Single-core mode for benchmarks

[MAA]

Hello,

1) CPU benchmarks have "Parameters" and we can select single-core mode (1 CPU). But Reference Results remain multi-core. Please add single-core Reference Results also.

2) Please add single-core mode also for Memory benchmarks.

[Fiery]

1) Single-core PC benchmarks aren't relevant in this day of age.  Also, it would grealty overcomplicate the maintenance of the reference benchmark results database.

2) It's not possible to limit the number of cores or CPU threads to be utilized by the AIDA64 multi-threaded cache and memory bandwidth benchmarks.  It's by design, since those benchmarks automatically calibrate themselves to use the optimum number of cores and optimum settings (code path) to squeeze the maximum bandwidth out of the caches and RAM.

[MAA]

1) Single-core benchmarks are very popular in reviews this year for comparing different microarchitectures (Sky lake vs. Kaby Lake vs. Coffee Lake and Ryzen vs. xxxLake). It will increase the reference benchmark results database only twice.

2) Maybe you'll write an additional benchmark specifically for "single-core mode"?

[Fiery]

On ‎2017‎. ‎12‎. ‎01‎. at 3:54 PM, MAA said:

1) Single-core benchmarks are very popular in reviews this year for comparing different microarchitectures (Sky lake vs. Kaby Lake vs. Coffee Lake and Ryzen vs. xxxLake). It will increase the reference benchmark results database only twice.

2) Maybe you'll write an additional benchmark specifically for "single-core mode"?

1) Reviewers do not rely on our reference results, but they use their own test results.  And apart from reviewers, not many folks are interested in single-core benchmarks.  Why to have 4, 6, 8 or 18 cores in your CPU and only use just one to check how great your CPU is?

2) No, we have no plans on that.

[Wild_Horse]

Single Core Benchmarks are very relevant even in this day and age. There are many applications which do not utilize multiple cores efficiently. This is where single core performance becomes of interest. For example, It is well-known that a lot of the topline CPUs are marketed as gaming chips and that gaming performance is heavily dependent on performance core. The fact that reviewers are interested in single core scores is because there is a strong market for it.

I am not writing this to convince Aida64 to include a single core benchmark. I am just saying that the argument being presented for not including a single core benchmark (i.e. no need and no demand) is not correct.

[Fiery]

16 hours ago, Wild_Horse said:

Single Core Benchmarks are very relevant even in this day and age. There are many applications which do not utilize multiple cores efficiently. This is where single core performance becomes of interest. For example, It is well-known that a lot of the topline CPUs are marketed as gaming chips and that gaming performance is heavily dependent on performance core. The fact that reviewers are interested in single core scores is because there is a strong market for it.

I am not writing this to convince Aida64 to include a single core benchmark. I am just saying that the argument being presented for not including a single core benchmark (i.e. no need and no demand) is not correct.

We never said single-core benchmarks are not included.  You can use the Parameters button on the AIDA64 tool bar to limit the CPU and FPU benchmarks to use only one CPU.

The argument was about including a list of reference benchmark scores for single-core performance.  That's what we're not doing, but you can still use the benchmarks with limited number of cores.