• 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.
  • The Politics forum has been nuked. Please do not bring political discussion to the rest of the site, or you will be removed. Thanks.

RedGamingTech: Xbox Series X Hot Chips Analysis Part 1 - GPU, CPU & Overview

Jun 13, 2020
1,761
3,375
400
but we found out XSX has 15% raytracing advantage (321 vs 380) not 80000%. Don’t listen to that fucking clown lol

What we found out is that
a) ray tracing is being done in CUs and not in any secret sauce undisclosed part,
and xbox has 44% more CUs than ps5, and
b) xbox does have hardware inference acceleration, which is a must if you are seriously going to discuss ray tracing in a game and not some theoretical crap,
while for ps5 we know didly squat about inference acceleration, and cerny failed to mention such a crucial detail in his long speech.

The other day I wrote a simpleton example for a case of a game that shoots for visual parity in everything but ray tracing between the two consoles, and then xbox has more than 40% advantage on ray tracing on top.

And this 40% was with all positive ps5 assumptions, like both cpu and gpu clock at full speed, ps5 having same hardware as xbox for machine learning and inference acceleration, and without compensating for VRS or anything else xbox has on top.
Just CUs and Ghz

So about that 15%... I would advise you to not bet on it
 

TBiddy

Member
Mar 16, 2015
4,790
6,966
830
Denmark
I'm trolling here because i've said nothing wrong? Well, that's a stretch. What you would say for then for TheGreatWhiteShark TheGreatWhiteShark who repeated over and over the same thing and trying to prove that X1 gamepad has a mechanism like Dualsense, but it doens't have it at all? Where is the logic in that the rumble motor which just create a vibrations and nothing else, can create a resistance? That's stupidity by a mile.
Or maybe he thinks that spring on the wrist inside of the trigger creates so called resistance. LOL. But purpose of that spring is actually and obviously is to bring back the trigger up ( just like on DualShock 4 or other gamepads with triggers).

Constantly changing the subject to PS5, console warring in a discussion thread and talking about peoples stupidity is trolling.

That tag of yours fits you perfectly.
 

Panajev2001a

GAF's Pleasant Genius
Jun 7, 2004
18,812
12,386
2,110
What we found out is that
a) ray tracing is being done in CUs and not in any secret sauce undisclosed part,
and xbox has 44% more CUs than ps5, and
b) xbox does have hardware inference acceleration, which is a must if you are seriously going to discuss ray tracing in a game and not some theoretical crap,
while for ps5 we know didly squat about inference acceleration, and cerny failed to mention such a crucial detail in his long speech.

The other day I wrote a simpleton example for a case of a game that shoots for visual parity in everything but ray tracing between the two consoles, and then xbox has more than 40% advantage on ray tracing on top.

And this 40% was with all positive ps5 assumptions, like both cpu and gpu clock at full speed, ps5 having same hardware as xbox for machine learning and inference acceleration, and without compensating for VRS or anything else xbox has on top.
Just CUs and Ghz

So about that 15%... I would advise you to not bet on it

I would not expect a game to try to maintain exact visual parity in terms of resolution and try with the same complexity in terms of rays being shot and the amount of bounces you want to trace (where a higher clocked CU would have an advantage over a slower clocked one). Use of the 4 and 8 bits mode for the CU’s trades compute power for ML acceleration which can improve denoising, so PS5 in the same scenario would need to balance with minor dips in the resolution of the rasterised portion of the screen and focus on making that renders path faster, likewise they have a per CU advantage with calculations that require multiple bounces and scattering rays post single bounce.

It also remains to be seen how much PS5 can flex its Rasterisation and pixel fill rate advantages (before being hit by memory bandwidth limitations) and geometry processing (the scenario of XSX having more ROPS and geometry engine resources turned out not to be the case, so PS5 flexes its clock speed advantage there).
 

jimbojim

Banned
May 7, 2020
1,196
3,811
480
Constantly changing the subject to PS5, console warring in a discussion thread and talking about peoples stupidity is trolling.

That tag of yours fits you perfectly.

But i didn't started it. Mr. "won't blame the green side". Cheers!
 
Last edited:
Jun 13, 2020
1,761
3,375
400
I would not expect a game to try to maintain exact visual parity in terms of resolution and try with the same complexity in terms of rays being shot and the amount of bounces you want to trace (where a higher clocked CU would have an advantage over a slower clocked one). Use of the 4 and 8 bits mode for the CU’s trades compute power for ML acceleration which can improve denoising, so PS5 in the same scenario would need to balance with minor dips in the resolution of the rasterised portion of the screen and focus on making that renders path faster, likewise they have a per CU advantage with calculations that require multiple bounces and scattering rays post single bounce.

It also remains to be seen how much PS5 can flex its Rasterisation and pixel fill rate advantages (before being hit by memory bandwidth limitations) and geometry processing (the scenario of XSX having more ROPS and geometry engine resources turned out not to be the case, so PS5 flexes its clock speed advantage there).
you already wrote it yourself, the ps5 fill rate advantage is bandwidth limited.

as I said, I tried to keep my example as simpleton as possible, as for example in such a case where you are working ps5 at full capacity, then cpu or gpu clock probably must give in.
also, before even the consoles are out, I wanted to shoot for visual parity in my example, and not a ps4pro-xbox x case.

and that >40% of CU leftover advantage on xbox, it can be used for anything
I just kept it for ray tracing, since that was the scenario I was presented with.
 
  • Like
Reactions: FritzJ92

jimbojim

Banned
May 7, 2020
1,196
3,811
480
Laughing at fanboy arguments isn’t fuelling fanboy wars, it’s laughing at them, but please tell me again how I’m the fanboy mr “report me for trolling Xbox threads” tag.

Your post surely didn't implied that you were laughing at fanboy wars arguments and also your post is surely pointed to one side and they are to blame. Is it now? Blue faithful will soon have no arguments and they will shut up, right? Right
 
Last edited:

jimbojim

Banned
May 7, 2020
1,196
3,811
480
But you’re not exactly stopping it either. On the contrary, actually.

Then others should stop too. You know, the ones you won't blame who fueled the discussion about triggers. But hey, lets blame me because the tag. My 1st post on 2nd page regarding triggers was quite normal and accurate. Yeah, problem is someone couldn't accept the right answer.
 
Last edited:
  • Like
Reactions: Yakuzakazuya

Panajev2001a

GAF's Pleasant Genius
Jun 7, 2004
18,812
12,386
2,110
you already wrote it yourself, the ps5 fill rate advantage is bandwidth limited.

I said it might be in some cases, but I might be wrong as well. This is not as clear cut as PS4 Pro filtrate to memory bandwidth (again Cerny himself in the interview stated they had more fill rate than bandwidth, refreshing when architects openly state their bottlenecks eh ;)?).

XSX and PS5 have very comparable bandwidth to FLOPS ratio and for PS5 to be clearly bandwidth limited in terms of fill rate you would have to assume that XSX is set to the maximum it can be without outstripping bandwidth.

There are scenarios where you may be fillrate bound and not have CU’s waiting to read that data and thus not stressing external VRAM bandwidth and would fillrate outstrip bandwidth in that scenario? I do not think so.

as I said, I tried to keep my example as simpleton as possible, as for example in such a case where you are working ps5 at full capacity, then cpu or gpu clock probably must give in.

It is not quite simple as that. First of all you need to have very very high utilisation of both CPU and GPU (not that common a scenario to be happening for each in isolation and even rarer to happen for both at the same time) and then you need to be running power hungry work too (not all work is the same, down to the instructions you run and which transistors you flip essentially... e.g.: AVX-256 instructions being very power hungry)... and then you need to be in a scenario where you cannot just transfer power “budget” over from the CPU and/or lower the CPU clockspeed.

First party games are trying to out the HW under the best light, third party games may trade off resolution differences (native 4K vs 99% of the time 4K native and 1% of the time reduced to slightly less than 4K is not something users will perceive that much if at all). Again realistic scenarios and outcomes vs “scoring console war points”.
 
Last edited:

Agnostic2020

Member
Oct 24, 2017
4,672
19,829
805
What we found out is that
a) ray tracing is being done in CUs and not in any secret sauce undisclosed part,
and xbox has 44% more CUs than ps5, and
b) xbox does have hardware inference acceleration, which is a must if you are seriously going to discuss ray tracing in a game and not some theoretical crap,
while for ps5 we know didly squat about inference acceleration, and cerny failed to mention such a crucial detail in his long speech.

The other day I wrote a simpleton example for a case of a game that shoots for visual parity in everything but ray tracing between the two consoles, and then xbox has more than 40% advantage on ray tracing on top.

And this 40% was with all positive ps5 assumptions, like both cpu and gpu clock at full speed, ps5 having same hardware as xbox for machine learning and inference acceleration, and without compensating for VRS or anything else xbox has on top.
Just CUs and Ghz

So about that 15%... I would advise you to not bet on it
Lol dude at least look up formula given by amd and ms in hot chip

4× cu × gpu clock in mhz

Ps5 321 giga rays
Xsx 380 giga rays

15% difference
 
Last edited:

Lysandros

Member
Jul 28, 2020
865
3,197
350
discussion derailed from comment #5

congrats!
What? It was Paul who mentioned Ps5 while discussing XSX's CPU L3 clusters in the context of 'unified or not'. I only shared the info. What is wrong in commenting on the video that op posted? There is no 'derailment' at all.
 
Jun 13, 2020
1,761
3,375
400
Lol dude at least look up formula given by amd and ms in hot chip

4× cu × gpu clock in mhz

Ps5 321 giga rays
Xsx 380 giga rays

15% difference
I am not an amd gpu guy, but from what we already know about the amd cards on pc, overclocking does not give them sufficient amounts of power. its more like a small fraction of the overclock percentage, rather unproportional also to heat and consumption.
instead, they get their big benefits from increasing CU numbers

I must have said this a hundred times already, theoretical numbers are fine, but they pretty much can differ a lot from reality. this goes both for the ps5 tflops, as well as the xbox's 25tf "combined" rt performance, among other things
 
Last edited:
  • Like
Reactions: Hendrick's

Agnostic2020

Member
Oct 24, 2017
4,672
19,829
805
I am not an amd gpu guy, but from what we already know about the amd cards on pc, overclocking does not give them sufficient amounts of power. its more like a small fraction of the overclock percentage, rather unproportional also to heat and consumption.
instead, they get their big benefits from increasing CU numbers

I must have said this a hundred times already, theoretical numbers are fine, but they pretty much can differ a lot from reality. this goes both for the ps5 tflops, as well as the xbox's 25tf "combined" rt performance, among other things
Lol u keep dropping theoritocal number like 25 Tf and bunch of PR non sense. U say amd card don't work great with over clocking ignoring the fact that no amd card has had raytracing in it which overclocking help tremendously with raytracing. Rdna 2 should theoretically work great with overclocking and we shall see.

Again we are given a formula by amd and MS for raytracing capabilities. This puts xsx 15.6% ahead of ps5 in raytracing capabilities. This is math not some theoritical number thrown out by any company.

Formula vs PR
 
Last edited:
Jun 13, 2020
1,761
3,375
400
Lol u keep dropping theoritocal number like 25 Tf and bunch of PR non sense. U say amd card don't work great with over clocking ignoring the fact that no amd card has had raytracing in it which overclocking help tremendously with raytracing. Rdna 2 should theoretically work great with overclocking and we shall see.

Again we are given a formula by amd and MS for raytracing capabilities. This puts xsx 15.6% ahead of ps5 in raytracing capabilities. This is math not some theoritical number thrown out by any company.

Formula vs PR
I am just saying what we already know.

now, if by any chance you want to bet that the difference will be "15,6%" between the two, if you insist I will gladly take your money
 

IDappa

Member
Nov 20, 2018
383
421
315
I sense some wishful thinking there... it could be as good as impulse triggers ;)... or better.

Jokes aside, the gimmick is as good as both how easy and intuitive it is to take advantage of as well as how versatile and useful it is. Haptic feedback is the next generation evolution of rumble and if the engine they pack is matched to the controller weight well it should offer a nice noticeable improvement without risking low third party adoption (similar API’s as what you have on iOS, Switch, etc... likely). Do you really wonder about rumble adoption?

Variable force feedback triggers may see some internal first party adoption dwarfing third party adoption, but they could catch on with third parties as well as they definitely seem far easier to make good use of in games than the touchpad (which would have seen more use had it been a screen too).

Any racing game or game with driving segments or any FPS can make basic force feedback trigger use for example and then you can build from there.
I wasn't saying anything thats not true, the support for the touchpad was abysmal this gen with few exceptions.

I do hope the features are supported this gen as what they have promised sound cool in theory, however there are a lot of unknowns. Hopefully Sony have tested these things rigorously to prevent any problems from prolonged use and also Microsoft has done the same, I've gone through 3 controllers this gen.
 

waxer

Member
Aug 24, 2006
929
231
1,325
the "from thin air" is for RGT though (this is RDNA 3 again)
Of course it please a particular audience
you know criticize people reactions of the video is also a thing that can happen in discussion.
now answer the question :
is 5 sec baseless speculation thought about ps5 the first and most important information a normal person would retain from a xbox hot chip video confirming lot of stuff ?
Depends on the person. No one said its the most important. Personally I think audio being great is. But for you it's probably something else.
What's your point?

Going to stalk around the forum telling people to shut up and only dicuss the most important things in videos. Who decides what's most important. Only positive xbox news perhaps.

Censorship is wonderful.
Are we the baddies gif
 

Agnostic2020

Member
Oct 24, 2017
4,672
19,829
805
Agnostic2020 Agnostic2020 and TheGreatWhiteShark TheGreatWhiteShark

What we all know is that RT tanks performance on these machines so I think this RT discussion is useless. Multiplats will use it in very small amounts, probably in some key areas and most of it will come from first party titles IMO.

I think it's not even worth the discussion 🤷‍♂️ both will suck at it.
I know man. I already said that but these guys were the ones saying xsx will have 80 times better raytracing which was all fake lies created by few "fans "🤣
 
Mar 27, 2020
12,211
33,666
720
Uncharted bunghole
What rumble? The trigger rumble on xbox one? Barely.

The improved haptics I mean. As for the resistive triggers I think shooters will use them the most. Plus there will be some games where you can customize how heavy the trigger is.

The DualSense doesn't have rumble in the triggers if you didn't know.
 
Last edited:

MrFunSocks

Banned
Jul 9, 2020
2,999
5,404
645
The improved haptics I mean. As for the resistive triggers I think shooters will use them the most. Plus there will be some games where you can customize how heavy the trigger is.

The DualSense doesn't have rumble in the triggers if you didn't know.
Don't expect any of the new haptic stuff to be used for anything meaningful outside of first party, it just won't happen.
 
Mar 27, 2020
12,211
33,666
720
Uncharted bunghole
So it'll be used for rumble, great. The new haptic feedback in the triggers won't.

Haptic Feedback =/= Adaptive Triggers


This is why we adopted haptic feedback, which adds a variety of powerful sensations you’ll feel when you play, such as the slow grittiness of driving a car through mud. We also incorporated adaptive triggers into the L2 and R2 buttons of DualSense so you can truly feel the tension of your actions, like when drawing a bow to shoot an arrow.

DualSense Haptics =\= DS4 Rumble.


Sorry that I wasn't clearer earlier.
 
Last edited:
  • Thoughtful
Reactions: Panajev2001a

MrFunSocks

Banned
Jul 9, 2020
2,999
5,404
645
Haptic Feedback =/= Adaptive Triggers


This is why we adopted haptic feedback, which adds a variety of powerful sensations you’ll feel when you play, such as the slow grittiness of driving a car through mud. We also incorporated adaptive triggers into the L2 and R2 buttons of DualSense so you can truly feel the tension of your actions, like when drawing a bow to shoot an arrow.

DualSense Haptics =\= DS4 Rumble.


Sorry that I wasn't clearer earlier.
But reading that it sounds like they can just be used for the usual rumble?
 
Mar 27, 2020
12,211
33,666
720
Uncharted bunghole
But reading that it sounds like they can just be used for the usual rumble?

Nope because the usual old school rumble uses a different technology.

Basically you will feel something similar to rumble but the range will be alot wider plus it will be alot more responsive. Developers have to use the new motors.

But it isn't vibration in the triggers. The triggers themselves have variable resistance.
 
Last edited:

MrFunSocks

Banned
Jul 9, 2020
2,999
5,404
645
Nope because the usual old school rumble uses a different technology.

Basically you will feel something similar to rumble but the range will be alot wider plus it will be alot more responsive. Developers have to use the new motors.

But it isn't vibration in the triggers. The triggers themselves have variable resistance.
Yeah and what I'm saying is you should just expect third party developers to make the controller rumble using whatever new things they have, and expect them to not use the variable resistance triggers for anything important.
 
Mar 27, 2020
12,211
33,666
720
Uncharted bunghole
Yeah and what I'm saying is you should just expect third party developers to make the controller rumble using whatever new things they have, and expect them to not use the variable resistance triggers for anything important.

The good thing is that both controllers have some sort of feedback in the triggers. Which means what they do with some things in the DualSense they can substitute it for rumble in the Xbox controller. Like if pulling the trigger half way does something for example. With the Xbox controller it will vibrate at the half way point and with the DualSense the resistance will just increase.
 

MarkMe2525

Member
Jul 16, 2017
1,257
1,575
445
35
Then others should stop too. You know, the ones you won't blame who fueled the discussion about triggers. But hey, lets blame me because the tag. My 1st post on 2nd page regarding triggers was quite normal and accurate. Yeah, problem is someone couldn't accept the right answer.
You mean someone wouldn't accept "your" answer
 

supernova8

Member
Jun 2, 2020
2,017
2,929
430
Paul's video's have a Sony bias to them though. He has way too many PS5 secret sauce videos. "Insiders" giving him the scoop on RDNA 3 and now Zen 3 nonsense. He does have that accent though that works for these type of videos. Even though he calls the SOC (system on chip) a sock.

I find a lot of his videos are padded out with bollocks to get his video length over 10 minute for that sweet Youtube ranking. There's a redgamingtech website and you'll often find you can read the information in a tenth of the time it takes Paul to go through stuff at snails pace (saying "truckload of salt" after every other sentence).

Don't get me wrong, generally good content but I play it at 1.5x speed.
 
Last edited:
  • Like
Reactions: Godfavor

TBiddy

Member
Mar 16, 2015
4,790
6,966
830
Denmark
Then others should stop too. You know, the ones you won't blame who fueled the discussion about triggers. But hey, lets blame me because the tag. My 1st post on 2nd page regarding triggers was quite normal and accurate. Yeah, problem is someone couldn't accept the right answer.

The 'others' aren't here to valiantly defend and talk about the PS5 in an Xbox thread.
 

SlimySnake

Member
Feb 5, 2013
10,549
28,788
1,230
I am not an amd gpu guy, but from what we already know about the amd cards on pc, overclocking does not give them sufficient amounts of power. its more like a small fraction of the overclock percentage, rather unproportional also to heat and consumption.
lets not pretend that overclocking only grants them a small fraction of performance. while it's not a 1:1 increase in perf per clock, it's by no means a fraction.

in fact, just flashing bios, and providing the 5700 enough power, these guys were able to outperform the 5700xt. thats a 7.9 tflops 36 cu card outperforming a 9.75 tflops 40 CU 5700xt simply by overclocking it and providing it enough power which is exactly what cerny said. the heat and power consumption should not be a concern to the gamers since the cooling solution should take care of the heat and the electric bill is the least of a hardcore gamer's concern.


then there are all these firestrike scores that clearly show that increasing clocks on the 5700 makes it hit the same high scores that you would for the 5700xt.

this one has an average clock of 2.12 ghz (9.7 tflops), and is hitting a 29.5k graphics score.

It matches the same graphics score of this 1.96 5700xt (10.03) below (29.6k), same CPU and fewer CUs and fewer tflops.


yes, it's not proportional, but its by no means a small fraction. thats fake news. Worst case scenario for the PS5, the extra xsx CUs offer performance thats directly proportional to tflops, and the PS5 offers slightly less performance as you boost clocks. so that 18% difference in tflops becomes 20-22% instead. again, this is the best case scenario, as we can see above, the graphics really do get a big boost from increasing clocks even with fewer CUs and fewer tflops. So best case scenario for sony might be reducing the 18% difference in tflops to a 12-15% difference in performance.
 
Last edited:
Aug 11, 2016
2,923
1,133
465
Haptic Feedback =/= Adaptive Triggers


This is why we adopted haptic feedback, which adds a variety of powerful sensations you’ll feel when you play, such as the slow grittiness of driving a car through mud. We also incorporated adaptive triggers into the L2 and R2 buttons of DualSense so you can truly feel the tension of your actions, like when drawing a bow to shoot an arrow.

DualSense Haptics =\= DS4 Rumble.


Sorry that I wasn't clearer earlier.

I had to double check to make sure I was in the XBOX Hot Chips thread. Haptic triggers? What are you guys on about?
 
Jun 13, 2020
1,761
3,375
400
its a small fraction of frequency
then you have to understand that voltage is proportionally cubic with frequency
and on top of that, you have to understand the thermals

that same dude on your video has tried to overclock 5700xt in all ways possible
see what he has to say about gpu core temperatures:

he dares you to touch it :)

now keep in mind that ps5 chip is smaller than xbox chip, therefore not only heat accumulation in higher frequencies will be much higher, but with the smaller surface it will be harder to cool off too.

I have written many times that the reason I want stripdown and hands-on with ps5 is not to see exactly how different (less powerful) it is compared to xbox, as I already know that xbox will be my main,
but to make sure my ps5 wont behave like a frying pan, like my ps4s do.
 
Last edited:

Eliciel

Member
Oct 30, 2009
1,718
869
1,095
its a small fraction of frequency
then you have to understand that voltage is proportionally cubic with frequency
and on top of that, you have to understand the thermals

that same dude on your video has tried to overclock 5700xt in all ways possible
see what he has to say about smd temperatures:

he dares you to touch it :)

now keep in mind that ps5 chip is smaller than xbox chip, therefore not only heat accumulation in higher frequencies will be much higher, but with the smaller surface it will be harder to cool off too.

I have written many times that the reason I want stripdown and hands-on with ps5 is not to see exactly how different (less powerful) it is compared to xbox, as I already know that xbox will be my main,
but to make sure my ps5 wont behave like a frying pan, like my ps4s do.

To be honest, I think at this time basically everyone wants to get a teardown of the consoles real interior. We want to know what the console is all about...finally..
 
  • Like
Reactions: TheGreatWhiteShark

SlimySnake

Member
Feb 5, 2013
10,549
28,788
1,230
its a small fraction of frequency
then you have to understand that voltage is proportionally cubic with frequency
and on top of that, you have to understand the thermals

that same dude on your video has tried to overclock 5700xt in all ways possible
see what he has to say about gpu core temperatures:

he dares you to touch it :)

now keep in mind that ps5 chip is smaller than xbox chip, therefore not only heat accumulation in higher frequencies will be much higher, but with the smaller surface it will be harder to cool off too.

I have written many times that the reason I want stripdown and hands-on with ps5 is not to see exactly how different (less powerful) it is compared to xbox, as I already know that xbox will be my main,
but to make sure my ps5 wont behave like a frying pan, like my ps4s do.
what are you talking about? small fraction of frequency? did you literally just ignore all the benchmarks to continue to spout your originally flawed argument? like wtf is the point of showing up on forums if you arent gonna look at evidence and change your point of view?

and yes, the heat goes up by a massive amount but if they can cool it then who gives a shit? rdna 2.0 offers 50% perf/watt. the 5700xt is rdna 1.0, the ps5 will NOT have those temperatures. they will be fine.
 
  • Like
Reactions: Rea
Jan 29, 2019
5,970
6,490
495
and that >40% of CU leftover advantage on xbox, it can be used for anything
The CU's computing power is dependent on their clocks--in similar architectures--this is why we compare TFs.

Obviously at equal clock speeds the machine with the most CUs would have a clear power advantage (raw 40%).... I suspect this is why Sony upped their clocks, as soon as there had been information about MS's machine being 12TF they went out and tried to look for solutions to raise the clock speeds to make the gap less jarring.

I still expect most games that are made by competent devs to run at dynamic resolutions, for a 4K game it would mean that the PS5 version would dip around 1800p at times... this or expensive effects would be run at lower resolution. Either way, it should be pretty hard to distinguish between the two (honestly at 4K upscaling has not bothered me nearly as much as it did on 1080p panels, pixel density took care of it for me, my TV is only 55inch after all).
 
Jun 13, 2020
1,761
3,375
400
what are you talking about? small fraction of frequency? did you literally just ignore all the benchmarks to continue to spout your originally flawed argument? like wtf is the point of showing up on forums if you arent gonna look at evidence and change your point of view?

and yes, the heat goes up by a massive amount but if they can cool it then who gives a shit? rdna 2.0 offers 50% perf/watt. the 5700xt is rdna 1.0, the ps5 will NOT have those temperatures. they will be fine.
when you up the clock by 15% or 18% , and you get 5% or 10% in performance, that is called a fraction.
depending on game case it can be a small fraction, like 18% overclocking for 5% gains

edit: about the part "if they can cool it who gives a shit?", I definitely give a shit if its going to be noisy for example
 
Last edited:

SlimySnake

Member
Feb 5, 2013
10,549
28,788
1,230
when you up the clock by 15% or 18% , and you get 5% or 10% in performance, that is called a fraction.
depending on game case it can be a small fraction, like 18% overclocking for 5% gains
you said small fraction.
like 18% overclocking for 5% gains
and no it's not that bad. you can see that from the examples i provided above.
 
Mar 27, 2020
12,211
33,666
720
Uncharted bunghole
I had to double check to make sure I was in the XBOX Hot Chips thread. Haptic triggers? What are you guys on about?

The DualSense doesn't have Haptics in the triggers. It has a mechanism that gives the triggers variable resistance. Like a dev can program them to be harder to pull with certain guns in a game for example. The Haptics is the new "rumble " in the DualSense. They don't function the same way that they do in the DualShock 4. Instead of using weights to generate vibration it uses sound waves instead.
 
  • Like
Reactions: Rea

Cyborg

Member
Aug 21, 2009
5,649
996
1,060
Netherlands
What we found out is that
a) ray tracing is being done in CUs and not in any secret sauce undisclosed part,
and xbox has 44% more CUs than ps5, and
b) xbox does have hardware inference acceleration, which is a must if you are seriously going to discuss ray tracing in a game and not some theoretical crap,
while for ps5 we know didly squat about inference acceleration, and cerny failed to mention such a crucial detail in his long speech.

The other day I wrote a simpleton example for a case of a game that shoots for visual parity in everything but ray tracing between the two consoles, and then xbox has more than 40% advantage on ray tracing on top.

And this 40% was with all positive ps5 assumptions, like both cpu and gpu clock at full speed, ps5 having same hardware as xbox for machine learning and inference acceleration, and without compensating for VRS or anything else xbox has on top.
Just CUs and Ghz

So about that 15%... I would advise you to not bet on it
I wish people would ignore this nonsense
 
Jun 13, 2020
1,761
3,375
400
you said small fraction.

and no it's not that bad. you can see that from the examples i provided above.
18% ovelclock for 5% gains is a fraction ~1/4. in fact anything under 1/2 is "small"
now, I already wrote before that you have to keep in mind that the relation between overclocking and voltage is cubic.
and voltage brings much heat. heat brings noise, or worse.
for you maybe its not that bad. for me it might be. I have a threshold where it becomes annoying and problematic.


I wish people would ignore this nonsense

for starters, I can ignore you.
 
Last edited:
Aug 11, 2016
2,923
1,133
465
The DualSense doesn't have Haptics in the triggers. It has a mechanism that gives the triggers variable resistance. Like a dev can program them to be harder to pull with certain guns in a game for example. The Haptics is the new "rumble " in the DualSense. They don't function the same way that they do in the DualShock 4. Instead of using weights to generate vibration it uses sound waves instead.

It sounds cool in theory. I wonder how many I'll break trying to "unjam my gun" after it locks up like an idiot.
 

SlimySnake

Member
Feb 5, 2013
10,549
28,788
1,230
18% ovelclock for 5% gains is a fraction ~1/4. in fact anything under 1/2 is "small"
now, I already wrote before that you have to keep in mind that the relation between overclocking and voltage is cubic.
and voltage brings much heat. heat brings noise, or worse.
for you maybe its not that bad. for me it might be. I have a threshold where it becomes annoying and problematic.
you are just making shit up now. 18% overlock for 5% gain is just made up shit.

We have realtime benchmarks that show that overclocks do not offer a small fraction. here you see a 19% boost in clocks offer a performance boost of 13% in actual benchmarks. Like I said earlier, it's not 1:1, but its not a small fraction here.



In the firestrike benchmarks i posted above, the higher clocked 5700 outperformed the lower clocked and higher tflops 5700xt anniversary edition.

and stop with the heat and noise nonsense. you dont know if the ps5 is gonna overheat. you dont know how noisy its going to be. it has fuck all to do with performance. Absolute nonsense.
 
Last edited:

Cyborg

Member
Aug 21, 2009
5,649
996
1,060
Netherlands
18% ovelclock for 5% gains is a fraction ~1/4. in fact anything under 1/2 is "small"
now, I already wrote before that you have to keep in mind that the relation between overclocking and voltage is cubic.
and voltage brings much heat. heat brings noise, or worse.
for you maybe its not that bad. for me it might be. I have a threshold where it becomes annoying and problematic.




for starters, I can ignore you.
Fine, hope people will do the same with the claims you make
 

Agnostic2020

Member
Oct 24, 2017
4,672
19,829
805
you are just making shit up now. 18% overlock for 5% gain is just made up shit.

We have realtime benchmarks that show that overclocks do not offer a small fraction. here you see a 19% boost in clocks offer a performance boost of 13% in actual benchmarks. Like I said earlier, it's not 1:1, but its not a small fraction here.



In the firestrike benchmarks i posted above, the higher clocked 5700 outperformed the lower clocked and higher tflops 5700xt anniversary edition.

and stop with the heat and noise nonsense. you dont know if the ps5 is gonna overheat. you dont know how noisy its going to be. it has fuck all to do with performance. Absolute nonsense.
Just had to click on ignored messages to see who u were replying to and I wasn't shocked it was him .

He is on a crusade to put out fake info on ps5 while having zero understanding of most simple things tech related. Its best to put him on ignore and let him lie to himself about ps5 .he might feel better playing x1 games on his new machine 👀🤣
 
  • LOL
Reactions: Rea

prinz_valium

Member
Oct 15, 2013
2,408
2,668
815
lets not pretend that overclocking only grants them a small fraction of performance. while it's not a 1:1 increase in perf per clock, it's by no means a fraction.

in fact, just flashing bios, and providing the 5700 enough power, these guys were able to outperform the 5700xt. thats a 7.9 tflops 36 cu card outperforming a 9.75 tflops 40 CU 5700xt simply by overclocking it and providing it enough power which is exactly what cerny said. the heat and power consumption should not be a concern to the gamers since the cooling solution should take care of the heat and the electric bill is the least of a hardcore gamer's concern.


then there are all these firestrike scores that clearly show that increasing clocks on the 5700 makes it hit the same high scores that you would for the 5700xt.

this one has an average clock of 2.12 ghz (9.7 tflops), and is hitting a 29.5k graphics score.

It matches the same graphics score of this 1.96 5700xt (10.03) below (29.6k), same CPU and fewer CUs and fewer tflops.


yes, it's not proportional, but its by no means a small fraction. thats fake news. Worst case scenario for the PS5, the extra xsx CUs offer performance thats directly proportional to tflops, and the PS5 offers slightly less performance as you boost clocks. so that 18% difference in tflops becomes 20-22% instead. again, this is the best case scenario, as we can see above, the graphics really do get a big boost from increasing clocks even with fewer CUs and fewer tflops. So best case scenario for sony might be reducing the 18% difference in tflops to a 12-15% difference in performance.


He increased a 1725MHz GPU to 2150MHz
you are just making shit up now. 18% overlock for 5% gain is just made up shit.

We have realtime benchmarks that show that overclocks do not offer a small fraction. here you see a 19% boost in clocks offer a performance boost of 13% in actual benchmarks. Like I said earlier, it's not 1:1, but its not a small fraction here.



In the firestrike benchmarks i posted above, the higher clocked 5700 outperformed the lower clocked and higher tflops 5700xt anniversary edition.

and stop with the heat and noise nonsense. you dont know if the ps5 is gonna overheat. you dont know how noisy its going to be. it has fuck all to do with performance. Absolute nonsense.
That's a 19.4% overclock for a 13.3% gain in once scenario at 85.6% more power draw.
That is quite proving is point.
 
  • Like
Reactions: FritzJ92

SlimySnake

Member
Feb 5, 2013
10,549
28,788
1,230
He increased a 1725MHz GPU to 2150MHz

That's a 19.4% overclock for a 13.3% gain in once scenario at 85.6% more power draw.
That is quite proving is point.
No it's not. I literally did the math in the post above. Go back and read his posts. He said a small fraction. How is 13.3% gain for 19.4% a SMALL fraction. He said 18% gives 5% more performance. That's a load of shit.

I cannot believe we are at a point where we are ignoring simple math.
 
  • Like
Reactions: Lysandros