Being specific Isn't spin...no matter how many times you say it. There's really nothing at stake here.
-
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.
Xbox Series X: What You Can Expect From the Next Generation of Gaming (12TF, RDNA2 & Zen 2, Hardware Raytracing, Variable Rate Shading) OFFICIAL SPECS
- Thread starter Spectromixer
- Start date
- Status
Being specific Isn't spin...no matter how many times you say it. There's really nothing at stake here.
360 was my primary that generation. I became a PC gamer in a lot of ways after that but PS4 was my main console this past generation. But given the long intervals between notable Sony first party games, I'm leaning towards starting out with a Series X for Game Pass and grabbing a PS5 on a Black Friday deal ~2-3 years in when it gets bundled with some exclusives
Microsoft's Halo servers will probably be down for a week with the amount of people trying to play. So yeah don't expect to play as soon as you get the game.
this forum is good for my english, i learn new sentences every day. =)
isn't this technically true about (almost) everything said about Ps5 in this thread that is about Series X?
LOL Pretty much. It's something that could be true or not.
RTX 2070 has ~8.5 Tflops of FP32 throughput.
RX 5700 XT has ~9 Tflops of FP32 throughput.
RDNA1 is just a tad less efficient in performance per flop than Turing.
So a 12 Tflops RDNA1 GPU should be more or less on par with RTX 2080 which is ~11 Tflops Turing GPU.
I think I would fall under this category, although I've owned the OG Xbox One, Xbox One X, PS4, and PS4 Pro this generation. I can't say for sure that the Series X will be my primary console though, as it matters which console my friends play on for multiplayer and party chat. Cross play will be the biggest deciding factor for me sense I plan on getting both consoles next generation.
Not yet. I need to see more of both consoles and their features for me to decide which one would be my go to.
The Series X sounds an absolute beast and I'm hyped for it (have a feeling Halo Infinite could be the best Halo since CE), but for me a console needs more than just raw power.
With those specs this is sounding like 799 to me.
But seriously I'd be surprised at 499
That depends on what you are comparing. If we're talking efficiency, then it would matter. If all you care about is performance (which is what most people want), then power consumption really doesn't provide an accurate comparison beyond maybe showing if 2 cards are even in the same ballpark.
I think my Rx 580 will be fine for 1080p 30 fps considering that at most the series x will be twice as powerful and next gen only games will be designed with the weaker ps5 in mind.
the PS5 isn't gonna be so much weaker and the 580 will be on borrowed time since cross-gen will end at some point
Games are scalable, pc requirements for games at 1080p 30 will continue to be much lower than either ps5 or Xbox for a long time.
They might require an ssd though.
Me but I want to see what Sony's got first. I do love Spiderman, God of War, and Naughty Dog games in general.
But tbh, I don't like fps games and competitive online games in general. I like a good single player story--open world or linear, doesn't matter. It's my escape from reality so I typically avoid online MP.
Last edited:
This is correct and to add to it the issue is that AMD puts out TF numbers based off of max peak/burst speed.(AMD way is as it should be.)AMD RX 5700 XT Max burst is 1905 and AMD says UP TO 9.75TF, typical in game is 100MHz lower.
Nvidia calculates TF numbers off of what they call boost clock, in RTX2070S is 1770MHz, but if you look at that same card in game clocks are typically 1850-1900MHz and actual peak without overclock I've seen at 2010MHz. So if you calculate TF for both on actual peak clocks then RTX 2070S is UP TO 10.29TF.
If you look at base cards in game performance you will see Nvidia card typically at 50-60MHz higher and 5-6FPS higher as well.
To me this all spells out that Nvidia cards are able to be sustainably overclocked to much higher clocks than AMD because their chips are better thermally designed.
Lock both to same MHz and you should get same TF and performance. People should really pay close attention to details of manufacturers trying to make their products look better in their own ways.
Nvidia to me is like Eneloop Pro batteries they say they are 2450mAH, but you buy the batteries and do the capacity test and you see 2700-2800mAH.
To me Nvidia is still better, but is potential overclock worth the money?
If the cheaper Lockhart is real and out at the same time, then I could see them going for $599.
If there is no cheaper Lockhart, then $499 seems the max because over that will be a harder sell for the mass market.
Nice specs, but I think I'm going to opt for a Nvidia 30 series GPU instead and wait some time before getting another console.
xCloud also kinda makes it an easier decision as I want to try that out.
xCloud also kinda makes it an easier decision as I want to try that out.
I did the opposite (PS3 only to XB1).
I'm inclined to go with XSX this gen, but there's a part of me that wants to go PS5 and gain back compat availability of all the exclusives I missed.
That said, I only ever really play Overwatch these days so I don't think it matters. :)
I know I'm not planning to upgrade my gtx 1080 until the series of cards after the 3000 series from Nvidia so I'm hoping it'll last me until then. Although I am still getting an XSX and relegating my PC to overwatch and World of Warcraft only.
what I mean is if games start to use the RT cores from the start rather than those effects being added on after the fact. how widespread it will be is up in the air but if 4A is going all in, I think there might be several others
My primary console was the 360 and I switched to the PS4 as my primary console. However I did get a Xbox One in June 2014, but my most of my game playing time was on the PS4. I also got a PS4 Pro at launch however when the Xbox One X came out I got it at launch. Now I usually only play my PS4 Pro for PlayStation exclusives and everything else is on the Xbox One X. For the next generation it will definitely be same where I will be playing almost all my third party and Xbox exclusives on the Series X and the PS5 will be used for PlayStation exclusives.
My PS4 was the first console I bought this gen, and my primary system until sometime in mid-2017. I got the Xbox One X, and it has overwhelmingly become my primary console since then - my PS4 Pro only really gets turned on when an exclusive I want to play comes out.
This time around I suspect I'll be a day one Series X buyer, followed by a PS5 once it's possible to find one in a store a few months later. I just really like what Microsoft are doing these days - Spencer has a really healthy view on the industry, and has pursued a lot of pro-consumer policies and programs that have turned me into a believer again.
It also helps that everything I hear and see about the new Halo excites me, and I'm also really interested to see what the new Fable from Playground looks like.
Because it's mostly the hardcore/core console gamers that buy launch consoles. You sell next gen launch machines to them.
Also, they aren't going to undercut the Ps5 with Series X in terms of price, so focus solely on the Series X now in terms of specs.
Once Sony reveals Ps5 and pricing, then they talk Series S/Lockhart which is basically designed to undercut Ps5 pricing.
No strategic value in revealing the Series S now, so why bother?
Agreed. In fact, they basically said as much.
They are touting GPU power now because they believe they have the advantage with that. The reason they aren't talking RAM or price is because these are the areas that Sony sandbagged them on last time around. They are clearly being more careful in certain areas this time around. They will want to get the Series X price as close to the Ps5 price as they can possibley get away with.
Both the Ps5 and Series X will start the gen selling at a loss, no doubt. With Microsoft willing to sell at much higher losses.
This is incorrect. NV's boost is both power and temperature limited meaning that you'll get higher clocks with more power at lower temperatures. Thus the actual sustained Turing boost clock is completely dependent on temperatures the chips are running on. 2010 MHz is a fairly high clock for your typical TU106 and certainly isn't indicative of the typical 2070 FE clock under load. Your typical non-FE 2070 will mostly stay at 1800 MHz, more advanced OCed models with better cooling - like the FE one - may hit 1850-1950 MHz:
Saying that 2070, even 2070S is "up to 10.29 TF" is completely misleading. There really is no "up to" on Turing, and thus there is no "way as it should be".
Of course not. Clocks has nothing to do with peak performance you're getting from a product since it is totally dependent on an architecture of said product. You may even out the theoretical Flops this way, making sure that two different chips have similar number of FP ALUs running at the same clocks - but this has little to do with actual performance since performance isn't a result of pure FP math throughput and even before going to things like ALU utilization (which is where this argument fails for GCN vs NV's several latest architectures) or ROPs/DCC we should consider the fact that Turing runs INT calculations on dedicated SIMDs while RDNA runs them on FP32 SIMDs meaning that your peak Flops figure doesn't even tell you how much work is actually being done inside the chip on these FP ALUs.
What overclock and what money?
The way NV declares boost clocks is safe since it's a worse case figure which is likely to be higher in practice, with no user intervention. This is a bit similar to how Intel states boost clocks for their CPUs these days.
The way AMD declares boost clocks for Navi is a bit misleading since it's a best case figure which is likely to be lower in practice, and a user will have to apply an overclock to actually keep the card at such clocks.
Yea that'll be the real test, everyone with a Series X and Game Pass will try to play that at launch, let's see how Azure gets leveraged.
People with PC and XB1 are going to try playing on launch as well. This will likely have the highest player count ever for a Halo game.
Yea that's why I said those with Game Pass, especially since they will bundle at least one month with the purchase of Series X.
And then there's gonna be hundreds of thousands buying the game separate, it's gonna be a bloodbath on the server side of things.
The Series X is more than twice the GPU of their last console (One X) and the Ps5 is more than twice the GPU power of their last console (ps4 pro)...sounds more reasonable and expected than "wtf!"
Yea I think this is a case of Sony being caught a little flatfooted when it comes to the future of AA/AAA 3rd party gaming platforms. (Which, the PlayStation/Xbox success mostly comes from).
Their 1st party will still be strong, but they are going to see their 3rd party sales from their services ecosystems decline going forward. Microsoft has been much forward thinking concerning this market. We should see a lot of pivoting by Sony in the next couple of years (similar to how we saw Microsoft regroup and shift gears halfway through this last gen)
Is there evidence that shows that it is mostly the hardcore audience that buy consoles at launch?
Actually, I thought about getting PS5 day 1 but that isn't a terrible idea. I wanted to upgrade my PC but I didn't feel the 2070 was that much of an upgrade over my 1070 (save a slight bump in fps and rtx). Sony first party has been my go to this gen but since the XBOX ONE X I've been noticing a trend on games i want to play on the sofa and not caring as much that I am not getting 100+ fps in 1440p over whatever the X port was doing.
My gut tells me something about XBOX's BC that may have been discussed but I am wondering if the way the BC team is treating BC this go round is more local than you downloading the pre-modified file. Meaning, even if a game is delisted but you have your disc, I think you will still be able to use it. I don't know why I have this feeling but it seems something in the way BC was discussed on the Major Nelson podcast that made me think this way. If true I will be playing JSRF day 1.
exactly. Technically "ps4 pro boost mode" is enhanced BC.
it could very well work like the ps4 pro handles ps4 games...some devs will really enhance the visuals like Shadow of the Tomb Raider (improved physically-based rendering, hardware tessellation, anisotropic filtering, additional dynamic foliage, draw distances, etc)....some devs will just use boost pro mode like Uncharted 4 (only increase resolution and steadier framerate)...some devs won't really bother like Bloodborne.
who knows? If it is just booting the Ps5 into ps4 pro mode, that would be a real shame though
Not really, expect for the fact that pretty much only the core/hardcore console gamer would spend $400+ for a dedicated gaming console at launch.
mainstream casuals who buy 1-2 games a year...at most...are not bothered to spend that much money at launch.
Technically "ps4 pro boost mode" is enhanced BC but if that's the context Matt was using it with PS5, then it was misleading as "enhanced" isn't used in that context.
I'm not sure what you mean specifically in your question. Do you mean, there would be some software update for a PS4 game that would only be applied when running it on a PS5?
Also small point of clarification, the XB1 absolutely has hardware features built in to make the "emulation" possible. It has hardware-based acceleration to decompress the Xbox 360's texture format, for example. So fom what I understand, it's not really emulation at all but rather recompiling the games to run on the XB1 architecture. But I have no idea if there are other translations or something going on.
I would expect XSX to XB1 BC to be substantially more straightforward, same for PS5 to PS4.
I think the fact that playstation is pretty big on social media is indicative that casual gamers are excited for new hardware this year as well. mainstream don't buy as much software but they'd be enticed to play games on the next generation rather than the generation that has been around for the past 7 years.
What I'm asking is, say a Ps4 game gets a software update. It's an update that doesn't affect the game being played on the Ps4 at all. But if the game was played on Ps5, that software update would go into effect to enhance the experience
Sorry I'm on mobile so my formatting isn't fancy.This is incorrect. NV's boost is both power and temperature limited meaning that you'll get higher clocks with more power at lower temperatures. Thus the actual sustained Turing boost clock is completely dependent on temperatures the chips are running on. 2010 MHz is a fairly high clock for your typical TU106 and certainly isn't indicative of the typical 2070 FE clock under load. Your typical non-FE 2070 will mostly stay at 1800 MHz, more advanced OCed models with better cooling - like the FE one - may hit 1850-1950 MHz:
Saying that 2070, even 2070S is "up to 10.29 TF" is completely misleading. There really is no "up to" on Turing, and thus there is no "way as it should be".
Of course not. Clocks has nothing to do with peak performance you're getting from a product since it is totally dependent on an architecture of said product. You may even out the theoretical Flops this way, making sure that two different chips have similar number of FP ALUs running at the same clocks - but this has little to do with actual performance since performance isn't a result of pure FP math throughput and even before going to things like ALU utilization (which is where this argument fails for GCN vs NV's several latest architectures) or ROPs/DCC we should consider the fact that Turing runs INT calculations on dedicated SIMDs while RDNA runs them on FP32 SIMDs meaning that your peak Flops figure doesn't even tell you how much work is actually being done inside the chip on these FP ALUs.
What overclock and what money?
The way NV declares boost clocks is safe since it's a worse case figure which is likely to be higher in practice, with no user intervention. This is a bit similar to how Intel states boost clocks for their CPUs these days.
The way AMD declares boost clocks for Navi is a bit misleading since it's a best case figure which is likely to be lower in practice, and a user will have to apply an overclock to actually keep the card at such clocks.
Sorry, I wasn't talking about sustained numbers because when people say that RX 5700 XT is 9.7TF card they are not talking about sustained numbers but actual maximum observed peak. Same should apply to Nvidia in that case and observed has been 2010, card can peak over 100MHz more than what it's overclocked to. AMDs game clock measurement is equivalent to Nvidia boost clock measurement so if we are saying AMD card is specific TF based on peak numbers then we must do the same for Nvidia card otherwise we are talking about apples vs oranges.
Of course there is up to, cards have varying clocks so TF is constantly changing so up to is legit. It is even more deceiving measuring TF numbers of your card at 1770MHz when your card is operating at 1850-1900MHz. I posted two videos in the other thread with cards claiming 1770MHz but running up to 2000MHz in games without OC.
You won't get peak performance when your clocks are running lower so of course that they have a lot to do with it.
I do agree that one company does estimates on best case scenario and the other at worst case and this is why we can't say that Touring TF is higher than Navi TF. Only way to measure performance differences between Nvidia TF and AMDs is to clock two similar cards to same clocks. In case of RX 5700 XT and RTX 2070 Super they both have same number of cores so same frequencies should give same TF at least in raw theoretical calculation of 2560 cores x 2 x MHz. This is when we see that Navi and Touring are actually on par 7.68TF vs 7.68TF performance in case of 1500MHz screenshot. There are always outliers where game engine favors one over the other too.
yea sorry, I didn't specifically mean the official "boost mode" toggle that helps even out framerates without any patching or dev input. My bad.
its my generic term for ps4 games whose only enhancement on ps4 pro are a higher resolution and better framerates. It's the very minimal of "enhancements", but still enhancements regardless.
yes Matt's response was nice, but still too vague seeing as it doesn't tell us anything new in this regard. Sony already confirmed BC to ps4 games (only ps4 games as of now). The rumors point to Sony using hardware to have the Ps5 function as ps4 pro hardware...which means its possible the BC to ps4 is the generally the same as what the ps4 pro does with ps4 games. Which means some have a lot of enhancements, some very minimal, some not at all.
I hope Sony puts more work into making their published games run on better hardware...way better than they treated the ps4 pro (see Uncharted 4 and Bloodborne). I didn't buy a ps4, but I totally would buy a Ps5 if I could get TLOU2 and Ghost running with and utilizing full Ps5 power. Something tells me I won't, really.
Anxiously waiting to hear more about the Ps5! The high of the new logo reveal is slowly wearing off
Oh ok, gotcha. I can only assume when Matt said enhanced BC that the PS5 will somehow enhance all of the PS4 games further than how they played on PS4 Pro.
I would love that to be true. But color me skeptical how much effort Sony is going to take on their own games. Some 3rd party devs/publishers choosing to go all out with enhancement patches? Yea I can see that.
I think it's a long time before anyone ship a pc game that has rt hardware as a minimum requirement.
We can, we know sustained under load clocks for both so we can get the ratio of performance per flop from similarly performing chips. With two RDNA chips out already this estimation is fairly precise too.
There's no point in running completely different GPU architectures on the same clocks. You're doing more than evening out the tflops this way and the results will be useless anyway since Flops aren't the only metric defining performance.
True, but just posting head to head perfomance numbers in % and FPS is not accurate measurement because in each of the tests Nvidia card runs at higher average clocks and thus operating at higher TF.
It's even more useless and deceiving running cards at different clocks and saying look this "lower" clock beats the higher when it's actually not backed up by facts. When both cards have 2560 cores running them both on same clocks you are comparing performance of same TF numbers. Nvidia thermal design is better so just running a maximum test doesn't give you a glimpse of Navi vs Touring TF performance differences.
Most accurate test would be at base clock of 1605MHz without any overclocking from manufacturers or users. But 1500MHz is much closer to that native clock than 1800MHz is. At higher clocks you are testing performance along with thermal design, to just test performance cards need to be at clocks where they aren't near thermal capacity.
I'm not talking about testing cards overal capability, we are talking about measuring 1TF navi vs 1TF Touring performance differences.
- Status