OK, now consider the RTX 3070 price and TDP. That kind of hardware will still be out of reach at console prices holiday 2020. The form factor of the Series X at least allows for a proper cooling solution, but I'm still very skeptical.
I would say that you are just well grounded in reality. :)
With NAVI 10, it's quad geometry, 64 ROPS with 4MB L2 unified cache follows TU106's quad GPC, 64 ROPS with 4MB MB L2 unified cache basic topology. Both top NAVI 10 and TU106 models has similar TFLOPS range and results.
RDNA with 12 TFLOPS, it depends on geometry and ROPS scaling.
TU104 has six GPC, 64 ROPS MB 4MB L2 unified cache basic topology. TU104 on RTX 2070 Super has five GPC units.
TU102 has six GPC, 96/88 ROPS MB ~6MB L2 unified cache backed by up to +17 TFLOPS FP32 compute.
I'm on Team Fake. 12TF is too much for a consumer priced console. Especially since the lower spec console "leaked" specs are 4TF. That seems very silly. That CPU clock speed is horseshit as well. You can't cool that with a console form factor and the TDP would then be far too low. Wake up folks.
For next year PC GPU products, 12 TFLOPS would be about RTX 3070 level GPUs.
RTX 2080 shift down to RTX 3070
OK, now consider the RTX 3070 price and TDP. That kind of hardware will still be out of reach at console prices holiday 2020. The form factor of the Series X at least allows for a proper cooling solution, but I'm still very skeptical.
NAVI is not GCN, and it's TFLOPS are nearly comparable to Turing's TFLOPS.
TU106 based RTX 2070's TFLOPS, quad GPC(geometry), 64 ROPS and 4MB L2 cache is similar to NAVI 10 RX 5700 XT results.
RDNA with 12 TFLOPS, it depends on geometry and ROPS scaling, but RTX 2080 Ti has 17 TFLOPS FP32 shader compute besides it's higher GPC and ROPS scaling.
In terms of raw TFLOPS power, NVIDIA has the high ground with Turing.
Both NAVI and Turing has Wave32 compute format with GCN's Wave64 losing GpGPU compute format war.
From https://news.xbox.com/en-us/2019/12/12/microsoft-unveils-xbox-series-x/
From a technical standpoint, this will manifest as world-class visuals in 4K at 60FPS, with possibility of up to 120FPS, including support for Variable Refresh Rate (VRR), and 8K capability. Powered by our custom-designed processor leveraging the latest Zen 2 and next generation RDNA architecture from our partners at AMD, Xbox Series X will deliver hardware accelerated ray tracing and a new level of performance never before seen in a console. Additionally, our patented Variable Rate Shading (VRS) technology will allow developers to get even more out of the Xbox Series X GPU
Scarlet GPU includes Variable Rate Shading (VRS) feature which is missing with NAVI and GCN GPUs.
NVIDIA Turing has Variable Rate Shading (VRS) feature to boost frame rates. Read https://developer.nvidia.com/vrworks/graphics/variablerateshading
Microsoft has patented Variable Rate Shading (VRS) technology.
Variable Rate Shading is better than Sony's hardware checkerboard semi-custom addition to PS4 Pro's GPU.
I see only couple of possibilities. First is that MS has decided to throw the profitability ratio out of the window to get ahead of Sony. Or these rumored specs are just that.... rumors with flaws ignoring cost and demands on maintaining the hardware.
Even on 7nm+ your still looking at around 55w TDP for the CPU at 3.5ghz boost and for the NAVI gpu, to be at 12 TFLOP(using NAVI performance metrics not GCN) would require some hefty cooling and power. Since even the RX 5700XT with its 9.7 TFLOP has a 220w max TDP. So even with 7nm+ your only shaving that 220w down to ~200w. But a 12 TFLOP NAVI would need more power and cooling than its ~9.7 TFLOP cousin.
AMD ussually applies higher voltages for it's PC GPUs while Microsoft has hardware auto-tuning under voltage tricks for it's X1X GPU.
X1X GPU's 6 TFLOPS power consumption (150 watts) is about half of Vega 64's 295 watts with 12 TFLOPS.
RX-580's 6 TFLOPS consumes 198 watts https://tpucdn.com/review/gigabyte-aorus-rx-580-xtr/images/power_average.png Vega 64 can not be built from Polaris silicon design maturity.
@sealionact: How far are you gonna reach? The case (Fractal Node 202) used in Bassman's PC partpicker build can fit GPUs up to 12.2 inches and can fit cards such as the GTX 1080ti and RTX 2080. Also they make mini sized variants of GPUs now so size for mini itx cases isn't something you have to worry about and you get more breathing room. But even then in this video the guy has PC build with the same case and GPU with the only major difference being that he has a faster CPU than the one used in Bassman's build and his average temp during gaming for both CPU and GPU was in the low 70s and he was using the stock cooler and only one fan added to the case. So there goes the loud and hot argument.
Loading Video...
Very few high demanding games on the Xbox One X run at true 4k without heavily compromising the graphics. Games like RDR2, The Division 2 and Battlefield 1/V use mostly a mix of a medium and low/lower than low settings to be able to run at 4k.
Not all of PC's AAA exclusives come from Xbox in fact PC likely has much more from other 3rd parties and it also gets access to games that aren't on the Xbox platform such as the majority of Japanese games for example. PC isn't as dependent on big budget AAA titles since even smaller titles sell many copies and receive critical reception like Disco Elysium for example. Nevermind that a fair amount of games revealed at TGA were only for PC or PC and other consoles not including Xbox.
@BassMan: Oh sweet jesus, look at that abomination of a pc, and try and compare it to the series x with a straight face. Please, I double dare you.
Gamecubes don't have to be that ugly and cost $2000.
It doesn't look bad and it is much smaller than the XSX. Also, that video is almost 2 years old. So, those parts are not worth as much now. Besides, the point of the video was to show the size of the case. All this talk about the importance of a small size and now that XSX is big and looks like a PC.... no talk of size, only deflections towards price and style. LOL
It is clear that consoles are becoming more and more like PCs, but without all the capabilities. So, you might as well just buy a PC.
@BassMan: Oh sweet jesus, look at that abomination of a pc, and try and compare it to the series x with a straight face. Please, I double dare you.
Gamecubes don't have to be that ugly and cost $2000.
It doesn't look bad and it is much smaller than the XSX. Also, that video is almost 2 years old. So, those parts are not worth as much now. Besides, the point of the video was to show the size of the case. All this talk about the importance of a small size and now that XSX is big and looks like a PC.... no talk of size, only deflections towards price and style. LOL
It is clear that consoles are becoming more and more like PCs, but without all the capabilities. So, you might as well just buy a PC.
As long as you're prepared to pay more for a bigger, louder, uglier box or a lot more for a similiar sized box and don't mind cheats, don't get fazed by how Windows update and Driver updates are handled, and don't mind that if you want to use a controller online, your opponents are probably using kb/m.
@sealionact: Since we are just going round in circles, I am not going to waste any more time. Enjoy your honey, but we all know what you are really eating.... shit. LOL
@sealionact: Since we are just going round in circles, I am not going to waste any more time. Enjoy your honey, but we all know what you are really eating.... shit. LOL
Enjoy your Personal Computer ... lol. Hey, maybe somebody will make a thread about PCs for you to comment on, instead of having to try and hijack Xbox threads eh?
@sealionact: How far are you gonna reach? The case (Fractal Node 202) used in Bassman's PC partpicker build can fit GPUs up to 12.2 inches and can fit cards such as the GTX 1080ti and RTX 2080. Also they make mini sized variants of GPUs now so size for mini itx cases isn't something you have to worry about and you get more breathing room. But even then in this video the guy has PC build with the same case and GPU with the only major difference being that he has a faster CPU than the one used in Bassman's build and his average temp during gaming for both CPU and GPU was in the low 70s and he was using the stock cooler and only one fan added to the case. So there goes the loud and hot argument.
Loading Video...
Very few high demanding games on the Xbox One X run at true 4k without heavily compromising the graphics. Games like RDR2, The Division 2 and Battlefield 1/V use mostly a mix of a medium and low/lower than low settings to be able to run at 4k.
Not all of PC's AAA exclusives come from Xbox in fact PC likely has much more from other 3rd parties and it also gets access to games that aren't on the Xbox platform such as the majority of Japanese games for example. PC isn't as dependent on big budget AAA titles since even smaller titles sell many copies and receive critical reception like Disco Elysium for example. Nevermind that a fair amount of games revealed at TGA were only for PC or PC and other consoles not including Xbox.
Absolutely not a reach. You want x1x level graphics on PC, you will pay more for a bigger, louder, arguably uglier box. You want a similiar sized pc with similiar performance with any attempt at aesthetics, you will pay a lot more. Heck, Bassman's attempt at a better, smaller PC went for $2400!
From https://news.xbox.com/en-us/2019/12/12/microsoft-unveils-xbox-series-x/
From a technical standpoint, this will manifest as world-class visuals in 4K at 60FPS, with possibility of up to 120FPS, including support for Variable Refresh Rate (VRR), and 8K capability. Powered by our custom-designed processor leveraging the latest Zen 2 and next generation RDNA architecture from our partners at AMD, Xbox Series X will deliver hardware accelerated ray tracing and a new level of performance never before seen in a console. Additionally, our patented Variable Rate Shading (VRS) technology will allow developers to get even more out of the Xbox Series X GPU
Scarlet GPU includes Variable Rate Shading (VRS) feature which is missing with NAVI and GCN GPUs.
NVIDIA Turing has Variable Rate Shading (VRS) feature to boost frame rates. Read https://developer.nvidia.com/vrworks/graphics/variablerateshading
Microsoft has patented Variable Rate Shading (VRS) technology.
Variable Rate Shading is better than Sony's hardware checkerboard semi-custom addition to PS4 Pro's GPU.
Another thing i notice they claimed twice the xbox one X power,that doesn't mean is 12TF as RDNA hardware is efficient than Polaris on xbox one X,so probably 9 or 10TF would double the xbox one X.
@sealionact: The point of that build he showed was how far you can go in building a small PC without compromising performance and being a better alternative to NUC for similar price point. That rig isn't top of the line anymore. Not to mention that the pcpartpicker build he linked I corrected you on your assumption that you couldn't fit high end cards into the Fractal Node 202 and it would be very hot and loud.
You keep exaggerating noise and size but have you ever seen a mini itx gaming build? I have one with a GTX 1080 and Ryzen 5 1600 with stock cooler and it never gets loud even when playing demanding games. Cooling fan tech has come a long way in effectively cooling hardware while not being loud. The fans on even high end GPUs and SLIs/crossfire set ups and stock CPU coolers at full load ranges from 38-46 DBa which is very reasonably quiet and in the same range of the Xbox One X noise levels. A lot of these GPUs are more powerful and run a bit quieter.
As for size you can easily see in the thumbnail of the video I posted that the Fractal Node 202 is only about 2 inches bigger than the PS4 and Xbox One if that is too big for you than that is some serious nitpicking That machine is still small enough to be carried around in a backpack and has no problem fitting into gaming set ups just like a console. Same with how much emphasis you put on aesthetic. I can understand if you don't like RGB but some of these mini ITX cases without it are still pretty sleek/minimalist and wouldn't look out of place in a living room set up. And aesthetics don't matter nearly as much as functionality since after all you're going to be looking at the game on your screen 95% of the time not your console. Need to get priorities straight if the appearance of the box matters far more than the functionality and capability of the components inside it.
For "Din't see this posted so what do you think system wars is 12TF RNDA enough for 4k 60FPS?" question.
1. VRS feature enables programmers to conserve shader/ROPS power at selective area while preserving native geometry edge resolution.
2. VRR feature reduce the need for locked 60 Hz. I have 4K FreeSync HDR LG Nanocell monitor and it's good for missing 60 Hz render target.
@tormentos said:
Another thing i notice they claimed twice the xbox one X power,that doesn't mean is 12TF as RDNA hardware is efficient than Polaris on xbox one X,so probably 9 or 10TF would double the xbox one X.
Microsoft's 4X CPU claim includes twice ALU count per core, hence it has raw math improvment in the mix.
Zen 2 has four ALUs while Jagaur has two ALUs, hence improvement is 2X at the same clock speed on this aspect.
Zen 2 has two 256bit FADD and two 256 bit FMUL while Jagaur has 128bit FADD and 128bit FMUL, hence improvement is 4X at the same clock speed on this aspect.
-----------
Polaris on X1X has Vega style 32 ROPS with 2MB render cache which is half of Vega 64 ROPS with 4MB L2 cache.
Pure Polaris RX-580's 32 ROPS doesn't have 2MB render cache, hence X1X GPU's ROPS with 2MB cache improvements place it between RX-580 and Vega 64/II/NAVI 10.
X1X GPU already has Variable Rate Shading (VRS) technology which is missing on all PC GCNs and PC NAVI 10!
X1X GPU already shown 2X rendering resolution despite of 1.42X raw TFLOPS increase over PS4 Pro/PS4/XBO's classic GCN resolution scaling vs TFLOPS increase e.g. Red Dead Redemption 2
Loading Video...
At X1X settings, RX 580 could NOT maintain locked 30 hz (down to 25 fps) which is different from X1X's locked 30 hz results. MSI Radeon RX 580 8GB ARMOR MK2 OC was used for the test which has 6.36 TFLOPS. RX-590 with 7.1 TFLOPS (Polaris) would be needed to match X1X's result with this game.
X1X GPU's 32 ROPS with 2MB render cache enables software micro-tile render methods which is impossible on RX-580's 32 ROPS.
Phil Spencer told GameSpot. “So when we do the math, we’re over eight times the GPU power of the Xbox One, and two times what an Xbox One X is.”
The context for "two times what an Xbox One X" is math, hence it's pure math compute.
Against XBO, over 10.1 TFLOPS
Against XBO S, over 11 TFLOPS
Against X1X, 12 TFLOPS.
RX 5700 XT/5700 XT AE has RX Vega II raster power with hardware culling bug fixed.
Unlike AMD, NVIDIA scales thier raster power with TFLOPS increase. Phil is just talking about GPU math power between X1X and XSX. There's nothing on geometry input and ROPS read/write units count (raster power).
Near all aspect 2X for math and raster NAVI type GPU config over X1X GPU.
CU = 12 TFLOPS vs 6 TFLOPS. ,Semi-custom.
ROPS = 64 pixel vs 32 pixels, NAVI 10 already provide this.
Geometry = 8 inputs vs 4 inputs, NAVI 10 already provide this.
TMU = twice the width per TMU, NAVI 10 already provide this.
L2 type Cache = 8 MB vs 4 MB (X1X GPU has 2MB L2 cache + 2MB render cache). Semi-custom.
@NoodleFighter: My point still stands, and is common knowledge. Consoles - in this case xbox - deliver a happy medium of performance, value and aesthetics. PC offers better performance at a price that I am not willing to pay...theres no point.
So basically all those rumors about the ridiculously powerful consoles were true. Alot of hermits on this forum look dumb as shit right now.
Don't worry they will argue that the next release of GPUs and CPUs is going to blow consoles out of the water.
Which is true. Nvidia hasn't even released 7nm GPUs yet. There are huge gains to be had. These consoles will be nothing special when they actually release. It's funny seeing console gamers hyping these 'powerful consoles' which are still a year away. What are they playing on now? Shit boxes that top out at 6 TFlops. Meanwhile, PC gamers are enjoying high end rigs with powerful hardware that people can buy now and even better tech on the horizon. There is no winning for consoles when it comes to hardware. They will always be shit compared to what is available on PC.
For "Din't see this posted so what do you think system wars is 12TF RNDA enough for 4k 60FPS?" question.
1. VRS feature enables programmers to conserve shader/ROPS power at selective area while preserving native geometry edge resolution.
2. VRR feature reduce the need for locked 60 Hz. I have 4K FreeSync HDR LG Nanocell monitor and it's good for missing 60 Hz render target.
@tormentos said:
Another thing i notice they claimed twice the xbox one X power,that doesn't mean is 12TF as RDNA hardware is efficient than Polaris on xbox one X,so probably 9 or 10TF would double the xbox one X.
Microsoft's 4X CPU claim includes twice ALU count per core, hence it has raw math improvment in the mix.
Zen 2 has four ALUs while Jagaur has two ALUs, hence improvement is 2X at the same clock speed on this aspect.
Zen 2 has two 256bit FADD and two 256 bit FMUL while Jagaur has 128bit FADD and 128bit FMUL, hence improvement is 4X at the same clock speed on this aspect.
-----------
Polaris on X1X has Vega style 32 ROPS with 2MB render cache which is half of Vega 64 ROPS with 4MB L2 cache.
Pure Polaris RX-580's 32 ROPS doesn't have 2MB render cache, hence X1X GPU's ROPS with 2MB cache improvements place it between RX-580 and Vega 64/II/NAVI 10.
X1X GPU already has Variable Rate Shading (VRS) technology which is missing on all PC GCNs and PC NAVI 10!
X1X GPU already shown 2X rendering resolution despite of 1.42X raw TFLOPS increase over PS4 Pro/PS4/XBO's classic GCN resolution scaling vs TFLOPS increase e.g. Red Dead Redemption 2
At X1X settings, RX 580 could NOT maintain locked 30 hz (down to 25 fps) which is different from X1X's locked 30 hz results. MSI Radeon RX 580 8GB ARMOR MK2 OC was used for the test which has 6.36 TFLOPS. RX-590 with 7.1 TFLOPS (Polaris) would be needed to match X1X's result with this game.
X1X GPU's 32 ROPS with 2MB render cache enables software micro-tile render methods which is impossible on RX-580's 32 ROPS.
Please don't start with your bullshit theories man the RX580 can beat the xbox one X in most games regardless of the xbox one X GPU having VRS,is juts another of your tile resources DX12,ESRAM buffoonery that never helped the xbox one in anything.
10TF RDNA would surely double the XBO X power,you don't need 12TF which will surely more than double it.
So when MS say double the xbox one X power that is the VERY best scenario on a test done with no AI or somehting like Forza,if MS says double it probably is little less than double.
And considering that YOU try endeless to pass the xbox one X as been better than a 1080TI and 1070TI then backtraqued on iti should not say any more.
Oh and rumors still point to the PS5 been stronger,and again MS for the second time didn't call scarlet (Now xbox Series x) the most powerful console ever built like they did with the xbox one X, they call it the most powerful xbox console THEY have built so you may end again crying about secret souces for some years to come i would not miss it..lol
Nobody takes @ronvalencia seriously with his awful track record.
Yep he already has start lobying for the xbox series X.
If MS claimed double the power of the xbox one X,it surely will not be 12TF as RDNA hardware is much more efficient that Polaris on xbox one X.
And that is taking MS word as been 100% true you know since they will picture the very best scenario not the worse.
Also is it me or 4X CPU power from the xbox one X doesn't seem impressive either considering that the CPU now is Ryzen.
I think it will be powerful just not super powerful it probably runs hot hance its PC style cooling system.
@r-gamer said:
So basically all those rumors about the ridiculously powerful consoles were true. Alot of hermits on this forum look dumb as shit right now.
Is will be powerful but we have to see,the cooling design is very good and alto i don't love the design is great that they choose one that benefit cooling over style.
@Pedro said:
@r-gamer said:
So basically all those rumors about the ridiculously powerful consoles were true. Alot of hermits on this forum look dumb as shit right now.
Don't worry they will argue that the next release of GPUs and CPUs is going to blow consoles out of the water.
And it will be true but again it will cost a pretty penny to get it.
I do think that if both MS and sony take a hit in profits in order to sell cheaper,they can greatly undercut what a PC with similar specs will cost you.
Only double? The Xbox One X was a failure so double the specs is still double the failure. This is the bare minimum a console should be right now and you're telling me xbox is to last to 2027? What kind of settings will you be playing the games around 2025, 720p@20fps at low settings with everything else turned off?
@Pedro: I can't put a number on the performance gains because I don't know the details of their next architecture. Nvidia does not always follow their roadmap. We were expecting Volta for their last consumer GPUs and then they came out of nowhere with Turing and their RTX line. So, we will have to wait and see.
@Pedro: I can't put a number on the performance gains because I don't know the details of their next architecture. Nvidia does not always follow their roadmap. We were expecting Volta for their last consumer GPUs and then they came out of nowhere with Turing and their RTX line. So, we will have to wait and see.
Well the fact than Nvidia are still on 14nm cough cough"12nm"... with next year, them going to 7nm will allow them to nearly double the transistor count with same die size or double the performance per watt . While AMD has been on 7nm this year with Radoen VII and Navi.
Even with them going with 7nm+ your only going to see 10-15% gains in transistor density or savings in power usage. So Nvidia will have some major gains just in that regard. Not even taking in account of new architecture improvements.
@BassMan: Exactly, everything I have seen you say is true. I like you as you speak facts. Thank goodness for smart people like you. Keep fighting the lies and ignorance around here, I appreciate what you do. People these days don't fight for the truth enough.
Why are you people counting teraflops between two different architecures?..
If it was 12 TFLOPS they would have been shouting it from the roof tops in the reveal, X1X 6TFLOPs compared to Navi is not the same so a 12 TFLOP Navi card is not 2x the power of X1X it's far more.
2x the performance of a X1X in raw power/performance is a aftermarket overclocked 5700 or a underclocked 5700XT.
It will be less than 10TFLOPs almost gaurenteed, do you guys not connect the dots?... Why do you think they have that box layout?... Navi runs hot and needs better cooling even at 36 CU at 1700Mhz a red dragon model hits 84c.
12 TFLOPs?... Stop it.
It won't be any more powerful than a 5700XT... Not even close, it will be running of RDNA2 which will be more efficient but not enough to add 4 TFLOPS of power and ray tracing cores with the same TDP with a one year release difference.
They are being vague and not mentioning TFLOP's because it won't look good for them on paper because people will see 6TFLOP on X1X and then 8-9.5 TFLOPs on a new console and be confused just like you are now.
A 5700 with 36 CU's with 8 TFLOPS is already giving you almost 2x the performance in games... Which is clearly what Phil was talking about.
One thing I would like to point out that people can't seem to grasp here and the non "technical" people in the gaming industry who are jumping on the 2x 6TFLOPs means a 12 TFLOP navi GPU!
First if he is saying that its 2x the power and it refers to performance then a 5700 overclocked is just that BUT if he is referring to TFLOPs... Then again a X1X is using GCN so 2x 6TFLOPs on GCN is a Vega 64 which performs the same as a 5700.
A 12 TFLOP Navi card is not 2x the performance of a 6 TFLOP GCN based X1X, a 12TFLOP Navi would be almost 3x the performance and would compete against a RTX 2080 S... Do you guy's not understand how ridiculously unrealistic that is?... A 5700XT runs HOT and draws more than 200w and is 10.5TFLOPs, you think in 1 year AMD has managed to get 12 TFLOPs with Ray Tracing cores from Navi and have it running 30% less power than it?
I understand why websites do this because it draws people to their sites and I understand why you guys believe it... Half of you don't understand a f***ing thing and find it too complex to apparently even game on PC.
They are being vague for a reason and they are focusing on the NVMe and 8 Core Zen 2 because those sound impressive... Remember when everyone said oh it will be faster than anything on PC because it uses Gen4 PCIe lanes?... Right rumour says slower than 2Gbps which is not even close to what a Gen3 NVMe max's out at and no mention of capacity?
The CPU will be MUCH slower than a desktop 8 core zen 2 by 1GHz or more
The NVMe is rumoured to be running at slower than 2Gbps?... I'm sorry that is the speed of a cheap NVMe.
2x the performance of a X1X in GPU?... That's 5700 territory.
12 TFLOPs... GEN4 NVME faster than any PC... 8 Core Zen two... Build me a PC like that?...
Sorry we slower than 2Gbps NVME's don't exist at good capacities and a Zen 2 on desktop doesn't run at 3.5GHz or less... We can't build it because not only does it not exist but its actually weaker than what we have already.
When they reveal the 8-9.5 TFLOP GPU you guys will pretend this never happened, just like no one is talking about GEN4 NVME speeds better than anything on PC... My NVMe runs at 3.3Gbps and is a year old.
Please don't start with your bullshit theories man the RX580 can beat the xbox one X in most games regardless of the xbox one X GPU having VRS,is juts another of your tile resources DX12,ESRAM buffoonery that never helped the xbox one in anything.
10TF RDNA would surely double the XBO X power,you don't need 12TF which will surely more than double it.
So when MS say double the xbox one X power that is the VERY best scenario on a test done with no AI or somehting like Forza,if MS says double it probably is little less than double.
Your arguments are real bullshit..
Xbox One X GPU has VRS like feature. https://www.eurogamer.net/articles/digitalfoundry-2017-the-scorpio-engine-in-depth
Andrew Goossen tells us that the GPU supports extensions that allow depth and ID buffers to be efficiently rendered at full native resolution, while colour buffers can be rendered at half resolution with full pixel shader efficiency.
-------------
This feature is missing in all AMD GCN and NAVI PC GPUS. NVIDIA included VRS feature with Turing for the PC market!
For my tile resource argument, it's faking W5000's 2GB GDDR5 like results. You can't handle both tile resource workaround and W5000 (12 TFLOPS GCN with 256 bit 2GB GDDR5) like upper limit aspects!
You can't grasp extra development work is needed for XBO's workaround to reach near W5000 results, hence it has less performance consistency when compared to flat 2GB memory architecture W5000 vs flat 2GB memory architecture R7-265.
The real bullshit comes from you.
@tormentos said:
And considering that YOU try endeless to pass the xbox one X as been better than a 1080TI and 1070TI then backtraqued on iti should not say any more.
That's a false narrative! You didn't understand the kindergarten IF conditional statement.
GTX 1070 Ti (GP104-300-A1, November 2017) did not exist during X1X's reveal and during my IF conditional X1X GPU includes XYZ feature forum postings. LOL.
X1X GPU included other features such as like VRS and ROPS with 2MB render cache instead!
You missed my GTX 980 Ti vs PS4 Pro with rapid pack math Mantis Burn Racing comparsion with entire GPU vs GPU context.
You purposely deleted my IF conditional statement with your bullshit narrative.
@tormentos said:
10TF RDNA would surely double the XBO X power,you don't need 12TF which will surely more than double it.
So when MS say double the xbox one X power that is the VERY best scenario on a test done with no AI or somehting like Forza,if MS says double it probably is little less than double.
10 TFLOPS RDNA being nearly doubled 6.1 TFLOPS RX-580 effectiveness comes with RX Vega II class rasterization power and memory bandwdith on gaming workloads.
When a game is optmised (e.g. skip ROPS limits with TMU I/O, wave64 compute format aware, geometry workload cap) for AMD, RX 580's 6.1 TFLOPS result is getting close GTX 1070's +6.5 TFLOPS.
With TMU read/write workaround for AMD ROPS bottleneck, both RX 580 and GTX 1070 has 2MB L2 cache storage for the ALUs.
Forza Motorspot 7's high alpha effects during wet scenes hammers ROPS, hence X1X GPU's (2M render cache with ROPS) superority over RX-580 (ROPS has no access to 2MB L2 cache). GTX 1070's ROPS has 2MB L2 cache access. TMU I/O raster workload doesn't have fix function alpha blending effects.
RX 5700 XT has near Vega II's raster/memory bandwdith.
On pure math context. NAVI 10 is a disappointing GPU, stick with RX Vega II. NVIDIA has TFLOPS high ground with Turing. NAVI 10 is not the best encryption currency cruncher.
NAVI 10 has better vector branching (single cycle latency wave32, double cycle latency for wave64) but less raw TFLOPS when compared to RX Vega II (this wave64 needs four wave64s to cover four cycle latency). Not all compute workloads needs heavy vector branching.
Loading Video...
On gaming workloads, RX Vega 56 at 1710 Mhz (12.2 TFLOPS) beaten ASUS Strix RX Vega 64 at 1590Mhz (13 TFLOPS) and beaten/rivaled RTX 2070.
Hint: Vega 56 at 1710 Mhz improved raster power for quad geometry (input and triangle processing) , 4MB L2 cache (fast temp storage) and ROPS (fix function effects, Z buffer and color buffer read-write I/O). ALU's TFLOPS by itself doesn't include I/O.
NAVI 10's improved raster behavior already shows from overclocked RX Vega 56 at 1710 Mhz beating RX Vega 64 at 1590 Mhz
I already argued for RX Vega II with reduced CU count (reduce power consumption) while increasing clock speed beyond 1710Mhz (improve raster) before RDNA was released!
Hint: (RX Vega II's 60 CU has 295 watts). (295 watts / 60 CU) x 40 CU = 196 watts (near RX 5700 XT's power consumption).
Note why NVIDIA Pascal GPUs has higher clock speed with less equalevent CU count!
Try again.
@tormentos said:
Oh and rumors still point to the PS5 been stronger,and again MS for the second time didn't call scarlet (Now xbox Series x) the most powerful console ever built like they did with the xbox one X, they call it the most powerful xbox console THEY have built so you may end again crying about secret souces for some years to come i would not miss it..lol
One thing I would like to point out that people can't seem to grasp here and the non "technical" people in the gaming industry who are jumping on the 2x 6TFLOPs means a 12 TFLOP navi GPU!
First if he is saying that its 2x the power and it refers to performance then a 5700 overclocked is just that BUT if he is referring to TFLOPs... Then again a X1X is using GCN so 2x 6TFLOPs on GCN is a Vega 64 which performs the same as a 5700.
A 12 TFLOP Navi card is not 2x the performance of a 6 TFLOP GCN based X1X, a 12TFLOP Navi would be almost 3x the performance and would compete against a RTX 2080 S... Do you guy's not understand how ridiculously unrealistic that is?... A 5700XT runs HOT and draws more than 200w and is 10.5TFLOPs, you think in 1 year AMD has managed to get 12 TFLOPs with Ray Tracing cores from Navi and have it running 30% less power than it?
I understand why websites do this because it draws people to their sites and I understand why you guys believe it... Half of you don't understand a f***ing thing and find it too complex to apparently even game on PC.
They are being vague for a reason and they are focusing on the NVMe and 8 Core Zen 2 because those sound impressive... Remember when everyone said oh it will be faster than anything on PC because it uses Gen4 PCIe lanes?... Right rumour says slower than 2Gbps which is not even close to what a Gen3 NVMe max's out at and no mention of capacity?
The CPU will be MUCH slower than a desktop 8 core zen 2 by 1GHz or more
The NVMe is rumoured to be running at slower than 2Gbps?... I'm sorry that is the speed of a cheap NVMe.
2x the performance of a X1X in GPU?... That's 5700 territory.
12 TFLOPs... GEN4 NVME faster than any PC... 8 Core Zen two... Build me a PC like that?...
Sorry we slower than 2Gbps NVME's don't exist at good capacities and a Zen 2 on desktop doesn't run at 3.5GHz or less... We can't build it because not only does it not exist but its actually weaker than what we have already.
When they reveal the 8-9.5 TFLOP GPU you guys will pretend this never happened, just like no one is talking about GEN4 NVME speeds better than anything on PC... My NVMe runs at 3.3Gbps and is a year old.
Does XSX has the same case as X1X?
Can ~300 watts GPU card installed into Corsair One case? Hint: RTX 2080 Ti can be installed.
Xbox Series X size is almost double the volume of Xbox One X
Using stills from the Xbox Series X announcement trailer, the console is an estimated 31cm tall, with 16cm width and depth, give or take a couple of millimeters for margin of error. That compares to the 30cm x 24cm x 6cm dimensions of Xbox One X, Microsoft's currently most powerful and smallest console on the market. Those dimensions provide an estimated volume of 7936 cubic centimeters for Xbox Series X, almost double of the Xbox One X's 4320.
Loading Video...
Xbox Series X is NOT Xbox One X size box with 7 nm procsssing parts
Xbox One X's vapor cooling and case solution already handles ~190 watts gaming with GPU around 150 to 160 watts.
Your 5700 argument is valid IF XSX's case is recycling X1X's cooling and case solution, but XSX's case is almost doubles the volume over X1X's case.
Based on XSX's near 2X volume increase, looking at a GPU at 250 to 300 watts range with 45 to 55 watts CPU (double X1X Jaguar's power consumption).
XSX is effectively a desktop gaming PC mini-tower dressed as a "Xbox game console".
XSX is like two X1X glued together with updated NAVI/RDNA 2 and Zen 2 IP parts.
XSX's size is similar to Apple's Mac Pro bucket.
Microsoft solves AMD's TDP problems with near-2X volume case from X1X by PC master race's brute force method.
Xbox Series X is effectively Microsoft's Corsair One like PC without "Surface" PC brand.
GPU, 52 CU x 1800 Mhz would do 11.980 TFLOPS.
RX 5700 XT AE as up to 1980 Mhz boost clock (10.14 TFLOPS) with 225 watts TDP (thermal solution).
Existing RX 5700 is ready for X1X's existing cooling and case solution, but XSX has ~1.83X volume of X1X's case.
Existing RX 5700 doesn't include 15% power efficiency improvments from TSMC 7nm+ EUV.
------
Phil Spencer claim
“So when we do the math, we’re over eight times the GPU power of the Xbox One, and two times what an Xbox One X is.”
From https://www.videogameschronicle.com/news/xbox-series-x-is-twice-as-powerful-as-xbox-one-x-says-phil-spencer/
-----
The difference between 12 TFLOPS NAVI and 10.1 TFLOPS NAVI is just 18 percent.
RDNA with 12 TFLOPS without any scaling on geometry inputs, geometry processing and ROPS would result in RX Vega II (13.8 TFLOPS) style bottlenecks! Such as design wouldn't be all aspect RTX 2080.
NVIDIA scaled RTX 2080's GPC (geometry) to 6 units for 11.2 TFLOPS with 64 ROPS backend.
RX-5700 XT AE has four equlavent GPC (geometry) units for 10.1 TFLOPS with 64 ROPS backend. RX Vega II's un-core parts
Log in to comment