Question How to simple test a CPU ?

[brucepascal]

This question will probably show how little knowledge I have of computers, so please, indulge me.... (and I am not using this desktop for gaming.)

I updated from an i5 4.Gen to an i7 4770. I was happy before and I am happy with it now, yet.... the i7 "feels" not really faster.
So I want to test it. I tried several CPU tests, none of which tells me what I want to know.

Basically I am looking for a number, or several numbers that tell me about my the shape of my CPU...
And shows me those numbers, that an i7 4770 had, when it was new, and what others users of the same cpu came up with...

If that is somehow possible, please tell me hwo...

Thank you very much

PS: I saw and used this: https://www.tomshardware.com/reviews/stress-test-cpu-pc-guide,5461.html
I found it not helpful for what I am looking for, which was not the fault of the article...

 

[CountMike]

This question will probably show how little knowledge I have of computers, so please, indulge me.... (and I am not using this desktop for gaming.)

I updated from an i5 4.Gen to an i7 4770. I was happy before and I am happy with it now, yet.... the i7 "feels" not really faster.
So I want to test it. I tried several CPU tests, none of which tells me what I want to know.

Basically I am looking for a number, or several numbers that tell me about my the shape of my CPU...
And shows me those numbers, that an i7 4770 had, when it was new, and what others users of the same cpu came up with...

If that is somehow possible, please tell me hwo...

Thank you very much

PS I saw and used: https://www.tomshardware.com/reviews/stress-test-cpu-pc-guide,5461.html and found it not helpful for what I am looking for, which was not the fault of the article...

What do you mean by "Shape" ? They usually work fully or not at all. Like in no "A little pregnant".
If it passes all stability tests they are fine. It's not gonna fade awy because of age, not in our life time anyway.
Use OCCT for instance.

OCBASE/OCCT : Free, all-in-one stability, stress test, benchmark and monitoring tool for your PC

Ocbase is the home of OCCT, the most popular all-in-one stability / stress testing / benchmarking / monitoring tool available for PC

www.ocbase.com

For performance there are many benchmark programs.
Those are mostly raw numbers but in a case like yoursit may not feel any faster unless a program can and does use more cores.

 

[brucepascal]

What do you mean by "Shape" ? They usually work fully or not at all. Like in no "A little pregnant".
If it passes all stability tests they are fine. It's not gonna fade awy because of age, not in our life time anyway.
Use OCCT for instance.

OCBASE/OCCT : Free, all-in-one stability, stress test, benchmark and monitoring tool for your PC

Ocbase is the home of OCCT, the most popular all-in-one stability / stress testing / benchmarking / monitoring tool available for PC

www.ocbase.com

For performance there are many benchmark programs.
Those are mostly raw numbers but in a case like yoursit may not feel any faster unless a program can and does use more cores.

Thank you. "They usually work fully or not at all, " well, I never heard that anywhere,. now I know, and I weill try Oc base.
So, any benchmark test will not tell me anything new at all ? Any computer with identical hardware, will show exactly the same numbers after ten years, as when it was new, and as all the other computers wu´ith the same configuration. If two such computer run the same stress test at the same time, they should come up with identical results? Has that been actually done, or is that just the theory?

Thank you, for taking the time...

 

[CountMike]

Thank you. "They usually work fully or not at all, " well, I never heard that anywhere,. now I know, and I weill try Oc base.
So, any benchmark test will not tell me anything new at all ? Any computer with identical hardware, will show exactly the same numbers after ten years, as when it was new, and as all the other computers wu´ith the same configuration. If two such computer run the same stress test at the same time, they should come up with identical results? Has that been actually done, or is that just the theory?

Thank you, for taking the time...

Stress tests do just what name implies, stress a component to it's limit to see if it would fail, if test fails it would let you know or part or system may shut down.
The results of stress tests are just Pass or Fail. During the tests, parameters like voltages, power, temperatures are monitored. OCCT for instance shows graphs of most important parameters.
Most components like CPU for instance can have some variations from specimen to specimen but that's because of variations when built but those are very small. They could be influenced by other components so no identical systems results from two otherwise similar systems but close enough.
Benchmarks on the other hand are to measure performance while under loads typical for programs/games they are supposed to approximate.

 

[TerryLaze]

Find any benchmark of the i7 4770 and run the exact same version of it on your CPU, if it shows roughly the same number then your CPU is ok. You would need the exact same setup including ram speeds and windows updates and so on to get the same numbers.

Since your CPU is not overclockable there is very little, to none, chance of degradation which is when the CPU starts needing more voltage to reach the same overclocked performance or starts to not be able to reach previous overclocked performance at all anymore.

Userbenchmark is one site with a lot of user input and it will show you at what percentage of performance you are of all the other users that ran that test on their CPUs.

 

[geofelt]

A simple test is to run the CPU-Z bench.
It will give you a rating for single thread performance as well as multithread.
For the I7-4770 you should get a single thread rating of 405:

CPU-Z Benchmark for Intel Core i7-4770 (1T) - CPU-Z VALIDATOR

Best CPU performance - 64-bit - November 2024

valid.x86.fr

You did not identify the previous I5, if it was a I5-4690, your cpu-z score would not have been much different.

CPU-Z Benchmark for Intel Core i5-4690 (1T) - CPU-Z VALIDATOR

Best CPU performance - 64-bit - November 2024

valid.x86.fr

The value of the i7 is that it has more processing threads and will benefit only when all threads are fully loaded.

 

[kanewolf]

This question will probably show how little knowledge I have of computers, so please, indulge me.... (and I am not using this desktop for gaming.)

I updated from an i5 4.Gen to an i7 4770. I was happy before and I am happy with it now, yet.... the i7 "feels" not really faster.
So I want to test it. I tried several CPU tests, none of which tells me what I want to know.

Basically I am looking for a number, or several numbers that tell me about my the shape of my CPU...
And shows me those numbers, that an i7 4770 had, when it was new, and what others users of the same cpu came up with...

If that is somehow possible, please tell me hwo...

Thank you very much

PS I saw and used: https://www.tomshardware.com/reviews/stress-test-cpu-pc-guide,5461.html and found it not helpful for what I am looking for, which was not the fault of the article...

Depending on which i5-4xxx CPU you started with, an i7-4770 is a mild upgrade. Hyperthreading just allows the OS to overload the physical CPU resources which are the same as your i5. So depending on the task, you may get little or no improvement.

 

[brucepascal]

Stress tests do just what name implies, stress a component to it's limit to see if it would fail, if test fails it would let you know or part or system may shut down.
The results of stress tests are just Pass or Fail. During the tests, parameters like voltages, power, temperatures are monitored. OCCT for instance shows graphs of most important parameters.
Most components like CPU for instance can have some variations from specimen to specimen but that's because of variations when built but those are very small. They could be influenced by other components so no identical systems results from two otherwise similar systems but close enough.
Benchmarks on the other hand are to measure performance while under loads typical for programs/games they are supposed to approximate.

Thank you again...
I ran several CPU benchmark tests, but never found one that will show me a result similar to the listed 405. The Core Speed between 3691 and 3890 is identical to what it shoud be...

 

[brucepascal]

A simple test is to run the CPU-Z bench.
It will give you a rating for single thread performance as well as multithread.
For the I7-4770 you should get a single thread rating of 405:

CPU-Z Benchmark for Intel Core i7-4770 (1T) - CPU-Z VALIDATOR

Best CPU performance - 64-bit - November 2024

valid.x86.fr

You did not identify the previous I5, if it was a I5-4690, your cpu-z score would not have been much different.

CPU-Z Benchmark for Intel Core i5-4690 (1T) - CPU-Z VALIDATOR

Best CPU performance - 64-bit - November 2024

valid.x86.fr

The value of the i7 is that it has more processing threads and will benefit only when all threads are fully loaded.

I have used the CPU Z for years, but never found that it tells me anything I want to know. The list of your link (thank you) shows a benchmark of 405, but where on CPU Z does it show that?

Depending on which i5-4xxx CPU you started with, an i7-4770 is a mild upgrade. Hyperthreading just allows the OS to overload the physical CPU resources which are the same as your i5. So depending on the task, you may get little or no improvement.

I have the i5 in a box with other computer parts, I will have a look at it. The benchmark of the 4690 would even be bigger than that of my i7....

Thank you for your help..

 

[geofelt]

When you select to do the cpu-Z bench test, it does a quick single threaded test followed by an all core utilized test.
These are not stress tests, but presumably a sampling of the instructions that you might be using every day.
The table I linked showing 405 is a compendium to tests that others have done.

 

[brucepascal]

Depending on which i5-4xxx CPU you started with, an i7-4770 is a mild upgrade. Hyperthreading just allows the OS to overload the physical CPU resources which are the same as your i5. So depending on the task, you may get little or no improvement.

Thank you, I have seen that now on the list.
In my world an i7 should be faster, actually a lot faster, than an i5 from the same generation.

I will chek on my old i5, as soon as I find it....

When CPU Z tests my i7 and comes up with speeds of 3691 and 3890 MHZ, which is what they are supposed to be, does it actually measure those or just gives the standard Intel numbers?

Thank you

 

[Math Geek]

for those cpu's even a fully loaded stress test would only see about a 10-15% gain moving to the i7 from an i5. hyperthreading is not an extra core but rather a way to utilize the little bit of cpu "downtime" that is there between it being tasked and when it completes that work. there is an inherent 10-15% of time where the cpu is waiting for work to do.

hyperthreading simply added an extra controller that could throw in work when the first controller is reloading. it's a bit more complicated than that but it's a decent way to think about what it is.

it's still the same today as well. hyperthreading from intel or amd is the same idea. an extra controller to feed work to the cpu so it is not waiting for a single controller all the time.

so when looking at benchmarks, don't expect any more than that extra 10-15% more work done since that's all there is available to gain.

 

[brucepascal]

for those cpu's even a fully loaded stress test would only see about a 10-15% gain moving to the i7 from an i5. hyperthreading is not an extra core but rather a way to utilize the little bit of cpu "downtime" that is there between it being tasked and when it completes that work. there is an inherent 10-15% of time where the cpu is waiting for work to do.

hyperthreading simply added an extra controller that could throw in work when the first controller is reloading. it's a bit more complicated than that but it's a decent way to think about what it is.

it's still the same today as well. hyperthreading from intel or amd is the same idea. an extra controller to feed work to the cpu so it is not waiting for a single controller all the time.

so when looking at benchmarks, don't expect any more than that extra 10-15% more work done since that's all there is available to gain.

Thank you..... learned some more....

 

[TerryLaze]

for those cpu's even a fully loaded stress test would only see about a 10-15% gain moving to the i7 from an i5. hyperthreading is not an extra core but rather a way to utilize the little bit of cpu "downtime" that is there between it being tasked and when it completes that work. there is an inherent 10-15% of time where the cpu is waiting for work to do.

hyperthreading simply added an extra controller that could throw in work when the first controller is reloading. it's a bit more complicated than that but it's a decent way to think about what it is.

it's still the same today as well. hyperthreading from intel or amd is the same idea. an extra controller to feed work to the cpu so it is not waiting for a single controller all the time.

so when looking at benchmarks, don't expect any more than that extra 10-15% more work done since that's all there is available to gain.

A fully loaded stress test would show the least amount of improvement because it would use close to all the resources of the core not leaving anything for the second thread.
Hyper-threading isn't great for server type workloads, but it is great for lighter threads that leave plenty of resources unutilized...like games.
As you can see it doesn't just work on stalls, the second thread runs on the gaps that the first one leaves unutilized.
It only runs just on stalls if the first thread is perfect and doesn't leave any resources free, like in stress tests and many benchmarks.

http://www.cslab.ece.ntua.gr/courses/advcomparch/2007/material/readings/Intel%20Hyper-Threading%20Technology.pdf

 

[brucepascal]

A fully loaded stress test would show the least amount of improvement because it would use close to all the resources of the core not leaving anything for the second thread.
Hyper-threading isn't great for server type workloads, but it is great for lighter threads that leave plenty of resources unutilized...like games.
As you can see it doesn't just work on stalls, the second thread runs on the gaps that the first one leaves unutilized.
It only runs just on stalls if the first thread is perfect and doesn't leave any resources free, like in stress tests and many benchmarks.

http://www.cslab.ece.ntua.gr/courses/advcomparch/2007/material/readings/Intel%20Hyper-Threading%20Technology.pdf

Thank you very much...
Seems I should have asked my question BEFORE deciding to buy the i7 and replace it.
I thought there would be a distinct and actually useful difference between an i5 and an i7, given the price difference. But I gues it will only make a difference for gamers.
Nut, the knowledge should help, when I get a new computer sometime in the future.

Thank you to everybody to understand a little more.