Question Geekbench 6 released and calibrated against Core i7-12700

Page 14 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Jul 27, 2020
26,829
18,480
146
Last edited:

Captante

Lifer
Oct 20, 2003
30,349
10,872
136
Yes, I guess so. It also shows the 5800X as faster than the 5900 (in MT). Well, I guess that's the limit of GB6. Can't use it for comparison of even small servers like this.

Much like the "X3D" 5/7-series AMD CPU's, the "vanilla" 8-core 5800X and 7700X have a few "architecture" advantages over the 12 core versions and match or exceed the 5/7900x's performance slightly in certain apps. (And they were nearly always pretty close or I would have upgraded myself... not worth the $ IMO)

The 7800X3d in some cases however is SIGNIFICANTLY faster in any app that actually uses the extra cache than the 7900X3d for the same reasons. (7900X3d = extra cache is only active on 6 cores vs a full 8 on the 7800X3d)
 
Last edited:

whoshere

Member
Feb 28, 2020
44
98
91
itvision.altervista.org
GeekBench 5 still works beautifully and it's my go-to benchmark.

No idea why people shy away from it and try to invent new benchmarks.

GB2/3/4 are exceptionally flawed and shouldn't be used. They emphasize RAM latency and throughput way too much and skew scores towards devices with fast/low-latency RAM.
 
Jul 27, 2020
26,829
18,480
146
They really should put in some animation or even just more interesting information about what is currently being tested during the benchmark. Would make it slightly more interesting to watch.
 

whoshere

Member
Feb 28, 2020
44
98
91
itvision.altervista.org
They really should put in some animation or even just more interesting information about what is currently being tested during the benchmark. Would make it slightly more interesting to watch.

That's unacceptable for a benchmarking application. That's an additional workload for the CPU which will negatively affect scores.

When I run GB to get proper scoring in Windows:
  • Set its priority to "Real Time" in the Task Manager
  • Exit/unload all other applications
  • Temporary disable real time protection in Windows Defender
  • Force minimize it (Win + D) immediately after starting benchmarking
  • Set Power Power Management Mode to Max Performance (in older versions of Windows via Control Panel, in new versions it's only possible for laptop users)
In Linux:
  • Exit a graphical desktop environment and switch to a virtual console (Ctrl + Alt + F2-F6)
 

Nothingness

Diamond Member
Jul 3, 2013
3,294
2,362
136
I'd also add that I know some people that do submitted runs of SPEC, reboot the system before launching the benchmark.
 
Jul 27, 2020
26,829
18,480
146
I suppose these are nice tricks to isolate the performance of the CPU/memory subsystem but in terms of system performance, I would prefer not to use these as I want to know how the system functions normally, that is, with the usual background tasks running etc. With Geekbench, an easier way to get a higher score is just to boot into a Linux bootable environment and run it from there (not as straightforward though, as I found out with Clear Linux. There was some command-line-fu involved).
 

Hitman928

Diamond Member
Apr 15, 2012
6,670
12,323
136
Jul 27, 2020
26,829
18,480
146

Introduce support for Arm Scalable Matrix Extensions (SME) instructions. Geekbench 6.3 includes SME implementations of the matrix multiplication kernels used by the Geekbench 6 machine learning workloads.

For systems without SME instructions, Geekbench 6.3 CPU Benchmark scores are comparable with Geekbench 6.1 and Geekbench 6.2 scores. Systems with SME instructions enabled will score higher in Geekbench 6.3 than in earlier Geekbench versions..

GB6 is now ready for WoA!
 

FlameTail

Diamond Member
Dec 15, 2021
4,384
2,761
106
Something must be coming up hard and fast. Otherwise, why bother releasing support for it if no current CPU uses it?
Well well well.
What could it be?

Apple isn't going to implement SME (they have their own custom AMX).

ARM already has implemented SME2 and SVE2 on their latest cores.

But Oryon is based on ARMv8 and doesn't have either SVE2 or SME2. Maybe.... Next gen Oryon core (Pegasus) adds support for ARMv9, SVE2, SME2!

Edit: On second thought, the next gen Oryon core wouldn't necessarily be the one used in X Elite G2/8G5. It will be the one used in 8G4, which is rumoured to be either Phoenix-L or Pegasus (According to 2 different rumours). For reference, the core used in X Elite is Phoenix. So Phoenix-L might be that core, but with ARMv9/SVE2/SME2 added.
 
Last edited: