100+ Linux Benchmarks Between The AMD Ryzen 7 4700U vs. Intel Core i7 1065G7

[rogowskys]

Looks like they are in line with my results, great! Has Renoir been working out well for you on Fedora 32? Haven't had a chance to test there yet.

I've only been running it a couple days, and it's my 2nd PC so I haven't had a good chance to really try it out. Graphic acceleration worked out of the box before the updates, and running on wayland has been smooth so far. I've tried a few games, Frostpunk via SteamPlay, and World of Warships via Lutris/Wine, and they launch to menus just fine. The ony game that gave me trouble so far is Civilization VI (native). It caused amdgpu to crash and reset. I haven't looked further into the issue yet.

Overall I'm very happy with this laptop so far, and the "out of the box" experience with Fedora has been great!

[cb88]

An exaggeration on top of an exaggeration on top of an exaggeration. AMD sect never fails to disappoint. I mean Intel still has undisputably the fastest single core CPUs using an arch from 2015 (Sky Lake) and AMD somehow "wins in almost all aspects". Bonkers!

95% of average people out there are more interested in single-core performance unlike what AMD fans have been touting since the multicore failure called AMD FX.



What???? Absolute most tasks are limited by single-threaded performance even the ones which scale perfectly because at the end of the day you need to synchronize threads or wait for the last of them to complete.

And this is what matters for absolute most people out there. Even among Phoronix readers barely anyone compiles tons of code daily or (re)encodes videos.

Why is it always multithreaded performance BS?

Biased are you, sir. Not only biased you don't understand the basic principles of desktop computing where single threaded performance is the end all.

What are you smoking... if single thread is within 10% and in some cases actually faster than Intel and on top of that you have 2x the cores.... that means even in workloads that have mixed single and multithreading (read games) AMD will win, as the single thread tasks wont' have to share any (or at least much less) CPU resources with the multithreaded work..... so even though peak single thread performance can by higher on Intel currently real world single thread is often not higher but much lower. And on top of all that... AMD is using less power than Intel finally....

I mean just look at the benchmarks, Intel is curently only winning in a few, and probably due to some AVX2 specifics, where whatever library in uses isn't optimized for AMD yet.

[CochainComplex]

An exaggeration on top of an exaggeration on top of an exaggeration. AMD sect never fails to disappoint. I mean Intel still has undisputably the fastest single core CPUs using an arch from 2015 (Sky Lake) and AMD somehow "wins in almost all aspects". Bonkers!

95% of average people out there are more interested in single-core performance unlike what AMD fans have been touting since the multicore failure called AMD FX.



What???? Absolute most tasks are limited by single-threaded performance even the ones which scale perfectly because at the end of the day you need to synchronize threads or wait for the last of them to complete.

And this is what matters for absolute most people out there. Even among Phoronix readers barely anyone compiles tons of code daily or (re)encodes videos.

Why is it always multithreaded performance BS?

Biased are you, sir. Not only biased you don't understand the basic principles of desktop computing where single threaded performance is the end all.

Ok, Im writing this from a Dell Precision 7540 Notebook with a Xeon 2286M .
I have always the Highest Clock frequency displayed in the upper corner. Im using this little https://extensions.gnome.org/extension/1082/cpufreq/ tool to set governor settings.
This Little Puppy reaches 5.0Ghz. If it is idleing or if I run a simple bash script what so ever it reaches 4.8 to 4.98 Ghz ..but if I only have anything else running ..4,4 ....4,8 Ghz (4- 6 Threads) using all threads or all Cores 8/16 Threads by one of my parallelized programs/or simply compiling a kernel I will have 3,2 to 3,4 Ghz. 12 Threads maybe 3.6Ghz on all cores.

Yes Singlecore performance is nice but only if all the other cores are idling ...and when is this the case on any modern system? I'm monitoring it _always_ because at the beginning there was a bug in the thermal management of this particular Device. With Dell Firmware prior 1.4.x - Idleing - I havent seen anything over 4,7 -4,8 Ghz.

What you mean would suggest to stop all other tasks to get back the 5Ghz ...this will definitively increase the overall performance sure. I hope you are working in the Linux scheduler Dev team.

Besides what superior Intel Device are you using which deflates all the Ryzens?

For now with one of the fastest Notebook Intel Processors available I can not win the Benchmark crown. Anything Michael is displaying here is valid.

p.s.:Btw I have changed the stock thermal paste to thermalgrizzly kryonaut and Im using a book to leverage the notebook a bit. Airflow is cooling it enough.

p.p.s: I have found in my browser history a geekbench result https://browser.geekbench.com/v5/cpu/459951 todays result https://browser.geekbench.com/v5/cpu/2194622

[HarlemSquirrel]

Here is an interesting bench by lab501. They put a 4800U and 10710U in the same chassis head to head at different TDP caps. The 4800U doesn't seem to run that much faster at 25 W TDP than at 15 W. It also pretty nothing but nukes the 10710U.

https://next.lab501.ro/notebook/english-lenovo-ideapad-s540-13are-vs-13iml-amd-ryzen-7-4800u-vs-intel-core-i7-10710u

Thanks for this link. I would like to see more battery life tests like Anandtech did for web and movies normalized for battery size.

[birdie]

AMD infestation of this website continues unabated.

Three people with a straight face argue that multithreaded performance is more important than single threaded despite the fact that absolute most people need single threaded performance a lot more because barely any tasks the average user runs are multithreaded by their nature. The average user does not compile, does not encode videos, does not render CAD scenes, nothing like that. I'm even sure these three individuals who argue about multithreaded performance barely run any multithreaded tasks because laptop is not a device to even consider for running them.

Others say they are not biased yet argue that some time in the past Intel behaved uncompetitively. LMAO. What else? Let's remember something else Intel did in the 80s and 90s? I mean how on Earth you can possibly be unbiased if you don't care about the current statuation and instead you remind us of Intel's past misdeeds, LMAO.

Intel is not innovating? WTF is this crap? Are you serious? Ice Lake has a much better better IPC than Zen 2. Tiger Lake (which is to be released later this year) is even faster. God, it's unbelievable how AMD fans believe the crap they are uttering.

Zen 2 has roughly the same IPC as Sky Lake from 2015. Zen 2 CPUs get better than Intel's only because AMD is running on top of the 7nm node while Intel still uses the old 14nm node for most of their high performance CPUs. In 2019 AMD finally reached performance parity with the Intel uArch from 2015 using TSMC factories (!) and suddenly AMD is all the rage. LMAO. Intel in its turn produces its own CPUs from the ground up.

[dave_boo]

AMD infestation of this website continues unabated.

Three people with a straight face argue that multithreaded performance is more important than single threaded despite the fact that absolute most people need single threaded performance a lot more because barely any tasks the average user runs are multithreaded by their nature. The average user does not compile, does not encode videos, does not render CAD scenes, nothing like that. I'm even sure these three individuals who argue about multithreaded performance barely run any multithreaded tasks because laptop is not a device to even consider for running them.

Others say they are not biased yet argue that some time in the past Intel behaved uncompetitively. LMAO. What else? Let's remember something else Intel did in the 80s and 90s? I mean how on Earth you can possibly be unbiased if you don't care about the current statuation and instead you remind us of Intel's past misdeeds, LMAO.

Intel is not innovating? WTF is this crap? Are you serious? Ice Lake has a much better better IPC than Zen 2. Tiger Lake (which is to be released later this year) is even faster. God, it's unbelievable how AMD fans believe the crap they are uttering.

Zen 2 has roughly the same IPC as Sky Lake from 2015. Zen 2 CPUs get better than Intel's only because AMD is running on top of the 7nm node while Intel still uses the old 14nm node for most of their high performance CPUs. In 2019 AMD finally reached performance parity with the Intel uArch from 2015 using TSMC factories (!) and suddenly AMD is all the rage. LMAO. Intel in its turn produces its own CPUs from the ground up.

So have you disabled all except one core so that it can reach maximum boost?

Interesting that you point out that "In 2019 AMD finally reached performance parity with the Intel uArch from 2015". Does it bother you that lack of competition from AMD allowed Intel to release products for 4 years with negligible increases in performance and yet significant increases in prices?

[JustinTurdeau]

Three people with a straight face argue that multithreaded performance is more important than single threaded despite the fact that absolute most people need single threaded performance a lot more because barely any tasks the average user runs are multithreaded by their nature.

This is an extremely moronic take. The only reason most desktop software runs single-threaded is because most desktop software is not even remotely performance-sensitive.

You don't benchmark CPU performance on mostly idle GUI apps for the same reason you don't test .50 BMG by shooting slugs. If nothing you do requires a high performing CPU then good for you. Just exit this thread instead of sperging out.

AMD infestation of this website continues unabated.
The average user does not compile, does not encode videos, does not render CAD scenes, nothing like that. I'm even sure these three individuals who argue about multithreaded performance barely run any multithreaded tasks because laptop is not a device to even consider for running them.

I run multi-threaded stuff all day. I bet a lot of other people on here do too.

You clearly don't know the first thing about any of these subjects. I'm not sure if you're trolling or really just that retarded. I hope for your sake it's the former.

[CochainComplex]

birdie the only reason why I have this Laptop for work because a need to be mobile and have multithreaded performance. For any desktop relevant click here and there I wouldn't recognize any difference to the far inferior i7-4720 and my fx8350.....nobody focus on single threaded anymore. Even gamers know st times are over

[blandbl]

New user. Had to make an account because it was frustrating how I constantly see these reviews that doesn't take power consumption into account at all on reddit.

The 4700u is consuming 30W while the 1065g7 is consuming 15W. You're comparing the 4700u with another cpu that is configured at half the tdp and concluding that it performs better. Is it true? Sure. But that's because of the power consumption not because of the cpu properties itself.

Where am I getting 30W and 15W? From a previous review the author made himself(Page 7 of "Initial AMD Ryzen 7 4700U Linux Performance Is Very Good")
There's another blaring error the author made there also. He is measuring cpu power consumption not system power consumption like he believes he is. System power consumption on notebookcheck of the same laptop shows it is consuming almost 40W on load. The cpu is listed as a 15W tdp cpu. His measurement shows the "system" is drawing 15W. Highly likely the author incorrectly configured phoronix's own test suite to "cpu.power" instead of "sys.power". If he really meant to measure cpu power, he incorrectly labelled his graph and he mentions the brightness of the display for no reason.

[Michael]

Highly likely the author incorrectly configured phoronix's own test suite to "cpu.power" instead of "sys.power". If he really meant to measure cpu power, he incorrectly labelled his graph and he mentions the brightness of the display for no reason.

System power was used not cpu.power. cpu.power doesn't even work on AMD at this point for Zen unless opting to manually go install Zenpower until the RAPL support gets added. Hence system power was used due to no official/out-of-the-box Zen CPU package power reporting in a similar manner to Intel (granted RAPL patches posted by Google now on Friday night). It's also not possible to accidentally mislabel a graph when they are auto-generated.

Additional system power numbers but with the displays off on Ice Lake and Renoir are also going to be published in the next few days.