Why Don't Dual/Quad CPU's Benefit CPU Mark Scores Much?

[bmichaels556]

One thing I've noticed, in both looking up benchmarks and running them myself, is that things like CPU Mark seem to show virtually zero improvement to what should be greatly improved CPU performance.

 

So for example... I know for a fact that say, an extra CPU probably won't benefit you in day to day office/desktop tasks, but as all know how much it can change something like video editing. Benchmarks like Cinebench accurately reflect this, at least to some extent. And then you look at CPU Mark, and (in my case) a Xeon X5675 gets about 8300. And yet my dual system is only showing a score of about 12612. Now, this would sound accurate if 50% of that score were weighed on single core performance next to multi-core performance. But within CPU Mark, a single-core test is only ONE out of NINE tests.

 

Do the other tests such as integer math, prime numbers, compression, physics, floating point math, extended instructions (SSE), encryption and sorting, really rely so much on single-threaded performance? It would sound like, at face value, they would benefit from the extra threads. Is this not the case? Or does CPU Mark simply put too much focus on single-core performance (50/50) therefore only showing a 50% increase?

[Livinloud]

I don't know that answer technically but also not sure what you are using. But for my example let's use my cpu which is a i7-8700k(model 2017) 6 cores and 12 threads just like the x5675(model 2011). X5675 uses ddr3 I believe and 8700k uses ddr4. Faster. The Cache on the 8700k is more than the x5675 and also higher clock speeds. Newer tech and also new instructions in bedded in the 8700k which benchmarking the 2 against each other the 8700k is going to be 100% faster.

 

Pulled from google:

 Cache memory is important because it provides data to a CPU faster than main memory, which increases the processor's speed. The alternative is to get the data from RAM, or random access memory, which is much slower.

[bmichaels556]

18 hours ago, Livinloud said:

I don't know that answer technically but also not sure what you are using. But for my example let's use my cpu which is a i7-8700k(model 2017) 6 cores and 12 threads just like the x5675(model 2011). X5675 uses ddr3 I believe and 8700k uses ddr4. Faster. The Cache on the 8700k is more than the x5675 and also higher clock speeds. Newer tech and also new instructions in bedded in the 8700k which benchmarking the 2 against each other the 8700k is going to be 100% faster.

 

Pulled from google:

 Cache memory is important because it provides data to a CPU faster than main memory, which increases the processor's speed. The alternative is to get the data from RAM, or random access memory, which is much slower.

Very good points, thanks for that info.  

[Neo-revo]

3 hours ago, bmichaels556 said:

Very good points, thanks for that info.  

cache speeds aside,   Focus on single core performance revolves around the fact that it is an 'intense' rendering test.   far more taxing than any real game would expect your computer to process for a fluid immersion.    And most games are optimized for single core thread performance,     Any thing beyond a quad core, hyper-threaded (or amd smt i think)  id not necessary if gaming is your most taxing event (even for hours)  unless you are gaming at 1440p, or 4k,  (60 or 120/144 hz)  with multiple things also happening at the same time.