Support NeoGAF

[Imtjnotu]

Its 2020 and back then you were bashing Xbox because of resolution. The is welcome. You can have a revisionist past all you want. You walked all high and mighty off klee and reiner rumors. Only diff is I always knew.

Tim you have me confused. I don't even know who Klee or Tim is lol I never venture into ree land. I've only stayed here and GameSpot.

 

[ManaByte]

Its 2020 and back then you were bashing Xbox because of resolution. The is welcome. You can have a revisionist past all you want. You walked all high and mighty off klee and reiner rumors. Only diff is I always knew.

Yea the funniest thing is all the revisionist history where people pretend they didn't care about 1080p vs 720p in 2013. It seemed to happen around the time of the Xbox One X launch when it was native 4K vs checkerboard 4K.

 

[XcloudTimdog]

Tim you have me confused. I don't even know who Klee or Tim is lol I never venture into ree land. I've only stayed here and GameSpot.

But you know me, lol. Thanks buddy.

 

[XcloudTimdog]

Yea the funniest thing is all the revisionist history where people pretend they didn't care about 1080p vs 720p in 2013. It seemed to happen around the time of the Xbox One X launch when it was native 4K vs checkerboard 4K.

Thats all they talked about lol!!!! Yep. It seems Xbox didnt forget going into next gen either.

 

[Imtjnotu]

But you know me, lol. Thanks buddy.

We were on GameSpot together lol of course I remember you

 

[Mass Shift]

Thats all they talked about lol!!!! Yep. It seems Xbox didnt forget going into next gen either.

Whenever you hear someone say, "I don't care", they reveal just how much they do.

 

[XcloudTimdog]

We were on GameSpot together lol of course I remember you

Systemwars???? Wow that takes me back

 

[XcloudTimdog]

Whenever you hear someone say, "I don't care", they reveal just how much they do.

Im sure you care lol anyway , make a 13tflop prediction or something

 

[Little Chicken]

Here’s another burn:

He should keep it cute considering how much the company he works for has an aversion to free speech and self-expression

 

[Mass Shift]

Im sure you care lol anyway , make a 13tflop prediction or something

I've never been a fan of these high clocks, heat and power are the traditional enemies of processors. I'm old enough to remember melting PS Ones and Egg Frying Heatsink 360s. Although MS' Series X form factor appears to be prepared for the stresses.

But since you asked

Series X

8x 3.6GHz Zen 2
56 CUs @ 1815MHz 13RDNA TFs Navi
20GBs G6 [16 Dedicated] + [4 Dash/OS]
1 TB NVMe SSD
256GB SSD Virtual Solution [mobo soldered]X

PS5

8x 3.2GHz Zen 2
48 CUs @ 2000GHz 12.2RDNA TFs Navi
24GBs G6 [20GBs Dedicated] + [4 Dash]
1 TB NVMe SSD [custom solution]

 

[Reindeer]

He should keep it cute considering how much the company he works for has an aversion to free speech and self-expression

He doesn't work for Microsoft anymore

 

[Agnostic2020]

XcloudTimdog you stayed respectful for one day and already switched back to old self ? What happened ? Why so soon ?

you prefer Xbox ecosystem and there is nothing Wrong with that . Whatever they do you will like it . if it’s low resolution you enjoy low resolution more than anything in the world , if its high resolution you will like that .What’s the point of trying fight the world with that attitude?
Life is short . Enjoy it more instead of bickering and wasting it online.

 

[Bogroll]

I kinda hope both come out with the exact specs. If the PS5 does come with 9.2tflops I don't think the internet is ready for meltdowns that will happen.

I feel as if the pressure is on Xbox to be more powerful even if it is 250 gigaflops more. But even if the rumors are true Sony will be absolutely fine with 9.2 tf.

 

[DJ12]

Or what if the AMD leak was correct and this means PS5 will be able to play Xbox/Arden games

Or what if this was for ms's servers so has BC modes for ps and Xbox games.

 

[Proelite]

Or what if this was for ms's servers so has BC modes for ps and Xbox games.

what if this intern made a typo in config file, so his script ended fetching data from Arden project and inserting them into Oberon test files.

Shocking and far fetched i know. How can an intern make a mistake. Surely it means MS / Sony is going third party.

 

[bitbydeath]

what if this intern made a typo in config file, so his script ended fetching data from Arden project and inserting them into Oberon test files.

Shocking and far fetched i know. How can an intern make a mistake. Surely it means MS / Sony is going third party.

So you believe everything was hand typed as opposed to a proper datamine?

That could mean the intern wrote everything incorrectly, or even that it wasn’t written by AMD at all and just a random person on the internet.

 

[-kb-]

The GitHub leaks gave us lots of data but its pretty sloppy and some of the results seem to be rather old (there's a navi10 test sheet with failing tests, indicating the original is from before navi10 launch), on the other hand theres also reasonably new tests in their from this year.

 

[Proelite]

So you believe everything was hand typed as opposed to a proper datamine?

The config file and the python scripts are hand written code. The xslm and csv files are results pulled from the data server.

 

[-kb-]

The config file and the python scripts are hand written code. The xslm and csv files are results pulled from the data server.

I believe the xslm files are also generated by the python scripts in the repo. Additionally theres some mentions of process nodes that no one seems to have picked up on yet.

 

[bitbydeath]

The config file and the python scripts are hand written code. The xslm and csv files are results pulled from the data server.

But it is the results from the ‘data server’ that appears wrong and not just in this instance.

 

[Proelite]

I believe the xslm files are also generated by the python scripts in the repo.

Yep. They were continuously updated too by these scripts.

Honestly it's a terrible repo by software standards.

 

[-kb-]

Yep. They were continuously updated too by these scripts.

Honestly it's a terrible repo by software standards.

Honestly running macro enabled spreadsheets for regression testing is terrible by any modern software engineering standards. But i feel its one of those we have to get it done ASAP things.

 

[DJ12]

what if this intern made a typo in config file, so his script ended fetching data from Arden project and inserting them into Oberon test files.

Shocking and far fetched i know. How can an intern make a mistake. Surely it means MS / Sony is going third party.

It's not a typo though is it. But you are welcome to believe he typed a completely different and unrelated codename by accident and didn't correct it if you like.

 

[Proelite]

Honestly running macro enabled spreadsheets for regression testing is terrible by any modern software engineering standards. But i feel its one of those we have to get it done ASAP things.

Could have switched to a data warehouse solution in a day.

 

[-kb-]

Could have switched to a data warehouse solution in a day.

Probably more related to the other tools that consume the spreadsheet data then the spreadsheets themselves. It is fun to know that most of AMDs internal web apps are entirely unencrypted though.

It's not a typo though is it. But you are welcome to believe he typed a completely different and unrelated codename by accident and didn't correct it if you like.

Its totally a typo mate.

 

[Proelite]

It's not a typo though is it. But you are welcome to believe he typed a completely different and unrelated codename by accident and didn't correct it if you like.

It's either a bug/mistake or one of MS / Sony is going third party.

Personally i make and see mistakes like this all time at work, (wrongly set config values). Anyone who programs for a living can relate.

 

[bitbydeath]

Its totally a typo mate.

If it’s a typo then that makes everything hand typed as the results are physical data.

It’s supposed to be the project result from the oberon.xlsm file.

And how far do we extend these typos too if we want to continue that it’s real AMD data? I’d imagine the dual GPU’s has to be another typo.

 

[-kb-]

If it’s a typo then that makes everything hand typed as the results are physical data.

It’s supposed to be the project result from the oberon.xlsm file.

That screenshot is from a configuration file. The xlsm files are auto generated by the python scripts that use the configuration files to pick what to run the test on.

 

[Raploz]

I have a theory: What if that 36CU 2Ghz is a mere "boost" mode for PS4 Pro titles (just like there were modes emulating PS4 Pro and base PS4 clocks). That would explain why it's the same number of CUs as the PS4 Pro and why there's no RT/VRS support.

With such a mode they could ensure all games targeting the Pro could run smoother on PS5 and with a better dynamic resolution. Maybe the chip can disable CUs as needed for compatibility mode (makes sense as a custom feature for Sony as I don't think any other GPU can do that) and the test was only for testing the BC.


Also, with CUs disabled they could target a high clock, but the full chip (when not in compatibility mode) would probably run with a slower clock and enable all available CUs.


What do you guys think?

 

[bitbydeath]

That screenshot is from a configuration file. The xlsm files are auto generated by the python scripts that use the configuration files to pick what to run the test on.

But the configuration data is pulled from the data server. And do you think the dual GPU’s Oberon or Ariel are typos too?

 

[Proelite]

K -kb- good luck. Peace.

 

[-kb-]

I have a theory: What if that 36CU 2Ghz is a mere "boost" mode for PS4 Pro titles (just like there were modes emulating PS4 Pro and base PS4 clocks). That would explain why it's the same number of CUs as the PS4 Pro and why there's no RT/VRS support.

With such a mode they could ensure all games targeting the Pro could run smoother on PS5 and with a better dynamic resolution. Maybe the chip can disable CUs as needed for compatibility mode (makes sense as a custom feature for Sony as I don't think any other GPU can do that) and the test was only for testing the BC.


Also, with CUs disabled they could target a high clock, but the full chip (when not in compatibility mode) would probably run with a slower clock and enable all available CUs.


What do you guys think?

It makes more sense in my mind to go double the pro clock and not 2Ghz. Additionally this feels like a worst of both worlds kinda situation where you have to bin chips that make the PS5 clock but not have enough CU's that make the PS4 Pro boost clock.

 

[Marlenus]

Systemwars???? Wow that takes me back

Only person I remember from back then is RonValencia the person who thought a 50% TFlop difference would not result in 900p Vs 1080p for Xbox one and PS4 respectively in the majority of titles.

 

[Rightisright]

I feel as if the pressure is on Xbox to be more powerful even if it is 250 gigaflops more. But even if the rumors are true Sony will be absolutely fine with 9.2 tf.

For sure. Sony can have the weaker console and still win the war. MS cant afford to.have any weakness.

 

[Raploz]

It makes more sense in my mind to go double the pro clock and not 2Ghz. Additionally this feels like a worst of both worlds kinda situation where you have to bin chips that make the PS5 clock but not have enough CU's that make the PS4 Pro boost clock.

Well, they could be just stress testing the chip, final clocks for that magic boost mode could be lower. The first leaks had a clock of 1.8Ghz, that's almost double PS4 Pro (911x2=1822)

 

[StreetsofBeige]

Thank god you guys can interpret that Oberon chart.

Pleas ensure whatever you guys decide upon to type your messages as dumbed down as possible for dumbasses like me.

 

[Mista]

For sure. Sony can have the weaker console and still win the war. MS cant afford to.have any weakness.

Correct. Sony this gen killed it so MS gotta step up next-gen. even if the new Xbox is more powerful than the PS5, we all know that the PS5 is the one selling more and we know why.

 

[Ovech-King]

I have a theory: What if that 36CU 2Ghz is a mere "boost" mode for PS4 Pro titles (just like there were modes emulating PS4 Pro and base PS4 clocks). That would explain why it's the same number of CUs as the PS4 Pro and why there's no RT/VRS support.

With such a mode they could ensure all games targeting the Pro could run smoother on PS5 and with a better dynamic resolution. Maybe the chip can disable CUs as needed for compatibility mode (makes sense as a custom feature for Sony as I don't think any other GPU can do that) and the test was only for testing the BC.


Also, with CUs disabled they could target a high clock, but the full chip (when not in compatibility mode) would probably run with a slower clock and enable all available CUs.


What do you guys think?

Funny you mentioned that I'm starting to think the same theory.... 36 cu tested at 2ghz for BC only . 56 cu total ,4 deactivated so 52 active. Only 36 used in ps4 BC but all of them available for ps5 games. Hmmm... That actually sounds very logic in a way

 

[TLZ]

I have a theory: What if that 36CU 2Ghz is a mere "boost" mode for PS4 Pro titles (just like there were modes emulating PS4 Pro and base PS4 clocks). That would explain why it's the same number of CUs as the PS4 Pro and why there's no RT/VRS support.

With such a mode they could ensure all games targeting the Pro could run smoother on PS5 and with a better dynamic resolution. Maybe the chip can disable CUs as needed for compatibility mode (makes sense as a custom feature for Sony as I don't think any other GPU can do that) and the test was only for testing the BC.


Also, with CUs disabled they could target a high clock, but the full chip (when not in compatibility mode) would probably run with a slower clock and enable all available CUs.


What do you guys think?

hmmm.....

HMMMMM....!

I like the way you think. Testing Navi CUs at PS4 and Pro speeds for any compatibility issuess. Then testing them at PS5 2ghz native speed to see how that works for BC as well; compatibility, bugs, heat, whatever. And obviously no RT and VRS in sight because those are nextgen components, and irrelevant in these tests!

Naisu Idea! I rike ito!



I hope it's true.

 

[Marlenus]

I have a theory: What if that 36CU 2Ghz is a mere "boost" mode for PS4 Pro titles (just like there were modes emulating PS4 Pro and base PS4 clocks). That would explain why it's the same number of CUs as the PS4 Pro and why there's no RT/VRS support.

With such a mode they could ensure all games targeting the Pro could run smoother on PS5 and with a better dynamic resolution. Maybe the chip can disable CUs as needed for compatibility mode (makes sense as a custom feature for Sony as I don't think any other GPU can do that) and the test was only for testing the BC.


Also, with CUs disabled they could target a high clock, but the full chip (when not in compatibility mode) would probably run with a slower clock and enable all available CUs.


What do you guys think?

Seems over engineered but possible I guess.

I just don't see 2GHz as a possible clock speed in a console tdp window from what we know about 7nm and RDNA.

It is always possible if there are changes in architecture for RDNA2 but it just seems very unlikely.

Not only would 2 GHz based on current RDNA and 7nm require a high voltage but not many chips can actually hit that speed so while going with a smaller die might mean more fully functional dies from the fab it does not mean they get more dies that work to required specs. Going bigger and lower clocked means that you get fewer dies from the fab but more will hit the desired spec so cost wise the difference is probably pretty small.

PS4 was clocked at 800MHz which was around 80% of the desktop versions of the chip available at the time. The same goes for PS4 pro as that was around 80% of the base clock of the rx480.

Current RDNA cards run at around 1.8GHz on the desktop so a reasonable clock would be around 1.5GHz for current tech. I just do not see a 500MHz uplift as possible in the same sort of tdp envelope.

 

[XcloudTimdog]

[/QUOTE]

XcloudTimdog you stayed respectful for one day and already switched back to old self ? What happened ? Why so soon ?

you prefer Xbox ecosystem and there is nothing Wrong with that . Whatever they do you will like it . if it’s low resolution you enjoy low resolution more than anything in the world , if its high resolution you will like that .What’s the point of trying fight the world with that attitude?
Life is short . Enjoy it more instead of bickering and wasting it online.

so i cant make a joke??? Maybe dont be so sensitive and yes i have been respectful. Ppl are putting up 2014 tweets ages ago and I laughed about it. Come on now.

 

[XcloudTimdog]

Honestly i been thinking. Since andrew reiner, colin , and klee all have said most power for Sony, maybe thats the truth since that they suggest it. Going by that, the github was always left open for info to get out there to counter what Xbox is doing, and they split the actual apu in half to mask the results. So we actually have 72 cu. that explains the lack of ray tracing, its in the half they arent testing. That puts them at 18.4 with a 2.0 clock. Crazy but it seems likely. So im on team 18

 

[-kb-]

Honestly i been thinking. Since andrew reiner, colin , and klee all have said most power for Sony, maybe thats the truth since that they suggest it. Going by that, the github was always left open for info to get out there to counter what Xbox is doing, and they split the actual apu in half to mask the results. So we actually have 72 cu. that explains the lack of ray tracing, its in the half they arent testing. That puts them at 18.4 with a 2.0 clock. Crazy but it seems likely. So im on team 18

I don't think 2.0Ghz is likely, its a bad decision especially at a part as large as 72 CU's. Yields would be absolutely terrible, heat would be huge. additionally if the raytracing is based off the same concept as AMD's it will be in the TMU's and there are TMU's in every SA so the raytracing would be in all SA's.

 

[DJ12]

Honestly i been thinking. Since andrew reiner, colin , and klee all have said most power for Sony, maybe thats the truth since that they suggest it. Going by that, the github was always left open for info to get out there to counter what Xbox is doing, and they split the actual apu in half to mask the results. So we actually have 72 cu. that explains the lack of ray tracing, its in the half they arent testing. That puts them at 18.4 with a 2.0 clock. Crazy but it seems likely. So im on team 18

Way to go the opposite way lol

18.4 isn't happening as that wouldn't be a slight advantage to ps5 which is what these people are saying.

The results are without context so no real conclusion can be drawn from them.

 

[XcloudTimdog]

Way to go the opposite way lol

18.4 isn't happening as that wouldn't be a slight advantage to ps5 which is what these people are saying.

The results are without context so no real conclusion can be drawn from them.

I hear ya but im sticking to maybe 15 with lower clocks.

 

[Darius87]

damm, this thread become clown shitshow

 

[chigstoke]

It is not me saying pr talk about 8k gaming, it's Sony itself. Now I expect that, even if checkboarded

Sony and MS are the masters at selecting specific buzzwords and phrases to appeal to the consumer who simply purchases on what looks 'good'.

Obviously us GAFers is more an audience where we know the 8K talk is bullshit, but we should be under no illusions, 8K will sell consoles.

 

[SenjutsuSage]

I have a theory: What if that 36CU 2Ghz is a mere "boost" mode for PS4 Pro titles (just like there were modes emulating PS4 Pro and base PS4 clocks). That would explain why it's the same number of CUs as the PS4 Pro and why there's no RT/VRS support.

With such a mode they could ensure all games targeting the Pro could run smoother on PS5 and with a better dynamic resolution. Maybe the chip can disable CUs as needed for compatibility mode (makes sense as a custom feature for Sony as I don't think any other GPU can do that) and the test was only for testing the BC.


Also, with CUs disabled they could target a high clock, but the full chip (when not in compatibility mode) would probably run with a slower clock and enable all available CUs.


What do you guys think?

Doesn't really explain that big ass 56 CU xbox chip with the same backwards compatibility modes that does have VRS and Hardware Ray Tracing. Microsoft has a much bigger, more advanced chip and it was impossible for Sony to get something along those lines in testing as of this year apparently? I think it's obvious that is the PS5 chip unless Sony reveals something else officially. They would have to decide they're ditching this thing to go for an entirely different chip.

Why test for a max clock speed on a chip that doesn't represent the total configuration and size you're going for? Any power or heat or performance results you would get would be completely useless because you would want to run that amazing clock on the full chip. Maybe if this chip showcased much slower speeds other than that 2GHz I might be willing to entertain the possibility that the chip is probably old. And I definitely don't expect it to launch at that clock speed. I expect it to be lowered to something more reasonable in a console unless Sony's cooling and heat solution is far better than I suspect and they're willing to let this thing loose in the watts department. I hope both consoles are able to exceed 300w on full load. PS5 should hit beyond 270-290 easy.

 

[sinnergy]

Honestly i been thinking. Since andrew reiner, colin , and klee all have said most power for Sony, maybe thats the truth since that they suggest it. Going by that, the github was always left open for info to get out there to counter what Xbox is doing, and they split the actual apu in half to mask the results. So we actually have 72 cu. that explains the lack of ray tracing, its in the half they arent testing. That puts them at 18.4 with a 2.0 clock. Crazy but it seems likely. So im on team 18

Secret sauce right here ... not doable btw. MS really shook things up.

 

[Reindeer]

9.2 tflops of RDNA (11.5 GCN) is still a substantial gain in comparison to 1.8 tflops of GCN,. Not sure why people are getting all worked up about it. This is over 6x of PS4 GPU with even greater difference in CPU advantage and tech like SSD rounding up the package. Sure it might not be 12 tflops, but the difference will likely be only seen in resolution and not much else. I think people are way too emotionally invested in this spec wars and should just rather focus on improvements in comparison to last gen.