New benchmarks of the Intel Xeon Platinum 8468 48-Core Sapphire Rapids CPU have leaked out which present aggressive efficiency versus AMD’s Milan.
Intel Xeon Platinum 8468 48-Core “Sapphire Rapids” CPU Trades Blows With AMD’s EPYC 7763 64-Core “Milan” Chips
The benchmark that has been found by HXL (@9550pro) reveals us an engineering pattern of the Intel Xeon Platinum 8468, one of many many Sapphire Rapids-SP CPUs. Based mostly on the specs sheet that leaked earlier, the Xeon Platinum 8468 goes to characteristic 48 cores and 96 threads whereas utilizing the 10nm ESF course of node and the Golden Cove core structure. The CPU is predicted to characteristic 105 MB of L3 cache & a 350W TDP. The CPU is listed with 2.1 GHz clocks inside Cinebench R23 and 800 MHz inside CPU-z which means that the clock speeds aren’t near the retail numbers.
The CPU examined is the sixth Stepping and we all know from earlier studies that the Sapphire Rapids-SP chips have gone by way of numerous steppings to repair numerous bugs that the chips have encountered to this point. It’s reported that the CPU was examined inside two benchmarks, one being Cinebench R23 & the opposite being V-Ray. The CPU was examined in a dual-socket configuration with 96 cores & 192 threads in complete.
Beginning with Cinebench R23, the chip scored 90411 factors in multi-core and 1351 factors in single-core. The CPU is certainly sooner than retail EPYC Milan chips in single-core and comparable within the multi-core assessments because the high EPYC 7763 scores round 90000-95000 factors in the identical take a look at. Because of this the EPYC 7763 is at finest 8.5% sooner in multi-threaded efficiency, a distinction that may simply be overcome by remaining retail clock speeds. The EPYC chip additionally contains 33% extra cores.
0
19357
38714
58071
77428
96785
116142
In comparison with an early ES EPYC Genoa chip, the Intel Xeon Platinum 8468 CPU with 48-cores and 96 threads or 96 cores and 192 threads in complete is round 18% slower however as soon as once more, the Genoa chip affords 33% extra threads. The complete potential of EPYC Genoa 96-core chips is restricted to 256 threads since that is the utmost thread restrict.
We even have V-Ray the place the Intel Xeon Platinum 8468 48-Core CPUs are listed in a 4-socket config but it surely was truly working a 2-socket config. The CPU clock velocity is rated a bit larger at 3.0 GHz & the rating is rated at 85,766 samples. Following is how the efficiency compares to different server chips:
0
16667
33334
50001
66668
83335
100002
Intel has moved the Sapphire Rapids-SP Xeon CPU launch to early 2023 so it appears to be like like they are going to be competing with EPYC Genoa-X & Bergamo CPUs by the point they’re extensively out there to x86 server clients. This sort of efficiency would have been aggressive a number of years again towards Milan however with Zen 4 on the horizon, issues look grim for Intel regardless of the key uplift over Ice Lake-SP which now appears too little.
Intel Xeon SP Households (Preliminary):
Household Branding | Skylake-SP | Cascade Lake-SP/AP | Ice Lake-SP | Sapphire Rapids | Emerald Rapids | Granite Rapids | Diamond Rapids | |
---|---|---|---|---|---|---|---|---|
Course of Node | 14nm+ | 14nm++ | 14nm++ | 10nm+ | Intel 7 | Intel 7 | Intel 3 | Intel 3? |
Platform Identify | Intel Purley | Intel Purley | Intel Cedar Island | Intel Whitley | Intel Eagle Stream | Intel Eagle Stream | Intel Mountain Stream Intel Birch Stream | Intel Mountain Stream Intel Birch Stream |
Core Structure | Skylake | Cascade Lake | Cascade Lake | Sunny Cove | Golden Cove | Raptor Cove | Redwood Cove? | Lion Cove? |
IPC Enchancment (Vs Prev Gen) | 10% | 0% | 0% | 20% | 19% | 8%? | 35%? | 39%? |
MCP (Multi-Chip Package deal) SKUs | No | Sure | No | No | Sure | Sure | TBD (Presumably Sure) | TBD (Presumably Sure) |
Socket | LGA 3647 | LGA 3647 | LGA 4189 | LGA 4189 | LGA 4677 | LGA 4677 | TBD | TBD |
Max Core Rely | Up To twenty-eight | Up To twenty-eight | Up To twenty-eight | Up To 40 | Up To 56 | Up To 64? | Up To 120? | Up To 144? |
Max Thread Rely | Up To 56 | Up To 56 | Up To 56 | Up To 80 | Up To 112 | Up To 128? | Up To 240? | Up To 288? |
Max L3 Cache | 38.5 MB L3 | 38.5 MB L3 | 38.5 MB L3 | 60 MB L3 | 105 MB L3 | 120 MB L3? | 240 MB L3? | 288 MB L3? |
Vector Engines | AVX-512/FMA2 | AVX-512/FMA2 | AVX-512/FMA2 | AVX-512/FMA2 | AVX-512/FMA2 | AVX-512/FMA2 | AVX-1024/FMA3? | AVX-1024/FMA3? |
Reminiscence Assist | DDR4-2666 6-Channel | DDR4-2933 6-Channel | Up To six-Channel DDR4-3200 | Up To eight-Channel DDR4-3200 | Up To eight-Channel DDR5-4800 | Up To eight-Channel DDR5-5600? | Up To 12-Channel DDR5-6400? | Up To 12-Channel DDR6-7200? |
PCIe Gen Assist | PCIe 3.0 (48 Lanes) | PCIe 3.0 (48 Lanes) | PCIe 3.0 (48 Lanes) | PCIe 4.0 (64 Lanes) | PCIe 5.0 (80 lanes) | PCIe 5.0 (80 Lanes) | PCIe 6.0 (128 Lanes)? | PCIe 6.0 (128 Lanes)? |
TDP Vary (PL1) | 140W-205W | 165W-205W | 150W-250W | 105-270W | Up To 350W | Up To 375W? | Up To 400W? | Up To 425W? |
3D Xpoint Optane DIMM | N/A | Apache Cross | Barlow Cross | Barlow Cross | Crow Cross | Crow Cross? | Donahue Cross? | Donahue Cross? |
Competitors | AMD EPYC Naples 14nm | AMD EPYC Rome 7nm | AMD EPYC Rome 7nm | AMD EPYC Milan 7nm+ | AMD EPYC Genoa ~5nm | AMD EPYC Bergamo | AMD EPYC Turin | AMD EPYC Venice |
Launch | 2017 | 2018 | 2020 | 2021 | 2022 | 2023? | 2024? | 2025? |