09:04PM EDT – Final discuss of the day is Xbox Series X System Architecture!
09:05PM EDT – Azure Silicon Architecture Team
09:06PM EDT – 3.eight GHz Zen2 Server cores
09:06PM EDT – DXR, VRS, Machine LEarning Acceleration
09:07PM EDT – 14 Gbps GDDR6, 320-bit = 560 GB/s
09:07PM EDT – Hardware accelerators in blue
09:07PM EDT – 120 Hz assist, VRR, Xbox Velocity Architecture for MSP Crypto/Decomp on NVMe SSD
09:07PM EDT – Acoustic acceleration
09:07PM EDT – HSP/Pluton RoT – safety
09:08PM EDT – 360.4mm2 TSMC N7 enhanced
09:08PM EDT – 15.3B transistors
09:08PM EDT – 2 4 core CPU clusters
09:08PM EDT – 10 GDDR6 controllers
09:09PM EDT – GPU 12 FLOPs
09:10PM EDT – AVX256 offers 972 GFLOP over CPU
09:10PM EDT – 16 GB of GDDR6 complete
09:11PM EDT – >Says Zen2 server class, however L3 cache is cell class?
09:11PM EDT – Display processing is stored off the shader engines
09:11PM EDT – IO hub helps PCIe 4.zero x8
09:11PM EDT – Operates on linear gentle values, not gamma gentle values
09:12PM EDT – ALLM – Auto Low Latency Mode
09:13PM EDT – Increased die value on this APU over earlier technology
09:13PM EDT – Significantly costlier!
09:13PM EDT – Trade off
09:13PM EDT – MS created Audio engines – Three engines, CFPU2, MOVAD, LOGAN
09:13PM EDT – CFPU2 for audio convolution, FFT, reverb
09:14PM EDT – akin to Project Acoustics to mannequin 3D audio sources
09:14PM EDT – MOVAD – hyper real-time {hardware} audio decoder
09:14PM EDT – >300x channels decode directly
09:14PM EDT – greatest commerce off codec, so made in {hardware}
09:15PM EDT – >100dB sign noise ratio
09:15PM EDT – HW realtime real-time matched to decode primarily based on sampling
09:15PM EDT – Logan is providing additionally higher offload in conventional modes
09:15PM EDT – HSP/Pluton: Root of belief, crypto, SHACK (crypto keys)
09:15PM EDT – MSP helps 5 GB/s high-bw crypto on the SSD
09:16PM EDT – DRAM to SSD steadiness wanted for refill
09:16PM EDT – Load occasions are all the time rising until SW-to-DRAM bw will increase, therefore NVMe SSDs
09:17PM EDT – Sampler Feedback System
09:17PM EDT – New metadata for texture parts to pre-load texture caches
09:17PM EDT – Direct Storage
09:17PM EDT – Manages knowledge areas forward of developer
09:18PM EDT – Distinct financial savings for many element texture maps financial savings
09:18PM EDT – Lossless MS XVA 2:1 compression
09:19PM EDT – Need massive GPU – get the tech out of the way in which
09:19PM EDT – Need uncooked ops/second enhance inside PPA and price
09:19PM EDT – 12.2 supported in HW
09:19PM EDT – 26 lively twin CUs (52 CUs)
09:20PM EDT – single geometry helps primatives
09:20PM EDT – DIrectly snoop CPU caches
09:20PM EDT – Dual stream multi-core command processor
09:20PM EDT – Double charge 16-bit math
09:20PM EDT – single cycle challenge charge to scale back stalls
09:21PM EDT – CUs have 25% higher perf/clock in comparison with final gen
09:21PM EDT – GPU Evolution: FLOPS have outpaced mem house and BW
09:21PM EDT – Screen pixels has elevated in th emiddle
09:22PM EDT – How to fill pixels higher with out blowing energy funds
09:22PM EDT – VRS
09:22PM EDT – helps as much as 2×2
09:22PM EDT – 10-30% perf achieve for tiny space value
09:23PM EDT – Full edge element
09:23PM EDT – SFS
09:24PM EDT – Previously very gradual to allow
09:24PM EDT – Two new HW constructions for tile-by-tile administration for in-DRAM textures
09:25PM EDT – clamps LOD
09:26PM EDT – Tilemaps ought to keep on die for greatest latency
09:27PM EDT – SFS: 60% IO/Mem financial savings for small die space value
09:27PM EDT – DX Ray Tracing Accel
09:27PM EDT – Not a whole alternative – RT might be utilized selectively primarily based on conventional fashions
09:28PM EDT – Custom ray-triangle items
09:28PM EDT – ML inference
09:29PM EDT – Two virtualized command streams – two VMs
09:29PM EDT – Main title OS vs system OS
09:29PM EDT – 32b HDR rendering, mixing show
09:29PM EDT – Optimized video games. Unable to point out on the occasion
09:30PM EDT – Q&A Time
09:31PM EDT – Q: TDP?…