'.
Can the passive aggressive platform wars posters please cease and desist?
This isn't the thread for that.
I'm at the point now where I think combining both Next-Gen consoles into one thread was a huge mistake. If folks can't play nice and be civil then maybe the mods should reconsider their position on it.
It becomes increasingly unbearable trying to sift through pages and pages of shitflinging posts just to try and find the few quality or insightful ones.
I would actually personally prefer a pure dedicated tech thread where if you're not there to contribute or learn you get banned.
This isn't the thread for that.
I'm at the point now where I think combining both Next-Gen consoles into one thread was a huge mistake. If folks can't play nice and be civil then maybe the mods should reconsider their position on it.
It becomes increasingly unbearable trying to sift through pages and pages of shitflinging posts just to try and find the few quality or insightful ones.
I would actually personally prefer a pure dedicated tech thread where if you're not there to contribute or learn you get banned.
10% higher performance at the same TDP is negation territory? The more pertinent question is if it justifies the die space. But if you already weren't going to clock it that high anyway, the perf gain is higher than 10%.
That's a lot of words to say you didn't read the article.
Welcome. You're a little late to the club, but we can read out the minutes.
Komachi posts a lot of interesting stuff, but its far too often to distinguish his personal speculation from actual information.
I fully expect him to be wrong on a lot more than just this.
Lol. But seriously tho, the mods should really reconsider.
The current thread isn't working. It's generating more bans than actual meaningful discourse.
I would at least consider it.
It is a shame to even have to think about doing something like that but here we are.
"nearly" as in "the expect gains nearly climbs back to the performance level of the original target". The RX 5700 XT has a peak performance of 10.13Tflops. 10.3 Tflops isn't 10% more. It's 2% more.
Erf nevermind, readed the 50th anniversary one by mistake.
9.75 for the regular one. So you get a 6% gain.
Already there it seems:
New Table is up
This is an updated table regarding the recent AMD drivers showing code for Navi 10,12,14,21 and "lite" variants spotted by apisak
This is an updated table regarding the recent AMD drivers showing code for Navi 10,12,14,21 and "lite" variants spotted by apisak
I'm curious about this chip, could it be an updated spec for PS5? or Anancoda/Date
I would go ahead and do that. It seems like a great idea.
Does anyone know what sort of cooling solution these machines will adopt?
From that reddit post:
" [–]dad2youAMD [score hidden] 29 minutes ago*
"P" is for performance. PCI ID for Navi10 LITE was 13e9 - Ariel (PS5 SOC), codename Gonzalo. Navi12 LITE and Navi21 LITE might actually be related to another console, Scarlett, which is rumored to actually be 2 consoles. One would be 4-5TF (Navi12 LITE) and another would be 10-11TF (Navi 21 LITE). Navi 14 is certainly mobile part, but the other three I would say are console parts.
BTW there was revision of NAVI10_LITE. First was committed in January, and another in April (1GHz v 1.8GHz)."
"Actually, both PS4 and PS4Pro chips have been committed in 3DMark11 for example back in 2013/2015 - https://d2skuhm0vrry40.cloudfront.n....jpg/EG11/resize/690x-1/quality/75/format/jpg
I am 100% Gonzalo leak was console APU (PCI ID matched Ariel, which is PS5 codename and ID decoding showed "G" and Navi10 LITE, meaning console, not desktop."
More or less what we discussed here already.
I found a fantastic tool to figure out the specs of the PS5.
talktotransformer.com
its an AI that creates a paragraph off a sentence that you write, now i can give everyone the real specs of the PS5!
InferKit
its an AI that creates a paragraph off a sentence that you write, now i can give everyone the real specs of the PS5!
- Status