Explain some CPU features to me.

[SwiftXShadow]

What exactly does Max Turbo Frequency mean?

For example if a cpu specs say Processor Base Frequency is 4GHz and the Turbo Frequency is 4,4Ghz does that mean the CPUs max overclocking is 4,4Ghz?

 

What exactly are threads? 

When the specs say a cpu has 8 threads what does it mean and what are they?

Are these things connected any way to the instruction thing? Like how many instructions can a CPU take on at the same time?

 

 

When CPU tests are run on games and it gives their FPS scores do they run these games on a CPU alone, like with out a video card?

[Godlygamer23]

1.) No. The max turbo frequency is how far the CPU will clock itself up on one core. If one core is being stressed(doesn't matter which one), it has the ability to clock itself up to 4.4GHz. It does not reflect any kind of overclocking capability or limit.

 

2.)If you open up Task Manager, you'll see Processes. Those processes are broken up into smaller tasks called threads, which get spread across the CPU(by default). Threads indicate how many of those smaller tasks the CPU can handle simultaneously. Here is a more in-depth look at what a thread is. http://stackoverflow.com/questions/5201852/what-is-a-thread-really

 

3.)No, they do not run on the CPU exclusively. What the benchmarks are trying to show are how much more or less powerful CPUs make a difference. Less CPU intensive games will show less of a difference between CPUs and more CPU intensive games(like Skyrim) will show a bigger difference.

[lufihengr]

These should answer the first two

 

[Prysin]

Check out this video for

Turbo boost

 

 

Threads:
The easiest way i can explain threads is this -

Imagine the CPU core as a human, and threads being a canvas. This guy needs to paint many pictures, so he paints two paintings at once to speed things up.

 

Yes and NO

The CPU is almost always involved in some sort of FPS benchmark, however programs such as Unigine Valley/Heaven, uses 98% Graphics card and 2% CPU. The CPU is there mostly to tell the Graphics card to do this and that, while it itself just sits there and watches the progress. A slow CPU would be slow to tell the GPU to do its job, so a better CPU could tell the GPU faster, thus improving score.

 

 

I may not have explained this well and i invite anyone to correct me if i am wrong.

[minibois]

-snip-

1. A CPU (I think it's Intel only? Not sure.) boosts up when doing demanding stuff (like playing a game), this way it runs faster, but hotter too.

Of course this frequency is not unlimited, so that is what the limit is about.

 

2. Threads are like virtual cores.

A quad core CPU has 4 cores and 4 threads. A dual core CPU has 2 cores, 2 threads.

Hyperthreading makes it so all cores have 2 threads. Hyperthreading is only on Intel Core i7 and i3 CPU's.

So with those, a dual core would have 4 cores, 8 threads or 2 cores, 4 threads respectively.

 

3. Usually they tell you about the config. Either it's a test on iGPU, or a test with a dedicated graphics card.

Usually with a dedicated video card, because not all CPU's have a GPU built in (for example AMD's FX CPU's don't have iGPU)

 

But usually tests with AMD A8/10 series CPU's (their APU's) are run without a GPU. But again, see the tests config to know for sure.

[Godlygamer23]

1. A CPU (I think it's Intel only? Not sure.) boosts up when doing demanding stuff (like playing a game), this way it runs faster, but hotter too.

Of course this frequency is not unlimited, so that is what the limit is about.

AMD has their own boosting technology as well which is called Turbo Core.

[LukaP]

turbo freqency is the the speed at which the CPU will work when its cooled properly, and only one core is working (according to intel, but motherboard makers can change how turbo works, usually so it can be over all cores)

 

Thread is like a pipe which ends at the part of the cpu that is executing instructions (simply put). a 4 core cpu will have 4 threads, meaning 4 pipes down which instructions flow to the execution part.

 

Hyperthreading adds another "pipe" to each execution unit, so that when its waiting for instructions from one pipe, it can execute those from the other one

 

Instructions are small packs of machine code that tell the CPU what operation to perform on what data that is saved in the data cache. Usually each core can perform at max 1 instruction each cycle (that is the limit, haswell currently does 0.84 on average iirc) but that can be faked to go faster through using a longer pipeline and a superscalar design, both of which dont speed up execution, but rather change how instructions are handled. with pipelines you break up execution into parts, and so each part can be worked on simoltaneously but on a different instruction

 

 

Superscalar

 

Another way to speed up the CPU is by using SIMD instructions like SSE or AVX (SIMD meaning single instruction, multiple data). you can use those instructions when you want to do one operation on many different datas, like add 8 to a vector of integers or something like that. the width of the simd instruction determines how many data packets you can affect at once. AVX does 128bit, so 4x 32bit float/int. AVX2 does 256bit and AVX3 does 512 bit.

 

PM me if you want more indepth info

 

@SwiftXShadow follow your own threads to see when people reply ^^ top right corner

[SwiftXShadow]

Thx for the quick answers i think i understand it now

[Nena Trinity]

Turbo boost is when the CPU needs more power like a fighter jet using afterburner! 

 

Treads is data outputs and inputs (more=better)