if you could change CPU ratings (speed, cores etc)

[JCBiggs]

How would you change it, or how would you advertise CPUS other than clock speed/core count. 

 

I think personally Id like to see them all rated in Pentium 4s.  

no seriously.  I was thinking that relating it to the OS would be a better way.  for instance, windows works using Calls,  So why not rate cpu's in "calls per second" it can process.  or something like that.  Raw power is one thing, which lots of benchmarks measure, but how realistic are they for most people?   would be interesting if windows itself clocked your speed and told you when you needed to upgrade. 

[Crunchy Dragon]

My operating system telling me when it's time to upgrade my CPU is probably the only thing it'll take to get me off Windows. That is just plain creepy. It's like your refrigerator telling you to get more milk. I just don't need that in my life, thanks.

[JCBiggs]

1 minute ago, Crunchy Dragon said:

My operating system telling me when it's time to upgrade my CPU is probably the only thing it'll take to get me off Windows. That is just plain creepy. It's like your refrigerator telling you to get more milk. I just don't need that in my life, thanks.

I don't think its  really "creepy."     I think its more of a math based situation in which your PC would be able to calculate that its on old hardware.  I mean, really it sort of already does this with the window experience score.  if all your scores aren't maxed out that's basically telling you,you have room to improve.   I'm just saying "that"  but more accurate / real time monitored. 

 

[tarfeef101]

Cinebench scores

[Crunchy Dragon]

Just now, JCBiggs said:

I don't think its  really "creepy."     I think its more of a math based situation in which your PC would be able to calculate that its on old hardware.  I mean, really it sort of already does this with the window experience score.  if all your scores aren't maxed out that's basically telling you,you have room to improve.   I'm just saying "that"  but more accurate / real time monitored. 

 

My Windows score has never been maxed out. I currently have rating of 5.6, always have on this laptop. It works just fine for me, I'll probably run it to the death before I upgrade it.

 

Sure, I've always had room for improvement, I'm 6 generations behind in CPU architecture. I didn't get this laptop until 5th Gen Core CPUs had been out for some time.

[Sauron]

I'm pretty sure you'll notice if you need an upgrade... YOU know if your machine feels slow for YOU, there is no algorythm that can tell you if the experience you're getting is enough for you. And of course there's always room to improve if you spent more... that would be a pretty useless metric.

2 minutes ago, JCBiggs said:

I think its more of a math based situation in which your PC would be able to calculate that its on old hardware.

In what situation wouldn't you already know your computer is old? Don't you know (at least roughly) when you bought it? If it's second hand, you can assume it's at least somewhat old, too... not to mention a quick google search will tell you exactly when your cpu was manufactured. And honestly, your hardware's age is a pretty useless metric all things considered - there are dozens of models with wildly different performance coming out every year and newer does not always mean faster. The only truly significant metric are benchmarks, one for every kind of operation you need; and even those can't decide for you if it's time to upgrade.

[spat55]

9 minutes ago, Crunchy Dragon said:

My Windows score has never been maxed out. I currently have rating of 5.6, always have on this laptop. It works just fine for me, I'll probably run it to the death before I upgrade it.

 

Sure, I've always had room for improvement, I'm 6 generations behind in CPU architecture. I didn't get this laptop until 5th Gen Core CPUs had been out for some time.

Mines running a Penryn CPU

[JCBiggs]

2 minutes ago, Sauron said:

I'm pretty sure you'll notice if you need an upgrade... YOU know if your machine feels slow for YOU, there is no algorythm that can tell you if the experience you're getting is enough for you. And of course there's always room to improve if you spent more... that would be a pretty useless metric.

In what situation wouldn't you already know your computer is old? Don't you know (at least roughly) when you bought it? If it's second hand, you can assume it's at least somewhat old, too... not to mention a quick google search will tell you exactly when your cpu was manufactured. And honestly, your hardware's age is a pretty useless metric all things considered - there are dozens of models with wildly different performance coming out every year and newer does not always mean faster. The only truly significant metric are benchmarks, one for every kind of operation you need; and even those can't decide for you if it's time to upgrade.

I agree with most of what you just said.  again, the point of the thread was ways to better rate cpus that reflect real world performance.  cinebench is nice and you can certainlyfind that info.  but I'm just saying...on the box it says  "18cores 3.5ghz" and that really tells you jack..

[Sauron]

4 minutes ago, JCBiggs said:

I agree with most of what you just said.  again, the point of the thread was ways to better rate cpus that reflect real world performance.  cinebench is nice and you can certainlyfind that info.  but I'm just saying...on the box it says  "18cores 3.5ghz" and that really tells you jack..

You could have cinebench on the box, but it would still tell you nothing about, say, gaming performance. Not to mention in a few years a new version of cinebench would make that result useless. The box tells you the hardware difference between cpus within the same line; 18 cores 3.5ghz on an i9 box tells someone with base knowledge of hardware that it's probably better at multithreaded tasks and worse at single threaded tasks than the 6 core 4ghz i7 next to it. It's just enough to know roughly which cpus you should be looking at, but it doesn't substitute a thorough research and looking at benchmarks - neither would cinebench.

[JCBiggs]

let just say, theoretically,  windows could monitor your usage.  And it basically looked at % usage,  time at 100%, and load across cores.   Then using your maximums, and average, it could tell you "something."  In my case I have lots of single threaded work, but I bought a 6 core cpu because when I do a few specific task, all 12 threads are 100% for hours.  Now its time for a new pc, so I open up some  app, and look at my usage history, And I see that my single core usage is  frequent but never staruated, and my multithread is high.  so its recommends "x" number of cores with "y" single thread performance, etc etc,.

  vice versa, if granny is using a quad core, and is never even saturating a single core, windows could tell her next time she could get by with a slower dual core.    of course both of these metrics are going to be subject to the basic "does it feel fast enough" question.  which also needs to be taken into account...  

 

   the thing is, to do this in any meaningful way, you need the chip maker, and the os maker to be on the same page when it comes to performance ratings.  a gig of ram is always a gig of ram, a cpu cycle however, is much more varied.  Perhaps at some point in the future there will be chips with microcode and architecture geared towards certain usage cases. if all you do is web browse, intel might make a 2.5ghz low power dual core that's optimized for it, while your 4.5ghz 18 core is obviously much faster, but simply doesn't execute windows calls as fast because its optimized for GP computing and math, not "just" running windows.  (ASIC vs gpu....same thing. ) 
 

[JCBiggs]

I really wish these admins would stop moving post. no one looks at jack if its not in the general thread. move it after 3 days or something. jeez

 

[MyName13]

4 hours ago, tarfeef101 said:

Cinebench scores

Can we stop using it as a measurement of performance?

[TahoeDust]

37 minutes ago, MyName13 said:

Can we stop using it as a measurement of performance?

Why?  If you look at single and multicore, the results are pretty telling.

[JCBiggs]

people optimize for benchmarks now.  id rather have a direct comparison

[Phate.exe]

8 hours ago, tarfeef101 said:

Cinebench scores

I'd say that some kind of benchmark results would be a pretty good way to advertise performance, but then you'd inevitably end up worrying about chips that are only good at that benchmark and might kinda suck in other workloads.  Like if you based the gaming performance of my old Phenom II X6 machine on some of the 3DMark scores, you'd assume it's completely unusable, especially since Timespy can't even run on it.

 

Because the benchmarks are currently only used by tech reviewers creating content for nerds, that's not a problem, but you'd quickly run into issues once things go into mainstream advertising.

[MyName13]

10 hours ago, TahoeDust said:

Why?  If you look at single and multicore, the results are pretty telling.

And they tell what?You just push a button, a number shows up and if it's bigger than the other it must be better, right?It doesn't say anything.

[TahoeDust]

9 hours ago, MyName13 said:

And they tell what?You just push a button, a number shows up and if it's bigger than the other it must be better, right?It doesn't say anything.

I'm not going to get into a whole "thing" about this...I am watching football.  I will just say that I disagree and find value in it.  There is a reason that a CPU that performs better in Cinebench also perform better in most other measurable tasks.