Sony PS6, Microsoft neXt Series - 10th gen console speculation [2020]

Well, going by the 16-month intervals that've been established so far for RDNA (granted it's not much of a pattern but it's something), in a best-case scenario RDNA 3 should be ready by early 2022. However, a few things could possibly that: the chip shortages, lower-than-anticipated sales due to lack of supply to satisfy demand with RDNA 2 cards, and ongoing strain on developments due to lockdowns backing up all of the channels in terms of manufacture, distribution etc.

I would just hope that those factors don't push RDNA 3 back to late 2022 because unless they're throwing out a ton of existing things through massive architecture redesigns or chiplets are proving to be a real pain to implement, there's nothing on the R&D front that seems it'd hold back release to late 2022/early 2023. Even if they're working on more hardware accelerators, that shouldn't increase the window between 2 and 3 too drastically (I'd hope).
Can't remember who it was, but one of AMD bosses confirmed some time ago (more like year than couple months) that they are committed to release new GPU generation every year again (though leaving the option open to count for example process shrink as new generation regardless of architectural features)
Also you can't really say we've established 16-month intervals, since there has been only one RDNA generation interval.
 
Can't remember who it was, but one of AMD bosses confirmed some time ago (more like year than couple months) that they are committed to release new GPU generation every year again (though leaving the option open to count for example process shrink as new generation regardless of architectural features)
Also you can't really say we've established 16-month intervals, since there has been only one RDNA generation interval.

Yeah I know we can't say there's a "pattern" per-se in release intervals because we only have one instance to look at, but it's a grounds from which we can use to go off from.

Here's to hoping AMD can keep to that ambition, as that means more competition from Nvidia, Intel and Apple.
 
2024 or whenever 3nm is in mass production for discrete gpus

PS5 slim:
Zen 3+ with unified l3 at ~5ghz
RDNA3/4 36CUs, 64 ROPs at 2.5-3ghz 10.6-12.7TFs
16GB of 16gb/s GDDR6 on 256bit bus, 512GB/s
1.5TB SSD 5-7gb/s
Checkerboarded 4k60 target with full RT
~150mm2 die size
$299

PS5 pro:
Zen 3+ with unified l3 at ~5ghz
RDNA3/4 72CUs, 128 ROPs at 2.5-3ghz 21.2-25TFs
16GB of 16gb/s GDDR6 on 256bit bus, 512GB/s
128MB infinity cache
3TB SSD 5-7gb/s
4k60 native target with full RT
250-300mm2 die size
$499
 
Last edited:
Xbox Series S2

Zen 3+ with unified l3 at ~5ghz
RDNA3/4 26CUs, 32 ROPs at 2-2.5ghz, 6.7-8.3TF
12GB of 14gb/s GDDR6 on 192bit bus, 336GB/s
1TB SSD 3-4GB/s
1080p-1440p native with DirectML upscaling to 4K. ABILITY TO PLAY 1X ENHANCED GAMES VIA BC. BC with series S.
~125mm2 die size
$199

Xbox series X2:
Zen 3+ with unified l3 at ~5ghz
RDNA3/4 56CUs, 64 ROPs at 2.25-2.5ghz, 16-18TF
16GB of 19gb/s GDDR6x on 256bit bus, 608GB/s
2TB SSD 3-4GB/s
4k native with DirectML upscaling to 8k
~200mm2 die size
$399

Xbox series Z:
Zen 3+ with unified l3 at ~5ghz
RDNA3/4 108CUs, 128ROPs, at 2.25-2.5ghz, 30-35TF
24GB of 21gb/s GDDR6x on 384bit bus, 1008GB/s
2TB SSD 3-4GB/s
4k native with best RT in class
~360mm2 die size
$599
 
Last edited:
Xbox Series S2

Zen 3+ with unified l3 at ~5ghz
RDNA3/4 26CUs, 32 ROPs at 2-2.5ghz, 6.7-8.3TF
12GB of 14gb/s GDDR6 on 192bit bus, 336GB/s
1TB SSD 3-4GB/s
1080p-1440p native with DirectML upscaling to 4K
~125mm2 die size
$199

Xbox series X2:
Zen 3+ with unified l3 at ~5ghz
RDNA3/4 56CUs, 64 ROPs at 2.25-2.5ghz, 16-18TF
20GB of 19gb/s GDDR6x on 320bit bus, 760GB/s
2TB SSD 3-4GB/s
4k native with DirectML upscaling to 8k
~200mm2 die size
$399

Xbox series Z:
Zen 3+ with unified l3 at ~5ghz
RDNA3/4 108CUs, 128ROPs, at 2.5-3ghz, 35-42TF
24GB of 21gb/s GDDR6x on 384bit bus, 1008GB/s
2TB SSD 3-4GB/s
4k native with best RT in class
~360mm2 die size
$599
Unified L3 cache on the CPU is all but confirmed if they stick to x86. Will be interesting if they remain with GDDR instead of HBM.
 
Unified L3 cache on the CPU is all but confirmed if they stick to x86. Will be interesting if they remain with GDDR instead of HBM.

Something about HBM not having multi-level cell.

"09:39PM EDT - Q: With 20 channels GDDR6, is that really cheaper than 2 stacks HBM? A: We're not religious about which DRAM tech to use. We needed the GPU to have a ton of bandwidth. Lots of channels allows for low latency requests to be serviced. HBM did have an MLC model thought about, but people voted with their feet and JEDEC decided not to go with it."

https://www.anandtech.com/show/1599...ft-xbox-series-x-system-architecture-600pm-pt
 
Something about HBM not having multi-level cell.

"09:39PM EDT - Q: With 20 channels GDDR6, is that really cheaper than 2 stacks HBM? A: We're not religious about which DRAM tech to use. We needed the GPU to have a ton of bandwidth. Lots of channels allows for low latency requests to be serviced. HBM did have an MLC model thought about, but people voted with their feet and JEDEC decided not to go with it."

https://www.anandtech.com/show/1599...ft-xbox-series-x-system-architecture-600pm-pt

Is MLC really that much a requirement for RAM? I know this relates directly to capacity but there are already 2 GB module capacities and 12-Hi stacks for HBM2E. Shouldn't be too much a stretch to expect 16-Hi stacks and 4 GB module capacities as standard by the time 10th-gen systems are around...hopefully relatively affordable too. HBM3 is supposed to be a pretty hefty architectural redesign anyhow (I really hope they incorporate FGDRAM ideas into it as those ideas would resolve a lot of current limitations).

The mid-gen and next-gen console specs sound pretty good. Some specific things I see differently though on are the RDNA spec for any mid-gen refreshes; MS and Sony probably want to keep costs down and maximize profit margins so if say, Sony goes for a 72 CU PS5 Pro, I don't think it'll be RDNA 3 or 4, but probably RDNA 2, with the benefit in your example being it'd be on 3nm process. I'm almost tempted to say the same for Microsoft, but they may have a vested interest in Azure to expand the featureset and that would just trickle down to a new mid-gen refresh console.
 
Is MLC really that much a requirement for RAM? I know this relates directly to capacity but there are already 2 GB module capacities and 12-Hi stacks for HBM2E. Shouldn't be too much a stretch to expect 16-Hi stacks and 4 GB module capacities as standard by the time 10th-gen systems are around...hopefully relatively affordable too. HBM3 is supposed to be a pretty hefty architectural redesign anyhow (I really hope they incorporate FGDRAM ideas into it as those ideas would resolve a lot of current limitations).

I think MLC is a requirement specifically for MS due to ECC needed for their Xcloud.

The mid-gen and next-gen console specs sound pretty good. Some specific things I see differently though on are the RDNA spec for any mid-gen refreshes; MS and Sony probably want to keep costs down and maximize profit margins so if say, Sony goes for a 72 CU PS5 Pro, I don't think it'll be RDNA 3 or 4, but probably RDNA 2, with the benefit in your example being it'd be on 3nm process. I'm almost tempted to say the same for Microsoft, but they may have a vested interest in Azure to expand the featureset and that would just trickle down to a new mid-gen refresh console.

What kind of architecture they use doesn't matter. Where if it's GCN, RDNA 1 or 17 won't add a penny in terms of fab cost. What only matters is the node, die size, and the frequencies you're pushing.
 
I think MLC is a requirement specifically for MS due to ECC needed for their Xcloud.

Good point; that's a need for it in the server side for sure.

What kind of architecture they use doesn't matter. Where if it's GCN, RDNA 1 or 17 won't add a penny in terms of fab cost. What only matters is the node, die size, and the frequencies you're pushing.

True, but what RDNA generation they go with will have some impact on the availability of driver support and development tools leveraging that specific generation. So there's still a "cost" associated with that in a sense, just not in a way of impacting direct pricing of wafers or components.

After the growing pains we're seeing with RDNA 2 feature uptake currently happening right now (certain features don't seem like they'll be widely available until early 2022, almost two years after the first wave of released cards), Microsoft and Sony might want to avoid that for what would basically be mid-gen refreshes.

Although since it seems it'll take a bit longer than usual to get current-gen fully rolling (due to COVID), if they want to prolong the generation by a year or so that could help them justify incorporating RDNA 3/4 feature sets and doing some design changes, since it would maybe take about a six months - one year for other things like tools and driver support to get up to par at large.
 
That's a good start, let's hope to see that trend continue with even faster SSD drives to come in the future at similar capacities. Especially if they can get those prices down while retaining 2:1 revenue/profit margins over BOM, assembly, packaging and distribution costs, because that means prices favorable for 10th-gen consoles in that department and also MSRP.
 
Here're my predictions for 10th-gen. Starting with Microsoft; I'll post the Sony one tomorrow.

----

[MICROSOFT XBOX SERIES Z/SURFACE Z]

>Stronger approach towards a PC-like modular design specification (Series Z)

>Future-proofed design with upclockable CPU

>Custom-built GPU card that is upgradable

>No standard VR or AR headset device included; compatible with helmets from
several brands, including Samsung and HTC

>Additional SKU could be provided with an included headset at a higher price point

>Positioned in marketing as a next-gen gaming, productivity & business high-end laptop
device (Surface Z.D)

>Two models; Series Z and Surface Z.D

>Series Z uses Xbox OS & has scaled-down port interface counts & some
reduced feature support. Sold at traditional loss-leading model

>Surface Z.D uses Windows Next Home OS & has expanded port interface*, slots, &
some additional QoL feature support. Sold at profit

* = Some of this via Docking Bay expansion module; MOTL

>Surface Z is intended for productivity, gaming and business work tasks,
while having same specs as Series Z

>Both utilize same default CPU, GPU, audio, SSD, RAM configurations

>Surface Z.D is a laptop-style device similar to Surface Book 3, but provided
in a single configuration with same processor and RAM components as Series Z.
However, Surface Z.D operates in two modes: Portable and Docked.

>Portable Mode uses it as a laptop, and the CPU & GPU clocks are lowered
to consume less power and provide suitable battery life.

>Docked Mode connects the Surface Z.D to a Docking Bay, with a special
cooling pass-through compartment that connects to a portion of the Surface
Z.D when a piece of the unit is removed, providing additional cooling
and thermal control. This allows the Surface Z.D to operate at full
clocks, i.e same clocks as Series Z's CPU & GPU. It's recommended to
keep the Series Z.D connected for charge via AC mode for this.

>The Docking Bay for Surface Z.D is not included in the default configuration,
but is included in another configuration for an additional fee. Surface Z.D
does not need the Docking Bay in order to function.

>The Docking Bay includes four additional PCIe slots, which the Surface Z.D
can use to address additional peripherals. This leverages use of a system
bus extension.

>Surface Z.D comes with a high-quality 17" AMOLED 6K display touchscreen
with up to 300 Hz refresh rate

>Both provide support for Optane-style "M.NVRAM" form-factor persistent storage,
PCIe 6.0-CXL links, 2 per slot (16 GB/s per slot), 2x slots

>Series Z does not come with included persistent memory storage. Can
support 2x 16 GB/s M.NVRAM cards (one per slot), 256 GB capacity each, for
512 GB @ 32 GB/s.

>Surface Z.D comes with 128 GB of Micron 3D X-point persistent memory as
2x 12 GB/s M.NVRAM cards providing 24 GB/s bandwidth. Can support 2x 16 GB/s
M.NVRAM cards (one per slot), 512 GB capacity each, for 1 TB @ 32 GB/s.

>Series Z supports 3P GPUs. However, the Series Z is only compatible
with a limited range of specific GPUs from Microsoft/AMD meant for the Z. Specific
GPUs will be made available at a later date; may require change of modular PSU.

Surface Z.D does not support any 3P GPUs, as its GPU is soldered onto the motherboard
and integrated into the package design.

>Series Z supports BIOS-level safe detection feature; the system is rated for installation
of specific peripheral components in PCIe slots that fall within a given power profile.
Devices exceeding this will not be authorized installation. CPU upclocking only
permitted at BIOS level, only on detection of new installed GPU. GPU installation
requires power to be off, also requires PSU replacement (PSU comes with GPU package).
If GPU is removed when power on, system shuts down and if new GPU is installed and
system turned on, system detects as unauthorized change and shuts down unless previous
GPU and PSU are reinstalled.

>Implementation of advanced SAM/BAR addressing scheme enforced through custom block that
handles data coherency and transfer of info from storage, NVRAM, GPU and CPU memory pools.
Hardware bandwidth limit of 80 GB/s

>NAND decompression I/O built into custom data management block. Supports 4:1 decompression ratio.

>Custom data management block supports ringbus arbitration to allow CPU, GPU, audio to share
access to storage and NVRAM​

[CPU]

>CORES: 16

>THREADS: 32

>CLOCK: 5.4 GHz

>L1$: 64 KB (per core), 1 MB (total)

>L2$: 1 MB (per core), 16 MB (total)

>L3$: 16 MB (all)

>SOCKET: N/A (embedded CPU)

**Series Z-X CPU socket only supports default CPU processor; upclock
possible only when new Microsoft/AMD GPU/PSU package authorized install.
Clock increase range varies depending on newly installed GPU/PSU metrics.​

[GPU]

>ARCHITECTURE: RDNA

>GENERATION: 7

>PROCESS: N3P

>CONFIG: Chiplet

>SHADER ENGINES: 2

>SHADER ARRAYS (per SE): 4 (HSE-1), 2 (HSE-2)

>COMPUTE UNITS: 80

>HSE-1: 56 CUs (4x Dual 7 CU SAs)

>HSE-2: 24 CUs (2x Dual 6 CU SAs)

>COMPUTE UNIT CONFIGURATION: Dual CU

>SHADER CORES (PER CU): 96

>SHADER CORES (TOTAL): 7680

>ROPs: 128

[* = In Series Z, both HSEs technically have 64 ROPs; however there
is additional logic built into the GPU allowing HSE-1 to use one
of the 32 ROP banks in HSE-2 as if it were a 3rd 32-block ROP bank.]

>TMUs: 6 (per CU), 480 (total)

>ALUs/SHADER UNITS: 7680

>STATE MODES (IF ANY): 4; FULL (80 CUs), HSE-1 (56 CUs), HSE-2 (24 CUs), DUAL-HALF (56 HSE-1, 24 HSE-2)

>CLOCK FREQUENCIES: 2723 MHz (FULL), 4291.388 MHz (HALF-SE 1), 4522.4968 MHz (HALF-SE 2), 2723 MHz (DUAL-HALF)

>IPC: 2

>IPS: 4875.92 million IPS (FULL), 8206.964 million IPS (HALF-SE 1), 9044.9936 million IPS (HALF-SE 2), 4875.92 million IPS (DUAL-HALF)

>THEORETICAL FLOATING POINT OPERATIONS PER SECOND: 41.825 TF (FULL), 46.141 TF (HSE-1), 20.839 TF (HSE-2), 41.825 TF (DUAL-HALF)

># PRIM UNITS: 4 (2 each SE, 1 each SA)

># PRIMs/CLOCK PER UNIT: 4

># TRI RAS/CLOCK PER UNIT: 2

>PRIMITIVES PER CLOCK: 16

>TRIANGLES PER CLOCK: 8

>GEOMETRY CULLING RATE: 43.568 billion (FULL), 34.331 billion (HSE-1), 36.179 billion (HSE-2), 43.568 billion (DUAL-HALF)

>TRIANGLE RASTERIZATION RATE: 21.784 billion (FULL), 17.1655 billion (HSE-1), 18 billion (HSE-2), 21.784 billion (DUAL-HALF)

>PIXEL FILL RATE: 348.544 Gpixels/s (FULL), 411.973 Gpixels/s (HALF HE-1), 289.439 Gpixels/s (HALF-HE 2), 348.544 Gpixels/s (DUAL-HALF)*

* = 261.408 Gpixels/s (HSE-1, 96 ROPs), 87.136 Gpixels/s (HSE-2 32 ROPs), or:

* = 174.272 Gpixels/s (HSE-1, 64 ROPs), 174.272 Gpixels/s (HSE-2, 64 ROPs)

>TEXTURE FILL RATE: 1307.04 Gtexels/s (FULL), 1441.906 Gtexels/s (HALF SE-1), 651.239 Gtexels/s (HALF-SE-2), 1307.04 Gtexels/s (DUAL-HALF)*

* = 914.928 Gtexels/s (SE1), 392.112 Gtexels/s (SE2)

>THREAD DEPLOYMENT RATE: 81,920 (FULL), 57,344 (HSE-1), 24,576 (HSE-2), 81,920 (DUAL-HALF)

>CACHE CAPACITY:

>L0$: 16 KB per CU (1.152 MB total)

>L1$: 64 KB per dual-CU/WGP (2.304 MB total)

>L2$: 512 KB per Array (2 MB total)

>L3$: 64 MB

>TOTAL: 69.456 MB

>CACHE BANDWIDTH:

>L0$: 41.4 TB/s (FULL), 45.679 TB/s (HALF SE-1), 20.63 TB/s (HALF SE-2), 41.4 TB/s (DUAL-HALF)*

* = 28.98 TB/s (HSE-1), 12.42 TB/s (HSE-2)

>L1$: 28.98 TB/s (FULL), 31.975 TB/s (HALF SE-1), 12.42 TB/s (HALF-SE-2), 28.98 TB/s (DUAL-HALF) (L0$ * .70)

>L2$: 17.388 TB/s (FULL), 19.185 TB/s (HALF SE-1), 7.452 TB/s (HALF SE-2), 17.388 TB/s (DUAL-HALF) (L1$ * .60)

>L3$: 8.694 TB/s (FULL), 9.592 TB/s (HALF SE-1), 3.76 TB/s (HALF SE-2), 8.694 TB/s (DUAL-HALF) (L1$ * .30)

>TOTAL: 96.462 TB/s (FULL), 106.431 TB/s (HALF SE-1), 44.262 TB/s (HALF SE-2), 96.462 TB/s (DUAL-HALF)
...2nd half below...
 
...continued...

[TASK ACCELERATION ENGINE]

**Heavily modified CU for various hardware-based RT, ML, IU, AI tasks

**Each Shader Engine is paired with a Task Acceleration Engine

**TAEs can house either 1 or 2 Task Acceleration Arrays

**TAAs can be of any given size when it comes to TAUs; however ideal
balanced designs generally call for TAAs that are 50% the core unit
count of a design's Shader Array (i.e a design with 5 Dual CU Shader
Arrays (10 CUs) having 5 TAU TAAs).

**TAUs can be thought of as pseudo-FPGA cores; each unit has internal
fixed-function hardware as well as a modified frontend compared
to normal CUs. TAUs also can range from having the following:

>2,000 To 8,000 logic cells

>16 Kb (2 KB) to 128 Kb (16 KB) distributed RAM (LUT) (acts as L0$)

>64 Kb (8 KB) to 512 Kb (64 KB) BRAM (acts as L1$)

>Internal Command Sync & Program Interface Configuration core
(directed by application to program the FPGA logic into
desired program state, manage unit state)

>Compute Thread Controller core (this is what programmers
interface with in dispatching code for the configured units
in the TAA to execute)

>1 Mb (128 KB) to 4 Mb (512 KB) SRAM (L2$, shared with all units in TAA;
meant for processed data locality storage)

>Shared 8 MB XIP (Execute-in-Place) Resource State Preset-allocated
MRAM (holds the microcode in bit-addressable/byte-addressable
format for immediate configuration)*

* = Will be able to be updated with future firmware changes

* = Shared between both TAAs

* = Simply used to configure the FPGA logic and fixed-function
aggregate unit states to specific profile presets (RT, ML, AI,
or IU) by storing the relevant set-up and configure/compile data


># TAEs: 2 (1 per SA)

># TASK ACCELERATION UNITS (TAU) PER TAE: 7 (HSE-1), 6 (HSE-2), 13 (TOTAL)

>LOGIC CELLS: 4500 (per TAU); 31,500 (TAE; HSE-1), 27,000 (TAE; HSE-2), 58,500 (total)

>LUT RAM (L0$): 8 KB (per TAU), 56 KB (TAE; HSE-1), 48 KB (TAE; HSE-2) 104 KB (total)

>BRAM (L1$): 32 KB (per TAU), 224 KB (TAE; HSE-1), 192 KB (TAE; HSE-2), 416 KB (total)

>SRAM (L2$): 256 KB (TAE; HSE-1), 128 KB (TAE; HSE-2), 384 KB (total)

>MRAM: 8 MB (total)

># CU: 1 (PER DUAL CU IN SA), 5 (PER SA), 10 (TOTAL)​

[AUDIO]

>ARCHITECTURE: Microsoft Next Audio Engine (NAE)

>GENERATION: 1

>DSPs: 4 (4x Logan)

>PROGRAMMABLE LOGIC: 2 (CFPU2 Gen 2)

>Can leverage up to 128 GB/s of system bandwidth
[MEMORY]

>TYPE: HBM4 (GPU/Audio), DDR6 (CPU)

>GENERATION: 1

>MODULE CAPACITY: 1 GB (GPU/Audio), 4 GB (CPU, per DIMM module)

>MODULE AMOUNT: 24x (HBM4), 2x (DDR6, DIMM modules)

>CONFIGURATION: 2x 12-HI (GPU/Audio), 2-Channel (DDR6)

>I/O PIN BANDWIDTH: 4 Gbps (500 MB/s) (GPU/Audio), 11.2 Gbps (1.7 GB/s) (DDR6)

>I/O PIN COUNT: 128 (GPU/Audio), 64 (DDR6)

>BUS WIDTH: 3072-bit (GPU/Audio), 128-bit (DDR6)

>MODULE BANDWIDTH: 64 GB/s (GPU/Audio), 89.6 GB/s (CPU, per DIMM module)

>TOTAL BANDWIDTH: 1536 GB/s (1.536 TB/s) (GPU/Audio), 179.2 GB/s (CPU), 1715.2 GB/s (total)

>TOTAL CAPACITY: 24 GB (GPU/Audio), 8 GB (CPU), 32 GB (total)

*Of GPU/Audio pool, suggested split is 22 GB (1408 GB/s) GPU, 2 GB (128 GB/s) Audio.
However, this partition amount is adjustable by the developer​

[I/O COMPRESSION/DECOMPRESSION/MEMORY CONTROLLER]

[SSD]

>FORM FACTOR: M.2

>INTERFACE: PCIe-CXL

>INTERFACE GENERATION: 6.0

>CONNECTION SLOT(S): 2x x2 (Series Z-X), 1x x2 & 1x x4 (Surface Z.D)

>NAND TYPE: ZNAND

>NAND MODULE CAPACITY: 2 Tb (256 GB)

>NAND MODULE BANDWIDTH: 4 GB/s

>NAND MODULE LATENCY: 1.8 ms

>NAND MODULE CHANNELS: 4 (per module), 16 (total)

># NAND MODULES: 4x

>DEVICE BANDWIDTH : 16 GB/s

>DEVICE BANDWIDTH SUPPORT: 16 GB/s (x2, Series Z-X), 32 GB/s (x4, Surface Z.D)

>DRIVE CAPACITY: 2 TB (2048 GB)

>MAXIMUM SUPPORTED DRIVE CAPACITY: 16 TB (16384 GB)(Series Z-X), 64 TB (Surface Z.D)

[FLASH MEMORY CONTROLLER]

>STORAGE CACHE: DDR6, 2 GB

>INTERFACE SUPPORT: x4 (full) M.2 PCIe-CXL 6.0 NVMe & x2 (half) M.2 PCIe-CXL 6.0 NVMe (Surface Z.D),
2x x2 (half) M.2 PCIe-CXL 6.0 NVMe (Series Z-X)

>BANDWIDTH CAPACITY (CACHE): 102 GB/s

>LATENCY THRESHOLD: Sub-100 ns
[COMPRESSOR/DECOMPRESSOR/MANAGEMENT]

>DECOMPRESSION BANDWIDTH LIMIT (HARDWARE): 64 GB/s

>DECOMPRESSION RATIO: 4:1

>COMPRESSION BANDWIDTH LIMIT (HARDWARE): 64 GB/s

>COMPRESSION RATIO: 4:1

># ENGINES: 6

1: SSD Device Core (manages both SSD devices)

2: NVRAM Path Block (PB)

3: CPU Path Block (PB)

4: GPU Path Block (PB)

5: Ethernet/Wifi Path Block (PB)

6: Management Core (MC)
[PERIPHERAL SUPPORT]

>USB GENERATION SUPPORT: GEN 1, GEN 2, GEN 3, GEN 4

>USB-C SUPPORT: YES

># USB PORTS: 4 (Series Z-X), 3 (Series Z-W)

(Series Z)

>FRONT: 2x regular USB Gen 4 Type-C 2x2
(controllers, charging kits, etc.)

>BACK: 1x USB Gen 4 Type-A 2x2, 1x USB Gen 3​
(Surface Z.D)

>1x regular USB Gen 4 Type-C 2x2, 1x USB Gen 3, 1x USB Gen 2​
>MICRO SD SUPPORT: YES

*Used for physical media delivery (64 GB/128 GB/192 GB Game Cards). Managed by FMC & Decompressor.

*Game Card bandwidth: 72 MB/s​
># MICRO SD SLOTS: 1 (Series Z-X) (FRONT, UHS-III, SDHC class), 2 (Surface Z.D) (1 FRONT, 1 SIDE)

>CF EXPRESS SUPPORT: YES

># CF EXPRESS SLOTS: 1 (Series Z-X), 1 (Surface Z.D)

*This is mainly for BC of Series X and Series S Expansion Storage Cards
>THUNDERBOLT SUPPORT: NO (Series Z-X), YES (Surface Z.D)

># THUNDERBOLT PORTS: 1 (Back)

>PCIe SUPPORT: YES

>PCIe SUPPORT GENERATION: 6.0-CXL

># PCIe SLOTS:

>SERIES Z (6): 1x x8 (64 GB/s) PCIe 6.0-CXL High-Speed slot (expandable GPU support), 2x x2 (16 GB/s each)
M.2 NVMe Gen 6-CXL slots(SSD/storage slots), 2x x2 (16 GB/s each) M.NVRAM slots (expandable NVRAM support),
1x x2 PCIe 3.0 (for included wireless module)

>SURFACE Z.D (4): 1x x16 (128 GB/s) PCIe 6.0-CXL High-Speed link (GPU), 1x x4 (32 GB/s) M.2 NVMe Gen 6-CXL slot
(SSD/storage slots), 2x x2 (16 GB/s each) M.NVRAM slots (expandable NVRAM support)

>SURFACE Z.D DOCKING BAY/BLOCK (4): 1x x2 (16 GB/s) M.2 NVMe Gen 6-CXL slot (SSD/storage slot), 2x x4
PCIe 6.0 (general peripherals), 1x x2 PCIe 3.0 (for optional devices like Series Z-branded wireless module)
[WIFI/ETHERNET/BLUETOOTH]

[ETHERNET]

>IMPLEMENTATION: Single Port

>ETHERNET PHYSICAL LAYER: 25 Gbit

>NAME: 25GBASE

>STANDARD: 802.3bq-2016

>CONNECTOR: RJ45

>LINE CODE: 64b/66b

>TRANSCEIVER MODULE: SFP28

>RECOMMENDED CABLE: 25GBASE-CR, 25GBASE-CR-S

>BANDWIDTH: 3.125 GB/s
[WIFI]

>IMPLEMENTATION: Single Module

>NAME: Wifi 7

>STANDARD: 801.11acx

>ENCODE RATE: 8b/10b encoding

>HARDWARE MU-MIMO STREAMS: 12

>BANDWIDTH: 2925.4275064 Mbps per MIMO, 14.627 Gbps theoretical, 1.8283921915 GB/s converted

>Specs for included wireless transmission module are:

>12 MU-MIMO antenna streams (4x 160 MHz + 4x 320 MHz)

>7 GHz band (7234 MHz)

>Compatible with 160 MHz + 80 MHz channel width

>1659.962 Mbps per MIMO (160 MHz), (60 MHz)

>19.91954489925 Gbps theoretical

>2.48994 GB/s converted

>~ 2.240946 GB/s effective (90% utilization)

>The ethernet/wifi processing block has the following specifications:

>Dual-core/quad-thread high-end ZEN-ARM processor block

>Embedded 256 MB PS-RAM

>25 Gbit ethernet PHY and LOG layer integration; ethernet port link off-chip
ethernet port module (where ethernet cable plugs in)​

[VR/AR SUPPORT]

**No included VR/AR headset

**Built-in support for Microsoft Mesh

**Microsoft will build upon partnerships with existing VR headset providers such as Samsung, HTC, Sony etc. to provide infrastructure for 3P headset compatibility on Series Z systems.​

[PRICING]

>SURFACE Z.D:

$2499.99 (Basic model with Windows Next Home Edition)

$2999.99 (Model with Docking Bay)

$2599.99 - $3099.99 (Models with bundled 15 months of Office suite subscriptions)

$2749.99 - $3349.99 (Models with 2 to 4-year Microsoft Protection Plans)

$549.99 (Docking Bay)
>SERIES Z: $499.99​

[RELEASE]

>November 2026 (Surface Z.D)

>Fall 2027 or Spring 2028 (Series Z-X)
 
Last edited:
Some people really have too much free time but instead of actually playing games on current consoles they write long gimmick lists fantasising about future consoles instead.....
 
Surface Z.D uses Windows Next Home OS & has expanded port interface*, slots, & some additional QoL feature support. Sold at profit
The very good point @Nisaaru made above aside, this falls already flat on it's face here. MS won't release "Windows Next Home OS", there will be just new versions of Windows 10

Also Micron already dropped 3D X-point development. Didn't read further if there's some other obvious faults too.
 
Some people really have too much free time but instead of actually playing games on current consoles they write long gimmick lists fantasising about future consoles instead.....

You can do both. I also do work in-between this. But why be snarky and passive-aggressive instead of contributing something valuable to the discussion?

The very good point @Nisaaru made above aside, this falls already flat on it's face here. MS won't release "Windows Next Home OS", there will be just new versions of Windows 10

That was just a made-up name for whatever comes after 10 they decide to do. They technically have "forks" of 10 (maybe that's not the best way to put it) for enterprise markets and such distinct from what's offered for the home. Same for education with the S Mode stuff.

Also Micron already dropped 3D X-point development. Didn't read further if there's some other obvious faults too.

Not really; they still have their X100 drives using the technology, they will probably make more of those. Intel still have Optane and I think I meant to put them there instead of Micron but it slipped by when proofing; they don't have plans for Optane dev in the consumer space currently far as iteration improvements but they have a new generation of it coming specifically to enterprise and server markets.

Dunno what "other obvious faults" would mean when this is me speculating, I don't claim these are what their 10th-gen specifications will be. But I use a lot of methodologies to try keeping speculation grounded.
 
The very good point @Nisaaru made above aside, this falls already flat on it's face here. MS won't release "Windows Next Home OS", there will be just new versions of Windows 10

Also Micron already dropped 3D X-point development. Didn't read further if there's some other obvious faults too.
Broadly agree, but there is a new version of windows 10 coming out, windows 10X its for more mobile form factors and low powered chromebook like devices at the moment but they will be expanding its target market into the future. Its also the first desktop OS to use the windows Core OS system that microsoft has been spending who knows how long developing. Windows Core OS is going to be a common operating system base layer onto which different UIs are built.

There are a couple products that already use implementations on top of Windows Core OS, the Surface Hub and the hololens, and when it is released the Surface Neo will also use it.
 
Broadly agree, but there is a new version of windows 10 coming out, windows 10X its for more mobile form factors and low powered chromebook like devices at the moment but they will be expanding its target market into the future. Its also the first desktop OS to use the windows Core OS system that microsoft has been spending who knows how long developing. Windows Core OS is going to be a common operating system base layer onto which different UIs are built.

There are a couple products that already use implementations on top of Windows Core OS, the Surface Hub and the hololens, and when it is released the Surface Neo will also use it.

Then replace Windows Next Home OS with Windows Core OS ver x.xx or something; roughly speaking these are things of similar wavelength. I just wanted to hard-distinguish a build intended for home consumer use, but we can't predict what Microsoft will call that several years from now.
 
Not really; they still have their X100 drives using the technology, they will probably make more of those. Intel still have Optane and I think I meant to put them there instead of Micron but it slipped by when proofing; they don't have plans for Optane dev in the consumer space currently far as iteration improvements but they have a new generation of it coming specifically to enterprise and server markets.
Micron won't.
 
Back
Top