Thursday, December 11th 2025

Another "Xeon 696X" Sample Gets Benched, PassMark Incorrectly Identifies 32-core Config

Late last month, an unreleased "Xeon 696X" processor was spotted in the SiSoftware Official Benchmark Ranker database. A recent uptick in "Granite Rapids-WS" leaks—including a long SKU stack—hints at an imminent launch. The latest leak, found within the PassMark CPU benchmark archive, seemingly reaffirms the Xeon 696X model's existence. Unfortunately, today's diagnostic points to a lower than anticipated core count: 32 instead 64. Based on older information, this appears to be a major mistake (on the PassMark side). Last month's SiSoftware-based info dump described a sample that leveraged 64 cores and 128 threads.

x86deadandback's initial social media post highlighted the PassMark suite's error. According to a quick benchmark score comparison (courtesy of Wccftech), the forthcoming sub-flagship workstation processor does better than the current "Sapphire Rapids-WS" range topper (Xeon W9-3595X) in multicore scenarios. A disappointing single-core tally likely comes from the next-gen candidate being disadvantaged as an early engineering sample. A rough equivalent from the Red Team—Ryzen Threadripper PRO 9985WX—enjoys a leading position on the comparative chart.
Sources: X86 is Dead & Back Tweet, CPU Benchmark, Wccftech
Add your own comment

3 Comments on Another "Xeon 696X" Sample Gets Benched, PassMark Incorrectly Identifies 32-core Config

#1
john_
Panther Lake is at 4100 in single thread which is about 22% faster than what 696X does here. If we assume that the chip can get a 20% extra performance until it gets released, then the scores will be around 4000 in ST and 135000 in MT. Maybe that will be something like "best case scenario"? In any case we can't expect Intel with all it's troubles in manufacturing to make the perfect chip with it's first serious try at 18A. But they are in the right path.
Posted on Reply
#2
AntiX3DGlazer
This is stupid late at this point. I don't get why they don't just release it.
Posted on Reply
#3
lemans24
john_Panther Lake is at 4100 in single thread which is about 22% faster than what 696X does here. If we assume that the chip can get a 20% extra performance until it gets released, then the scores will be around 4000 in ST and 135000 in MT. Maybe that will be something like "best case scenario"? In any case we can't expect Intel with all it's troubles in manufacturing to make the perfect chip with it's first serious try at 18A. But they are in the right path.
I don't think 696X is 18A. The first Xeon chip manufactired in 18A is Diamond Rapids, which is coming out 2H 2026. The node for 696X is Intel 3 as far as I recall.
Posted on Reply
Nov 11th, 2026 02:18 CST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts