If your workloads are limited by single thread performance you need better software. It's why vulkan and dx12 are a thing. (The single thread limitation of committing a frame to the GPU has been reduced by an order of magnitude) It's why C++ has the parallel algorithms library baked into the language.
I get it, threading is hard. But it's honestly not that hard. It's only hard when you're maintaining some super old program with single threadedness engineered into its core architecture. (note: this is my day job) Greenfield applications since 2009 should have had threading built in as a core assumption.
AMD is doing the right thing by optimizing for multithread performance over single thread performance. Moore's Law is dead for single cores. It has been for a decade and a half.
Depends on the program. If it requires, say, to synchronize hundreds of thousands of entities every 16ms, it's probably better to go with single threaded instead...
in virtually all cases, "single-threaded performance" should be read as "serial performance" - the performance of a specific, single thread, within a larger program. While confusing to a layman, nobody uses the term to refer to performance when there is only a single thread running on the entire processor - we indeed left that time behind in the 90s, so that is not a sensible metric.
However, Amdahl's law means that serial/single-threaded performance remains an important determinant of total performance even in multi-threaded applications. Locking is a great example, even in a heavily multithreaded program you can be limited by lock contention, that specific part of the program is serial.
Gustafson's Law is of course a thing, eventually we will come up with additional work to fill those cores, but single-threaded performance remains a critical factor for CPU performance.
>If your workloads are limited by single thread performance you need better software.
Most of the time we don't get to choose which software we run. Hell, most core components of Windows get stuck in a single core still. I'll take single-core performance over multi-core performance any day. Maybe that will change in 10 years, but I live in the present.
If your workloads are limited by single thread performance you need better software. It's why vulkan and dx12 are a thing. (The single thread limitation of committing a frame to the GPU has been reduced by an order of magnitude) It's why C++ has the parallel algorithms library baked into the language.
I get it, threading is hard. But it's honestly not that hard. It's only hard when you're maintaining some super old program with single threadedness engineered into its core architecture. (note: this is my day job) Greenfield applications since 2009 should have had threading built in as a core assumption.
AMD is doing the right thing by optimizing for multithread performance over single thread performance. Moore's Law is dead for single cores. It has been for a decade and a half.