Geekbench 6 - Geekbench Blog
www.geekbench.com
Weird choice of baseline CPU and even weird is that the baseline score is 2500.
i7-12700 does hardly 2000 in GB5 with the fastest DDR5.
What he writes also partially applies to SPECrate which is lacking against GB6 in a way: there's no sharing of data, just running multiple instances of the same benchmark. It's a good test for shared resources like L3 and main memory, but not very good for performance of cross cores exchanges, which as far as I know some of the GB6 workloads test.People apparently still need to be told this:
A Reminder That Geekbench 6 is NOT for Big CPUs
A quick reminder that Geekbench 6 is not designed for big CPUs, espeically the newest generation of server CPUswww.servethehome.com
Very valuable result, but unsurprising. Only the render subtest in GB6 seems to scale well with large number of cores. I guess each test has a different limit in scaling (didn't someone make a post about this in the past?).Yeah i did alittle test run on my 9950X to check the scaling.
Seems like i'm getting ~30% higher MT score when i double the corecount from 8 to 16
First with CCD1 disabled -> 8core and 16 threads active on CCD0
View attachment 109361
Both CCDs enabled -> 16 core and 32 thread
View attachment 109362
I think that depends on the BIOS. On my 12700K, I can disable whatever core I want as long as I keep one P-core active. On the Epyc ES BIOS, there are limited options available such as keep minimum two cores active per CCD out of the eight CCDs. Don't remember if it's possible to disable all CCDs except one.Is it possible to disable cores in a finer way than a whole cluster?
Yes you can enable/disable cores one by one in the biosIs it possible to disable cores in a finer way than a whole cluster?
Out of curiosity did you try GB6 with 4 cores on each CCD (8 total)? I would like to know how much that hurts MT score.Yes you can enable/disable cores one by one in the bios
Should serve as a nice comparison point to the 9700X.Out of curiosity did you try GB6 with 4 cores on each CCD (8 total)? I would like to know how much that hurts MT score.
Employing *nix operating systems isn’t a cheat, it’s the performance you expect that Windows doesn’t give you by default.Got 20K finally but a cheat was employed. (Linux) It's the first and only run and maybe I can improve on it.
105W - Geekbench
Benchmark results for an ASUS System Product Name with an AMD Ryzen 7 9700X processor.browser.geekbench.com
Here my runs with 9700X. (+200MHz auto OC, CO -20 all core) For the life of me I cannot get multi 20K.
Wonder if Zen 6 will easily beat this in 2026.265K @6.8GHz!
ASUS System Product Name - Geekbench
Benchmark results for an ASUS System Product Name with an Intel Core Ultra 7 265K processor.browser.geekbench.com
What is worth looking at for compilation workloads are the results of the clang test.Text processing seems to be the forte of Arrow Lake (good for compilation workloads I guess?).
Oh yeah. Dumb meWhat is worth looking at for compilation workloads are the results of the clang test.