Peformance Overview for Intel Meltdown after Windows is Patched - Is Performance a lot worse or just a little?

[AlTech]

So, after the whole debacle with Intel CPUs being vulnerable to security issues, I decided to test performance before and after the patch to see if performance has been affected on my laptop.

 

CPU Tested: Intel core i7-6650U 2.2GHz-3.2GHz Officially although a few times it turboed to 3.39GHz.

High Performance mode was used.

 

Programs I've used:

• Geekbench 4 (Tryout Mode)
• 7Zip Benchmark
• UserBenchmark
• Blender Render Test
• Cinebench R15
• CSMarkLib Test 1.4.1

 

Windows version used Not Patched against Meltdown: Windows 10 Build 16299.125

Windows version used Patched against Meltdown: Windows 10 Build 16299.192

 

For each program, I ran the test several times and got the arithmetic mean.

 

Quick Intro for those not familiar with these programs:

The Blender Render Test I used renders the AMD Ryzen Graphic in Blender and I record the time taken from Blender itself.

Geekbench 4 and Cinebench R15 are CPU benchmarks.

Userbenchmark is an overall benchmark which produces a standalone CPU score in points.

CSMarkLib Test 1.4.1 is a program which runs the benchmarks in CSMarkLib which are usually used in CSMark but I didn't have time to finish working on the new version before this happened so I've improvised in a way which should not compromise the integrity of the scores.

 

In Geekbench we can see performance is marginally better and this is clearly within margin of error.

 

In 7Zip we can see a tiny bit of performance decrease but again it's really within margin of error.

 

In userbenchmark we can see mostly the same results and all are within margin of error.

 

In Blender, we can see a larger gap of 5%. A higher number in this case is worse.

 

In Cinebench we can see that there is a 4.5% difference with higher results being better.

 

In CSMarkLib Test we see that the results are worse on the patched Windows but the results are within margin of error.

 

So, In conclusion Microsoft's patch to Intel Meltdown does appear to provide a fairly minimal impact in my testing. In some cases it is around 5% but it is nothing major.

 

When Ubuntu is patched with Linux Kernel 4.15, I'll do testing on that as well.

 

Thanks for reading, sorry this was rushed. Wanted to get this out there. Hope you enjoyed this.

Edited January 5, 2018 by AluminiumTech
sorry for loads of typos, 12am lol

[AlTech]

Oh and it's worth noting that anything older than Haswell is likely to have a far bigger impact on performance.

 

Oh and here are the patched versions of Windows which have mainstream support:

Windows 10 Build 16299.192

Windows 10 Build 15063.850

Windows 10 Build 14393.2007

[stlwrx]

Good write up and testing, man. I saw the same results when reading this write up from guru3d.

 

http://www.guru3d.com/articles-pages/windows-vulnerability-cpu-meltdown-patch-benchmarked,1.html

 

 

[Sauron]

Thanks for the information!

21 minutes ago, AluminiumTech said:

Oh and it's worth noting that anything older than Haswell is likely to have a far bigger impact on performance.

 

Oh and here are the patched versions of Windows which have mainstream support:

Windows 10 Build 16299.192

Windows 10 Build 15063.850

Windows 10 Build 14393.2007

Hopefully someone will make a round up of older cpus and compare them at some point.

[Fonograph]

What SSD do you have,can you try ATTO?

[Napalm360]

37 minutes ago, Sauron said:

Thanks for the information!

Hopefully someone will make a round up of older cpus and compare them at some point.

ive got a 5820k and a q6600 and q9400, might do some testing if there is interest