• The GiftBot 2.0 Launch Giveaway Extravaganza has come to a close with an astounding 8073 games given away to the community by 696 members, a huge success thanks to you! The gifting now continues with more official prizes in the new Gaming Giveaways |OT|. Leftover Steam codes are also being given away to the PC Gaming Era community.

Next-gen PS5 and next Xbox speculation launch thread |OT7| - nm

Mecha Meister

Member
Oct 25, 2017
980


Previous threads:
PS5 and next Xbox launch speculation - timing and specifications

PS5 and next Xbox launch speculation - Post E3 2018

Next gen PS5 and next Xbox launch speculation - Secret sauces spicing 2019

Next-gen PS5 and next Xbox speculation launch thread - MY ANACONDA DON'T WANT NONE

Next-gen PS5 and next Xbox speculation launch thread |OT5| - It's in RDNA

Next-gen PS5 and next Xbox speculation launch thread |OT6| - Mostly Fan Noise and Hot Air

Threadmarks
Kleegamefan - Next Generation Console Information (OT6) - Verifed by ZhugeEx

Official PlayStation 5 Specifications Revealed so far:
  • Zen 2 - 8 Cores 16 Threads - 7nm (Unknown clock-speeds)
  • AMD Radeon NAVI GPU - 7nm (Unknown clock-speeds and core count)
  • Ray tracing support (Hardware-accelerated)
  • 8K Output support
  • SSD (allegedly faster than PC solutions available at the time of publication)
  • PS4 Backwards compatibility
  • Coming Holiday 2020
  • 4K Bluray player
  • 100GB optical disc support for games
  • Having an SSD removes the need for data duplication, this was used to allow Hard Disk Drives to read the data faster. This consumes more disc space than necessarily. With an SSD, data duplication is no longer needed, so game developers can save space or use it for other things.
  • Devkit design is confirmed to be real, no acknowledgement whether the final console will resemble it.
  • Due to having an SSD, booting and loading times will be faster. World streaming in games will also be faster, and more data can be streamed in.
DualShock 5 Features Revealed so far:
  • Controller Features:
  • Adaptive triggers - offering varying levels of resistance which can be used to express tension when using weapons, etc.
  • Haptic Feedback (Highly programmable voice-coil actuators)
    Can convey the feel of traversing through different terrain.
    Sand can feel slow and sluggish, while mud can feel slow and soggy. (as mentioned in the Wired Interview).
    The difference between driving on dirt and on a track can also be conveyed.
  • Improved speaker
  • USB Type-C Connector
  • Larger battery capacity
  • Allegedly lighter than the current Xbox controller with batteries in it.
  • The one used in the second Wired Interview looks like the DualShock 4
Operating System Improvements Revealed so far:
  • Can choose to install the single player portion of a game, and install the multiplayer later, or install the entire game and delete the portion of the game you want after.
  • Multiplayer game servers will provide information such as joinable activities.
  • Single-player games will provide information such as the available missions you can play and the rewards you can obtain when you complete them. The user interface will also providing the choices of rewards you will have available to you.
Wired Exclusives - Mark Cerny on the PlayStation 5
First article - Exclusive: What to Expect From Sony's Next-Gen PlayStation
Second article - Exclusive: A Deeper Look at the PlayStation 5

PlayStation Job Listing - Senior Cloud Engineering Manager
PlayStation is growing rapidly and needs your help to build next generation cloud infrastructure and build awesome tools for our team. In this position, you will be part of a top-notch engineering team focused on delivering our container orchestration solution (Kubernetes) to the organization.

You will be managing distributed systems that are powering 100+ million PS4 consoles that deliver immersive gaming experiences. You will also be one of the leaders of an elite team that is super excited to launch the upcoming world’s fastest console(PS5) in 2020. You will love working at PlayStation if you have a strong passion for systems, availability, and resiliency.
Senior Cloud Engineering Manager
at Sony Interactive Entertainment PlayStation

🤔

Found by kostacurtas
Official Xbox Scarlet Specifications Revealed so far:
  • Zen 2 CPU (Unknown clock-speeds, and whether it will have SMT)
  • NAVI GPU (Unknown clock-speeds and core count)
  • GDDR6 Memory: Scarlet is speculated to have around 10 memory chips based on images of a PCB shown in the Project Scarlet reveal video, potentially giving it a 320-bit memory bus, ram capacity could range between 10 and 20GB.
    On the topic of this, in 2016 Microsoft showcased a render of the Xbox One X's board while the system was in development and the number of chips were able to match up with the number of chips the retail system has. Of-course, things may be subject to change so this is somewhat speculation included with officially revealed information.
  • Up to 8K resolution support
  • Up to 120 fps support
  • Ray tracing support - Hardware-accelerated
  • Variable Refresh Rate Support
  • SSD (Can apparently be used as virtual memory)
  • Backwards Compatible
  • Xbox One Controllers are Forwards Compatible
Revealed at E3 2019: Xbox Project Scarlett - E3 2019 - Reveal Trailer

RDNA Architecture:
RDNA Whitepaper

Patents
Becareful with patents, don't take everything you read in a patent to mean that it will be implemented in a company's next product, as some things that companies patent don't always come to fruition.

PlayStation 5
PS5 - a patent dive into what might be the tech behind Sony's SSD customisations (technical!)
V has come to (PS5 dev kit design?)
Expandable Storage

New PlayStation VR
Possible PSVR2 patents emerge (inside-out tracking, wireless)

DualShock 5
DualShock 5 Patent
DualShock 5 Microphone
DualShock 5 vs DualShock 4




This patent has been discovered under YUSUHIRO OOTOR's name, showcasing a design for an electronic device which was rumoured to be the PlayStation 5's Development KIt.



The existence of this development kit has been confirmed in the second Wired Interview.

Next, a version of Gran Turismo Sport that Sony had ported over to a PS5 devkit—a devkit that on quick glance looks a lot like the one Gizmodo reported on last week. (The company refused to comment on questions about how the devkit's form factor might compare to what's being considered for the consumer product.)
PlayStation 5 SSD customisations

Gofreak found a patent possibly relating to the PS5's SSD. They then proceed to break it down in this thread:
PS5 - a patent dive into what might be the tech behind Sony's SSD customisations (technical!)

This will be one for people interested in some potentially more technical speculation. I posted in the next-gen speculation thread, but was encouraged to spin it off into its own thread.

I did some patent diving to see if I could dig up any likely candidates for what Sony's SSD solution might be.

I found several Japanese SIE patents from Saito Hideyuki along with a single (combined?) US application that appear to be relevant.

The patents were filed across 2015 and 2016.

Caveat: This is an illustrative embodiment in a patent application. i.e. Maybe parts of it will make it into a product, maybe all of it, maybe none of it. Approach it speculatively.

That said, it perhaps gives an idea of what Sony has been researching. And does seem in line with what Cerny talked about in terms of customisations across the stack to optimise performance.

http://www.freepatentsonline.com/y2017/0097897.html
The TLDR is

- some hardware changes vs the typical inside the SSD (SRAM for housekeeping and data buffering instead of DRAM)
- some extra hardware and accelerators in the system for handling file IO tasks independent of the main CPU
- at the OS layer, a second file system customised for these changes

all primarily aimed at higher read performance and removing potential bottlenecks for data that is written less often than it is read, like data installed from a game disc or download.
Next Xbox rumours:
There was rumoured to be two models, Codenamed Lockhart and Anaconda, with one being more powerful than the other.
Now only one model known as "Scarlet" has been publicly revealed, while nothing has been mentioned about any other other models.

(Pre E3 2019) - Windows Central: Xbox 'Scarlett,' 'Anaconda' and 'Lockhart:' Everything (we think) we know


What we know about RDNA:
  • A new 7nm GPU architecture
  • New Compute Unit Design with improved efficiency and increased IPC offering 1.25x performance per clock
  • Features higher clock speeds and gaming performance at lower power requirements
  • First RDNA GPUs available in July, starting with the RX 5700 series GPUs
It is important to note that The PS5 and Project Scarlet are not necessarily going to be using GPUs based on the RX 5700 Series.
The RX 5700 Series GPUs have no hardware ray tracing capabilities, while the PlayStation 5 has been confirmed to have hardware ray tracing capabilities.

RDNA Details:


















Zen 2 - features many improvements such as:
  • Improved Branch Prediction
  • Single operation AVX2
  • Larger L3 Cache (2x size of Zen and Zen+)
  • 15% higher IPC
Zen 2 Details











Digital Foundry: We built a 'next-gen' Zen 2/Navi-based PC - how much faster is it than current-gen consoles?

Digital Foundry (Article): We built a 'next-gen' Zen 2/Navi-based PC - how much faster is it than current-gen consoles?
Digital Foundry (Video): We built a 'next-gen' Zen 2/Navi-based PC - how much faster is it than current-gen consoles?

Richard tests a PC that has been built with Zen 2 and NAVI hardware, this serves as a concept of theoretical Next Generation console performance.
Bare in mind that consoles are designed with strict power, thermal and size constraints, as a result of this they maybe unlikely to have the luxury of being able to power hardware that significantly exceeds a 300W envelope.

There's much we don't know about PS5 and Project Scarlett: special GPU features, shader counts, ray-tracing implementation etc. But we do know that the machine is based on Zen 2 CPU and Navi GPU architecture... and with Ryzen 7 3700X, Radeon RX 5700 and RX 5700 XT we can deliver a generational comparison... with some fascinating results. Many thanks to Asus ROG for building this PC for us, and to IO Interactive for sharing the PC equivalent settings for Hitman 2 on every console!
Next Gen Console PC Concept Build
  • CPU - Ryzen 7 3700X at 3.2GHz (Underclocked)
  • CPU Cooler - Wraith Prism
  • Motherboard - Asus ROG Strix B450F
  • Memory: 2x 8GB DDR4 3600MHz
  • GPU (Config 1) - RX 5700 XT
  • GPU (Config 2) - RX 5700
  • Storage - 1TB NVME SSD
  • Power Supply - Asus ROG 650W
  • Optical Drive: Pioneer 4K UHD Blu-ray
  • Case - Coolermaster N300
Richard conducts a test using Cinebench R15, and includes projected performance of the PlayStation 4 and Xbox One X's Jaguar CPUs in his testing results.



Cinebench R15

Athlon 5370 at 1.6GHz (Jaguar CPU Core x 4)
  • 1T (Single thread) - 35
  • MT (Multi thread) - 128
  • 8 core (Projected PS4 equivalent, 100% scaling assumption) - 256
Athlon 5370 at 2.3GHz (Jaguar CPU Core x 4)
  • 1T (Single thread) - 49
  • MT (Multi thread) - 183
  • 8 core (Projected Xbox One X equivalent, 100% scaling assumption) - 366
Ryzen 7 3700X at 2.3GHz
  • 1T (Single thread) - 110 (2.24X faster than 1.6GHz Jaguar)
  • MT (Multi thread) - 618 (3.37x faster than 1.6GHz Jaguar)
  • Octo-Core Score - 1220 (4.76x faster than 1.6GHz Jaguar)
In the event that a next generation console features a CPU at 2.3GHz:
Single-threaded performance gains are around 2.24x higher than 2.3GHz Jaguar, and 3.14x faster than 1.6GHz Jaguar.
Multi-thread performance of 8 cores is around 3.3x faster than 2.3GHz Jaguar, and 4.76x faster than 1.6GHz Jaguar.

Ryzen 7 3700X at 3.2GHz
  • 1T (Single thread) - 152 (3.1x faster than 2.3GHz Jaguar)
  • MT (Multi thread) - 868 (4.74x faster than 2.3GHz Jaguar)
  • Octo-Core Score - 1702 (4.65x faster than 2.3GHz Jaguar)
In the event that a next generation console features a CPU at 3.2GHz:
Single-threaded performance gains are around 3.1x faster than 2.3GHz Jaguar, and 4.34x faster than 1.6GHz Jaguar.
Multi-threaded performance of 8 cores is around 4.65x faster than 2.3GHz Jaguar, and 6.64x faster than Jaguar at 1.6GHz.

Currently, we only know that the PlayStation 5 is going to have a Zen 2 CPU with 8 cores and 16 threads, the clock speed of this CPU is unknown at this point in time.

Gaming Performance
This was performed by using console equivalent settings, Alex contributed in finding the console equivalent settings for Hitman 2, and DF also reached out to IO Interactive (the developers of the game) who provided them with the settings that are equivalent to the console versions.
  • PlayStation 4 PRO - 36 Compute Units at 911MHz (4.2 TF)
  • Xbox One X - 40 Compute Units at 1172 MHz (6 Teraflops)
  • 5700 - NAVI GPU featuring 36 Compute Units at 1800MHz (8.29 TF)
  • 5700 XT - NAVI GPU featuring 40 Compute Units at 1800MHz (9.2 Teraflops)
Hitman 2
  • 5700 vs PS4 Pro - The 5700 is 126% faster at 1440p (around 2.26x faster)
  • 5700 XT vs Xbox One X - The 5700 XT is 83% faster at 4K (1.83x faster)
Xbox One X vs RX 5700 XT PC Configuration



PlayStation 4 Pro vs RX 5700 PC Configuration



Wolfenstein 2: The New Colossus
  • 5700 vs PS4 Pro - The 5700 is around 3x faster at 1440p
  • 5700 XT vs Xbox One X - The 5700 XT is around 2x faster at 4K
PC configurations vs PlayStation 4 Pro and Xbox One X



PlayStation 4 vs RX 5700 Configuration




Xbox One X vs RX 5700 XT PC Configuration




Power and thermal thermal constraints
I must stress that Digital Foundry have tested a machine with PC components, and consoles are designed around strict power, thermal and size limitations, so it would be unrealistic to expect console hardware to have clock speeds matching their PC derivatives.

I think it is likely that that the CPUs will be clocked within the 2.7-3.2GHz range, or even the 2.3-2.7GHz range.
Higher clock speeds often require more power and more capable cooling solutions, this can be more expensive or may require a larger surface area.
Thermodynamics isn't my forte but an example of this would be the use of copper heatsinks over aluminum ones, as copper is a better conductor of heat, but can be more expensive to implement.

I think it would be fascinating to see further exploration of the subject of PC hardware power consumption. Dictator what do you think about this?

In previous reviews of GPUs, the peak system power consumption was something that was observed but I haven't seen much of this lately, perhaps I have missed it? However, power consumption is something that has been explored recently in CPU reviews, such as the Ryzen 5 3600X review.



Here's an example of peak system power consumption from Digital Foundry's RX 480 review:
Crysis 3 was used to test the power consumption of a system outfitted with a i7-6700K and one of 4 GPUs, these were the RX 480, the R9 390, the GTX 970 and the GTX 1070.
The RX 480 system had a peak power consumption of 271 watts, whilst the GTX 1070 system had a peak of 263 watts.
The GTX 1070 is a more powerful GPU, however it is more efficient than the RX 480.


EuroGamer - AMD Radeon RX 480 review

In the more recent NVIDIA Geforce GTX 1660 TI review, Digital Foundry observed that a system equipped with a 1660 Ti and a overclocked i7-8700K had a peak power draw within the 220-230W range.

Anandtech also touched upon the subject of power consumption in their RX 5700 and 5700 XT review.
Their test system has these specs:

On the topic of power consumption, I mentioned earlier that consoles maybe unlikely to have the luxury of being able to power hardware that signficantly exceeds a 300W envelope. I'll use the PlayStation 4 Pro and Xbox One X as an example of this.

Gamers Nexus and iFixit did teardowns of the Xbox One X and found that it had a 245W PSU.
Gamers Nexus - Xbox One X Tear-Down
iFixit - Feast Your Eyes on the Xbox One X Teardown

Xbox One X's Power Supply


(Image from iFixit teardown)
Output:
20.42 Amps x 12 Volts = 245 Watts

iFixit also did a teardown of the PlayStation 4 Pro and found that it had a 289W PSU

PlayStation 4 Pro's Power Supply


(Image from iFixit teardown)
Output:
23.5 Amps x 12 Volts = 282 Watts
1.5 Amps x 4.8V = 7.2 Watts
282 + 7.2 = 289.2
Total = 289.2 Watts

However, even with these PSUs that are rated to output upwards of 245 Watts, Digital Foundry has observed power consumption for the PlayStation 4 Pro and the Xbox One X in the sub 200W range. I think it's possible that there may be some variance to these numbers depending on the game and the load it puts on the hardware.

PlayStation 4 Power Consumption Comparison


EuroGamer - Sony PlayStation 4 Pro review

Xbox One Power Consumption Comparison


EuroGamer - Microsoft Xbox One X review

Looking at the past generation consoles, some PlayStation 3 models had PSUs that were capable of outputting power over 350 watts, but didn't necessarily consume that much power when in use. This was reduced over time with newer revisions of the console, like the PlayStation 3 Slim with it's 216W PSU, and the Super Slim with it's 156W PSU.

I suppose a console with a 300-400W PSU would be the most we could see, but having more power hungry components means that there is a higher heat output, this will require a capable cooling system which may be more expensive to implement, and/or consume a larger surface area.

iFixit have also performed teardowns of these two machines.

iFixit PlayStation 3 Slim Teardown
18 Amps x 12 Volts = 216 Watts

iFixit PlayStation 3 Super Slim Teardown
13 Amps x 12 Volts = 156 Watts

RX 5700 Series Performance

These GPUs are based on AMD RDNA's architecture, they may be a solid basis of what to expect from Next Generation Consoles.

Specifications

5700 XT - 40 Compute Units (2560 Cores)
  • Core/Base Clock: 1605 MHz = 8.21 TF
  • Game Clock: 1755 MHz = 8.98 TF
  • Boost Clock: 1725MHz = 7.94 TF

5700 - 36 Compute Units (2304 Cores)
  • Core/Base Clock: 1425 MHz = 7.29 TF
  • Game Clock: 1625 MHz = 7.48 TF
  • Boost Clock: 1905 MHz = 9.75 TF
From the reviews I have seen, it seems that these GPUs are more likely to hold their game clocks than their boost clocks.

TechPowerUp
5700 XT Review
5700 XT Performance Summary
AMD Radeon RX 5700 XT Review - Clock Speeds and Power Limit
AMD Radeon RX 5700 Review - Clock Speeds and Power Limit

Anandtech
RX 5700 XT and 5700
5700 XT and 5700 Review - Power, Noise and Temperatures

Due to power and thermal constraints, console GPUs may prioritize higher core counts over higher clock speeds.

RX 5700 Series GPU Specifications:

Official Clock Speeds
  • 5700 XT - 40 Compute Units at 1755Hz (Game Clock - 8.98 Teraflops)
  • 5700 - 36 Compute Units at 1625MHz (Game Clock - 7.48 TF)
Digital Foundry's clock speeds
  • 5700 XT - 40 Compute Units at 1800MHz (9.2 Teraflops)
  • 5700 - 36 Compute Units at 1800MHz (8.29 TF)
Examples of this core count prioritization for a GPU would be these:

3072 cores (48 Compute Units)
  • 48 Compute Units at 1218MHz = 7.48 Teraflops (3072 x 1218 x 2)
  • 48 Compute Units at 1252MHz = 7.69 Teraflops (3072 x 1252 x 2)
  • 48 Compute Units at 1350MHz = 8.29 Teraflops (3072 x 1350 x 2)
  • 48 Compute Units at 1462MHz = 8.98 Teraflops (3072 x 1462 x 2
  • 48 Compute Units at 1514MHz = 9.3 Teraflops (3072 x 1514 x 2)
2816 cores (44 Compute Units)
  • 44 Compute Units at 1329MHz = 7.48 Teraflops (2816 x 1329 x 2)
  • 44 Compute Units at 1366MHz = 7.69 Teraflops (2816 x 1366 x 2)
  • 44 Compute Units at 1472MHz = 8.29 Teraflops (2816 x 1472 x 2)
  • 44 Compute Units at 1595MHz = 8.98 Teraflops (2816 x 1595 x 2)
  • 44 Compute Units at 1652MHz = 9.3 Teraflops (2816 x 1652 x 2)
2688 cores (42 Compute Units)
  • 42 Compute Units at 1392MHz = 7.48 Teraflops (2688 x 1392 x 2)
  • 42 Compute Units at 1431MHz = 7.69 Teraflops (2688 x 1431 x 2)
  • 42 Compute Units at 1543MHz = 8.29 Teraflops (2688 x 1543 x 2)
  • 42 Compute Units at 1670MHz = 8.98 Teraflops (2688 x 1670 x 2)
  • 42 Compute Units at 1730MHz = 9.3 Teraflops (2688 x 1730 x 2)
As you can see, the theoretical teraflops are dependent on the clock speed and core count of the GPU.
In these examples, compared to the desktop NAVI parts I have reduced the clock speeds of these hypothetical GPUs and increased their number of compute units, I did this to obtain a target number of teraflops similar to those that are found in the desktop NAVI parts.

Here are examples of how the RX 5700 series GPUs perform compared to other GPUs:

EuroGamer - The GPU power ladder: all current graphics cards ranked



What can we expect to see from Next Generation Games?

Next Generation consoles are going to have substantially more powerful hardware, and will no longer be shackled to the limitations of the base machines, much the like PlayStation 4 Pro and Xbox One X were.
With significantly more powerful CPUs we can expect to see things such as more advanced simulations taking place in game worlds, these could be in the form of more advanced physics such more advanced cloth, fluid and destruction simulations. As well as advanced physics based animations.
Longer draw distances are another thing that we are likely to see, as well as NPCs with more advanced AI behaviour which offer more emergent gameplay scenarios.

A technique used to free CPU cycles is to reduce the update rate of animations of characters in the distance, games like Halo 5 and Resident Evil 2 do this.
Theoretically, with more powerful CPUs there would be more headroom to run animations at higher update rates so hopefully we will see less of this, but this depends on the developers as they may still use this technique to free CPU cycles.

Alex from Digital Foundry recently created a video about Star Citizen:
Star Citizen: A Next-Gen Experience In The Making... And You Can Play-Test It Now

This is a controversial game, but that's not the focus of this video.
I would argue that Star Citizen is undoubtedly the most ambitious and advanced game knowingly in development, it has impressive visuals but what I would argue is most impressive about it would be the gameplay and the technological feats it has accomplished.

This game is very expensive to make, and has acquired over 200 million USD in funding, however I would argue that this game is representative of what can be accomplished when targeting hardware that transcends current generation consoles, and when a development team have the funds to push the boundaries.
This game, even in its current unfinished state is simply in a league of its own on a technological level.

Alex expresses in the video that he's excited about the concept of next generation consoles having Zen 2 CPUs with 8 cores and 16 threads as well as SSDs because of the possibilities that the new minimum specifications of consoles will enable, Star Citizen is currently the shining example of this.

Star Citizen has features such as:
  • 64-bit coordinate system
  • Space and planetary combat
  • Highly detailed environments
  • Meticulously detailed ships
  • Comprehensive simulation of ships and their components, each component consumes power and outputs heat, it is important to manage your ship's power distribution and heat output. Ships have components such as thrusters, power plants, coolers and shield generators.
  • Bounty hunting
  • Mining
  • Synchronised first and third person animations
The game is achieving this stuff while striving to be an MMO! The list just goes on!

There are colossal ships which you can enter and move seamlessly around, featuring stunningly detailed rooms and a variety of interactive elements, these can be things such as turrets and terminals.

Every ship has their their own frame of reference, simply put:

You can be on a ship that's traveling 100s of miles per hour and walk around it smoothly, while interacting with other players or even engaging in on-foot combat!
This can be happening all while the ship is rotating, or a space battle takes place around you. There are so many different scenarios that can take place!

Alex talks about the concept of travel in Star Citizen at 5:48 on wards, and frames of references at 7:05 but I recommend watching the entire video!

To get an idea of the scale of things in Star Citizen, here's a ship size comparison:



(Click to enlarge)

You can land something like a Reclaimer (bottom left) on a planet, this is a huge salvage ship that features many rooms and even elevators!

Star Citizen - The Reclaimer over the Hurston Savannah (Video)

Star Citizen is also an example of a game that has been designed for and greatly benefits from SSDs. As a result of this it suffers from long loading times and performance issues on Hard Disk Drives.

Alex showcases the loading times on an NVME SSD and a 7200 RPM RAID 0 HDD configuration, in his video it took almost 11x more time to load into the game with the Hard Drives. The game is also constantly streaming in data for the highly detailed world.
(With RAID 0 reads and writes occur concurrently on multiple drives, this increases throughput)

Star Citizen loading times (Universe mode)
RAID 0 HDD - 3:24.47 seconds
XPG SX8200 PRO NVME SSD - 19:32 seconds.

Having seen what Star Citizen is doing makes me incredibly excitedly for Next Generation games, there is so much potential!

Current affairs regarding Zen 2's performance and Intel's offerings.

Intel's current architecture which has seen widespread use across mobile and desktops is Coffee-Lake, Intel have introduced other architectures to market which are refinements of Coffee-Lake, however these are chips which are focused on the mobile segment and to my knowledge do not reach the performance standards of Coffee-Lake to replace it in the desktop segment.

With Zen 2, it seems that AMD has rectified many of the performance bottlenecks from the first generation Zen CPUs. As a result of this, AMD has pretty much caught up to Intel in the majority of workloads, and are even offering CPUs that deliver competitive performance in their price range, some of which undercut Intel's own chips. However, Intel still maintain a clock speed advantage for their CPUs, this leads to higher performance in gaming workloads in the currently available software.

I found TechReport's review of the 3700X and 3900X to be very telling of this, it's a very comprehensive review and they did a fantastic job reviewing these CPUs, check it out here:
Tech Report - AMD’s Ryzen 7 3700X and Ryzen 9 3900X CPUs reviewed

From this review I wanted to bring this section to attention, check out GTA V's performance on the Ryzen 3700X and 2700X here:




That's a 24.7% gain in average fps and significantly lower frame times, this is great to see as this game has been something the Zen architecture has struggled with since it's inception, and Zen 2 seems to have closed the gap significantly between it and Intel's offerings.
However, regardless of this improvement in GTA V, Intel still have a sizable lead in gaming performance in other games such as Deus Ex: Mankind Divided and Hitman 2 as featured in this review.




Here we see Intel's 9900K leading by 26.6% against the 3700X, and 34.5% against the 3900X.
Where the third-generation Ryzens traded blows with their Intel competitors in Crysis, Deus Ex is a different story altogether. Both of the new CPUs take a back seat to all three of our Intel CPUs in this title. I’m not qualified to pass judgment on why, but if you forced me to guess I might suspect that it has something to do with memory latency.
Anandtech

Anandtech conducted a test with the SPEC2006 and SPEC2017 benchmark suite in their Ryzen 3700X and 3900X review, these are industry standard benchmarks.

The Ryzen 9 3900X (Zen 2) is compared against AMD's own Ryzen 7 2700X (Zen+) and Intel's i9-9900K (Coffee-Lake)
AMD Ryzen 9 3900x.
The 3900X and i9-9900K used DDR4 3200MHz CL16 memory, whilst the Ryzen 7 2700X used DDR4 2933MHz memory with "similar" CL16 timings.

Anandtech: The AMD 3rd Gen Ryzen Deep Dive Review: 3700X and 3900X Raising The Bar

One big talking point around the new Ryzen 3000 series is the new augmented single-threaded performance of the new Zen 2 core. In order to investigate the topic in a more controlled manner with better documented workloads, we’ve fallen back to the industry standard SPEC benchmark suite.

We’ll be investigating the previous generation SPEC CPU2006 test suite giving us some better context to past platforms, as well as introducing the new SPEC CPU2017 suite. We have to note that SPEC2006 has been deprecated in favour of 2017, and we must also mention that the scores posted today are noted as estimates as they’re not officially submitted to the SPEC organisation.
This testing suite features a variety of different tests, these bring to light to the different improvements that have been brought to Zen 2.
From these tests, the reviewers were able to deduce the impacts of things such as Zen 2's improved branch prediction capabilities from things such as the "445.gobmk" benchmark as well as the cache improvements. They have also shown how it compares to Intel's i9-9900K and their Coffee-Lake architecture and showcase Zen 2's strengths and weaknesses.

From this testing suite, they concluded that Zen 2 has a higher overall IPC when compared to Intel's Coffee-Lake architecture. IPC is basically the performance of the CPU at a given clock speed, a simplified example of this would be:

CPU A at 4GHz scores - 100
CPU B at 4GHz scores - 130

This shows that CPU B has a 30% higher IPC



Normalising the scores for frequency, we see that AMD has achieved something that the company hasn’t been able to claim in over 15 years: It has beat Intel in terms of overall IPC. Overall here, the IPC improvements over Zen+ are 15%, which is a bit lower than the 17% figure for SPEC2006.
This is just a short overview of their findings in the SPEC benchmarks, there is more information available at the source.

TechSpot

TechSpot (Hardware Unboxed) conducted a test featuring the Ryzen 7 1700X, Ryzen 7 2700X, the Ryzen 7 3700X and the Ryzen 9 3900X with 8 cores enabled (down from 12) and tested it against Intel 9900K. Their highest performing mainstream socket desktop CPU. All CPUs were tested at 4GHz.
To my knowledge, the i9-9900K has an all-core turbo of 4.8GHz so this is theoretically leaving perhaps 20% performance on the table, but the main purpose of this test was to showcase performance differences between the CPUs at the same clock speed.

This is a really interesting test, as it gives us an insight to how AMD's Zen 2 CPUs fair against Intel's own i9-9900K at the same clock speeds.

Techspot (Hardware Unboxed)
Article: 4GHz CPU Battle
Video: 3rd Gen Ryzen IPC Test, 3900X & 3700X vs. Core i9-9900K

I have included some of TechSpot's tests below, be sure to check out the article or even the video for more!https://www.youtube.com/watch?v=RmxkpTtwx1k

In their tests, most of the results show the Ryzen 7 3700X performing within 10% of Intel i9-9900K, losing in the gaming test and performing better than i9-9900K in the application tests. AMD have a highly competitive product on their hands.

So far, it seems that Intel still has an advantage in gaming performance in the current software that is available.
I have seen up-to 20-30% performance advantages go to Intel's 9900K when tested against AMD's current best mainstream CPU, the 3900X (some of this could be due to Windows scheduler issues as I've seen smaller performance advantages when disabling SMT on AMD CPUs) this is a 12 core 24 thread CPU with a base clock of 3.8GHz and a boost clock of 4.6GHz, while the 9900K is an 8 core 16 thread CPU, with a base clock of 3.6GHz and a boost clock of 5GHz.

Intel does have a clock-speed advantage, as well as more headroom to reach higher overclocks.
The i9-9900K in particular has the ability to boost all 8 of it's cores to 4.8GHz, while the AMD CPUs typically max out at around 4.4GHz. There have been improvements to how AMD CPUs boost due to new BIOs releases, this helps these CPUs to maintain their boost clock speeds.

4GHz CPU Battle

Cinebench R20


Cinebench R20 shows the AMD 3700X leading by 13.4% in multi-core performance, and the 3900X with 8 cores enabled leading by 13.6%



For single-core performance, the 3900X (4 cores disabled) leads by 9.5% against the i9-9900K, and the 3700X leads by 9%



Gaming Performance

In Battlefield V, the I9-9900k leads by 2.5% in average frame rates versus the 3900X (4 cores disabled), and 7.3% versus the 3700X.
For minimum frame rates, the 9900K leads by 8.7% versus the 3900X and 3700X.
Comparing the 3700X to the 1700X sees the Zen 2 CPU leading by 15.5% for average frame rates, and 10.7% for minimum frame rates.



In Far Cry New Dawn, the i9-9900K leads by 5.3% versus the 3900X (4 cores disabled), and 9.2% versus the 3700X in average frame rates.
For minimum frame rates, the i9-9900K leads by 9.5% versus the 3900X and 3700X.



In this AIDA64's latency test, we see that the Zen 2 CPUs have higher latency when compared to Intel's i9-9900K.



In SiSoftware's Multi-Thread Latency test we see that while Zen 2 has higher latency when comparing "Worst Matched Cores" to the i9-9900K, it has lower latency when compared against the Best Matched Cores. However, we also see that the latency between the "Best Matched Cores" is lower than the i9-9900K.

Zen 2 has made notable improvements with in regard to multi-thread latency, as the first generation Zen CPUs see 53% higher latency when comparing the "Worst Matched Cores" latency to the Ryzen 7 3700X, while the Ryzen 7 2700X sees 35% higher latency when comparing the latency of the "Worst Matched Cores" to the 3700X.
For "Best Matched Cores" the Ryzen 7 1700X has 43.9% higher latency when compared to the Ryzen 7 3700X, and the 2700X has 42% higher latency when compared to the 3700X as well.



Some more Ryzen 3000 CPU Reviews:
Tom's Hardware - AMD Ryzen 9 3900X and Ryzen 7 3700X Review - Tom's Hardware even test the i9-9900K at 5GHz!
Ryzen 3000 and Zen 2 Review Thread

Edits and Updates:
Grammar and spelling corrections.
OP is occasionally updated with more information.
 
Last edited:

Noctis114

Member
Jan 25, 2019
1,009
The speed of this OT's gonna highly depend on what we see during X019 & TGA this year, whether there's any next gen news or not.

or PSX 2019 if that damn things gonna happen....
 

PianoBlack

Member
May 24, 2018
271
New thread, new hypothetical:

Sony's custom SSD implementation has a 25% performance advantage over Scarlett's bog standard PCIe 4.0 SSD. But Microsoft supports expansion in (presumably overpriced) 512 GB, 1TB, or 2TB increments using 360-style custom "cartridges" that plug into some custom PCIe port and contain approved flash storage of sufficient performance, whereas PS5 makes you move things between an external HDD and the 1TB SSD if you want to play them.

Which do you prefer?
 

jroc74

Member
Oct 27, 2017
6,254
New thread, new hypothetical:

Sony's custom SSD implementation has a 25% performance advantage over Scarlett's bog standard PCIe 4.0 SSD. But Microsoft supports expansion in (presumably overpriced) 512 GB, 1TB, or 2TB increments using 360-style custom "cartridges" that plug into some custom PCIe port and contain approved flash storage of sufficient performance, whereas PS5 makes you move things between an external HDD and the 1TB SSD if you want to play them.

Which do you prefer?
Hmmm. This is interesting, lol.

I honestly can't say right now, but I like the expansion scenario.

How fast is that transfer speed for PS5....
 

Oluasc

Be more dog!
Member
Oct 25, 2017
2,720
...
Great OP :)

My prediction:

Xbox Scarlett

CPU:
Zen 2 8C/16T @ 3.4GHz
GPU: RDNA2 @ 1.5GHz~1.8GHz and 10.3TF~12TF
RAM: 16GB GDDR6 @ 560GB/s
SSD: 1TB NVMe @ 5GB/s

$499
 

BiTheWay

Member
May 24, 2019
109
Goddamn that is a monster of a thread. Anyways, with a new thread, comes new predictions.

Pessimistic:
9.3 TF PS5 (40CU @ 1.8GHz) / 8.9 TF Scarlett (40 CU @ 1.75GHz)
16 GB GDDR6
2.8 GHz 8c/16t Zen 2 CPU

Realistic:
10.4 TF PS5 (44CU @ 1.85GHz) / 10.1 Scarlett (44CU @ 1.8 GHz)
16GB GDDR6
3.2 GHz 8c/16t Zen 2 CPU

Optimistic
12.288 TF PS5 (48CU @ 2GHz) / 11.98 TF Scarlett (48CU @ 1.95GHz)
20GB GDDR6/18 GB GDDR6 + 2 GB DDR4
3.6 GHz 8c/16t Zen 2 CPU
 

statham

Member
Oct 25, 2017
2,272
FloRida


Previous threads:
PS5 and next Xbox launch speculation - timing and specifications

PS5 and next Xbox launch speculation - Post E3 2018

Next gen PS5 and next Xbox launch speculation - Secret sauces spicing 2019

Next-gen PS5 and next Xbox speculation launch thread - MY ANACONDA DON'T WANT NONE

Next-gen PS5 and next Xbox speculation launch thread |OT5| - It's in RDNA

Next-gen PS5 and next Xbox speculation launch thread |OT6| - Mostly Fan Noise and Hot Air

Threadmarks
Kleegamefan - Next Generation Console Information (OT6) - Verifed by ZhugeEx

Official PlayStation 5 Specifications Revealed so far:
  • Zen 2 - 8 Cores 16 Threads - 7nm (Unknown clock-speeds)
  • AMD Radeon NAVI GPU - 7nm (Unknown clock-speeds and core count)
  • Ray tracing support (Hardware-accelerated)
  • 8K Output support
  • SSD (allegedly faster than PC solutions available at the time of publication)
  • PS4 Backwards compatibility
  • Coming Holiday 2020
  • 4K Bluray player
  • 100GB optical disc support for games
  • Having an SSD removes the need for data duplication, this was used to allow Hard Disc Drives to read the data faster. This consumes more disc space than neccessarily. With an SSD, data duplication is no longer needed, so game developers can save space or use it for other things.
  • Devkit design is confirmed to be real, no acknowledgement whether the final console will resemble it.
  • Due to having an SSD, booting and loading times will be faster. World streaming in games will also be faster, and more data can be streamed in.
DualShock 5 Features Revealed so far:
  • Controller Features:
  • Adaptive triggers - offering varying levels of resistance which can be used to express tension when uising weapons, etc.
  • Haptic Feedback (Highly programmable voice-coil actuators)
    Can convey the feel of traversing through different terrain.
    Sand can feel slow and sluggish, while mud slow and soggy. (as mentioned in the Wired Interview).
    The difference between driving on dirt and on a track can also be conveyed.
  • Improved speaker
  • USB Type-C Connector
  • Larger battery capacity
  • Allegedly lighter than the current Xbox controller with batteries in it.
  • The one the used in the second Wired Interview looks like the DualShock 4
Operating System Improvements Revealed so far:
  • Can choose to intall the single player portion of a game, and install the multiplayer later, or install the entire game and delete the portion of the game you want after.
  • Multiplayer game servers will provide information such as joinable activies.
  • Single-player cames will provide information such as the available missions you can play and the rewards you can obtain when you complete them. The user interface will also providing the choices of rewards you will have available to you.
Revealed by Mark Cerny in two Wired Interviews.
First Interview - Exclusive: What to Expect From Sony's Next-Gen PlayStation
Second Interview - Exclusive: A Deeper Look at the PlayStation 5

Official Xbox Scarlet Specifications Revealed so far:
  • Zen 2 CPU (Unknown clock-speeds, and whether it will have SMT)
  • NAVI GPU (Unknown clock-speeds and core count)
  • GDDR6 Memory: Scarlet is speculated to have around 10 memory chips based on images of a PCB shown in the Project Scarlet reveal video, potentially giving it a 320-bit memory bus, ram capacity could range between 10 and 20GB.
    On the topic of this, in 2016 Microsoft showcased a render of the Xbox One X's board while the system was in development and the number of chips were able to match up with the number of chips the retail system has. Of-course, things may be subject to change so this is somewhat speculation included with officially revealed information.
  • Up to 8K resolution support
  • Up to 120 fps support
  • Ray tracing support - Hardware-accelerated
  • Variable Refresh Rate Support
  • SSD (Can apparently be used as virtual memory)
  • Backwards Compatible
  • Xbox One Controllers are Forwards Compatible
Revealed at E3 2019: Xbox Project Scarlett - E3 2019 - Reveal Trailer

RDNA Architecture:
RDNA Whitepaper

Patents
Becareful with patents, don't take everything you read in a patent to mean that it will be implemented in a company's next product, as some things that companies patent don't always come to fruition.

PS5 - a patent dive into what might be the tech behind Sony's SSD customisations (technical!)
V has come to (PS5 dev kit design?)
Possible PSVR2 patents emerge (inside-out tracking, wireless)
DualShock 5 Patent

This patent has been discovered under YUSUHIRO OOTOR's name, showcasing a design for an electronic device which is rumoured to be the PlayStation 5's Development KIt.

The existence of this development kit has been confirmed in the second Wired Interview.








Next Xbox rumours:
There was rumoured to be two models, Codenamed Lockhart and Anaconda, with one being more powerful than the other.
Now only one model known as "Scarlet" has been publicly revealed, while nothing has been mentioned about any other other models.

(Pre E3 2019) - Windows Central: Xbox 'Scarlett,' 'Anaconda' and 'Lockhart:' Everything (we think) we know


What we know about RDNA:
  • A new 7nm GPU architecture
  • New Compute Unit Design with improved efficiency and increased IPC offering 1.25x performance per clock
  • Features higher clock speeds and gaming performance at lower power requirements
  • First RDNA GPUs available in July, starting with the RX 5700 series GPUs
It is important to note that The PS5 and Project Scarlet are not necessarily going to be using GPUs based on the RX 5700 Series.
The RX 5700 Series GPUs have no hardware ray tracing capabilties, while the PlayStation 5 has been confirmed to have hardware ray tracing capabilties.

RDNA Details:


















Zen 2 - features many improvements such as:
  • Improved Branch Prediction
  • Single operation AVX2
  • Larger L3 Cache (2x size of Zen and Zen+)
  • 15% higher IPC
Zen 2 Details











Digital Foundry: We built a 'next-gen' Zen 2/Navi-based PC - how much faster is it than current-gen consoles?

Digital Foundry (Article): We built a 'next-gen' Zen 2/Navi-based PC - how much faster is it than current-gen consoles?
Digital Foundry (Video): We built a 'next-gen' Zen 2/Navi-based PC - how much faster is it than current-gen consoles?

Richard tests a PC that has been built with Zen 2 and NAVI hardware, this serves as a concept of theorhetical Next Generation console performance.
Bare in mind that consoles are designed with strict power, thermal and size constraints, they are unlikely to have the luxury of being able to power hardware that exceeds a 300W envelope.



Next Gen Console PC Concept Build
  • CPU - Ryzen 7 3700X at 3.2GHz (Underclocked)
  • CPU Cooler - Wraith Prism
  • Motherboard - Asus ROG Strix B450F
  • Memory: 2x 8GB DDR4 3600MHz
  • GPU (Config 1) - RX 5700 XT
  • GPU (Config 2) - RX 5700
  • Storage - 1TB NVME SSD
  • Power Supply - Asus ROG 650W
  • Optical Drive: Pioneer 4K UHD Blu-ray
  • Case - Coolermaster N300
Richard conducts a test using Cinebench R15, and includes projected performance of the PlayStation 4 and Xbox One X's Jaguar CPUs in his testing results.



Cinebench R15

Athlon 5370 at 1.6GHz (Jaguar CPU Core x 4)
  • 1T (Single thread) - 35
  • MT (Multi thread) - 128
  • 8 core (Projected PS4 equivalent, 100% scaling assumption) - 256
Athlon 5370 at 2.3GHz (Jaguar CPU Core x 4)
  • 1T (Single thread) - 49
  • MT (Multi thread) - 183
  • 8 core (Projected Xbox One X equivalent, 100% scaling assumption) - 366
Ryzen 7 3700X at 2.3GHz
  • 1T (Single thread) - 110 (2.24X faster than 1.6GHz Jaguar)
  • MT (Multi thread) - 618 (3.37x faster than 1.6GHz Jaguar)
  • Octo-Core Score - 1220 (4.76x faster than 1.6GHz Jaguar)
In the event that a next generation console features a CPU at 2.3GHz:
Single-threaded performance gains are around 2.24x higher than 2.3GHz Jaguar, and 3.14x faster than 1.6GHz Jaguar.
Multi-thread performance of 8 cores is around 3.3x faster than 2.3GHz Jaguar, and 4.76x faster than 1.6GHz Jaguar.

Ryzen 7 3700X at 3.2GHz
  • 1T (Single thread) - 152 (3.1x faster than 2.3GHz Jaguar)
  • MT (Multi thread) - 868 (4.74x faster than 2.3GHz Jaguar)
  • Octo-Core Score - 1702 (4.65x faster than 2.3GHz Jaguar)
In the event that a next generation console features a CPU at 3.2GHz:
Single-threaded performance gains are around 3.1x faster than 2.3GHz Jaguar, and 4.34x faster than 1.6GHz Jaguar.
Multi-threaded performance of 8 cores is around 4.65x faster than 2.3GHz Jaguar, and 6.64x faster than Jaguar at 1.6GHz.

Currently, we only know that the PlayStation 5 is going to have a Zen 2 CPU with 8 cores and 16 threads, the clock speed of this CPU is unknown at this point in time.

Gaming Performance
This was performed by using console equivalent settings, Alex contributed in finding the console equivalent settings for Hitman 2, IO Interactive (the developers of the game) helped with this also.
  • PlayStation 4 PRO - 36 Compute Units at 911MHz (4.2 TF)
  • Xbox One X - 40 Compute Units at 1172 MHz (6 Teraflops)
  • 5700 - NAVI GPU featuring 36 Compute Units at 1800MHz (8.29 TF)
  • 5700 XT - NAVI GPU featuring 40 Compute Units at 1800MHz (9.2 Teraflops)
Hitman 2
  • 5700 vs PS4 Pro - The 5700 is 126% faster at 1440p (around 2.26x faster)
  • 5700 XT vs Xbox One X - The 5700 XT is 83% faster at 4K (1.83x faster)
Xbox One X vs RX 5700 XT PC Configuration



PlayStation 4 Pro vs RX 5700 PC Configuration



Wolfenstein 2: The New Colossus
  • 5700 vs PS4 Pro - The 5700 is around 3x faster at 1440p
  • 5700 XT vs Xbox One X - The 5700 XT is around 2x faster at 4K
PC configurations vs PlayStation 4 Pro and Xbox One X



PlayStation 4 vs RX 5700 Configuration




Xbox One X vs RX 5700 XT PC Configuration




Power and thermal thermal constraints
I must stress that Digital Foundry have tested a machine with PC components, and consoles are designed around strict power, thermal and size limitations, so it would be unrealistic to expect console hardware to have clock speeds matching their PC derivatives.

I think it is likely that they will be within the range of 2.7-3.2GHz, or even 2.3-2.7GHz as higher clock speeds often require more power and more capable cooling solutions, this can be more expensive or may require a larger surface area.
Thermodynamics isn't my forte but an example of this would be the use of copper heatsinks over aluminum ones, as copper is a better conducter of heat, but can be more expensive to implement.

I think it would be facinating to see someone from Digital Foundry to explore more about the subject of power consumption for PC hardware.
Dictator what do you think about this?

In previous reviews of GPUs, the peak system power consumption was something that was observed but I haven't seen much of this lately, perhaps I have missed it? However, power consumption is something that has been explored recently in CPU reviews, such as the Ryzen 5 3600X review.



Here's an example of peak system power consumption the RX 480 review:
Crysis 3 was used to test the power consumption of a system outffited with a i7-6700K and one of 4 GPUs, these were the RX 480, the R9 390, the GTX 970 and the GTX 1070.
The RX 480 system had a peak power consumption of 271 watts, whilst the GTX 1070 system had a peak of 263 watts.
The GTX 1070 is a more powerful GPU, however it is more efficient than the RX 480.


EuroGamer - AMD Radeon RX 480 review

In the more recent NVIDIA Geforce GTX 1660 TI review, Digital Foundry observed that a system equipped with a 1660 Ti and a overclocked i7-8700K had a peak power draw within the 220-230W range.

Anandtech also touched upon the subject of power consumption in their RX 5700 and 5700 XT review.
Their test system has these specs:

On the topic of power consumption, I mentioned earlier that consoles are unlikely to have the luxury of being able to power hardware that exceeds a 300W envelope. I'll use the PlayStation 4 Pro and Xbox One X as an example of this.

Gamers Nexus and iFixit did teardowns of the Xbox One X and found that it had a 245W PSU.
Gamers Nexus - Xbox One X Tear-Down
iFixit - Feast Your Eyes on the Xbox One X Teardown

Xbox One X's Power Supply



Output:
20.42 Amps x 12 Volts = 245 Watts

iFixit also did a teardown of the PlayStation 4 Pro and found that it had a 289W PSU

PlayStation 4 Pro's Power Supply



Output:
23.5 Amps x 12 Volts = 282 Watts
1.5 Amps x 4.8V = 7.2 Watts
282 + 7.2 = 289.2
Total = 289.2 Watts

However, even with these PSUs that are rated to output upwards of 245 Watts, Digital Foundry has observed power consumption for the PlayStation 4 Pro and the Xbox One X in the sub 200W range. I think it's possible that there may be some variance to these numbers depending on the game and the load it puts on the hardware.

PlayStation 4 Power Consumption Comparison


EuroGamer - Sony PlayStation 4 Pro review

Xbox One Power Consumption Comparison


EuroGamer - Microsoft Xbox One X review

Looking at the past generation consoles, some PlayStation 3 models had PSUs that were capable of outputting power over 350 watts, but didn't neccessraily consume that much power when in use. This was reduced over time with newer revisions of the console, like the PlayStation 3 Slim with it's 216W PSU, and the Super Slim with it's 156W PSU.

I suppose a console with a 300-400 PSU would be the most we could see, but having more power hungry components means that there is a higher heat output, this will require a capable cooling system which may be more expensive to implement, and/or consume a larger surface area.

iFixit have also performed teardowns of these two machines.

iFixit PlayStation 3 Slim Teardown
18 Amps x 12 Volts = 216 Watts

iFixit PlayStation 3 Super Slim
13 Amps x 12 Volts = 156 Watts

RX 5700 Series Performance

These GPUs are based on AMD RDNA's architecture, they may be a solid basis of what to expect from Next Generation Consoles.

Specifications (Clockspeeds taken from Anandtech RX 5700 XT and RX 5700 review):
  • 5700 XT - 40 Compute Units at 1823MHz (9.3 Teraflops)
  • 5700 - 36 Compute Units at 1671MHz (7.69 TF)
The 5700 XT has 40 Compute Units, while the RX 5700 has 36.
RX 5700 XT (2560 x 1823 x 2 = 9.3 Teraflops)
RX 5700 (2304 x 1671 x 2 = 7.69 Teraflops)

TechPowerUp
5700 XT Review
5700 XT Performance Summary
AMD Radeon RX 5700 XT Review - Clock Speeds and Power Limit
AMD Radeon RX 5700 Review - Clock Speeds and Power Limit

Anandtech
RX 5700 XT and 5700
5700 XT and 5700 Review - Power, Noise and Temperatures

Due to power and thermal constraints, GPUs may priorize higher core counts over higher clock speeds.

Desktop NAVI GPUs:
  • 5700 XT - 40 Compute Units at 1800MHz (9.2 Teraflops)
  • 5700 - 36 Compute Units at 1800MHz (8.29 TF)
An example of this core count prioritization for a GPU would be this:
  • 3072 cores (48 Compute Units)
  • 48 Compute Units at 1400MHz = 8.6 Teraflops (3072 x 1400 x 2)
  • 48 Compute Units at 1500MHz = 9.2 Teraflops (3072 x 1500 x 2)
  • 48 Compute Units at 1500MHz = 10 Teraflops (3072 x 1630 x 2)
As you can see, the theorhetical teraflops are dependent on the clock speed and core count of the GPU, in this example I have reduced the clock speed of this GPU and increased the number of compute units to obtain a target number of teraflops.

Here is are examples of how the RX 5700 series GPUs perform compared to other GPUs:

EuroGamer - The GPU power ladder: all current graphics cards ranked



What can we expect to see from Next Generation Games?

Next Generation consoles are going to have substantially more powerful hardware, and will no longer be shackled to the limitations of the base machines, much the like PlayStation 4 Pro and Xbox One X were.
With signficantly more powerful CPUs we can expect to see more advanced simulations taking place in game worlds, these could be in the form of more advanced physics such more advanced cloth, fluid and destruction simulations. As well as advanced physics based animations.
Longer draw distances are another thing that we are likely to see, as well as NPCs with more advanced AI behaviour which offer more emergent gameplay scenarios.

A technique used to free CPU cycles is to reduce the update rate of animations of characters in the distance, games like Halo 5 and Resident Evil 2 do this.
Theorhetically, with more powerful CPUs there would be more headroom to run animations at higher update rates.

Alex from Digital Foundry recently created a video about Star Citizen:
Star Citizen: A Next-Gen Experience In The Making... And You Can Play-Test It Now

This is a controversial game, but that's not the focus of this video.
I would argue that Star Citizen is undoubtedly the most ambitious and advanced game knowingly in development, it has impressive visuals but what I would argue is most impressive about it would be the gameplay.

This game is very expensive to make, and has acquired over 200 million USD in funding, however I would argue that this game is representative of what can be accomplished when targeting hardware that trascends current generation consoles, and when a development team have the funds to push the boundaries.
This game, even in its current unfinished state is simply in a league of its own on a technological level.

Alex expresses in the video that he's excited about the concept of next generation consoles having Zen 2 CPUs with 8 cores and 16 threads as well as SSDs because of the possiblites that the new minimum specifications of consoles will enable, Star Citizen is currently the shining example of this.

Star Citizen has features such as:
  • 64-bit coordinate system
  • Space and planetary combat
  • Highly detailed enviroments
  • Meticulously detailed ships
  • Comprehensive simulation of ships and their components, each component consumes power and outputs heat, it is important to manage your ship's power distribution and heat output. Ships have components such as thrusters, power plants, coolers and shield generators.
  • Bounty hunting
  • Mining
  • Synchronised first and third person animations
The list just goes on!

There are colossal ships that you can traverse, these can have many different rooms which you can traverse seamlessly, they are also exceptionally detailed.
Every ship have their their own physics grids simply put:
You can be on a ship that's traveling 1000s of miles per hour and walk around it smoothly, and even interact with other players, as well as engage in combat.
There's so many different scenarios!

Alex talks about the concept of travel in Star Citizen at 5:48 onwards, and physics grids at 7:05 but I recommend watching the entire video!

To get an idea of the scale of things in Star Citizen, here's a ship size comparison:



(Click to enlarge)

You can land something like a Reclaimer (bottom left) on a planet, this is a huge salvage ship that features many rooms and even elevators!

Star Citizen - The Reclaimer over the Hurston Savannah (Video)

Star Citizen is also an example of a game that has been designed for and greatly benefits from SSDs. As a result of this it suffers from long loading times and performance issues on Hard Disk Drives.

Alex showcases the loading times on an NVME SSD and a 7200 RPM RAID 0 HDD configuration, in his video it took almost 11x more time to load into the game with the Hard Drives. The game is constantly streaming in data for the highly detailed game world and simulations that take place.
(With RAID 0 reads and writes occur concurrently on multiple drives, this increases throughput)

Star Citizen loading times (Universe mode)
RAID 0 HDD - 3:24.47 seconds
XPG SX8200 PRO NVME SSD - 19:32 seconds.

Having seen what Star Citizen is doing makes me incredibly excitedly for Next Generation games, there is so much pottential!

Current affairs regarding Zen 2's performance and Intel's offerings.

Intel's current architecture which has seen widespread use across mobile and desktops is Coffee-Lake, Intel have introduced other architectures to market which are refinements of Coffee-Lake, however these are chips which are focused on the mobile segment and to my knowledge do not reach the performance standards of Coffee-Lake to replace it in the desktop segment.

With Zen 2, it seems that AMD has rectified many of the performance bottlenecks from the first generation Zen CPUs. As a result of this, AMD has pretty much caught up to Intel in the majority of workloads, and are even offering CPUs that deliver competitive performance in their price range, some of which undercut Intel's own chips. However, Intel still maintain a clock speed advantage for their CPUs, this leads to higher performance in gaming workloads in the currently available software.

I found TechReport's review of the 3700X and 3900X to be very telling of this, it's a very comprehensive review and they did a fantastic job reviewing these CPUs, check it out here:
Tech Report - AMD’s Ryzen 7 3700X and Ryzen 9 3900X CPUs reviewed

From this review I wanted to bring this section to attention, check out GTA V's performance on the Ryzen 3700X and 2700X here:





That's a 24.7% gain in average fps and signficantly lower frame times, this is great to see as this game has been something the Zen architecture has struggled with since it's inception, and Zen 2 seems to have closed the gap signficantly between it and Intel's offerings.
However, regardless of this improvement in GTA V, Intel still have a sizable lead in gaming performance in other games such as Deus Ex: Mankind Divided and Hitman 2 as featured in this review.




Here we see Intel's 9900K leading by 26.6% against the 3700X, and 34.5% against the 3900X.


Anandtech

Anandtech conducted a test with the SPEC2006 and SPEC2017 benchmark suite in their Ryzen 3700X and 3900X review, these are industry standard benchmarks.

The Ryzen 9 3900X (Zen 2) is compared against AMD's own Ryzen 7 2700X (Zen+) and Intel's i9-9900K (Coffee-Lake)
AMD Ryzen 9 3900x.
The 3900X and i9-9900K used DDR4 3200MHz CL16 memory, whilst the Ryzen 7 2700X used DDR4 2933MHz memory with "similar" CL16 timings.

Anandtech: The AMD 3rd Gen Ryzen Deep Dive Review: 3700X and 3900X Raising The Bar



This testing suite features a variety of different tests, these bring to light to the different improvements that have been brought to Zen 2.
From these tests, the reviewers were able to deduce the impacts of things such as Zen 2's improved branch prediction capabilties from things such as the 445.gobmk benchmark as well as the cache improvements. They have also shown how it compares to Intel's i9-9900K and their Coffee-Lake architecture and showcase Zen 2's strengths and weaknesses.

From this testing suite, they concluded that Zen 2 has a higher overall IPC when compared to Intel's Coffee-Lake architecture. IPC is basically the performance of the CPU at a given clock speeds, a simplified example of this would be:

CPU A at 4GHz scores - 100
CPU B at 4GHz scores - 130

This shows that CPU B has a 30% higher IPC





This is just a short overview of their findings in the SPEC benchmarks, there is more information available at the source.

TechSpot

TechSpot (Hardware Unboxed) conducted a test featuring the Ryzen 7 1700X, Ryzen 7 2700X, the Ryzen 7 3700X and the Ryzen 9 3900X with 8 cores enabled (down from 12) and tested it against Intel 9900K. Their highest performing mainstream socket desktop CPU. All CPUs were tested at 4GHz.
To my knowledge, the i9-9900K has an all-core turbo of 4.8GHz so this is theorhetically leaving perhaps 20% performance on the table, but the main purpose of this test was to showcase performance differences between the CPUs at the same clock speed.

This is a really interesting test, as it gives us an insight to how AMD's Zen 2 CPUs fair against Intel's own i9-9900K at the same clock speeds.

Techspot (Hardware Unboxed)
Article: 4GHz CPU Battle
Video: 3rd Gen Ryzen IPC Test, 3900X & 3700X vs. Core i9-9900K

I have included some of TechSpot's tests below, be sure to check out the article or even the video for more!https://www.youtube.com/watch?v=RmxkpTtwx1k

In their tests, most of the results show the Ryzen 7 3700X performing within 10% of Intel i9-9900K, losing in the gaming test and performing better than i9-9900K in the application tests. AMD have a highly competitive product on their hands.

So far, it seems that Intel still has an advantage in gaming performance in the current software that is available.
I have seen up-to 20-30% performance advantages go to Intel's 9900K when tested against AMD's current best mainstream CPU, the 3900X (some of this could be due to Windows schedular issues as I've seen smaller performance advantages when disabling SMT on AMD CPUs) this is a 12 core 24 thread CPU with a base clock of 3.8GHz and a boost clock of 4.6GHz, while the 9900K is an 8 core 16 thread CPU, with a base clock of 3.6GHz and a boost clock of 5GHz.

Intel does have a clock-speed advantage, as well as more headroom to reach higher overclocks.
The i9-9900K in particular has the ability to boost all 8 of it's cores to 4.8GHz, while the AMD CPUs typically max out at around 4.4GHz. There have been improvements to how AMD CPUs boost due to new BIOs releases, this helps these CPUs to maintain their boost clock speeds.

4GHz CPU Battle

Cinebench R20


Cinebench R20 shows the AMD 3700X leading by 13.4% in multi-core performance, and the 3900X with 8 cores enabled leading by 13.6%



For single-core performance, the 3900X (4 cores disabled) leads by 9.5% against the i9-9900K, and the 3700X leads by 9%



Gaming Performance

In Battlefield V, the I9-9900k leads by 2.5% in average framerates versus the 3900X (4 cores disabled), and 7.3% versus the 3700X.
For minimum framerates, the 9900K leads by 8.7% versus the 3900X and 3700X.
Comparing the 3700X to the 1700X sees the Zen 2 CPU leading by 15.5% for average framerates, and 10.7% for minimum framerates.



In Far Cry New Dawn, the i9-9900K leads by 5.3% versus the 3900X (4 cores disabled), and 9.2% versus the 3700X in average framerates.
For minimum framerates, the i9-9900K leads by 9.5% versus the 3900X and 3700X.



In this AIDA64's latency test, we see that the Zen 2 CPUs have higher latency when compared to Intel's i9-9900K.



In SiSoftware's Multi-Thread Latency test we see that while Zen 2 has higher latency when comparing "Worst Matched Cores" to the i9-9900K, it has lower latency when compared against the Best Matched Cores. However, we also see that the latency between the "Best Matched Cores" is lower than the i9-9900K.

Zen 2 has made notable improvements with in regard to multi-thread latency, as the first generation Zen CPUs see 53% higher latency when comparing the "Worst Matched Cores" latency to the Ryzen 7 3700X, while the Ryzen 7 2700X sees 35% higher latency when comparing the latency of the "Worst Matched Cores" to the 3700X.
For "Best Matched Cores" the Ryzen 7 1700X has 43.9% higher latency when compared to the Ryzen 7 3700X, and the 2700X has 42% higher latency when compared to the 3700X as well.



Some more Ryzen 3000 CPU Reviews:
Tom's Hardware - AMD Ryzen 9 3900X and Ryzen 7 3700X Review - Tom's Hardware even test the i9-9900K at 5GHz!
Ryzen 3000 and Zen 2 Review Thread
I just typed up the same thing, thats 10 years im not gettin back.
 

EsqBob

Member
Nov 7, 2017
113
From previous thread:
Because 1995 Toy Story CG has yet to be 100% achieved in a real-time demo, much less a video game. Although, we’ve seen strides made, like in Kingdom Hearts III, but still not yet
“there, boom, TS in real-time!”
These Toy Story comparisons are not fair because the Toy Story we see on Blu-ray and UHD Blu-ray is from a new higher resolution render that probably also changed a few small things, and people usually don't know that.

We don't have a pristine digital file of the original 922p version. If we did I think we could maybe conclude that it does some things better because of higher compute power but overall it looks worse than current games on powerful machines because rendered is much more efficient now, the CG art is more developed and the artists are better.
 

Flutter

Member
Oct 25, 2017
5,913
From previous thread:

These Toy Story comparisons are not fair because the Toy Story we see on Blu-ray and UHD Blu-ray is from a new higher resolution render that probably also changed a few small things, and people usually don't know that.

We don't have a pristine digital file of the original 922p version. If we did I think we could maybe conclude that it does some things better because of higher compute power but overall it looks worse than current games on powerful machines because rendered is much more efficient now, the CG art is more developed and the artists are better.
Huh 922p? I didnt know that.
 
Oct 25, 2017
1,052
the only problem i had with that is they couldve done that for the ps4 but chose not to. they went wide and slow. now granted their cooling solution will be much more expensive and advanced compared to the ps4, but why change now?

300 vs 350 mm2 is roughly 20%.
AFAIK, the cost of production per mm2 at 7nm is greater than at 28nm or 14nm.
So that percentage difference in wafer size likely doesn't result in the same financial impact this time around.
 

Deeke777

Member
Oct 25, 2017
279
United States
I heard it'll launch December 2020. Would it hurt Sony to miss a November launch and get all that extra time to soak up store purchases? Pre-orders will be crazy I'm sure.
 

Kyser73

The Fallen
Oct 25, 2017
2,516
QLD, Australia
Doubt they want to miss out on those Black Friday sales.
Why would you discount a brand new product for BF? Unless something goes badly wrong, both consoles will sell out their launch shipments in pre-orders alone, probably still keep selling on pre-orders through until Jan/Feb depending on stock availability. BF is pretty much irrelevant for this launch.

I expect the launch windows to be similar to this gen, Sony possibly going for a full WW launch instead of US/EU staggered over 2 weeks with JPN later.
 

FarisR

Member
Oct 25, 2017
1,440
Why would you discount a brand new product for BF? Unless something goes badly wrong, both consoles will sell out their launch shipments in pre-orders alone, probably still keep selling on pre-orders through until Jan/Feb depending on stock availability. BF is pretty much irrelevant for this launch.
People buy a lot of stuff on black Friday even not on sale. And no, there's a ton of stock that makes it to store shelves, they aren't going to be selling only via preorders until Jan/Feb.