Futuremark PCMark 10

[VSG]

PCMark 10 is the latest iteration of Futuremark's popular full system benchmarking suite. It supports Windows 10 and runs quicker and more efficiently than the previous version. Futuremark also updated the tests to better reflect real world scenarios used by office and home professionals alike, including digital content creation, a new gaming test group, and 3D modeling/simulations.

Show full review

 

[SARVAMANGALAM]

• No support for more than a thread/CPU core and more than a GPU ... can you explain more this...?? for 30 hmm

 

[VSG]

• No support for more than a thread/CPU core and more than a GPU ... can you explain more this...?? for 30 hmm

The test programs use only one thread per core upto however many cores are supported, and similarly only one GPU. I am not sure what else to say

 

[EarthDog]

It doesnt support HT/SMT.. but will scale with logic cores.

 

[VSG]

It doesnt support HT/SMT.. but will scale with logic cores.

Yes, and here too up to however many each program will make use of. Video conferencing, for example, won't make use of 8 cores.

 

[BarbaricSoul]

so basically it will score CPUs based on how many cores they have, not how many threads it has? So a 4c/8t i7 7700k at 4.5GHz will score the same as a 4c/4t i5 7600k at 4.5GHz?

 

[birdie]

I wonder how much AMD paid to Futumark to skew the results towards CPUs with more cores. Compare these two results:

@W1zzard Also the bars are not to scale.

And the final scores:

I'm sorry but for 95% of people out there the last four tests are meaningless, while they actually power on their PCs and launch applications every day.

 

[VSG]

so basically it will score CPUs based on how many cores they have, not how many threads it has? So a 4c/8t i7 7700k at 4.5GHz will score the same as a 4c/4t i5 7600k at 4.5GHz?

Yes, and that is more to do with the test programs which in turn are actual real-work programs people use. So it says more about the state of applications today than anything else.

@birdie : These are as vendor neutral as can be. You will see that in upcoming hardware reviews that use these benchmarks. It's also funny since a lot of the earlier tests do not use more than 2-4 cores anyway, so the Ryzen system is not getting any advantage in the tests you mentioned were more useful. As for scale, each individual chart scales itself based on the max and min values.

 

[Italia1]

I don't ever more use bench from Aquamark 3...

 

[Mistral]

I wonder how much AMD paid to Futumark to skew the results towards CPUs with more cores.

They paid nothing. Everyone knows AMD only cares about Cinebench.

 

[Dave65]

I wonder how much AMD paid to Futumark to skew the results towards CPUs with more cores. Compare these two results:

@W1zzard Also the bars are not to scale.

And the final scores:

I'm sorry but for 95% of people out there the last four tests are meaningless, while they actually power on their PCs and launch applications every day.

Oh well yeah, this surely proves it.. Right?

 

[Vayra86]

I wonder how much AMD paid to Futumark to skew the results towards CPUs with more cores. Compare these two results:

@W1zzard Also the bars are not to scale.

And the final scores:

I'm sorry but for 95% of people out there the last four tests are meaningless, while they actually power on their PCs and launch applications every day.

95% of the people have the Celeron-class CPU in some form or another.

So that problem just solved itself. As for the rest of your comment, well, others have said it before me. This is what bench runs look like when Intel sits on its ass for five years progressing nothing. If anyone is to blame for core/thread performance scaling, its Intel.

 

[birdie]

>80% of tasks out there are single threaded where IPC performance is paramount but you can imagine whatever you want. There are too many rabid fanboys nowadays and very little rational thinking. AMD has introduced lots of cores and suddenly IPC performance has lost its relevance. What a load of utter BS.

While for most users out there the last four test are practically meaningless, still let's inflate multicore CPUs scores just because Intel has stagnated. What a load of pure horseshat.

 

[Vayra86]

>80% of tasks out there are single threaded where IPC performance is paramount but you can imagine whatever you want. There are too many rabid fanboys nowadays and very little rational thinking. AMD has introduced lots of cores and suddenly IPC performance has lost its relevance. What a load of utter BS.

While for most users out there the last four test are practically meaningless, still let's inflate multicore CPUs scores just because Intel has stagnated. What a load of pure horseshat.

With Ryzen being largely on-par with Intel's IPC, how is this still an argument in the first place?
Ryzen simply has more cores at reasonable prices, it is what it is, and it will only show an advantage when more cores are in use. So explain to us again how IPC is not weighted in this benchmark? Intel still gets very close to Ryzen, and when Ryzen is forced to use one core, it doesn't fall apart either.

And wouldn't it make sense to run this bench and find the CPU that excels on your use case anyway? If you just take the total scores and only look at those, that's hardly good research.

 

[birdie]

Ryzen is not on par with Intel CPUs in regard to IPC. It has a very high AES encryption/decryption throughput but other than that it loses by a large margin even to Sandy Bridge CPUs.

My six years old Intel Core i5 2500 (non-K, i.e. stock frequencies) vs Ryzen 7 1700: Geekbench 4.1 - pure embarrassment.https://browser.geekbench.com/v4/cpu/compare/2979799?baseline=3060943

AMD fanboys must be smoking some very strong stuff to claim that Ryzen is somewhere near Intel CPUs in regard to IPC. At best they are competing with the Westmere architecture from 2010. Yeah, suddenly AMD has reached the performance level of 7 years old Intel architecture.

 

[Caring1]

Ryzen is not on par with Intel CPUs in regard to IPC. It has a very high AES encryption/decryption throughput but other than that it loses by a large margin even to Sandy Bridge CPUs.

Are you reading that chart backwards

 

[birdie]

Are you reading that chart backwards

Wow, at a technical forum we'll be arguing about reading charts?

OK,

Ryzen 1700 vs Intel Core i5 2500:

AES: +70%
N-Body Physics: +26%
Ray Tracing: +38%
Memory Copy: (DDR4 2666 vs DDR3 1600): +29% (LOL, memory throughput must be at least 66% higher)
Memory Bandwidth: +46% (much better but still not even close to a theoretical improvement)

LLVM: -49%
SFFT: -25%
Memory Latency: -24%
Lua: -11%
Dijkstra: -17%
HTML5 Parse: -17%
Histogram Equalization: -11%
PDF Rendering: -13%

More or less a draw in all others.

MOAR COARS it's what Ryzen is. And luckily their SMT(HT) implementation is great as well. So if one needs MORE COARS then Ryzen shines but such use cases are primarily for professionals (rendering, encoding).

Strangely AMD fanatics scream that Ryzen is better overall while clearly it's not the case.

 

[Caring1]

Wow, at a technical forum we'll be arguing about reading charts?

Actually you have yet to prove you're not an idiot as you admitted with that picture.
I wasn't arguing anything, merely questioning your ability to comprehend something written in simple English.
Anyone with a brain can look at the chart and see it is not a case of apples V's apples when they are totally different systems running different O.S's.
Come back when the systems are equitable.

 

[birdie]

OMG, this is getting ridiculous. The stupidity is way over the top now. Geekbench for Windows and Linux is compiled using the same compiler and its code for these two OSes is very similar except for the parts which interact with the OS - for instance the Linux version doesn't have any GUI whatsoever.

So you want a test done in Windows? I will give you one in a short while. Too bad it will match my Linux result.

Actually I won't bother. Like Mark Twain said, "Never argue with stupid people, they will drag you down to their level and then beat you with experience".

Edit: my Windows results - you were right, Windows is 6% slower. OMG, what a difference it makes. Suddenly it invalidates all my previous claims.

 

[RCoon]

The pooslinging isn't needed here. When you devolve into calling each other stupid it tends to make your arguments look weak and lacking. Tone it down a bit, we're all for hotly debated topics.

 

[DRDNA]

look at the chart and see it is not a case of apples V's apples when they are totally different systems running different O.S's.
Come back when the systems are equitable.

Also different versions of the bench app too!


Ryzen Geekbench 4.1.0 Tryout / Intel Geekbench 4.0.0 Tryout

 

[birdie]

Also different versions of the bench app too!


Ryzen Geekbench 4.1.0 Tryout / Intel Geekbench 4.0.0 Tryout

All the results that you need: https://browser.geekbench.com/v4/cpu/search?q=AMD+Ryzen+7+1700

Now what? What other excuses will you conjure up? Maybe there's something stupendously wrong with my system and my results are wrong?

 

[80-watt Hamster]

I wonder how much AMD paid to Futumark to skew the results towards CPUs with more cores. Compare these two results:

<chart images snipped>

I'm sorry but for 95% of people out there the last four tests are meaningless, while they actually power on their PCs and launch applications every day.

I don't understand why you feel these results are in AMD's favor. Essentials group: Intel wins by a score of 9759 to AMD's 8077. Productivity: 8127 vs. 6605. Which brings us to content creation, where the AMD system comes out on top by a almost exactly a factor of two (3313 to 6570). The AMD system also happens to have double the cores, so that result is entirely predictable. In the overall score, you have the 980 paired with the 1700X against the HD 530 in the 6700K, so the 3DMark portion of the test skews that result a bit. If AMD paid anything for altered results, it sure doesn't look like they got their money's worth to me.