Intel CPU issues

[Doctorwarthog]

I can't say I have experienced any slowdowns in CMANO after the Windows 10 fix.

[thewood1]

Ran another benchmark, this time on my MSI. Did a benchmark prior to the Win10 patch and one post-patch. Ran each three times, Modified Op Bass drum at 5x continuous with all range circles off and hi fi off.

Average for pre-patch - 29 seconds of game time for each real world second
Average for post-patch - 27 seconds of game time for each real world second

So on two separate machines, I am seeing slightly less than 10% impact from the patch. So regardless of what is going on with the chip, MS's changes in the OS patch do have a measurable impact. Most likely not noticeable to the human eye, but it is measurable. I did a test with project cars and Combat mission. I lost 5% to 10% with Combat Mission but nothing on Project Cars. My hypothesis is that CPU-intensive apps get hit more than games that are GPU-intensive.

I ran this by one of my analysts and he said that the Win10 patch will have an impact on performance with no changes on the chip. In fact, he said the changes on the chip are almost negligible. The impact is on the OS performance. Now, could there have been other changes in that patch beyond the chip issue? Of course there could. But, it doesn't matter. The Win10 patch does have some impact.

I am going to rerun the post-patch MSI and Surface Book tests and make sure I didn't change anything by mistake. I did it in a hurry and want to just make sure.

[kevinkins]

Missed this yesterday regarding chip replacement: http://money.cnn.com/2018/01/03/technol ... index.html

"And a U.S. government-backed body warned that the chips themselves need to be replaced to completely fix the problems."

Never heard of the U.S. Computer Emergency Readiness Team: https://www.us-cert.gov/

Their web site could be useful.

Kevin

[stilesw]

Here is some more information:

https://www.bleepingcomputer.com/news/s ... tre-flaws/

[thewood1]

One last follow up on this...

A tech group I work with just ran some benchmarks on the pre-patch and post-patch Intel performance. What they found wasn't that it wasn't so much CPU-dependent applications as much as drive access that gets hit. Hence all the hand-wringing about servers and datacenters. For some reason, it also seems that SSDs get hit a little worse that HDs. Although the project leader felt that they hadn't done enough testing to confirm that absolutely.

I talked to him a little about Combat Mission and Command. He felt that maybe they access the HD a little more than most modern games. He also said if I am running on 8Mb of RAM, and there is a lot of file swapping to HD, getting more RAM might help mitigate the new issues. Of course, that is the advice we always get.

On a side note, all the office apps they tested actually had slightly improved performance, except where they got complicated enough to be hitting the hard drives. Just thought it was interesting about standard office apps. They also tested Battlefront and Call of Duty and found slightly improved performance post-patch...just interesting.

[BDukes]

Interesting. Its great you bring stuff like this up so we know what to worry about[:)]

[thewood1]

Yeah, never fun bringing bad news. But I thought it would be good that instead of just listening to hysterical people or reading reposts of news reports, it might be good to see what quantified impact it might have on the game(s) we play.

[BDukes]

Yes. You are cool for doing so. Thank You!

[Maromak]

ORIGINAL: thewood1

Yeah, never fun bringing bad news. But I thought it would be good that instead of just listening to hysterical people or reading reposts of news reports, it might be good to see what quantified impact it might have on the game(s) we play.

Are there other games???

[Cik]

sims my man

sims

[kevinkins]

Surprised the malicious code can be written in JS and it's short. Two articles for those interested in going under the hood:

https://www.react-etc.net/entry/exploit ... javascript

https://react-etc.net/page/meltdown-spe ... it-example

[mavfin]

I've never had any issues with CPU usage with Command. Any issues I've had involve GUI freezing, and not because it's CPU intensive. But that's still being worked on.

My old (2011) 6-core Xeon works great.

[thewood1]

This thread isn't about absolute CPU usage. Its about impact on Command by any patch that addresses the new chip security issues.

[stilesw]

Here is some more news re

Warning: Microsoft's Meltdown and Spectre patch is bricking some AMD PCs

https://betanews.com/2018/01/08/microso ... s-amd-pcs/

[zakblood]

ORIGINAL: thewood1

This thread isn't about absolute CPU usage. Its about impact on Command by any patch that addresses the new chip security issues.

atm it's effecting more SDD's than CPU's in real world tests

Code: Select all

https://www.techspot.com/article/1556-meltdown-and-spectre-cpu-performance-windows/

just look at the SDD's returns

[thewood1]

I think that is what I said was the result on the previous page a couple days ago. Both my machines have SSDs. The analyst I work with running tests confirmed that SSDs seem to take a bigger hit than HDDs, but they both take some kind of performance hit in Command.

[thewood1]

The main thing I got out of that link was this...

"On the GPU front, Nvidia is reportedly also affected, so there will be loads of additional tests to be done when time comes. Our interpretation from Nvidia's blog is that they rely on CPU-like aggressive branch prediction on their GPU architectures. It's part of their performance gains over consecutive generations. The flaw appears to be the same as Intel CPUs, in that speculative operations occur without security checks first, as a secure design should be done."

That is the main concern I have right now. While I think we are starting to get a handle on the Intel issue, this little nugget makes me biggly concerned about the eventual impact on nVidia chipsets.

edit: Got an nVidia driver update this morning and accepted it for install. After reading above article, I did some digging and that new driver was released specifically to address Spectre bug.

[stilesw]

And, some more information for anyone interested:

https://www.theverge.com/2018/1/9/16868 ... c-slowdown

[stilesw]

And some possible mitigation from "Windows Secrets":

How to Protect Against the Meltdown and Spectre Vulnerabilities


https://windowssecrets.com/windows-secr ... abilities/

[HalfLifeExpert]

ORIGINAL: Dimitris


The coming CMANO update includes a number of performance improvements that should likely compensate for the raw CPU throughput reduction.

Any idea on a rough ETA?


I played some Command today on my main PC (Windows 7 64bit with a Haswell i5), along side a couple other games. The only problem I had with Command was in one of the new CSP scenarios, Halloween Horror, 1991, where the map seemed to get very slow. The drop down menus seemed okay, as did other scenarios, so I don't know if this was an issue with the scenario or if it had to do with this patch worries.

The other two games I fired up today on my main PC were Half Life 2 and Cold Waters. No issues with the former, the latter seemed to load a little longer than usual.

I should also mention that Command, as well as all my games plus my Steam install, are on a secondary internal HDD, with my OS on an SSD. Cone to think of it I don't think I noticed any issues until today with the Halloween Horror Scenario, which I had only briefly looked at before.